xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EN3DVIIBO_KM5JNE INGEST_TIME 2017-07-11T22:07:03Z PACKAGE UF00101036_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
CONSUMER SATISFACTION: AN EXTENDED RESEARCH CONCEPTUALIZATION DOUGLAS R. HAUSKNECHT A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1988
This dissertation is dedicated to the memory of my friend, Michael Ray Albaugh. Mike was among my first students; one who made teaching worthwhile. He displayed a keen interest and ready aptitude for marketing. These traits were demonstrated further when he assisted in the collection of the main study data, reported in these pages, as part of an independent study course. I deeply regret the fact that this document was not completed at the time of his death in an Akron, Ohio, rooming house fire May 14, 1987. Mike was graduated from the University of Akron as Bachelor of Science in Business Administration/ Marketing on May 31, 1987. He had completed all of the requirements for the degree.
ACKNOWLEDGMENTS Any undertaking of this magnitude is completed only with the assistance and cooperation of others. So many people have been so helpful over such a long period, I hesitate to identify any in fear of omitting others. The most direct influence on a dissertation is, of course, the committee chairman. Rich Lutz provided not only guidance and direction, but also the motivation and encouragement that were necessary to keep things moving. Most importantly, he helped me to develop the confidence and perception of self-worth that enabled me to achieve closure at last. Joe Alba and Sam Ahmed, the other official committee members, kept me from going astray in designing and conducting the research. For this, and their willingness to adapt to tight schedules and too-oftenmissed deadlines, I am grateful. In a very real sense, each member of the marketing faculty has contributed to the dissertation over the years. Whether as the result of in-class instruction, seminar presentation, or informal discussion, each has influenced the research compiled herein. I am truly the product of the entire faculty. Some deserve special recognition. Joel Cohen planted the seed that caused me to consider doctoral studies. Even before that, he set education and career matters in their proper perspective by helping me
learn an unforgettable lesson about life's priorities. Dipankar Chakravarti and Chezy Ofir introduced me to the details of research and the academic life. Without this, I would likely never have pursued this career. Bill Wilkie first acquainted me with and encouraged my perspectives on postpurchase topics in consumer behavior. I should also acknowledge the contributions of academics with whom I had no formal connection, but who were willing to provide the feedback and advice that I needed. Many fall into this category, but Rich Oliver stands out for the effort expended on my behalf. Professional debts are only part of the story. When a dissertation takes over life, it affects and is affected by all of the individuals who have provided encouragement along the way and have responded with patience when that was most needed. My family, immediate and extended, bore the brunt of this burden. In addition, Jim and Sunny Flavin are the friends who started me on this path and have walked with me along the way. Ron and Judy Albaugh provided incentive for the final stretch. Finally, understanding, patience, and assistance were supplied at the University of Akron Â— where I began teaching midway through the research. Faculty colleagues, especially Judy Wilkinson, have given advice and encouragement. Staff assistance, especially the typing of Pat Johnson, made the feat manageable. Both the University of Akron and the University of Florida provided partial financial support of the research. There are others who have contributed much. They may be left off the printed page, but they are etched indelibly in memory.
TABLE OF CONTENTS page ACKNOWLEDGMENTS iii LIST OF TABLES vii LIST OF FIGURES ix ABSTRACT x CHAPTERS 1 INTRODUCTION 1 Purpose 1 Importance of Research in Consumer Satisfaction 2 Audiences Addressed by the Research 4 CS/D Literature to Date 6 Contribution of Dissertation 7 Dissertation Overview 8 2 LITERATURE REVIEW 10 Introduction 10 Definitions of Satisfaction 11 The Disconf irmation Paradigm 21 Other Perspectives on Satisfaction 33 Types of Consumer Satisfaction 33 Attitudinal-Behavioral Consistency 44 Measurement Issues in the CS/D Literature 47 Behavioral Outcomes of Satisfaction 64 Summary 78 3 SATISFACTION: AN EXTENDED CONCEPTUALIZATION OF THE GENERAL CONSTRUCT 80 Introduction 80 Definition and Theory of Emotions 81 Motivation 83 Lack of a Satisfaction Feeling 91 Other Types of Satisfaction 95 The Relationship Between Satisfaction and Attitude 98 The Extended Conceptualization 102 Purpose of the Extended Conceptualization 142 Summary and Conclusions 165
4 RESEARCH HYPOTHESES Introduction Main Hypotheses Secondary Hypotheses Supplemental Hypotheses Summary 5 EXPERIMENTAL METHOD Introduction Pilot Study--Measure Development Research Conducted Summary 6 STUDY RESULTS Introduction Pretest Findings Main Study Results Summary and Discussion 7 CONCLUSION Introduction Summary of Findings Implications of Findings Future Outlook APPENDIX A PILOT STUDY QUESTIONNAIRE APPENDIX B INSTRUCTIONS FOR ADMINISTRATION OF EACH EXPERIMENTAL CONDITION APPENDIX C QUESTIONNAIRE AND SUBJECT FORMS FROM MAIN STUDY REFERENCES BIOGRAPHICAL SKETCH 167 167 167 169 170 172 174 174 174 195 234 235 235 235 241 339 342 342 342 351 354 357 365 397 413 452
LIST OF TABLES Table Page 1 Perspectives on Consumer Satisfaction 29 2. Types of Consumer Satisfaction 40 3 Actions Following Dissatisfaction 76 4 Measures of Fundamental Emotions 85 5. Patterns of Emotions in Automobile Ownership 86 6 "Triggers" of Response Involvement 132 7. Prior CS/D Research and the Extended Conceptualization... 146 8. Results of Factor Analysis of Emotion Items 180 9. Behavioral Intent Scales from Factor Analysis 183 10. Reliability Coefficients for Scales Used in Pilot Study 186 11. Adjectives Describing Ends of Satisfaction Continuum 137 12. Mean Responses by Condition and Classification 190 13. Mean Responses to Likert Items by Condition and Classification 192 1 4. Manipulation Summary 199 15. Preparation of Peanuts for Coffee Blend 212 16. Coffee Preparation Instructions 220 17. Measure Summary 223 18. ANOVA Summary of Decision 243 19. ANOVA Summary of Coffeemaker Decision 244 20. Selection of Coffee Discrepancy Scale 245 21. Selection of Coffeemaker Combined Discrepancy Scale 246 22. ANOVA Summary of Coffee Discrepancy 248 23. ANOVA Summary of Coffeemaker Combined Discrepancy 249 24. ANOVA Summary of Coffeemaker Discrepancy 250 25. ANOVA Summary of Coffeemaker Expectancy Disconf irmation. 251 26. Selection of Coffee Expectations Scale 252 27. ANOVA Summary of Coffee Bitterness Expectations 253 28. Selection of Coffeemaker Expectations Scale 255 29. ANOVA Summary of Composite Coffee Expectations 257 30. ANOVA Summary of Coffeemaker Expectations 258 31 Selection of Coffee Perception Scales 260 32. Correlations Among Coffeemaker Perception Scales 262 33. ANOVA Summary of Coffee Quality Perception 263 34. ANOVA Summary of Coffee Color Perception 264 35. ANOVA Summary of Coffee Flavor Perception 267 36. ANOVA Summary of Coffee Bitterness Scale 263 37. ANOVA Summary of Coffee Flavor (single) Scale 269 38. ANOVA Summary of Coffeemaker Speed Perception 271 39. ANOVA Summary of Coffeemaker Filling Perception 272 40. ANOVA Summary of Coffeemaker Use Perception 273 41 ANOVA Summary of Coffeemaker Appearance Perception 274
List of Tables Â— continued [Â•able Page 42. Selection of Coffee Involvement Scale Combinations 277 43. Selection of Coffeemaker Involvement Scale Combinations 278 44. Results of Response Involvement Analyses 280 45. Reliability of Coffee Attitude Scale 283 46. ANOVA Summary of Coffee Attitude Ratings 284 47. Reliability of Coffeemaker Attitude Scale 288 48. ANOVA Summary of Coffeemaker Attitude Ratings 289 49. Correlations Between Perception and Attitude Measures.... 291 50. Significance Tests for Perceived Discrepancy Measures.... 294 51. ANOVA Summary of Coffee Satisfaction Ratings 302 52 ANOVA Summary of Coffee Odds Scale 303 53. ANOVA Summary of Coffee Delighted-Terrible Scale 304 54. ANOVA Summary of Coffee Faces Scale 305 55 ANOVA Summary of Oliver s Six Item Scale 305 56. Reliability of Oliver's Six Item Scale 309 57. ANOVA Summary of Coffeemaker Satisfaction Ratings 310 58. ANOVA Summary of Coffeemaker Odds Scale 311 59. ANOVA Summary of Coffeemaker Delighted-Terrible Scale.... 313 60. ANOVA Summary of Coffeemaker Faces Scale 314 61 Number of Subjects Contacted by Telephone at One Week Delay 315 62. Significant Linear Correlations Between Measures of Satisfaction, Dissatisfaction and Attitude with Hypothesized Behavioral Outcomes 318 63. Significant Linear Correlations Between Measures of Satisfaction, Dissatisfaction and Attitude with Hypothesized Behavioral Outcomes Positive Discrepancy Conditions Only 319 64. Significant Linear Correlations Between Measures of Satisfaction, Dissatisfaction and Attitude with Hypothesized Behavioral Outcomes Negative Discrepancy Conditions Only 320 65. Correlations Between Involvement and Emotion Measures.... 323 66. Number of Subjects Reaching Criterion Emotion Levels 326 67. ANOVA Summary of Satisfaction Emotions 327 63. ANOVA Summary of Dissatisfaction Emotions 328 69. Reliability of Emotion Measures 330 70. ANOVA Summary of Interest 333 71 ANOVA Summary of Joy 334 72. ANOVA Summary of Surprise 335 73. ANOVA Summary of Anger 335 74. ANOVA Summary of Disgust 337
LIST OF FIGURES Figure Page 1 Cardozo's Model of Satisfaction 23 2. Anderson and Hair Treatment of Satisfaction 24 3. Model of the Satisfaction Process 27 4. Measures Used in Consumer Satisfaction Research 49 5 Post Evaluation Responses 66 6. Satisfaction Responses 67 7. Satisfaction As Emotional Continua 90 8. Relative Influence of Satisfaction and Attitude on Certain Consumer Behaviors 100 9. An Extended Conceptualization of the General Satisfaction Construct 103 10. Descriptive Stimuli 203 1 1 Hypothesized Relationships 230 12. Perceptions of Coffee Strength Interaction Among Experimental Conditions 266 13. Attitude Toward Coffee Responses by Experimental Condition 286 14. Mean Responses to Standard Satisfaction Scales 296
Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy CONSUMER SATISFACTION: AN EXTENDED RESEARCH CONCEPTUALIZATION By DOUGLAS R. HAUSKNECHT December 1988 Chairman: Richard J. Lutz Major Department: Marketing Consumer satisfaction/dissatisfaction (CS/D) is conceptualized as the antecedent of specific post-consumption behaviors. The conceptualization presents a general construct, satisfaction, as the feeling that results from six necessary and sufficient conditions: expectations decision, personal experience perception response involvement and conf irmat ion/ dis confirmation of the expectations and decision concerning the object. A measurement technique designed to tap the emotional nature of the satisfaction construct is derived from the Differential Emotions Scale (DES) using a critical incident methodology. A satisfaction feeling is defined as the emotional pattern of interest, joy and surprise; dissatisfaction is defined as anger, disgust and surprise. An experiment was conducted using 203 subjects from convenience intercepts randomly assigned to the cells of a 2x2x2 design. Factorial manipulations of positive/negative discrepancy between performance and
expectations, presence/absence of opportunity to make a decision, and low/high response involvement with the test product were operationalized using ground roast coffee. Dependent variables included the DES, other commonly used satisfaction scales, attitudes, behavioral intent and actual behavior. Attitude and typical satisfaction scores and behavioral measures were consistent with the discrepancy manipulation; the better coffee averaged better scores than the poorer quality coffee. These measures also tended to indicate an effect of the decision manipulation, with free choice making subjects more critical. A low level of response involvement was evidenced in the lack of effects on the dependent measures and contributed to weakness in the measures of emotions. The DES measures of CS/D showed none of the predicted patterns. Correlation analyses between emotion measures and behavior or behavioral intent measures also displayed little consistency. The importance of a decision in CS/D studies was demonstrated by its influence on most measures. Among the measures attempted, the behavioral self-reports indicated promise for future investigation. Measures of perceptions versus expectations at the attribute level indicated that even a single, brief consumption experience may lead to a more complex cognitive structure.
CHAPTER 1 INTRODUCTION Purpose The study of consumer satisfaction/dissatisfaction (CS/D) has been recently criticized for the lack of shared constructs (Day 1983, 1980b; Swan 1983; Woodruff, Cadotte and Jenkins 1983b) and insufficient attention to experimental realism (LaTour and Peat 1979; Russo 1979). These problems lead to difficulty in making comparisons across studies or even in drawing conclusions from a single study. The dissertation addresses these concerns in several ways. The general construct of "satisfaction" is defined to be a feeling, distinct from attitude, with separate and distinguishable influences on behavior. The application of this construct to marketing situations defines the area of interest known as consumer satisfaction. Within this field of study, the object of the feeling may be a particular product/use occasion, 1 the brand in general, or the product class in general (that is, the object can be defined at any level of specificity Â— as long as it is defined consistently across the conceptualization and measurement of the study) Satisfaction feelings do not, however, invariably result from all consumption situations. Rather, there must be certain key factors 1 The word "product" is used generically throughout the dissertation and will generally encompass services and other consumer goods 1
which serve to generate, or at least activate, the mechanisms that arouse the emotions. When one or more of these factors is missing, the measurement of satisfaction as a distinct feeling should automatically be suspect. The dissertation presents these situational factors as necessary and collectively sufficient antecedent conditions for the arousal of the satisfaction feeling. The primary contribution of the conceptualization is the delineation of a theoretical network which integrates (1) the construct described by Oliver (1981); (2) the model proposed by Day (1983); (3) a measurement approach (Westbrook 1983) that operationalizes the construct as defined; and (4) the influence on the consumer behaviors that have been determined to be cf interest. One useful result of this theorizing and research is a template against which to compare empirical work on ''consumer satisfaction" to see whether the construct is justifiably invoked. Finally, the satisfaction research in consumer behavior is examined vis-a-vis satisfaction research in other disciplines and a case is built for a general construct and a general theory of satisfaction in which the field of consumer satisfaction is positioned and to which the definition and framework presented apply. Importance of Research in Consumer Satisfaction "Customer satisfaction with a product presumably leads to repeat purchases, acceptance of other products in the same product line, and favorable word-of-mouth publicity. If this assumption is correct, then knowledge about factors affecting customer satisfaction is essential to marketers" (Cardozo 1965, p. 244). This is the premise with which one of the seminal papers in consumer satisfaction began. That is, there
is some construct referred to as satisfaction, that has unique effects on behaviors that are of great interest to marketers. In fact, Marketing has been defined as ". the profitable creation of customer satisfactions" (Risley 1972, p. 10). This investigation falls within a growing literature that deals explicitly with postpurchase, consumption behavior (Belk 1984b). The Marketing Science Institute's 1976 conference on Consumer Satisfaction/Dissatisfaction (CS/D) was, in part, an attempt to address the need for rigorous conceptualization and definition (Hunt 1977a). Yet, such features are still lacking in the scholarly research in the area (Day 1983). If satisfaction is to have any value as a separate construct, then its definition and measurement must be distinct from other related constructs such as brand attitude, and it should be shown to have separable influences on phenomena of interest (MacCorquodale and Meehl 1948) The dissertation focuses on satisfaction as a unique construct and bases its conceptualization and measurement on this philosophy. The equating of satisfaction and attitude (e.g., LaTour and Peat 1979) and the operationalism approach (e.g., Aiello and Czepiel 1979) are rejected in favor of a more integrative framework in which satisfaction intervenes between its antecedents and the behaviors it is presumed to affect (Day 1983). Day's framework, however, implicitly assumes that the feeling of satisfaction Â— at some levelÂ— always results from a consumption situation. The dissertation argues that other antecedent conditions must be present before evaluations are made and feelings result. Accurate measurement of satisfaction, which includes the
ability to detect its absence, is therefore crucial to the conceptualization being proposed. Toward this end, a measurement technique is developed based on the emotional measures used by Westbrook (1983; 1987) and is used to predict behaviors such as those originally proposed by Cardozo (1965). This measurement scheme is (a) bound to the definition and conceptualization of satisfaction that are developed herein and (b) less susceptible to the experimental demand effects raised by more conventional techniques. The definition of satisfaction that is adhered to throughout the dissertation is the one proposed by Day: "an emotional response manifested in feelings and is conceptually distinct from cognitive responses, brand affect and behavioral responses. In particular, it is not a 'kind of attitude' as suggested by some authors" (1983, p. 113). The conceptualization that is proposed elaborates on this emotional response as resulting from the presence of certain necessary and collectively sufficient antecedents: (1) expectations, (2) decision, (3) personal experience, (4) perception, (5) response involvement, and (6) conf irmation/disconf irmation. Each of these antecedents is defined and discussed in greater detail in the context of deriving a general conceptualization that is derived from and applies to a range of human behaviors (Belk 1984c). Audiences Addressed by the Research Many businesses rely to a large extent on repeat purchasers and positive word-of-mouth among customers for their continued success. Unfortunately, a very common method of assessing (and thereby making
decisions about) satisfaction has been to monitor complaints and/or the absence thereof (Bohl 1987). Interest in consumer satisfaction by the federal government also has focussed primarily on complaint behavior (Technical Assistance Research Programs Â— hereafter TARP Â— 1979) and methods to improve the effectiveness of consumer complaining (Knauer 1983). This is not too surprising inasmuch as consumers who feel they have not gotten an adequate response from the company or industry involved can and sometimes do turn to state or federal agencies and legislators (Smith and Bloom 1984) But as will be shown later, complaining Â— especially to a third party after prior attempts at redress have failed Â— is a relatively extreme response that depends on more than just the level of dissatisfaction (Richins 1932). Moreover, the absence of dissatisfaction sufficient to prompt a complaint is not evidence for the presence of satisfaction sufficient to encourage word-of-mouth or repurchase. Complaining behavior, although not an adequate surrogate for dissatisfaction (or satisfaction) measurement, does offer the advantage of visibility. The abundance of complaints helps to maintain interest in consumerism among academics (Bloom 1982a; Bloom and Greyser 1981), authors of popular "How to Complain" books (Charell 1973; Eisenberger 1977; Horowitz 1979) and members of grassroots consumer groups (Barnes and Kelloway 1980) It should also be noted that dissatisfaction and the resultant negative behaviors are not the sole focus. Equivalent positive
behaviors (brand loyalty, complimenting, etc.) are also of interest. The seemingly more urgent complaint behavior, however, has garnered the bulk of the attention directed toward post-satisfaction behavior. The research reported in these pages should be of interest to marketing practitioners, regulators, consumers, and consumer advisors. Moreover, a consistent research paradigm can clarify the interpretability of satisfaction reports by third-party research suppliers (Serafin 1987). CS/D Literature to Date The last decade or so has witnessed a great deal of interest and scholarly research activity focused on consumer satisfaction, dissatisfaction and complaining behavior (Hunt 1982; 1983). The output resulting from this effort ranges across a spectrum of topics and details. Yet, despite the sheer volume (Hunt lists nearly 600 relevant references), there is only qualified agreement on the most basic of issues. No real consensus is approached concerning the definition of satisfaction, its measurement, or the psychological states or processes involved (Day 1983; Swan 1983; Woodruff, Cadotte and Jenkins 1 983a) The major consistencies within the field have been (1) the treatment of dissatisfactionÂ—satisfaction as a single continuum and (2) the adoption of the disconf irmation of expectations paradigm. A few authors have proposed two separate satisfaction and dissatisfaction constructs, arguing that the attributes that determine each are different (Swan and Combs 1976). Research that tests this proposition has generated, at best, mixed support in limited circumstances.
Similarly, alternatives and modifications to the consensus paradigm (i.e., that satisfaction is the result of perceived reality exceeding or matching expectations) have been discussed in limited contexts. As noted by Day (1983), disconf irmation-of-expectations models have been applied and tested at both attribute-specific and summary-evaluation levels. The literature has not addressed government concerns of how to increase satisfaction or decrease dissatisfaction (Leavitt 1977), industry questions regarding the seemingly conflicting needs to raise expectations to generate sales but lower expectations to generate satisfaction (Lutz 1980), or even how the research might aid individual consumers (Haines 1979). These broader questions can not be adequately answered until a broader framework for satisfaction is understood. Contribution of Dissertation Although we do not yet know enough about satisfaction to propose a general theory that can specify the influence of a multitude of variables across a range of situations (Russo 1979), the field is wellserved by attempts such as that by Day (1983) to circumscribe what is known within a general framework. The dissertation continues this effort by applying his definition and an extended version of the framework to the derivation of a prototypical experimental protocol. Such a protocol must include each of the necessary antecedent conditions so that the satisfaction construct may truly be tapped. This represents one step in a movement toward realism and viewing
consumers as subjects who instigate behavior rather than as objects affected by the marketer or experimenter (Olander 1980). Further, satisfaction is defined in emotional terms, requiring a measurement methodology consistent with such a definition. A set of emotions that comprise satisfaction is measured and its relationships to behaviors that have been theorized to result from satisfaction are explored. Finally, these measures are used to test the necessity of three of the antecedent conditions by testing for the presence or absence of the crucial emotions when the presence of the antecedents is manipulated. Dissertation Overview Chapter 2 reviews the current state of the CS/D literature. The chapter begins with a definition of the satisfaction construct and discussion of the paradigm that dominates the research in the area. Other approaches to the topic are then discussed, followed by a review of the various types of satisfaction that have been considered under the guise of consumer satisfaction. The importance of recognition of the interrelationships of types of satisfaction is emphasized in discussions of measurement issues and behavioral outcomes of satisfaction that conclude the chapter. In Chapter 3 an extended conceptualization of the construct is presented. General discussions of the nature of emotions and motivation introduce the chapter, followed by the description of satisfaction as an emotional construct not limited to the domain of consumer behavior. This latter point is highlighted by a discussion of the treatment of satisfaction in other literatures and the related
concept of attitude. Each of the necessary antecedents is then defined and its role in the overall process discussed. Finally, prior research in CS/D is evaluated with respect to conformity with the present conceptualization and the likely impact on interpretability of the findings. Chapter 4 presents explicit hypotheses based on the arguments previously detailed. The major hypotheses discussed represent tests of the necessity of the two of three antecedents that are introduced by the dissertation. A measurement technique capable of assessing the emotional nature of satisfaction is derived in the first half of Chapter 5. Following this, a research design is presented that uses emotion and other measures to test the hypotheses. Chapter 6 details the results of the study. Chapter 7 discusses the implications of the study for the new conceptualization and outlines future research that builds upon the findings
CHAPTER 2 LITERATURE REVIEW Introduction The topic of consumer satisfaction/dissatisfaction (CS/D) has been the subject of much study over the past decade. In this chapter, the literature resulting from this attention is reviewed. Although the term "satisfaction" has connotative meaning to most people, this is not adequate for examining and establishing the existence of a theoretical construct. This chapter, therefore, begin? with a look at the various definitions of satisfaction that have been proffered in the CS/D literature. Next, the disconf irmation paradigm Â— generally considered to dominate CS/D research--is outlined and its development is discussed. Other viewpoints on satisfaction are also mentioned briefly. Because the dissertation treats satisfaction as a general feeling that may arise from or focus on a variety of concepts or objects, it is necessary to show the diversity of objects that may evoke this feeling. These objects of satisfaction are arrayed by level of specificity (e.g., particular item versus product class), and implications for measurement and theory are discussed. One serious shortcoming in the CS/D area that has limited its ability to generate a more integrated and better received literature has been the lack of an agreed upon measurement methodology. The 10
measures that have been brought to bear are examined and their failings with respect to satisfaction theory and experimental demand artifacts are discussed. Finally, CS/D researchers are also implicitly or explicitly interested in behaviors that are thought to result from satisfaction. The chapter concludes by examining the link between the satisfaction feeling and these behaviors (or behavioral intentions). The large body of research that examines the consumer complaining phenomenon is considered, however, to be a separate literature and is discussed only as necessary. Definitions of Satisfaction Researchers in CS/D have been haunted since the early days of the area by the lack of a consensus definition of satisfaction: As a concept, at the level of thought, the understanding of satisfaction becomes weaker. In a simple sense, it lacks substantive meaning Â— that immediate sense of being able to transfer to another exactly what it is that is known and experienced, how it can be experienced, why it is important, and how it is a thought worth sharing and doing something about. As a concept, dissatisfaction has substantive meaning far exceeding satisfaction's. Dissatisfaction implies many actions: angry consumers, worried managers, protesting consumer advocates, and rule-making government officials. (Czepiel and Rosenberg 1977, p. 93) This early uncertainty did not dissipate. During the first CS/D conference, satisfaction was defined as (a) the happiness resulting from consumption, (b) a cognitive state resulting from a process of evaluation relative to previously established standards, (c) a subjective evaluation of experiences and outcomes associated with acquiring and consuming a product relative to a set of subjectively
12 determined expectations, (d) a two-factor process of evaluating a set of satisfiers and a set of dissatisf iers, and (e) one step in a process including prior attitudes, experience, post attitude and future purchase (Day 1980a, Hunt 1977a). Even more recent overviews and commentaries on the CS/D literature cite a persistent confusion concerning possible definitions (Woodruff, Cadotte and Jenkins 1983b) and attempt to provide consensual definitions that would be useful to the majority of researchers in the area (Day 1983, Swan 1983). As mentioned in Chapter 1, Day's definition has been adopted for use within the dissertation. Before elaborating the reasons why that definition was chosen, it seems appropriate to discuss the types of definition that were eliminated from consideration. Many authors seem to eschew theory and proceed directly to operationalizing and testing specific hypotheses of interest. This approach evades the question of definition and how the definition may fit a larger theoretical framework. The CS/D literature is rife with papers that view satisfaction through such an operationism perspective. The authors implicitly accept the operationalization of satisfaction and its measures as the definition of the construct and provide no additional conceptualization. Aiello and Czepiel explicitly define satisfaction as, "the responses given to simple 'How satisfied are you with .' questionnaire items with five response categories ranging from 'Completely satisfied' to 'Not satisfied'" (1979, p. 129). This method is quite successful at removing concerns about the correspondence between the definition of a construct and its measures. However, such
13 an approach leaves something to be desired in the construction of a theory. One use of "consumer satisfaction" that has dropped from vogue is as a social indicator (Hunt 1977a; Swan 1983; Withey 1977). The behavioral orientation of many of the researchers in CS/D has re-directed interest in the phenomenon from such attempts to examine general consumer welfare to elaborations on individual personal experience and behavior. Although the need to measure consumer welfare still exists, the satisfaction construct and the theory surrounding it have complicated the "Are consumers sufficiently satisfied?" question. Thus, the issue of how best to aggregate satisfaction across a population will not be dealt with here (see Pfaff 1977 for a discussion of some of these issues). Another controversy that arose from the desire to optimize consumer welfare concerned whether to define the goal as "maximizing satisfaction" or "minimizing dissatisfaction" (Leavitt 1976, 1977). Although these policy questions are not discussed in the dissertation, the issue of whether or not to separate satisfaction from dissatisfaction def initionally and theoretically is a valid concern. The so-called two-factor theory (attributed to Herzberg, Mausner and Snyderman 1959) is a direct outgrowth of the job satisfaction literature and is detailed by Leavitt (1977). In essence, it is felt that satisfaction is the result of some critical incident or process involving "intrinsic" factors whereas dissatisfaction results from "extrinsic" factors. In the job satisfaction literature, intrinsic factors included such attributes as achievement, recognition,
advancement and responsibility. Extrinsic factors, on the other hand, are such things as salary, interpersonal relations with superiors, subordinates and peers, status and job security. A particularly salient experience involving one of these factors evokes the satisfaction or dissatisfaction response. In investigations of satisfaction, descriptions of such "critical incidents" are used to assess the individual's overall response. The difficulty with applying this rationale to consumer satisfaction lies in defining the types of factors a priori. That is, the discriminations made in the job satisfaction literature do not map well onto consumer situations. Leavitt attempted to classify factors affecting CS/D as intrinsic or extrinsic based on surface similarity to the classification of factors in the job satisfaction literature. The research failed to generate the unique pattern of influences that were postulated. He would have been better served by relying on or extending the more theoretically driven distinctions used by Swan and Combs (1976). In their paper, Swan and Combs propose that satisfaction results from evaluations of expressive (nonmaterial, psychological) dimensions and dissatisfaction results from instrumental (physical) dimensions of a product. They theorized that satisfaction can result only when both instrumental and expressive performance expectations are fulfilled. They admitted to difficulty in categorizing some dimensions (e.g., comfort) in their preliminary study involving clothing. Although they found some support for their operationalization of two-factor theory, it can be argued that the findings were bound to the critical incident
15 methodology. They extended the use of this technique in which people discuss specific occurrences that they recall to the consumer arena, but without arguing why these most salient events are necessarily representative of the majority of consumer satisfaction or dissatisfaction responses. In order to support clearly a two-factor theory of consumer satisfaction, better conceptual definitions from which unambiguous operational definitions of the two types of factors can be derived must be provided. Further, factors defined in this manner must be shown to be uniquely linked to the feeling (satisfaction or dissatisfaction) which is said to result. In the final analysis, Swan and Combs did not go beyond the disconf irmation of expectations paradigm except to order individual levels of expectations from physical to psychological needs (similar to the hierarchy attributed to Maslow 1943). This interpretation mirrors that by Scherf (1974) in which he applied Alderfer's hierarchy to consumer satisfaction. In an attempt to replicate the Swan and Combs findings, Maddox (1981) reported a similar discomfort with the discrimination between types of factors and the difficulties that arose in the attempt to operationalize the distinction. His research found different results from intrinsic and extrinsic factors across products. He concluded that the interaction with product class is probably more damaging to two-factor theory than supportive. The greatest difficulty that this perspective encounters is with justification as a theory versus simply being a measurement device. The conf irmation/disconf irmation evaluative process appears to remain
16 intact with both satisfiers and dissatisf iers ; no new process is proposed. The most likely result of this particular stream of research will be to identify what it is that makes the critical discrepancy (i.e., between expectations and perceptions) salient. In some product classes, it is conceivable that expressive dimensions will be more noticeable than instrumental dimensions when each exceeds expectations. For example, in the Swan and Combs case, compliments from friends would be more noticeable and lead to more positive responses than "additional" durability in clothing of college students. Similarly, lack of durability Â— fading, tears Â— would be more noticeable than lack of negative comments (assuming that students are not given to brutal frankness, whereas the reverse may be true if mothers were asked about critical incidents regarding the clothing of young children) This notion of salience of the expectations-perception discrepancy will be discussed in Chapter 3. Westbrook, using similar reasoning, reports different good and bad emotional dimensions arising within the same process (1987). His measures, however, do not reduce necessarily to definitions of separate satisfaction and dissatisfaction constructs. Even having accepted dissatisfaction/satisfaction as a single continuum (that will, henceforth, usually be referred to simply as satisfaction), we must still distinguish among the cognitive, attitudinal and emotional definitions of the construct. Again, these must be evaluated in light of the underlying theory of satisfaction that each definition represents
17 The disconf irmation of expectations paradigm (hereafter, the disconf irmation paradigm), which will be discussed in more detail presently, posits a relationship between (1) the discrepancy between expectations and perception and (2) certain resultant attitudes and behaviors. Some authors have defined satisfaction to be this discrepancy Â— or the judgment which identifies the discrepancy Â— rather than a separate construct (Aiello, Czepiel and Rosenberg 1977; Day 1980a). Note that the concern here is with the equivalence by definition of satisfaction and the discrepancy judgment, not with the use of one to predict (or measure) the other. This definition is rejected in the belief that satisfaction and the discrepancy judgment are conceptually distinct, the former carrying more motivational implications, and should remain separate in the lexicon derived in the study of postpurchase processes (Oliver and Westbrook 1982; Woodruff, Cadotte and Jenkins 1983b). The studies that initiated interest in postpurchase consumer behavior were concerned with perceptions of product performance (Anderson 1973; Cardozo 1965). Many marketing practitioners appear to equate performance with satisfaction (e.g., Ford Motor Company's "Number One in Customer Satisfaction" advertising campaign based on reports of the service requirements of new cars.) But this point of view ignores the potential for divergence under situations posited by the disconf irmation paradigm. For example, a consumer who experiences product performance which would be evaluated as merely fair on some absolute scale (indicating a low positive or even negative attitude) may in fact be satisfied because the performance exceeded expectations.
(A realistic example might be a lowincome consumer who has purchased an inexpensive used automobile. Although the car does need occasional repair and is not given to quick starts and high speeds, it performs reliably and lasts beyond its expected life.) Thus, it is probably best to maintain the conceptual distinction between "performance perception" and "satisfaction" and not attempt to implement a satisfaction strategy via performance engineering. (In the Ford case, improved performance may have been matched or even exceeded by expectations heightened by the promotional campaign.) A similar discrimination is attempted by Bahr (1982) although he makes the distinction using the terms "dissatisfaction" and "disappointment" Â— adding yet another concept to the literature. Another concept distinct from but related to satisfaction is attitude. Again there is a tendency to confound, by measurement devices and by definitions, constructs that are separate in theory (see Hartley 1967 for an interesting discussion of such definitional fallacies). Some authors have recognized this inconsistency and characterize satisfaction as "a special kind of attitude" that is limited to a post-purchase, post-consumption existence (Czepiel and Rosenberg 1977; LaTour and Peat 1979). Without complicating matters any further, satisfaction can be maintained as a separate concept from attitude by acknowledging the differences in motivation level (Weiner 1985) that each implies and the likelihood of changes over time. LaBarbera and Mazursky (1983) cite Oliver's (1981) contention that the surprise or excitement elements of satisfaction are of finite duration; as these decay, satisfaction
19 decays into attitude. In view of this argument, however, it was not clear whether they measured attitude or satisfaction in their study. The relationship between satisfaction and attitude that was hypothesized by Oliver is compelling and is examined further in Chapter 3. This conceptualization, however, requires definitions that allow one to discriminate between satisfaction and attitude and order them within a theoretical network. Moreover, Oliver appears to require satisfaction to be a construct which intervenes between emotion and attitude, "it is the summary psychological state resulting when the emotion surrounding disconfirmed expectations coupled with the consumer's prior feelings about the consumption experience satisfaction soon decays into one's overall attitude toward purchasing products" (1981, p. 27). If we accept the definition of attitude as "a learned implicit response that varies in intensity and tends to guide an individual's overt responses" (Shaw and Costanzo 1982, p. 285), then we are left to identify the distinguishing characteristics of the satisfaction construct. Swan (1983) makes some advances by tying cognitive and affective elements of satisfaction together and arguing that there are potential differences in the type of information used to derive satisfaction than that used in deriving (or learning) attitude. Taken as a whole, the foregoing arguments sum to a conceptualization of satisfaction as something that mediates between a disconf irmation judgment and some post-experience attitude (Oliver 1977; Oliver and Westbrook 1982). Cardozo's (1965) original concept also proposed that satisfaction led to or at least had distinct
20 influence upon particular behaviors. What is required then, is a more dynamic and more motivational definition of the construct than that provided for attitude. Such a definition has been proposed by Day: an emotional response manifested in feelings and is conceptually distinct from cognitive responses, brand affect and behavioral responses (1983, p. 113). Thus, Day explicitly defines satisfaction as emotion rather than a precursor (as with perceptual or attitudinal definitions) or a product (as Oliver 1981 describes). Chapter 3 introduces and discusses theories of emotion and their relationship to more general motivation. The emotional character of this definition is amplified by developmental work conducted by Westbrook (1983, 1987). He used measures derived from Izard's Differential Emotions Scale (1972) to describe emotional profiles of post-purchase consumers. (This theory of emotions is explicated further in Chapter 3.) By defining satisfaction as an emotional phenomenon and using measures of emotion in its assessment, Westbrook was better able to describe the construct directly, albeit as a result of emotion (1987). Moreover, as a concept founded on emotions, satisfaction is unique in the marketing literature. Thus, the overlap between related constructs, such as disconf irmation and attitude, is eliminated and the potential for measuring each individually is enhanced. The adopted definition should also be held to be distinct from one final class of definitions: the (cognitive) evaluation of emotions (Day 1980a; Oliver 1977; Westbrook 1981). This has been characterized as a "stepping away from an experience and evaluating it" (Hunt 1977a,
21 p. 459). This formulation seems to require that emotion precede cognition (and, presumably, the behaviors that have been posited). Whether or not other cognitions precede these emotions is not clear. These definitions have not been pursued vigorously in the CS/D literature (their proponents having moved to more emotional and less cognitive versions) and are therefore also rejected in this dissertation. In summary, it is perhaps best to recall the argument raised by Swan (1983) that the question of what satisfaction is "in essence" is not answerable. Rather, we wish to achieve a consensus accepting a rigorous definition that is useful in a larger theoretical framework/ an approach that is more consistent with a functionalist approach (Alderson 1965). These requirements are met by Day's definition presented above and the framework and measurement techniques presented below. The Disconf irmation Paradigm As was stated previously, the one major agreement within CS/D has been the acceptance of the disconf irmation paradigm. Essentially all of the theory and research in the area derives in part from an assumed interaction between expectations and perceived outcomes, and a strong natural preference for the expected over the unexpected (Silverman 1968). This recognition of the crucial role of expectations traces to the work by Cardozo (1965). He introduced to marketing the notion that prior expectations may influence one's perception of reality (Aronson and Carlsmith 1962; Watts 1968). Prior to this, physical performance
22 of a product was thought to drive directly consumer response. Cardozo suggested that expectations moderate the perception of this physical performance. It is this modified perception, which he called satisfaction, that drives the post-purchase responses (Figure 1 ) For a time, interest in the post-purchase consumer continued to focus on this interplay between expectations and perception (generally, perceived product performance). The next major study in the area attributes the conf irmation/disconf irmation episode to the decision made by the purchaser rather than to a simple confirmation of particular expectations (Cohen and Goldberg 1970). This formulation is particularly useful because it avoids the complications that arise from attempts to specify whether expectations and performance react at an attribute specific or summary level. This issue and others raised by Cohen and Goldberg which have not as yet been pursued in CS/D are discussed in more detail in Chapter 3. Anderson and Hair (1972) provided an interesting and useful perspective on the role of expectations. They argued that consumer satisfaction decreases as the consumer's experience with products of improving quality increases because improvements in quality raise expectations. They also point out the potentially self-defeating nature of promotion--i e the possibility of raising expectations too high Â— which was alluded to in Chapter 1 In their paper, consumer satisfaction and product perception were treated as conceptually distinct (Figure 2). Expectations were still thought to affect product perception (through the same sort of mechanism that Cardozo had proposed), and this relationship became the focus of the empirical
23 Expectations Physical Performance Perceived -> Performance Â— "Satisfaction' Repeat Purchase Â•> Product Line Acceptance Word-of-Mouth Figure 1 Cardozo's Model of Satisfaction based on Cardozo (1965)
24 Expectations Product -(Objective) Product Perception > Satisfaction Dissatisfaction Figure 2 Anderson and Hair Treatment of Satisfaction Based on Anderson and Hair 1972
25 study. The distinction between perception and satisfaction was lost in the study design. Anderson restates the conceptualization in his later, more frequently cited paper (1973), in which he proposes that although satisfaction has no literal definition it "might be measured by the degree of disparity between expectations and perceived product performance" (p. 38). Operationally, however, he still treats perceived performance as the variable of interest and does not further elaborate the relationship with satisfaction. Other investigations of post-purchase evaluations have not invoked the satisfaction construct and thus do not perpetuate the ambiguity in terms (Oliver 1977; Olshavsky and Miller 1972; Winter 1974), although each is occasionally cited as having been a satisfaction study. LaTour and Peat (1979) reviewed this "product performance" literature and were careful to point out that "perceived product performance cr quality is not necessarily the same as satisfaction" (p. 432). They detailed three psychological explanations of how expectations and reality may interact to produce perceived performance: 1) contrast; 2) assimilation-contrast; and 3) cognitive consistency theories. They conclude that expectations are really only likely to affect perception when objective performance is ambiguous or otherwise difficult to assess (for example, when dealing with providers of services; Zeithaml 1981). Such modifications, when they occur, would resemble an assimilation effect (i.e., perception would be closer to expectations than it would in objective reality) and would presumably be driven by the assumed value of the information provided by prior
26 expectations versus that provided by the physical product. This is explicated more fully in their presentation of a comparison level theory of satisfaction discussed below (see also LaTour and Peat 1978; 1980). LaTour and Peat conceived of satisfaction as equivalent to post-consumption attitude. This is incompatible both with the orientation taken in the dissertation and with more recent comprehensive satisfaction models that include attitude as a separate (but related) construct. Again, the most compelling of these, and the one on which the dissertation builds, was proposed by Day (1983) (Figure 3) In this model, perception, disconf irmation, attitude (affect) and the satisfaction feeling are separate constructs. The comparison of perception to expectation may result in positive or negative disconf irmation or in confirmation of the expectation. 1 The disconf irmation judgment results in the feeling that has been defined as "satisfaction." This satisfaction feeling, in turn, drives specific responses, including word-of-mouth or complaining, changes in repurchase intention for the brand and/or the product class. Satisfaction feelings also influence brand and/or product class attitude (which Day called affect) after the evaluation has occurred. As presented, this is not intended to be a causal model because some relationships may be more important for certain product-usage situations than others (e.g., a link between expectations and 1 For simplicity, the comparison and its result are hereafter referred to as disconf irmation.
28 perception as just discussed). What the figure does provide, however, is a diagram of how the flow would proceed in the prototypical situation. Day has taken a crucial step in developing the CS/D literature. In attempting to derive a consensus, he has proposed a framework for all consumer satisfaction which could be restated as a set of lawlike generalizations, each of which can form the basis of sub-theories. Relationships posited here that have been examined (e.g., disconf irmation causes satisfaction versus perception causes satisfaction) have not been presented as part of a grander theory or nomological network (S. Hunt 1976). The increased level of generality that is represented by Day's model is carried further by the conceptualization of a general satisfaction construct that is explicated in Chapter 3 (note that, even for Day's model, no clear discrimination is made between product class level and brand level constructs) Other comprehensive satisfaction theories have recently been proposed (Table 1). Woodruff, Cadotte and Jenkins (1983a) describe a model very similar to Day's. They, like Day, separate norms from expectations and define satisfaction as an emotion. A key element in this model is the attempt to specify the relationship between the cognitive disconf irmation process and the emotional result (satisfaction). Specifically, they predict that negative disconf irmation (performance falls below norms) will lead to negative feelings (dissatisfaction); positive disconf irmation (performance exceeds norms) will lead to positive feelings (satisfaction); and
29 Table 1 Perspectives on Consumer Satisfaction Paradigm Disconf irmation Variants two factor mood experience (also comparison level and value percept disparity) Dependent variables immediate postpurchase satisfaction, intentions, repurchase postpurchase satisfaction postpurchase satisfaction postpurchase satisfaction indifference postpurchase satisfaction attribution Equity Normative deficit Composite postpurchase satisfaction, complaining distress or postpurchase satisfaction satisfaction with goods and services as comprising a standard of living overall (global) satisfaction Causal mechanism Satisfaction covaries with difference between performance and expectations satisfaction results from performance on instrumental dimensions, dissatisfaction results form performance on expressive dimensions satisfaction results in part from mood state (or business attitude) at purchase expectations are discrete from post product experience, both affect satisfaction, expectations derive from many sources satisfaction occurs under usual disconf irmation conditions but no such feeling results when confirmation occurs problems attributed to external (i.e., marketercontrolled) sources are more likely to arouse dissatisfaction and complaints distress increases as consumer's outcome becomes less favorable than (1) the marketer's or (2) another consumer' s satisfaction occurs when the quality of goods and services equals social norms overall satisfaction = summation of satisfaction with attributes Note: Based on Swan (1983)
30 simple confirmation (performance functionally equivalent to norms) will lead to a neutral result. The perspective offered by Woodruff et al. is rejected on a number of grounds. The limiting of expectations to brand specific predictions of performance is unsupported in the CS/D literature and an unnecessary complication. Rather than separate brand and product prior experience, it is more parsimonious to consider these as separate influences on a single expectation construct. Similarly, although the suggestion that imperfect perceptions and comparison processes lead to discrete, step function disconf irmation judgments and resultant satisfaction feelings does provide a methodological advance over the previously assumed continuous relationships, this is not a development specific to satisfaction theory. Rather, Woodruff et al. have proposed technical approaches to examining perceptual discrimination of performance or performance versus expectation discrepancies (following a kind of Signal Detection Theory logic) but no real modification to the disconf irmation paradigm. Other authors have sought to evaluate versions of the disconf irmation paradigm through causal modeling. Churchill and Surprenant (1982) report evidence of different processes in the cases of satisfaction with a nondurable (chrysanthemum) and a durable product (video disc player), with the former depending on both performance and disconf irmation, and the latter dependent only on product performance. In a similar vein but with contradictory results, Wilton and Tse (1983) report that satisfaction with a compact disk
31 player (a durable product) was dependent on both performance and dis confirmation. While the intent of each of these studies (i.e., testing the complete process model) is laudable, it can be argued that both suffer from conceptual and methodological defects that render the findings equivocal. At this point it should suffice to note that neither experiment operationalized a purchase situation. Because both are interested in consumer satisfaction and both define this construct as post-purchase they possess a fatal flaw. This issue and others are discussed in more detail in Chapter 3 which presents a more comprehensive description of the prototypical satisfaction evoking situation and discusses the conditions necessary to evoke the feeling. Finally, some modifications to the disconf irmation paradigm have been proposed in the CS/D literature that are beyond the scope of Day's conceptualization of satisfaction and are not central to the dissertation. These embellishments do however reflect how the general paradigm can be elaborated by detailed study within certain portions of the model. The first of these has already been alluded to; that is, the possibility that a "zone of indifference" exists which may result in no particular satisfaction feeling resulting from a given consumer situation. Woodruff, Cadotte and Jenkins (1983a) and Swan and Trawick (1980a) assert that satisfaction feelings do not result when performance simply meets expectations (or the discrepancy between them is not detectable) and include this confirmation effect as part of the theory. Both Dav (1977) and the current conceptualization include a
32 neutral state or lack of satisfaction feeling as a potential outcome; however, it is not felt to be specifically linked to a confirmatory judgment (i.e., a lack of a discrepancy). Other authors have investigated the influence of general attitudes toward business (Anderson, Engledow and Becker 1979) or the consumer's overall emotional state or mood (Westbrook and Cote 1980) on the satisfaction process. Although each reports support for relationships in the directions hypothesized, it is conceivable that the dependent variable in each study was attitude rather than satisfaction. While it is not argued that attitudes toward business and/or mood will not influence satisfaction, neither construct is central to the process. As mentioned above in the section concerning definitions of satisfaction, the Two-factor Theory (Swan and Combs 1976; Swan and Trawick 1980b) can be considered an extension of the basic disconf irmation paradigm. Essentially, Swan and his colleagues (and others--see Day 1983) have chosen to focus attention on the expectations portion of the model. Although presented as separate constructs affecting satisfaction, each of the variables proposed can be considered to be an influence on the satisfaction construct. That is, the expectations that become important in the disconf irmation experience may reflect attitude toward business, mood, or only certain physical product dimensions (categorized as instrumental or expressive), depending on the situation. Finally, the CS/D literature is so young that the investigation of a single end-user's satisfaction with consumer products is sufficiently complex to challenge the field. Lambert, Dornoff and Kernan (1977),
33 however, open the way to more expanded horizons by looking at the post-purchase industrial buyer. Although only an exploratory study involving post-choice attitudes, the paper does raise the question of whether or not the disconf irmation paradigm would be amenable to applications in which buying centers or other group decision-makers are involved as well as situations in which the buyer is not the user. Given the perspective involving emotions that this dissertation has adopted, it is difficult to address these situations within the classical satisfaction framework. It is argued in Chapter 3 that, among other causes, satisfaction emotions result from the decision to commit to and personal experience with the product. Thus, the situations suggested by Lambert et al. do not translate directly into the satisfaction paradigm (note that they do not invoke this paradigm Â— they are interested in attitudes after a choice is made) Other Perspectives on Satisfaction Although the disconf irmation paradigm dominates the CS/D literature, some theorists have proposed other views purportedly distinct from this perspective. Swan (1983) briefly summarizes three paradigms in addition to disconf irmation. Of the three, only the equity approach considers specific satisfaction with a good or service at the same, summary, level as the current form of the disconf irmation paradigm. Equity theory, borrowed from social psychology, basically posits that a person involved in an exchange relationship compares the ratio of his outcomes from : inputs to the exchange with a similar ratio for another person involved in the relationship. When the ratios are
34 equal, the exchange is said to be equitable; when unequal, the exchange is inequitable. Inequitable relationships create distress which results in attempts by the individuals to restore equity (Shaw and Costanzo 1982). Huppertz (1979) introduced equity theory into the CS/D literature, followed by Mowen and Grove (1983) and Evans (1983) (see also Huppertz, Arenson and Evans 1978). None of the authors has dealt with the critical conceptual issue of specifying how the comparison other will be identified. The theory, as presented in social psychology, generally assumes that the comparison other is the one with whom the exchange relationship is forged (Shaw and Costanzo 1982--e.g., a retail merchant). Mowen and Grove used another consumer who got a better or worse deal in a similar exchange as the comparison in order to generate distress; Huppertz et al. studied memorable exchanges subjects had actually experienced with merchants. Another problem faced by equity theory analyses, even in social psychology, is the quantification of the inputs and outcomes of each party. Even if this can be handled in devising an experimental protocol, the issue of how the evaluation process operates in the typical consumer situation remains unexplicated. In summary, application of equity theory to satisfaction is still in its infancy, as suggested by Swan (1983). Before this approach will gain acceptance, its particular conceptual questions must be addressed along with the issues of definition and measurement with which the disconf irmation paradigm is still wrestling. Evans, for example, from
35 the perspective taken here conducted an attitude study, rather than a true satisfaction study. Swan discussed a second paradigm that he labelled the Composite approach. This perspective does not posit a process different from disconf irmation, but suggests that the judgment occurs at attribute specific levels and satisfaction with the product is a composite of these separate satisfactions. As an example, Swan cites Westbrook's (1981) investigation which explained department store satisfaction "in terms of satisfaction with sales personnel, store appearance, merchandise selection, etc." (1983, p. 124). Further development of Composite approaches is awaited. For example, it would be useful to understand how one might discriminate between a "summing" of discrepancy cognitions and a summing of satisfactions, each resulting in an overall satisfaction. In particular, the summing of emotions (as satisfaction is defined in Chapters 3 and 5) is a difficult concept to reconcile. The next section of the present chapter builds its discussion of Types of Satisfaction on a similar hierarchical logic. However, discriminations are at the expectation and perception levels and the conceptualization remains within the d iscc-nf irmation paradigm. The last paradigm, the Normative Deficit approach, is more concerned with the qualitv of life definition of satisfaction that was dismissed earlier in this chapter (Morris, Winter and Beutler 1976). This perspective assumes that societies possess norms regarding what an individual or household should expect from the marketplace. To the extent that one's current state does not reach this level,
56 dissatisfaction occurs. Swan uses the example of housing status that requires older children of opposite sexes to share a bedroom. As outlined, Normative Deficit shares some concepts with the Mowen and Grove operationalization of equity theory, i.e., "I should get what the other guy is getting." This view, however is less concerned with relative comparisons (inputs versus outcomes) than with absolute comparisons (specific ownership per individual or household) (Morris 1977; Withey 1977). This version of consumer satisfaction may be better captured in other terms such as "anomie" (Stein 1980) or "consumer discontent" (Lundstrom and Lamont 1976), that attempt to describe the totality of the consumer X marketplace interaction. It is not clear, for example, from Swan's discussion whether a Normative Surplus would create satisfaction beyond that achieved by Normative Parity. Approaches to satisfaction other than those discussed by Swan may be considered extra-paradigmmatic by some in the field. Two-factor theory (Maddox 1981; Swan and Combs 1976) has already been determined to be more a theory of expectations than a theory of satisfaction. Similarly, Comparison level theories (Barbeau and Quails 1984; LaTour and Peat 1978, 1979, 1980) are also concerned with specifying what it is that is compared against absolute performance. These too, can be considered expectations under the disconf irmation paradigm (see the section in Chapter 3 on Expectations). Other models have been proposed and discussed as "alternatives" to the disconf irmation paradigm. These generally have been theories borrowed from Psychology and applied to expectation formation. Principal among these are applications of the
37 Brunswik Lens model (Tapp 1984) and Value percept disparity; i.e., conformity with a person's value also influences satisfaction (Westbrook and Reilly 1983). As expectation theories, these can be subsumed within the disconf irmation framework. 2 Attribution theory has been introduced into the CS/D literature as an explanation for consumers' choice of alternative responses to a dissatisfying experience (Folkes 1984a, 1984b; Valle and Wallendorf 1977). In this context, it is really a part of the consumer complaining literature which is discussed briefly below in the section concerning Results of Satisfaction. More recent theory and research by Folkes and her colleagues (Folkes, Koletsky and Graham 1987; Folkes and Kotsos 1986) have addressed the influence of attribution on satisfaction itself. They, and others (e.g., Westbrook 1987), argue that the source of disconf irmation influences the arousal of emotions. This formulation reflects the concept of locus of control or responsibility (Carlsmith and Freedman 1968; Weiner, Russell and Lerman 1978) that posits different responses to disconf irmation attributable to one's self, another party or an uncontrolled situation. Westbrook describes the likely emotional profile that would result in each instance. Oliva and Burns (1978) present an innovative mathematical approach, Catastrophe Theory, to discriminate between those who will and those who will not engage in complaining behavior. Although 2 Another theory, constraint-reinforcement (Shelly 1972) reflects more of an operant conditioning approach to the concept. This behaviorist tradition is sufficiently far afield of the other theories discussed that it is not investigated further.
38 neither a psychological approach nor even a satisfaction paper per se, this work is mentioned here because the mathematics proposed could present an interesting way of discriminating satisfied from dissatisfied consumers. This theory has some promise in integrating attempts to sum attributes (Composite approach) with those separating attributes into categories (Two-factor theory) and with the more dominant, gestalt approaches. In essence, the theory offers a mathematical interpretation of discontinuous phenomena that are dependent on direction (e.g., it may be a greater distance from left to right than vice versa). The mathematics may provide solutions to such problems as the apparently uncertain levels of disconf irmation that are necessary to reach "satisfaction" or to prompt complaining. The further application of such mathematics, by suitably sophisticated users, could provide general models to define such states. As indicated before, the CS/D literature is sorely in need of basic generalizations Types of Consumer Satis faction CS/D theories or paradigms have developed predicated on the assumption that "consumer satisfaction" was the construct under scrutiny. In this section, the various levels of abstraction at which the concept has been operationalized are discussed and the necessity of specifying the relationships among these is demonstrated through a rationale derived from the literature that examines the consistency between attitude and behavior. Consumer satisfaction is a term that has been applied to consumers' responses to target stimuli ranging from a particular
39 product usage occasion to the totality of life (Table 2). Of these, the most often discussed and operationalized in the literature has been satisfaction with a particular item, generally a product (Westbrook 1980c). This is perhaps a vestige of surveys that had focused on individual items that prompted a satisfaction (usually negative) response that was of interest to marketers (e.g. Day and Landon 1976; Summers and Granbois 1977). An example of Individual Item satisfaction can be drawn from one condition of a study reported by Churchill and Surprenant (1982; Surprenant and Churchill 1984). The study concerned modelling the satisfaction process for two types of goods: durables (a video disc player) and nondurabies (a chrysanthemum plant). Subjects in one condition of the design were given a chrysanthemum and interviewed as to their satisfaction with the plant after four weeks. Thus, they were exposed to this one example of the new hybrid and the questioning focused on the plant they had been given, and on the experience they had with it over the four week "trial" period. In contrast, a corresponding group of subjects was merely shown a four-week old plant and asked to respond to a similar questionnaire. 3 This latter condition mirrors the designs of the early product performance studies (Anderson 1973; Cardozo 1965) in which only a single experience with a single product was allowed. This roughly corresponds to satisfaction with the Purchase/Usage Occasion, the 3 Problems with the design of this study and implications for its usefulness as an examination of satisfaction as defined here are discussed in Chapter 3. At issue presently is simply the level of abstraction of the target of the satisfaction feeling.
40 Table 2 Types of Consumer Satisfaction Type Cited in Purchase/Usage Occasion Individual Item Brand/Manufacturer Product Class Store/Retailer Marketplace (Consumer Discontent) (Surprenant and Churchill 1984) (Churchill and Surprenant 1982; Shuptrine & Wenglorz 1980) (Diener 1977) (Gibson 1981; Hunt 1976; Ortinau 1982) (Westbrook 1981 ) (Aiello and Czepiel 1979; Lundstrom and Lamont 1976; Hunt 1976) Life (Sherman and Schiffman 1984; Withey 1977)
41 classification which heads Table 2. In many contexts, satisfaction with the Item is resultant from satisfaction with the series of Usage Occasions involving it. To a degree, the satisfaction types in Table 2 are arrayed along this dimension of cumulativeness or generality (although this is not strictly true). Specifying the exact relationships among these is well beyond the scope of the dissertation but two varieties of relationship are suggested. As just discussed, a number of more specific responses may accumulate in some way (additive, Bayesian, random walk?) and generate a more general response (Andreasen 1977a, b). On the other hand, a specific stimulus may be interpreted as representative of the more general stimuli (e.g., the use obtained on this sample Occasion is judged to be typical for this particular Item) and the response will be attributed to the broader classification (using this Item will always make me feel this good/bad). This latter prediction is based on the same rationale that predicts that more general norms (e.g. how this Item performed in the past) influence a more specific expectation (how this Item will perform on this Usage Occasion). Similarly, users may choose to use the Occasion as case information in order to make judgments about the Item, as discussed in the decision literature (Ofir 1982). The same logic carries through to generalizing beyond satisfaction with the Individual Item. Whereas Westbrook (1987; 1933; 1980a) examines consumers' responses to their individual automobile, Ortinau (1982) asks consumers to use personal experience with their recentlypurchased automobile in responding to questions about the model
42 (Pontiac Grand Prix) in general. By similar extrapolation, one might wish to consider satisfaction with the entire product category (Hunt 1976; Shuptrine and Wenglorz 1980), which may later result in more generalized behaviors (e.g. commenting to a friend about American cars or automobiles as a whole). Similarly, a satisfaction response to the Item can result in generalization to the Manufacturer or Brand even though they are no longer responsible for the product. Diener (1977) reports negative attributions to manufacturers of cosmetic products that were re-packaged and re-sold as samples by an unscrupulous and unauthorized vendor. In this particular case, the actual retailer was virtually unidentified and did not become the target of dissatisfaction feelings or responses. Satisfaction with the Retailer has, however, been the subject of some CS/D research. Westbrook (1981) discusses Retailer satisfaction as resulting from a set of component satisfactions (e.g., with salespersons, store environment, etc.) that distinguish it from Product satisfaction (see also Rodgers and Sweeney 1980). Other authors have discussed Retailer satisfaction implicitly as resulting from satisfaction with products carried (e.g., Bearden and Teel 1983) or follow-up service (Vredenburg and Wee 1986). This conceptualization has also been extended by authors interested in complaints about products and how consumers' satisfaction with complaint handling affects satisfaction with the Retailer (Diener 1980; Gilly and Gelb 1982; Lippert and MacDonald 1981; Resnik, Gnauck and Aldrich 1977; Resnik and Harmon 1983; Summers and Granbois 1977).
43 Aiello and Czepiel (1979) posit a similar hierarchy in which Product/Service satisfaction generalizes to Enterprise Satisfaction which similarly generalizes to System satisfaction (see also Aiello, Czepiel and Rosenberg 1977). This last concept, referring to "consumers' subjective evaluation of the total benefits received from the operations of the institutional marketing system" (Aiello and Czepiel 1979, p. 129), closely resembles the Consumer Discontent definition of consumer satisfaction discussed previously. This level of analysis has quite a history in the early CS/D literature (Hunt 1976; Lundstrom and Lamont 1976; Scherf 1974; Strumpel 1974) but is of less concern in present theorizing (but see Liefeld 1980; Olander 1980). This very generalized satisfaction with the Marketplace has also been implicated as one determinant of satisfaction with Life (Leavitt 1976; Sherman and Schiffman 1984; Witney 1977). As an ultimate generalization of the concept, Life satisfaction also includes Job satisfaction and other non-consumer behavior types that are discussed briefly in Chapter 3. Presumably, Life satisfaction represents the result of a cumulative rather than an attr ibutional process as had been described previousiv. Such processes are also posited in relationships among Individual Item satisfaction and satisfaction with (a) the purchase process (Day 1977; Wanous and Lawler 1972), (b) warranty service (Bernacchi, Kono and Willette 1930), and (c) other indirect aspects of the product such as a restaurant's use of consumer surveys (Swan, Trawick and Carroll
44 1981) or (d) the systems used to deliver health care services (Langmeyer and Miaoulis 1981; Locker and Dunt 1978). At titudinal -Behavioral Consistency As the discussion of types of satisfaction becomes more fine-grained, the necessity of specifying all constructs in a given model or operationalization at the same level of generality becomes evident. This issue has been discussed at length in the attitude literature. A brief review of the conclusions provides a basis for evaluating the usefulness of the concept of levels of specificity for CS/D. In the attitude literature, the most direct analogy to the preceding discussion is Fishbein's (1966) discrimination between attitude toward an object and attitude toward a class which includes that particular object. He argued that, although these may be related concepts, "It seems fairly obvious that our chances of predicting behavior from attitude are practically nil until we at least start measuring attitudes toward the appropriate stimulus." (p. 222). On this basis, he critiqued LaPiere's (1934) study that reported apparently contradictory attitudes and behaviors toward Orientals. Fishbein argued that the Oriental couple in LaPiere's study so mismatched the stereotypic Oriental that attitude toward Orientals would be essentially irrelevant (i.e. attitude toward this couple would dominate) in determining behavior and, therefore, the finding of no correlation is not counterintuitive. 4 4 There are other grounds raised by Fishbein on which to criticize the LaPiere study. The argument just cited, however, remains valid even allowing for these other difficulties.
45 Fishbein extended this work in later writings. In an article discussing attitudinal-behavioral consistency (1972), he revised a prior theory (attributed to Dulanv) and introduced the notion of attitude toward the act as one of the determinants of action (rather than the more general attitude toward the object(s) affected by the action). Again, his argument was based upon the appropriateness of making predictions about specific behaviors (or behavioral intentions) based on specific attitudes. This is similar to the concept of attitude toward the situation introduced by Rokeach, "an attitude maybe focused on either an object or a situation behavior is always a function of at least these two types of attitudes" (1970, p. 135). Finally, Fishbein and Ajzen pose a similar question about consistency between behavioral intention and behavior in a discussion of behavior toward the church, "correlations between intention and behavior were higher when the levels of specificity tended to correspond than when they did not" (1975, p. 350). It is this concept of levels of specificity that is most relevant for the CS/D literature The studies discussed in the prior section each purport to investigate some aspect of "Consumer Satisfaction". Drawing on the rationale developed from the attitude literature, we can conclude that there is no inherent "consumer satisfaction" construct but, more appropriately, separate satisfactions with certain target concepts (e.g. particular items or services, marketing entities, etc.) that are part of a consumer context. Furthermore, these satisfactions are
46 interrelated within a level of specificity hierarchy that will enable predictions of effects on attitudes and behaviors. More concretely, consider the following example. A consumer finds that, on using her new calculator for the first time, the multiplication button fails to function. Her feeling of dissatisfaction on this occasion may have as its target (a) the particular calculator, (b) the brand of calculator, or (c) calculators in general (among other targets). Prediction of which type of satisfaction would demonstrate the greatest effects would depend on such factors as the consumers' specific expectations and/or attribution of the cause of the poor performance. Adequate measurement, then, requires that such specificity be built into the techniques used. Moreover, hypotheses concerning subsequent behaviors will also be required to be drawn at the same level of specificity (e.g., (a) exchange for a different unit of the same model, (b) change brands, or (c) find another alternative for doing arithmetic, respectively). On the basis of the foregoing discussion, the term "Consumer Satisfaction" is rejected as being indicative of a particular construct or concept. Rather, the term "satisfaction", as defined previously, will be used to refer to a feeling that results from or has as "Targets" particular stimuli. Consumer satisfaction (or CS/D), then refers to situations or the literature that discusses the feeling with respect to stimuli generally regarded as part of the consumer domain (e.g., products, retailers). This generalization of the satisfaction construct is extended beyond CS/D in the next chapter.
47 Measurement Issues in the CS/'D Literature Before introducing a new conceptualization and new measures of satisfaction in Chapter 3, it is useful to consider the measurement techniques that have been developed and discussed at some length in prior publications. The earliest discussions of measurement focused on evaluations of product performance (Anderson 1973; Miller 1977; Oliver 1977) and on what we are now calling consumer discontent (Lundstrom and Lamont 1976; Pfaff 1977). Since these are now thought to comprise a separate construct in the satisfaction process and a more general attitudinal phenomenon, respectively, and are distinct from the feeling and theory of interest these measures are not reviewed in depth here. At the opposite level of specificity from attribute specific measures are scales intended to evaluate one's satisfaction with life (Andrews and Witney 1976) or the marketplace (Lundstrom and Lamont 1976; Pfaff 1977). As constructed, these resemble the multi-attribute measures developed to investigate job satisfaction (e.g., Wanous and Lawler 1972). In the organizational behavior literature, this treatment of job satisfaction is intended to reflect a more enduring attitude than a specific, acute feeling (compared, for example, with Herzberg's interest in critical incidents). Similarly, Consumer Discontent and its measures, while useful in many contexts (e.g., economics, public policy questions) does not provide a viable alternative to the measurement techniques that are discussed presently. If we agree that perception of performance comprises a construct that precedes satisfaction, then it is useful to have good measures of
48 this. These could take the form of measures of overall value such as used by Anderson (1973) or measures of performance on an attribute specific level (Oliver 1977; Oliver and Linda 1981). The Oliver (1977) article does, however, suggest that caution is necessary in order to estimate perception as separate from disconf irmation. Specific Measures Use d Although considered to be separate from the feeling of satisfaction in the conceptualization adopted for the dissertation (Figure 3), the disconf irmation cognition does play a role and has historically been used as an operationalization of satisfaction. Thus measures of disconf irmation are included as the first entry in Figure 4a. Figure 4 comprises a taxonomy of measures of satisfaction that differs somewhat from Andreasen's (1977a:b). Whereas Andreasen was concerned with the influence of time (which is discussed in the next chapter; and the relative "objectivity" of measures, the present taxonomy is based more on the nature of the measures themselves than on any implicit theory of satisfaction or theory of measurement. As such, two dimensions evolved: (1) a cogni tive-af f ective-conative dimension reflecting what is being measured, and (2) a verbal-graphic dimension reflecting how the measures are collected. 5 Evaluative verbal measures of satisfaction, such as those presented in Figure 4a, are the most commonly used of the techniques 5 At this point, no distinction is being made between these measures which treat satisfaction/dissatisfaction as a single continuum and those which posit separate constructs, as discussed previously. For the most part, the scales presented could be adapted for either conceptualization: they are shown in as close to original form as is feasible
49 (a) Evaluative/Cognitive Measures in Consumer Satisfaction Verbal Disconf irmation measures 1. My expectations were: Too high: Accurate: Too low: It was poorer It was just as It was better than I thought I had expected than I thought (Oliver 1977; was much better (worse)' than I expected. Very Strong Strong Strong Very Strong Yes Yes Yes ? No No No (Swan, Trawick and Carroll 1931) Much more Somewhat About Somewhat Much less than I more than what I less than than I expected I expected expected I expected expected 1 2 3 4 5 (Aiello, Czepiel, and Rosenberg 197 7^ Derived measure for attribute levels (Level currently provided) (Level ideally desired) = Disconf irmation (Westbrook and Oliver 1981) Degree of satisfaction measures 5. Overall, how satisfied have you been with this ? 100% 90 80 70 60 50 40 30 20 10 0% Completely (Half & Not at Satisfied Half) all Satisfied (Mocre & Shuptrine 1984: Oliver & Eearden 1933; Oliver & Westbrook 1982; Westbrook 1980b; Westbrook 1981). Figure 4 Measures Used in Consumer Satisfaction Research
50 was very satisfactory (unsatisfactory) Very Strong Strong Strong Very Strong Yes Yes Yes ? No No No (Swan, Trawick and Carroll 1981 7. How satisfied were you with Very Somewhat Slightly Slightly Dissatisfied Dissatisfied Dissatisfied Neither Satisfied Somewhat Very Satisfied Satisfied (Oliver and Bearden 1983; Oliver and Linda 1981) 8. Were you satisfied/dissatisfied? (Choose one.) (Gronhaug and Arndt 1980) I am always or I am I am I am always or almost always sometimes sometimes almost always satisfied satisfied dissatisfied dissatisfied with. with. with. with. (Dav and Bodur 1978; 1979; 10. I am quite I am somewhat I am somewhat I am quite satisfied satisfied dissatisfied dissatisfied with... with... with... with... (Day and Bodur 1979) Figure 4 (a) --continued
51 11. Completely Very Somewhat Satisfied Satisfied Satisfied Satisfied (Dissatisfied) (Dissatisfied) (Dissatisfied) (Dissatisfied) 12 3 4 Not Satisfied (Dissatisfied) 5 (Aiello & Czepiel 1979; Aiello, Czepiel and Rosenberg 1977) 12. Very Somewhat Neither Satisfied Somewhat Very Satisfied Satisfied Nor Dissatisfied Dissatisfied Dissatisfied 12 3 4 5 (Aiello, Czepiel and Rosenberg 1977; Mowen and Grove 1983) 13. Now that you've actually used the product, how satisfied with it are you? Dissatisfied Satisfied (Bahr 1982) 14. Completely Fairly Not too Satisfied Satisfied Satisfied (Hughes 1977. 5. I am satisfied with Agree : : : Â• : Â• : Â• Disagree (Oliver and Bearden 1983) Other evaluations 16. To what extent does this __ meet your needs at this time? Extremely Extremely Well __:__:__:_ : : : __Â•'__ : Poorly (Oliver and Westbrook 1982; Westbrook 1980b) Figure 4(a)--continued
17. Summed semantic differential scales of satisfaction. (Oliver and Bearden 1983; Oliver and Westbrook 1982; Westbrook and Oliver 1981) 52 18. Likert Scales (Oliver 1980a) 19. Satisfactory or Dissatisfactory occasions/products as judged by respondent. (Day and Bodur 1978; Locker and Dunt 1978; Richins 1983a) Graphic 20 Imagine that the following circles represent the satisfaction of different people with Circle has all minuses in it, to represent a person who is completely dissatisfied with Circle 8 has all pluses in it, to represent a person who is completely satisfied with Other circles are in between. Which circle do you think comes closest to matching your satisfaction with ? Write the circle number here: (Oliver and Bearden 1983; Oliver and Westbrook 1982: Westbrook 1983; Westbrook and Oliver 1981) 21. Here is a picture of a ladder. At the bottom of the ladder is the worst you might reasonably expect to have. At the top is the best you might expect to have. On which rung would you put 9 Best I could expect to have 8 7 6 5 4 3 2 1 Worst I could expect to have (Andrews and Withey 1976) Figure 4(a) Â— continued
53 Verbal 22. Likert scales a. I am satisfied with b. If I had it to do all over again, I would c. My choice to was a good one. d. I feel bad about my decision concerning I think that I did the right thing when I decided f. I am not happy that I did what I did about Agree ...(9) ... (7) ... (5) ... Disagree Strongly Agree Strongly Disagree (Moore & Shuptrine 1984; Oliver 1980a; Oliver and Bearden 1983; Oliver and Westbrook 1982; Westbrook and Oliver 1981) 23. Mark on one of the nine blanks below the position which most closely reflects your satisfaction with Delighted Pleased Mostly Mixed Mostly Satisfied Dissatisfied Unhappy Terrible Neutral Never Thought About It (Jordan and Leigh 1984 [7 items] ; Moore and Shuptrine 1984: Oliver and Bearden 1983; Oliver and Westbrook 1982: Westbrook and Oliver 1981] 24. Content analysis of subject-provided protocols with scoring for satisfaction or emotional statements. (Locker and Dunt 1978; Westbrook 1980b) 25. Scales measuring separate emotions. Please indicate the extent to which each word describes the way you feel with respect to 12 3 4 5 Very Slightly Slightly Moderately Considerably Very or not at all Strongly (b) Emotional/Affective Measures in Consumer Satisfaction
54 Adjectives "loading" on each of the ten emotional dimensions d 1) Interest Excitement, 2) Enjoyment Joy, 3) Surprise Startle, 4) Sadness Anguish, 5) Anger Rage, 6) Disgusted Revulsion, 7) Contempt Scorn, 8) Fear Terror, 9) Shame Shyness and 10) Guilt Remorse (Westbrook 1987; Westbrook and Oliver 1984) Graphic 26. How do you feel about I feel: 7 3 Delighted Pleased Mostly Mixed Mostly Unhappy Terrible Satisfied (about Dissatisfied equally satisfied and dissatisfied A Neutral (neither satisfied nor dissatisfied) I never thought about it. (Andrews and Withey 1976; Westbrook 1980b) Figure 4 (b)--continued
55 27. "Feeling" Thermometer Where would you put on the feeling thermometer? WARM 100 Â— Very warm or favorable feeling 85 Â— Good warm or favorable feeling 70 Â— Fairly warm or favorable feeling 60 -A bit more warm or favorable than cold feeling 50 -No feeling at all 40 -A bit more cold or unfavorable feeling 30 -Fairly cold or unfavorable feeling 15 -Quite cold or unfavorable feeling COLD 0o Â— Very cold or unfavorable feeling (Andrews and Withey 1976; Oliver and Westbrook 1982; Westbrook and Oliver 1981) 28. Faces scale Here are some faces expressing various feelings. Below each is a letter. A B C D E F Which face comes closest to expressing how you feel about (Churchill and Surprenant 1982; Oliver and Westbrook 1982; Westbrook and Oliver 1981) Figure 4 (b) --continued
56 Verbal Behavioral intentions 29. Because of I would come (shop) here again. Very Strong Strong Strong Very Strong Yes Yes Yes ? No No No (Swan7 Trawick~and Carroll 1981) 30. How likely are you to play with (use) in the future? Very Very Unlikely Unlikely Likely Likely -2 -1 +1 +2 (Jordan and Leigh 1984] 31. Knowing what you know now, what are the chances in ten (10) that you would choose to use the again? 0123456 789 10 No Chance Certain (Oliver and Bearden 1983; Westbrook and Oliver 1 98 1 ) Graphic (Observational) 32. Measures of time and extent of use. (Bjorklund and Bjorklund 1979) 33. Filing complaint as sign of dissatisfaction. (TARP 1979) 34. Loyalty, repurchase as sign of satisfaction. (LaBarbera and Mazursky 1983) Notes to accompany Figure 4: a Citations provided are meant to serve as examples of scale use, not to provide an exhaustive bibliography. b Some parenthesized modifications of what was essentially the same scale are presented for the sake of simplicity. c In this case, the scale was compressed by omitting the neutral responses d Again, for the sake of simplicity, all cf the adjectives which were used are not presented in this Figure. (c) Behavioral/Conative Measures in Consumer Satisfaction
57 depicted in both survey and experimental research. As such, it is important to note several issues which arise in their use: (1) Each requires an interpretation by the respondent and questioner of the word "satisfaction" or some variant thereof. As noted in the beginning of this chapter, the word has a variety of meanings even in academic research. (2) Where a midpoint is included, it is often not marked (e.g., measure #13), marked in an ambiguous manner (e.g., measure #6), or marked in apparent conflict with other measures (e.g., measure #5 hypothesizes a mixed evaluation whereas measure #7 suggests a lack cf evaluation). (3) Not providing for a neutral or mixed response (e.g., measure #9) forces a possibly spurious response. (4) The range of gradations provided, from dichotomous (e.g., measure #8) to a 13-point scale (e.g., measure #13) suggests different evaluation processes. The measures classed as "other" provide a slightly different perspective. (1) The need scale (measure #16) is somewhat more evaluative than the expectation disconf irmation scales in the first group. (2) Although the specific adjectives used for the semantic differentials (measure #17) were not provided by the original authors, related work by Westbrook and by Oliver suggests that this particular technique may alleviate some of the definitional ambiguity mentioned above by the use of synonyms in the measurement of aspects of "satisfaction". (3) The last type of "measurement" in this group encompasses those open-ended survey techniques that ask the respondent to first identify a satisfactory or dissatisfactory critical incident and then provide more details about it. The graphic measures of satisfaction as an evaluation, presented
58 at the end of Figure 4a, probe for essentially the same information as the verbal measures. The graphic format, however, is better able to communicate the concept of quantities of satisfaction (dissatisfaction), thus removing some of the ambiguity of the verbal measures. This technique also makes the distinction between mixed evaluations and the absence of evaluation more evident. Figure 4b presents measures that are more consistent with the emotional definition of satisfaction that has been adopted for the dissertation. The Likert-type scales (measure #22) were defined to be "emotional in content" when presented by Oliver (1980a, p. 463) despite some resemblance to the evaluative scales in part (a) of the Figure. The next scale (measure #23) was derived from Andrews and Withey's (1976) set of indicators of well-being (as were a number of the measures being reviewed) This single scale seems to capture both the evaluative and emotional aspects of satisfaction in its labels and seems to provide ample options for neutral responses (although this is sometimes defeated by omission of the off-scale response categories; e.g., Jordan and Leigh 1984). It is not clear from the CS/D literature, however, whether or not the scale possesses more than ordinal properties (e.g., interval which has been presumed for some of the other scales) A recent innovation in measurement within the CS/D literature is the technique being developed by Westbrook (1983) to assess individual emotions at times or occasions that are critical in the satisfaction process. Although still in preliminary stages, these measures hold promise as more direct, but less obtrusive measures of satisfaction
59 than the evaluative scales presented previously. Figure 4b shows only a portion of the measurement scheme (measure #25) that has been adapted from Izard's (1972) Differential Emotions Scale. The original scale utilized responses to 69 separate items describing a person's emotional profile at a particular time. Responses to specific items are summed and the resultant profile is a subset of the ten emotional dimensions listed in the Figure. Westbrook has used a truncated, 31-item version of the scale to describe consumers' emotions with respect to their automobiles. Factor analyses have revealed four (westbrook 1983) to six (Westbrook and Oliver 1984) distinct profiles of emotional responses. How these findings relate to the emotional theory developed in the dissertation is explored in the beginning of Chapter 3. Measurement of emotions is difficult because emotions are difficult to verbalize and degrees of emotion are hard to represent. Thus, the graphic emotion instruments (measures #26-28) offer some simplification in representing a gestalt reaction. Finally, satisfaction has been defined as the precursor to specific behaviors (Cardozo 1965) and is being treated in the dissertation as theoretically linked to such behaviors (Figure 3). Figure 4c outlines behavioral measures that have been used as measures of satisfaction. Even if these are not justified as direct measures of feeling, the causal flow that has been proposed suggests that behavior may be indicative of strong feeling. Note that in the case of Figure 4c, the graphic measures are observations of actual behavior (measures #32, #33).. This categorization preserves the composite, less dependent
60 on connotation, characteristics of previous graphic measures. Review and Usefulness of Measures There have been prior efforts at reviewing and empirically comparing satisfaction measures in the CS/D literature. Hunt (1977a;b) commented on the discussions at the first two CS/D conferences and identified the need for a common measurement methodology as a key to deriving integrated CS/D theories. He seemed to favor conative verbal measures as "a composite measure getting at all the influences affecting the decision without having to identify those influences If a repeat purchase is intended, that says that all things considered, the purchase was critical enough that it has to be repeated and the choice was good enough that the respondent doesn't think any better choice is worth making." (1977b, p. 39). Note that the use of behavioral intentions rather than actual behaviors as measures reduces concerns about influences external to the satisfaction situation that could modify behavior (Fishbein 1966; Fishbein and Ajzen 1975). Westbrook and Oliver (1981; Oliver and Westbrook 1982) have compared a number of the measures depicted in Figure 4 empirically (the dimensions in the Figure are based, in part, on discriminations by these authors). From two analyses of pilot data they conclude (1) Likert, semantic differential and a composite verbal scale (corresponding to measures #22, #17 and #5 + #20b + #23 from Figure 4, respectively) performed best on convergence versus divergence criteria; (2) discriminability of various scales seemed to be product class dependent (automobiles versus calculators); and (3) as a whole, the measures did not succeed very well in discriminating satisfaction from
61 attitude. The first finding is encouraging for those who wish to identify measures which will be generally useful. The latter two findings threaten the universality of both the measures and theories of satisfaction. The conclusions by Westbrook and Oliver are tempered greatly by their methodology and the ensuing discussion. Assuming that a student sample (n=160) is sufficiently representative for this type of research (theory testing), the manner of questioning still leaves room for criticism. Essentially, the students were asked to give their present reaction to durables which they may have possessed for some time and with which they are likely to have varying histories of ownership and usage experience. It is not surprising, then, that the measures reflect some difference in response to product class and some confounding with attitude. The authors allude to an explanation "satisfaction is, in effect, a response to disconf irmat ion and is expressed in affective terms .... In a sense, satisfaction may be seen as a disturbance acting on an attitude system" (1982, p. 13). They go on to suggest that temporally distinct measures may help to separate satisfaction and attitude. Given Oliver's (1981) argument that satisfaction is likely to rapidlv "decay" into attitude. it would seem that temporally distinct methodology would be necessary to uniquelv identifv the constructs. In the absence of such temporal distinction, the cautionary conclusions by Westbrook and Oliver are suspect. This inattention to the theory relating satisfaction and attitude is continued in the work by the same authors in which they attempt to calibrate the emotional
62 measures (measure #25) derived from the Differential Emotions Scale. In this research (Westbrook 1983; Westbrook and Oliver 1984) they repeat the methodology comprised of collecting contemporaneous reactions to products (i.e. automobiles) which have been owned and used over a range of time periods, and usage patterns. The resulting emotional dimensions are correlated with more typical satisfaction measures (e.g., measures #7, #19, #20, #33) with mixed results. Again, the theoretical distinction between satisfaction and attitude is not maintained and the influence of time is disregarded. 6 Similar concerns about timing versus attitude formation have also been expressed outside the "mainstream' 1 CS/D literature (Locker and Dunt 1978). In this review of British studies concerning the measurement of patient satisfaction with medical care, the authors identify many of the same techniques that are listed in Figure 4. They also are concerned about the necessity of developing universal measures of satisfaction to allow for inter-study comparability. The interviewtechnique seems to be much more prevalent in this literature, and the findings suggest that global evaluations exposed by direct measurement scales seriously mask the separate underlying attribute evaluations. This argument returns us to the levels-of-specif icity question that was discussed previously. In order to be useful in theory testing, measures must be applicable at the same level of specificity as the operational theory. Locker and Dunt (1978) argue for examining 6 The measurement technique proposed here is used in conjunct ior with the critical incident methodology, measures #19 a. id #24, in an attempt to alleviate this time problem in the derivation of measures for the dissertation. This is discussed in more detail as the Pilot Study in Chapter 5.
63 various aspects of the patient's experience with medical care in order to explain his satisfaction, much the same as Wanous and Lawler (1972) examine the influence of various aspects of a job on job satisfaction. Although these are likely to be useful for understanding which factors influence satisfaction (maybe in a specific situation), it is less likely that a respondent will be able to discriminate the separate influences on an inventory of emotions such as that presented by Westbrook (1983) in order to build a "net" satisfaction. The final result of this may be that the emotional measures (Figure 4b) will best reveal the presence and degree of satisfaction whereas the evaluative measures (Figure 4a) will be more useful for explaining the cause of the feeling (i.e. it may be difficult to make emotional measures salient at more than one level in a given measurement situation). Even so, these measures should not be used without additional cautions. Miller (1977) argued that attempts to measure a process may actually impact on the process being studied. This problem of reactive measures and concomitant demand artifacts has been discussed elsewhere in consumer behavior (Sawver 1975) but the arguments do not seem to have effected the design of less obtrusive measures (as suggested by Webb, Campbell, Schwartz, and Sechrest 1956 and Sechrest 1979). Sobei and McGuire (1977) have gone so far as to argue that disconf irmation (as the difference between expectations and perceptions--discussed here in a slightly different context) mav be simply a measurement artifaci. The possibility of this, given current operat ionalizat ions of the satisfaction theory that has been presented, should not be overly discounted. Some authors have reported the necessity of forcing a
64 satisfaction response by eliminating the opportunity for neutrality (e.g. Day and Bodur 1979; Jordan and Leigh 1984). Although this may be viewed as forcing more careful responses, it is argued in Chapter 3 that such a procedure obscures important theoretical differences between a mild response and no response. For our purposes, it is important to note that the treatment of neutral options such as "Don't Know" or "Never Thought About It" has been shown to bias recall data (Mizerski, Freiden and Greene 1983) and can be presumed to have similar effects on attitudinal or satisfaction responses. For these reasons, observational measures such as those suggested in Figure 4c may actually be better for some purposes than the more direct satisfaction measures identified elsewhere in the Figure. Although these behavioral measures may be influenced by other externa] factors, the causal flowmay be sufficiently strong to drive discernible, theoretically important differences. The influence of satisfaction on such behaviors as' word-of-mouth, complaining, and repurchase is examined in detail in these separate literatures. The next section of this chapter describes how the concepts central to these literatures relate to CS/D as the products of the emotions. Behavior al Outcomes of Satisfaction As discussed previously, it was the observation of certain marketplace behaviors that first aroused interest in satisfaction (Cardozo 1965) and has continued to fuel the growth and development of the CS/D literature. In this section, the theories and research that examine these behaviors and implicate the satisfaction feeling as a cause are introduced. Further, it is suggested that the continuum of
65 behaviors proposed is sufficiently stable within the CS/D literature to be useful as a gross measure of relative satisfaction. Day (1977) presented the most integrated discussion of behaviors thought to follow feelings of satisfaction. Although he presented separate taxonomies for satisfaction responses and dissatisfaction responses, the distinctions made were the same and it is more in keeping with the unidimensional nature of the construct to summarize his taxonomy in a single hierarchy (Figure 5). Day's presentation of the hierarchy in a form reminiscent of a decision tree is an apt metaphor for describing the research that examines these behaviors. That is, the literature tends to view these behaviors as resulting from rational choice processes among economically evaluated alternatives. The behaviors are generally treated as spanning a continuum which seems to suggest that, all else equal, a stronger satisfaction feeling (positive or negative) motivates a stronger response that may require more effort or other forms of "cost" from the individual (Figure 6). For example, in most cases it is less convenient for a consumer to contact either the marketer or a third-party complaint-handler than it is to express his/her dissatisfaction to a friend. This Figure also expresses a concept that is frequently overlooked in the CS/D literature; responses to satisfaction are analogous to the responses to dissatisfaction (negative satisfaction). The latter behaviors (especially complaining) are more visible and more troublesome to marketers and have garnered the majority of the research interest. Yet, consumers definitely do exhibit overt positive responses to feelings of satisfaction. Day (1977) reported that many companies
Max imum Max imum Dissatisfaction Satisfaction Complaining Negative Exit Loyalty Positive Compliment (seek Word-of(Dis(Degree Word-ofredress) Mouth position) of Use) Mouth Figure 6 Satisfaction Responses 6'
63 that operate "consumer hot lines" routinely receive complimentary calls about their products and services. Extreme actions are comparatively uncommon, however. In this section, the analysis of the literature on complimenting and complaining follows discussions of loyalty and exit, and positive and negative word-of -mouth. Some articles have examined relatively abnormal behavior such as destruction of property and merchandise (Mil is 1981), certain forms of consumer fraud (Wilkes 1978), as well as shoplifting (Herndon 1972; Klemke 1982; Moschis, Cox and Kellaris 1987). While these behaviors have been characterized as retribution for dissatisfaction and may result from the process under discussion, they are not sufficiently common or well-understood to be relevant for this discussion. The behaviors which are most likely to be affected by weak feelings of satisfaction are identified in the center of the continuum (Figure 6) as Loyalty and Exit. The specific behaviors that are encompassed by these classifications (e.g. repeat purchase) tend to arise naturally as part of a consumer's experience. Satisfaction is thought to influence the choice of behavior, perhaps through an attitudinal process, rather than to initiate the performance of specific acts. The consumer decisions that might be influenced at the Loyalty versus Exit level can be categorized under three classifications: (1) Use versus Do Not Use; (2) Retain versus Dispose; and (3) Repurchase versus Do Not Repurchase. Choices among these classes of behavior are generally thought to be determined by attitude which, in turn, has been
69 argued to be related to satisfaction (Oliver 1 98 1 Â— the nature of this relationship is explicated more fully in Chapter 3). Usage decisions have been related to satisfaction generally in the context of repeatedly used services such as electrical utilities (Heberlein, Linz and Ortiz 1982), public radio station broadcasts (Swan and Trawick 1983) and leisure activities (Hawes, Blackwell and Talarzyk 1976). Arguably, such consumption actually entails a series of repeated "purchase" decisions that are affected more by attitude than by satisfaction. These papers serve to introduce the argument that a feeling of satisfaction on one usage occasion may prompt additional usage whereas a feeling of dissatisfaction may reduce usage. For example, the extent of use of a new home computer may be related to the satisfaction feelings generated by the first few trials. Similarly, the decision to retain or dispose of a product can hinge on one's attitude toward the item and on the feelings of satisfaction which may have contributed to that attitude (or, even, dissonance after purchase but before delivery-Donnelly and Ivancevich 1970). DeBell and Dardis (1979) conveyed similar thoughts in their argument that product failure may hasten the perception of obsolescence and disposal of major appliances despite the possibility of or actual repair of the item (also, Burke, Conn and Lutz 1978a,b). Many of the items placed for disposal are still serviceable as evidenced by the popularity of flea markets ("swap meets" and other terms are also used --Belk, Sherry and Wallendorf 1988; Wallendorf and Arnould 1988). Unfortunately, there is little other research concerning disposal decisions on which to base further argument.
70 In the consumer context, the term "loyalty" most often suggests some form of repeat purchase behavior and is generally associated with the purchase of a particular brand in a particular product form (Bernardo and Tarpey 1980; Goering 1985). The literature on brand loyalty is copious (see Jacoby and Chestnut 1978 for a review) and is not detailed here. It is useful to note that consumers' prior satisfaction experiences have been linked explicitly to repurchase (Oliver 1980a; Oliver and Linda 1981) and current evaluation (Cohen and Houston 1972). This can result even when the target of the satisfaction feeling and the subsequent purchase are related, but not identical, as seen with choice of housing type (Leigh 1984), warranty service and automobile purchase (Bernacchi, Kono and Willette 1980), as well as the purchase of different small appliances from the same manufacturer (Smead, Wilcox and Wilkes 1980). Direct repeat purchase decisions have been described for diverse products from automobile repair (Biehal 1983) to children's videogames (Jordan and Leigh 1984). Schmalensee (1978.) suggested a phenomenon that is essentially identical to the disconf irmation paradigm when he proposed a method of modeling advertising effects on repeat purchase. The upshot of this, then, is that experience with prior purchases is likely to influence future purchases, whether of identical or merely related items. LaBarbera and Mazursky (1933) for example, reported that much of the decision making for commonly purchased products was completed before the actual shopping situation (see also Hover 1984).
The satisfaction feelings that result from repeated use of such products can act to maintain or adjust the consumer's attitude which, in turn, influences the next purchase decision. Finally, it should again be noted that although the preceding discussion has centered on discrete products and brand loyalty, there is no reason to suspect that the effects of satisfaction are so limited. As discussed previously, there are a number of levels of specificity at which the target of the satisfaction feeling may be identified. Given the appropriate opportunity, the consumer may respond by remaining loyal (repurchasing, maintaining a business relationship) or by exiting (switching to another supplier, product form, etc.). Carey, Clicque, Leighton and Milton (1976), for example, suggested that managing customer satisfaction may be an effective means of garnering repurchase for a retailer. Whereas the opportunity to select between Loyalty and Exit behavior arises in the conduct of the consumer's regular life, as mentioned in the preceding discussion, other responses to satisfaction r eelings require the consumer to initiate specific actions. The next class of behaviors represents an increase in the specific motivation required. These behaviors have been termed Wordof -Mouth, and again, can entail either positive or negative reactions. Marketers first became interested in the positive word-of-mouth which was identified as helpful in enhancing the diffusion of innovative products (Arndt 1967; Engel, Kegerreis and Blackwell 1969; Sheth 1971). This diffusion literature eventually yielded the concept of opinion leaders (individual consumers who heavily influenced the
"2 attitudes and behaviors of their associates) that eventually became the more general concept of social influence within reference groups (Burnkrant and Cousineau 1975; Moschis 1976; Park and Lessig 1977). The conclusion from this research is that consumers exchange information about consumption experiences among themselves and that the subsequent behaviors of the recipients of such information show an influence attributable to the communication. 7 Moreover, it has been suggested that evaluative information about the product is more influential than signals of social acceptability or group preference (Burnkrant and Cousineau 1975). This suggests that the product expectations and perceptions central to the disconf irmation paradigm are related to communications among consumers (Coca-Cola 1982). Researchers in CS/D eventually became more interested in investigating the causes cf word-of-mouth behavior rather than its effects on the recipients of the communications. Day's (1977) taxonomy (Figure 5) categorizes word-of-mouth behavior as "private action" although it may be viewed as considerably more public than one's own internal resolve to increase or decrease purchases or use cf an item (except in cases where use constitutes a form of word-ofmouthÂ— -Koepp 1985). No matter how this particular distinction is ultimately resolved, the placement on the continuum of responses (Figure 6) is quite consistent with the treatment of word-of-mouth in CS/D (e.g. Lewis 1983). Richins (1983a, b) is more specifically interested in negative word-of-mouth as a response to dissatisfaction. 7 Westbrook (1987) treats W-O-M and public opinion rumors as essentially similar. The emotional treatment of satisfaction/ dissatisfaction suggests different motivations, however.
73 She reported that (1) the likelihood of any response increases with degree of dissatisfaction, and (2) word-of-mouth tends to be more common than complaining to the seller. Furthermore, behaviors categorized as word of mouth can range from a simple mention to an extensive, animated discourse. The kev criterion is the consumer's motivation to act this way. Of course, word-of-mouth behaviors are not solely dependent on the valence and extent of the satisfaction feeling but are also influenced by other aspects of the situation. Research attempting to describe the role of these other determinants has focused on the extremes of the continuum presented in Figure 6, Compliments and Complaining (sometimes called Voice in other satisfaction literatures). In contrast to the classes of behaviors just discussed, the literature concerned with this last class of behaviors has originated with and concentrated on negative rather than positive behaviors. Although ccmpliments arise in measurable amounts (Day 1977; Lewis 1983) it is likely that they are viewed as financially or economically less useful to the firm than loyalty or positive word-of-mouth. On the other hand, complaints imply a greater extent of switching and negative word-of-mouth and generally include requests for redress which require immediate attention. Thus, it is not too surprising that there is a voluminous literature on complaining behavior itself. This literature addresses such diverse topics as the extent of complaining (TARP 1979), the influence of complaints on business practices (Cobb, Walgreen and Hollowed 1987; Fornell and Wernerfelt 1987; Resnik and Harmon 1983; Ross and Gardner 1985) and the subsequent
effects cf business response on the attitudes and behaviors of the complainants (Bearden and Oliver 1985; Gilly and Gelb 1982; Wilkinson and Mason 1979). At issue in this review however, is the research that seeks to identify the causes of complaining or, more correctly, the selection of complaining as a response to dissatisfaction (Figure 5). Consumer complaining behavior is not a recent phenomenon; for example, legend has it that the young Michelangelo once counterfeited a statue and that, on learning of the deception, the buyer returned the item and demanded a refund (MacDougall 1958). The necessity of avoiding such negative impact on revenues has fueled a great deal of interest among marketers in complaining behavior. 3 This type of behavior can become even more extreme and more costly if reparations are sought via the legal system (Best 1981; Brown and Swartz 1984; Goodwin, Mohajan and Bhatt 1979; Morgan and Avrunin 1982). A number of attempts have been made to identify causes or factors leading to complaining. In the typical study, groups of dissatisfied consumers are divided into those who did complain and those who did net (e.g., Bernacchi, Kono and ivillette 1980). The groups have been compared on such dimensions as demographics Â— age, education, sex, income, etc --(Bearden and Mason 1984; Morganosky and Buckley 1987), personality (Fornell and Westbrook 1979), prior experience with complaining (Gronhaug and Zaltman 1981) and a belief-based attitude b "Complaining" will generally include returns for repair, replacement or refund, communicating displeasure to the seller or otherchannel member without a specific request or desire for redress; and similar communications addressed to some "neutral" third party such as government agencies or newspaper "Action Lines"--Feick 1984.
75 model that implicates specific beliefs about the consequences of complaining (Richins 1982). The literature has not provided firm conclusions concerning the influences of such variables (Bearden 1983; Oliver 1987); the results are mixed for each. Some problems have been definitional. For example, there is little to distinguish a personality variable labelled "propensity to complain" from a cognitive variable labelled "complaining experience" when each is defined by past complaint behavior (Robinson, Trebbi and Adler 1983). Moreover, it is difficult to account for the effects of the degree of dissatisfaction, which is sometimes significant (Zaltman. Srivastava and Deshpande 1978) arid sometimes not significant (Gilly and Gelb 1982) in determining complaint behavior (see also Oliver 1987). Another possible reason for these inconsistencies is the failure to view complaining as one element on a continuum of possible responses (Figure 6). Although not all authors have treated the choice to complain as dichotomous (Jacoby and Jaccard 1981), some have (Fornell and Westbrook 1979, 1934; Gronhaug 1977) which may be contributing to the discrepant findings in the literature (Richins 1987; Singh 1988). Iable 3 is a somewhat more complete list of alternative behaviors than that provided by Day's taxonomy (Figure 5). The list was compiled from surveys that were concerned with consumers' post dissatisfaction responses. Although there has been interest in compliments and other positive behavior (Robinson and Berl 1 980), a similar inventory of responses to satisfaction has not been reported. Treatment of the
76 Table 3 Actions Following Dissatisfaction Consumer Action Stopped payment or refused to pay Decided not to buy the product or service or deal with that compam again Complained to family or friends Asked for replacement or refund Complained to the person who sold me the product or service Complained to the company or store Complained to a consumer agency Complained to a public agency or my congressman Complained in a letter to a newspaper or magazine (for publicity rather than redress) Considered taking legal action Consulted or hired a lawyer to protect my interests Took no action at all Note: Table based on Dav 1983 and Krishnan and Valle 1979.
77 relationships among possible behaviors and the similarities of positive and negative responses has recently been integrated into models that encompass both the satisfaction and complaining processes (Bearden and Teel 1983; Day 1977; Lippert and MacDonald 1981). The result of the integration of these theories is a view of postevaluative behavior that posits that the degree of satisfaction emotion which is felt determines the level of drive or motivation to respond; the relative costs and benefits that are relevant for the particular individual in the particular situation determine the choice of response (Day 1984; Feick 1987; Fornell and Didow 1980; Richins 1932). Thus the factors identified previously (demographics, personality", etc.) can be reinterpreted as affecting choice of behavior by first affecting the expectation of costs or benefits (e.g. prior experience may increase the expected gain from complaining in certain situations and decrease the cost of learning how, where and to whom to complain). Consumer handbooks designed to improve the public's ability to complain typically capitalize on just such an economic concept byillustrating how much can be "won" through complaining (Charell 1973: Horowitz 1979) and/or bv making complaint handlers more accessible (Eisenberger '977; Feinman 1979; Knauer 1983). Finally, this economic approach also accounts for the discrepancy in levels of positive versus negative behaviors. Although some measure of the degree of satisfaction may indicate equivalent levels, the cost:benefit ratio of complaining in a given situation is not likely to equal that of conveying a compliment, for example. The continuum
presented in Figure 6 incorporates this cost versus benefit concept at a very gross level. It is the assumption that this behavioral continuum holds across situations that underlies the usefulness of complaint data as a surrogate for measuring dissatisfaction. Comparisons across products and individuals have shown that this is inappropriate. In a particular situation, with better defined or constant costs and benefits (such as a laboratory experiment) the use of behavior as a measure of satisfaction may be more justified. The loss of precision may, in many cases, be balanced by the gain in the unobtrusiveness of collecting behavioral (versus rating scale) data. S ummary This chapter has defined the satisfaction construct as an emotional response and adopted one model of the satisfaction process from the CS/D literature. The types of consumer satisfaction that have been addressed by researchers v:ere shewn to be specific manifestations of the general construct and model, made specific by identifying the target of the emotions. The advantages and disadvantages of various measurement techniques were discussed, especially their conformity to the definition and model. Finally, behaviors which result from satisfaction were identified and the question of their feasibility as measures was raised. The next chapter extends this satisfaction construct, as defined and modeled here, bevond the consumer behavior framework. It is argued that a basic human behavioral process results in a basic feeling of satisfaction, which has been observed in consumer situations. The
79 present framework is extended to identify other factors that are necessary to evoke the emotion. This extended conceptualization is then specifically applied to consumer behavior and the CS/D literature.
CHAPTER 3 SATISFACTION: AN EXTENDED CONCEPTUALIZATION OF THE GENERAL CONSTRUCT Introduction In this chapter, the CS/D literature is augmented with theories and research findings from other fields. The concepts that are borrowed clarify the nature of the pattern of emotions which defines satisfaction and which can be aroused in a variety of contexts, only a subset of which involve behavior as a consumer. It is further argued that satisfaction is not aroused by all behavior. As with attitudes, satisfaction with a given target frequentlv may not exist for a particular individual. As an emotional construct, satisfaction is thought to dissipate relatively rapidly but not without affecting attitude toward the target object. This relationship is explored and implications for the time-sensitivity of satisfaction measurement are discussed. Given the uncertainty regarding arousal of the satisfaction feeling and its fleeting nature once present, it is necessary to understand and be able to predict its occurrence. The major contribution of this chapter, and the dissertation as a whole, is the presentation cf six necessary and collectively sufficient conditions for the arousal of the general feeling of satisfaction. The explication of these antecedent conditions is followed by a discussion of consumer situations in which they are likely to be present and an 30
examination of CS/D research in which the likelihood that the satisfaction feeling was aroused in a given study is assessed in light of the antecedent conditions and the measures used for satisfaction. The remainder of the dissertation examines two of the antecedent conditions empirically. The philosophy that has been adopted views satisfaction as one construct (MacCorquodale and Meehl 1948) in an integrated network. The theory that is developed concerns the general satisfaction construct. The empirical portion of the dissertation addresses the more specific consumer oriented construct and theory, derived from these theories (S. Hunt 1976). It is crucial that the distinction be maintained between the term "satisfaction," signifying the general emotional construct, and the labels "CS/D" or "consumer satisfaction", signifying a literature that applies the theory to arange of consumer behaviors. Definition and Theory of Emotions Psychologists have wrestled with defining and conceptualizing emotions for some time (see, for example, James 1890; Reymert 1928). These issues are not yet resolved, but some major views and theories have been outlined. Westbrook (1983) introduced the emotion literature to CS/D research by adopting Izard's (1972; 1977) conceptualization of emotions. 1 This perspective has been used elsewhere in consumer research (Allen, Machle'it and Marine 1987) and also is the basic emotional theory for the dissertation (versus those used by others, e.g., Averill 1980; Mehrabian and Russell 1974). 1 Other emotion models have been employed in consumer research (e.g., Havlena and Holbrook 1986; Peterson, Hoyer and Wilson 1986), but Westbrook provides the most lucid theory.
82 Izard defined a fundamental or discrete emotion (as opposed to a combination or pattern of emotions) as "a complex phenomenon having neurophysiological, motor-expressive and experiential components. The intraindividual process whereby these components interact to produce the emotion is an evolutionary-biogene tic phenomenon" (1977, p. 64). Of the dimensions mentioned in the definition, the motor-expressive and experiential components of emotion are the most relevant for understanding the emotional nature of satisfaction. Viewed within the perspective of Izard's framework, satisfaction would be considered a pattern of emotions. He defined a pattern to be "a combination of two or more fundamental emotions which under particular conditions tend to occur together (either simultaneously or in a repeating sequence) and to interact in such a way that all of the emotions in the pattern have some motivational impact on the organism and its behavior" (1977, p. 64). The fundamental emotions identified by Izard and a procedure for identifving which ones combine to constitute satisfaction are discussed below. One issue raised in the measurement section of Chapter 2 was that a major difference between attitude and emotion is the motivational dimension attributed to the latter. This aspect of the conceptualization of emotion is not unique to Izard, but rather has a long tradition in the literature dealing with emotions. Young (1961) suggested that emotions had come to be considered a subset of all possible motivations. The definition that he supplied reflects this, "an acutely disturbed affective state of the individual that is
psychological in origin and revealed in behavior, conscious experience, and visceral functioning" (1961, p. 355). Taken together, the definitions by Young and Izard present a foundation for the conceptualization of satisfaction as an emotional state that is part of a more integrated process. Young states, "An emphasis upon the psychological situation in the origin of emotions is important because it is the 'psychological' origin that distinguishes emotions from other affective states" (1961, p. 346). That is, emotions are separate from feelings which originate from organic states (e.g., hunger, thirst) or relief from those states. Izard does not take so strong a position on this point. Although emotions and cognitions tend to interact, he proposes no strict unidirectional flow of causality (see also Zajonc 1980). Izard also postulates a discrimination between emotion and the cognition of emotion. That is, a person's ability to "objectively" label his or her feelings reflects a different phenomenon than simply reacting to those fee lings On the other hand, Izard bases his measurement technique on an individual's ability to recognize not only the nature but also the magnitude of feelings. Izard's Differential Emotions Scale (DES) (1972; 1977) is, essentially, a list of adjectives representing the ten fundamental emotions. Subjects are asked to indicate on a five-point scale whether each word presented describes how they feel: 1) Very Slightly or not at all; 2) Slightly; 3) Moderately; 4) Considerably; or
84 5) Very Strongly. 2 The original (1972) version of the DES used 69 adjectives to describe the ten emotions but shorter instruments have been used by Izard (1977) and Westbrook (1983). The ten emotions and the adjectives used in each of these later versions are presented in Table 4. The Item/Emotion correlations (factor loadings) that are reported were provided by Izard (1977, p. 126) to indicate reliability of the scale. Westbrook used his version of the DES to evaluate consumers' responses to their automobile. Using a factor analysis (1983) he identified four patterns of emotions. A subsequent cluster analysis of different data (Westbrook and Oliver 1984) yielded six patterns (Table 5 lists the emotions that were most characteristic of each group, i.e., the dominant pattern). The latter analysis provides descriptions of the emotional profiles of specific segments of consumers and thus lends itself well to examining the definition of satisfaction as one or more of these profiles. The disconf irmation paradigm suggests that Surprise should be one emotion in the pattern defined to be satisfaction. Based en the Westbrook and Oliver (1984) analysis Cluster 2 and Cluster 4, which were characterized by the highest levels of surprise, would represent the dissatisfied and satisfied ends of a continuum, respectively (Eahr 1982 also cites surprise as an important dimension of the dissatisf action--or disappointment--feeling. ) Contrary to the position 2 For the purposes of the dissertation, only emotions as transient states are of interest. Thus, the DES II which measures emotional traits by collecting responses based on frequency rather than intensity is not discussed.
Table 4 Measures of Fundamental Emotions 35
Table 5 Patterns of Emotions in Automobile Ownership 86 a) Westbrook (1983; /estbrook and Oliver (1984; Factor I Anger Sadness Disgust Factor II Interest Enjoyment Factor III Shame /Shvness Guilt Factor IV Surprise Cluster 1 "Contented" Enjoyment Interest Cluster 2 "Mad as hell" Anger Distress Disgust Contempt Surprise Cluster 3 "Apprehension" Interest Fear Enjoyment Cluster 4 "Delighted" Interest Enjoyment Surprise Cluster 5 "Unemotional" all low responses Cluster 6 "Guilty/Sheepish' Guilt Shame /Shyness
taken by these authors, the remaining clusters are not considered to describe intermediate points of that continuum, however. Rather, these are considered to be emotion patterns which simply are not addressed by the satisfaction theory that is being explicated in the dissertation. This interpretation is consistent with the scores on satisfaction scales reported by the authors (for five scales with ten possible extremeÂ—highest or lowest Â— scores, eight of the extremes were attributed to Clusters 2 and 4; all in the appropriate direction.) It should be noted at some point that the DES is necessarily a unidirectional scale. That is, the five-point scale which is used could be scaled only positively; negative "quantities" of emotions would not be interpretable by respondents or researchers. The ends of the satisfaction continuum are described by patterns of different emotions, but the emotions arise from the same process. Thus, although it may later prove useful to have discontinuous measures of satisfaction, such measurement should not be taken as an argument for separate constructs. Westbrook's treatment of the entire post purchase phase of consumer behavior under the satisfaction rubric ignores the necessity of engaging in a specific process to effect a state of satisfaction. That is, the present conceptualization explicitly outlines certain antecedent factors that are necessary to the arousal of satisfaction feelings. The process described by the antecedents clarifies the discrimination that theoreticians have sought to make between satisfaction and the lack of a satisfaction response (this issue is discussed in detail below). Westbrook's research procedure omits
reference tc, and, in some cases, the presence of these factors. This omission makes further attempts to define the construct in terms of specific patterns of emotions problematic, especially since these were derived using his data. In addition, other aspects of the Westbrook studies raise concerns about basing the derivation of a measurement scheme on his findings. For example, the Differential Emotions Scale was modified but without a clear indication of the rationale of item selection or the necessity of diverging from Izard's own abbreviated scale. Westbrook' s implicit assumption that the modified scale will exhibit the same properties as the UE5 may not be borne out. Although these concerns do not condemn the data for the use to which they were applied, it is evident that the results are not adequate for the development of measures. Pilot Study I (Chapter 5) addresses these issues in more detail and is used to assess measures of the satisfaction construct which are based upon the DES. Motivation The importance of the distinction between satisfaction and non-satisfaction emotions lies in the notion that the motivational components of each of these classes of emotion are likelv to prompt different behaviors. The behaviors that are thought to result from satisfaction responses were identified in Chapter 2 and were even proposed as possible measures of the feeling. The validity of this measurement logic depends on how well the motivational relationship between the emotion pattern and the behaviors is defined.
89 Young describes motivation as "the 'energy' aspect of experience and reaction; a basic motivational principle is that varying degrees of stimulation liberate different quantities of energy" (1961, p. 16). Thus, the continuum that was presented in Figure 6 is actually reflecting this motivational concept of intensity (Atkinson and Birch 1978; Mook 1987; Steers and Porter 1975). Moreover, it is more consistent with Young's treatment of motivation and emotion and with Izard's measures of strength of emotion to assume that the emotional continuum that underlies the behavioral responses reflects changes in response strength (Figure 7b) rather than changes in composition (Figure 7a). Weiner (1980) presents similar arguments in an interpersonal context. Finally, it is useful to note that motivation is a familiar concept in consumer research. CS./D research has borrowed a number of concepts from the dissonance literature, among these motivation. Dissonance was typically treated as a tension which people were motivated to reduce (Cummings and Venkatesan 1976; Oshikawa 1969). Although reduction was sometimes effected through choice of subsequent behavior (Cohen and Goldberg 1970), dissonance was typically assumed to initiate new behaviors (Kiesler and Pallak 1976). This discrimination is similar to the selection versus persistence effects of motivation in general (Atkinson 1957; Vroom 1964). As presented in Figure 5, exactly parallel decisions are thought to follow satisfaction or dissatisfaction when it occurs. In fact, it has been argued that arousal may be sufficiently non-specific that the individual is unable to label the response CZillman 1978).
90 Apprehensive Unemotional Angry Guiltv Surprised Contented Minimum Satisfaction (Likert Scale) -Maximum a) from Westbrook and Oliver (1984) Low High' Interest Joy 'Surprise Anger Distress Disgust .^Contempt Interest\^ Joy ^>High Surprise..^ Anger ^ Distress \ Disgust )>Low Contempt/^ Minimum Satisfaction b) based on Strength of Pattern of Emotions Maximum Figure 7 Satisfaction As Emotional Continua
Carlsmith and Freedman (1968) present an interesting discussion that linked dissonance and satisfaction (although they did not use the latter term). In essence, they indicate that dissonance arises when a disconf irmation can be attributed to internal factors ("I should have known better"). External factors ("You did not tell me that") are more likely to instill what has come to be called satisfaction. While this is compelling (see the attribution discussion in Chapter 2), most CS D researchers have assumed relatively complete information availability. Lack of a Satisfaction Feeling The discussion of motivation returns us to an issue that was briefly alluded to in the measurement section of Chapter 2; the occurrence versus non-occurrence of satisfaction. This question has been raised repeatedly since the earliest CS/D literature focusing on theory development (Day 1977; Hunt 1977a; Olander 1977). Empirical reports have been much less likely to acknowledge and control for such contingencies in descriptions of research designs and measurement procedures. This section of the dissertation integrates theoretical explanations of the occurrence of satisfaction with research designs and measures more sensitive to its non-occurrence. Given the current lack of detail within the disconf irmation paradigm, it is difficult to maintain a firm concept of how or when satisfaction occurs. It is not unusual to find theory and measurement techniques that are essentially in opposition on the issue. At one point, Day argued eloquently for added sensitivity to the detection of the feeling: It seems more plausible to assume that the consumer usualiy enters the consumption experience without even thinking about
being satisfied or dissatisfied. In general, something out of the ordinary must occur either prior to the purchase, during the purchase process, or during consumption to alert the consumer or call attention to some aspect of the purchase situation. Research procedures which assume that an evaluation always takes place may produce one as an artifact of the research'." (1977, p. 150) Yet, only a short time later, he reported forcing satisfaction responses where, in fact, they may not have occurred. It was found that a five-point scale with a neutral midpoint appeared to provide a ccpout for a small number of respondents. When respondents were not offered a neutral response they seemed comfortable with making a choice between 'somewhat satisfied' and 'somewhat dissatisfied'. (Day and Bodur 1979, p. 184) Swan and Trawick (1979) reported similar discomfort with their own methodology: "As an important methodological implication, we found that if you ask for an evaluation you will get one Conceptually, if satisfaction is the product of a conscious evaluation, it follows that the research must first determine if an evaluation occurred before asking the consumer about his satisfaction" (p. 60). Comments such as these and acknowledgement of the issue in reviews of the CS/D literature (e.g., Dickson and Wilkie 1978) are laudable for raising an important issue but do not advance our understanding of the phenomenon. Some authors, on the other hand, have attempted to explain and predict the occurrence or non-occurrence of satisfaction. The Two-Factor theory discussed earlier posited separate causes of satisfaction and dissatisfaction. This implicitly theorizes a state in which neither occurs, although the techniques typically used (e.g.,
93 critical incident) tend to avoid such situations (Maddox 1981; Swan and Combs 1976). A more recent cperationalization of the disconf irmat ion paradigm suggested that although performance is appraised against norms in all consumption situations, the discrepancy may not always be detectable. When performance falls in this "zone of indifference", no satisfaction feelings result (Woodruff, Cadotte and Jenkins 1983a). The width of the zone is postulated to vary among consumers and situations, although precise determinants of this dimension are not hypothesized. The authors propose that the factors which influence the width of the zone be studied and further suggest that unobtrusive measures may be necessary to avoid demand artifacts. 3 Thus, the measurement techniques discussed in Chapter 2 gain added importance. If, as Day (1977) suggested, the typical "cognitive" measures of satisfaction produce measurement artifacts, then the less obtrusive emotional and behavioral measures should be considered. Behavioral measures were argued to be the least obtrusive (i.e., they can be collected by observation) but also the least sensitive (i.e., motivation must be relatively high to generate any response). The emotional measures proposed by Westbrook (1983) appear to be a reasonable trade-off between obtrusiveness (i.e., still a paper and pencil methodology, but with a less obvious research intent) and sensitivity (i.e., the responses are scalable). 3 Note that these authors implicitly assume that only disconf irmat ion will produce a satisfaction response. As adapted in the dissertation, the disconf irmation paradigm does not make so strong an assumption. Given the appropriate circumstances, satisfaction can result from a confirmatory experience.
Westbrook (1984) recognized the necessity of establishing the presence or absence of "true" satisfaction: "The study fails to provide a convincing answer as to whether indeed the presumed sentiments of satisfaction actually took place, and were not simply the product of the researcher's instruments" (p. 311). Yet, his use of the DES (1983; Westbrook and Oliver 1984) does not reflect a similar concern with the distinction. Based on the research reported by these authors, positive satisfaction can be defined tentatively as characterized by the presence of a pattern of emotions including Interest, Joy, and Surprise. Similarly, negative satisfaction can be characterized by a pattern including Anger, Distress, Disgust, Contempt, and Surprise (Figure 7b). This definition should not be interpreted as a revival of the separate constructs view of satisfaction; no difference in process is proposed. 4 The satisfaction feeling as defined is considered to be a continuous range of emotional patterns in which the middle range is defined by a pattern of emotions that combine positive and negative elements. As one's feelings approach the ends of the continuum, both valence and strength increase which affects the increase in motivation that was cited earlier. The absence of any satisfaction feeling (positive or negative) is defined by patterns of emotions that lie outside the continuum. 4 This definition, again, is tentative. It is based on studies which were not conducted with the objective of constructing a definition; Westbrook 1983; Westbrook and Oliver 1984.
95 Finally, it is appropriate at this point to recognize that CS/D is not the only literature to wrestle with this issue. Over the years, the questions of whether attitudes are as intense as had been presumed (Guttman and Suchman 1947) or even present at all (e.g., Converse 1970; Lastovicka and Bonfield 1982; Pierce and Rose 1974) have been raised and debated. Again, the arguments arise from the methodology that is typically employed in attitude studies. In that period I had been struck by two facts in particular. One was that very many respondents could not understand that a battery of pure opinion items had no objective 'right-wrong' scoring, or that don't know responses were not a confession of the most abject ignorance, tc be avoided at all cost. The other was the frequency with which respondents chose a response alternative dutifully but accompanied the choice with side cues--shoulder-shrugging, eye-rolling, giggles and even sotto voce commentsÂ— indicating that they were very much out of their element and would pick any alternative haphazardly by way of helping me out. (Converse 1974, p. 650) Lastovicka and Bonfield (1982) also noted this problem with demand artifacts and proposed that less leading measures (e.g., open-ended questions) be used to assess attitude or opinion (see also Heller 1982; Norpoth and Lodge 1985). Of course, difficulties arise in quantifying such techniques and estimating their reliability. But in some applications, these are less problematic than the over-estimation cf attitude presence that is likely to accompany forced-response questionnaires. Nevertheless, such questionnaires have dominated the attitude as well as CS/D literatures. Other Types of Satisfaction Th''s existential question is not limited to the consumer domain. Consumer satisfaction is not thought to be a unique feeling, but rather one manifestation of a more general pattern of emotions. However, the
pattern of emotions is aroused only under certain circumstances, such as described by the enabling conditions that are described below. Various literatures have identified other targets of a satisfaction feeling which may be addressable under the paradigm presented here. Locke (1967) proposed a general satisfaction "affect", for example, that is cited in both the job satisfaction and CS/D literatures. These similarities have been noted by others working in the consumer domain. Westbrook and Oliver (1981) derived satisfaction measures from other literatures (e.g., marital satisfaction, life satisfaction) and theories of "consumer satisfaction" have been adapted directly from the job satisfaction literature (Adler and Robinson 1980: Maddox 1981; Swan and Combs 1976; Westbrook and Reilly 1983). Belk quotes William James thusly, A man's self is the sum total of all that he CAN call his, not only his body and his psychic powers, but his clothes and his house, his wife and his children, his ancestors and friends, his reputation and work, his lands, and yacht and bank-account. All these things give him the same emotions. If they wax and prosper, he feels triumphant; if they dwindle and die away, he feels cast down--not necessarily in the same degree for each thing, but in much the same way for all. (1986, p. 6) The job satisfaction literature dates back at least to the 1 930 s and, as such, is much more extensive than the research in CS/D (Locke 1969). This has also allowed deeper development of individual theories and constructs than has thus far been the case in the consumer domain. One such theorv, Herzberg's Two-Factor theory (Herzberg, Mausner and Snyderman 1959), has been discussed as the antecedent to Two-Factor approaches in CS/D.
97 Although the specific contention that the attributes that cause satisfaction are separate from the ones that cause dissatisfaction has been rejected in Chapter 2 and has been discounted strongly even in the job satisfaction literature (Carroll and Tosi 1977), the general approach has been shown to be consistent with the disconf irmation paradigm as operat ionalized in the dissertation. That is, the emotions (similar to Locke's conceptualization) are evoked by disconf irmation of certain expectations about individual aspects of a job. The critical incident research technique suggests that Herzberg and his colleagues were, in fact, more concerned with specific disconf irmations than with general attitude toward the job. In contrast, the majority of the job satisfaction literature has focused on more deeplv ingrained, longer-held responses that seem more consistent with the attitude construct than with satisfaction (e.g., Steers and Porter 1975), as each was defined in Chapter 2. Typical measurement techniques consist of multi-item scales for rating various attributes of one's job (Smith, Kendall and Hulin 1969; Wanous and Lawler 1972). Recent organizational behavior texts also define job satisfaction in terms of attitude (Feldman and Arnold 1983; Reitz 1981) and report that the relationship between satisfaction and presumably resultant behaviors (Absenteeism and TurnoverÂ— equivalent to Loyalty and Exit in Figure 6) is weak. This apparent discrepancy between theory and findings may be the result of mis-specification of motivations. Vroom (1964), in a landmark book dealing with job satisfaction and motivation, made the distinction between motivation to act and motivation to act in a particular way. He chose to concentrate
98 on the latter and operationalized satisfaction with job as an attitude construct. The discrimination made seems to have been omitted in much of the later research that attempts to predict satisfaction responses from measures of the attitude. 5 The Relationship Between Satisfaction and Attitude In Chapter 2, satisfaction and attitude were defined as separate constructs that should not be confounded by theory or measurement technique. However, the willingness of some authors (e.g., Evans 1983; LaTour and Peat 1979) to treat satisfaction as synonymous with postpurchase attitude reflects the strong links that do exist between the concepts. As an emotional response, satisfaction is experienced for a very limited time. Oliver (1980a; 1981) has recently proposed that the response does not merely dissipate, but rather "decays into" overall attitude. In the process, one's prior attitude is "updated" by this new experience. The process by which satisfaction decays and what factors cause or mediate the process are net clearly specified. Oliver (1981) hypothesizes an opponent process that exists to keep emoticnal levels within some "normal" homeostatic range (based on the work by Solomon and Corbit 1974). A theory such as this, however, does not suggest why attitude is affected but only that satisfaction will dissipate Satisfaction with marketing related posit ions--salesman, etc. --also has been discussed in the marketing literature as an attitude phenomenon, Becherer, Morgan and Richard 1982; Churchill and Pecotich 1982; Futrell and Parasuraman 1984.
9 The short-lived nature of the satisfaction feeling is an important factor in the study of satisfaction effects. For example, if complaining is influenced by satisfaction but not by attitude as some findings suggest (Bearden and Teel 1983), then the motivation to engage in such behavior may dissipate along with the emotional pattern. Indications that this may occur arise from surveys concerning dissatisfaction and complaining behavior in which "I wanted to do something about it but never got around to it" is offered as a reason why a consumer did not complain (Day and Ash 1979). That is, there may be a discrete interval after a satisfying experience in which the range of satisfaction-related behaviors may result but after which the net effect is witnessed only through attitude. This is similar to the discrete interval in which mood seems to affect other social behaviors (Isen, Clark and Schwartz 1976; Walster .1964). This "window" may be regenerated in further experience with the item or in conjunction with related behaviors such as actual complaining (Andreasen 1977b; Diener 198C). (On the other hand, as argued previously, certain behaviors such as repurchase are thought to be related more to attitude than to satisfaction Â— e.g. Bearden and Teel 1983; Oliver 1977.) The relative influence of attitude and satisfaction on consumer behavior, as proposed, is depicted in Figure 8. The representation of the horizontal axis, Time, assumes that there is already some prior attitude which is affected only by the consumption experience at time t (the satisfaction occasion). The vertical axis, Resulting Behavior, is the continuum from Figure 6 rotated on one end. The figure shows that in the absence of a particular satisfaction occasion Resulting Behaviors
100 Compliment Positive Word-of-Mouth Loyalty Exit Negative Word-of-Mouth Complaining Emotion or Satisfaction Attitude >-Ti ^/ Emotion or Satisfaction Stronger Influence on Behavior Weaker Influence on Behavior Figure 8 Relative Influence of Satisfaction and Attitude on Certain Consumer Behaviors
101 (i.e, prior to t ) attitude guides behavior in the middle range of the scale. When satisfaction emotions are aroused (either positive or negative, as represented by the mirror image) the more extreme behaviors may be motivated. However, the motivation decays until attitude is again dominant ( t n ) and eventually there is no residual motivation. 6 Specifying the nature of the reaction of satisfaction over time is an issue of critical importance to the researchers interested in the construct. Studies which are designed to examine satisfaction must measure the construct within the appropriate period or at least (as with Herzberg's critical incident technique) attempt to target the interval with the procedure and measurement techniques. Within the CS/D literature, studies typically have neglected this time dimension and created unnecessary ambiguity in their interpretation (e.g., LaBarbera and Mazursky 1983; Westbrook 1983). This confusion persists in part because satisfaction and attitude are related (as suggested in Figure 3), but the precise nature of this relationship has not been documented nor even well-conceptualized. Oliver's "decay" theory has sometimes garnered only weak support (Oliver and Linda 1981) and does not contain much explanatory power. Kennedy and Thirkell (1982) considered only the cognitive aspect of satisfaction and suggested that attribute level disconf irmation may increase the perceived importance or salience of that attribute which 5 The decay of motivation does not stop at t n because there is no motivational component of attitude hypothesized. In fact satisfaction and attitude processes may be simultaneous but separate, as noted below. The figure however, is not limited to theories utilizing this argument, it serves merely to represent observed change over time.
102 will persist through future attitudes. Similarly, Fishbein and Ajzeri (1975) had proposed, in the more general case, that satisfaction and attitude processes occur in parallel. Rather than the decay of satisfaction causing attitude change, this theory posits the decay of satisfaction leaving attitude change as the sole trace of separate processes. Olson and Dover (1976) report changes in cognitive structure resulting from product trial that are consistent with this formulation (i.e., the beliefs about product attributes were different after usage experience). Discrimination and selection among these theories is well beyond the scope of the dissertation. However, the general framework provides for more discriminable conceptualization and measurement of the two constructs. The temporal relationship suggested is fairly well established (Westbrook and Cote 1980 found little evidence for a direct effect of prior attitude on satisfaction) and the measurement techniques that have been discussed are useful tools for the future examination of the process. The next section introduces a conceptualization of the general satisfaction process that posits certain antecedent conditions which must occur to evoke this more emotional, motivational feeling (as contrasted with simple attitude). The Extended Conceptualization Day's model of the satisfaction process (Figure 3) is an adequate representation of the CS/D literature at the time the model was published (1983). A revised version (Figure 9) incorporates much of the preceding theoretical discussion defining the construct and its properties
!G4 In contrast to the Day model, the new conceptualization describes a generic satisfaction process that operates to evoke the critical pattern of emotions in many situations. Thus, although Day strove to integrate all of the CS/D literature, the current model spans CS/D, job satisfaction (Locke 1969), marital or relationship satisfaction (Gurwitz and Dodge 1977; Rusbult 1983; Rusbult and Zembrodt 1983), interpersonal satisfaction (Brewer 1977), satisfaction with decisions (Kourilsky and Murray 1981), and satisfaction with particular types of consumer situations (leisure activities Â— Bloch and Bruce 1984; Sobel and McGuire 1977; patient satisfaction with healthcare--Swan and Carroll 1980; or satisfaction with legal services Â— LaTour 1978). The new conceptualization can be summarized as follows: Satisfaction is the feeling that results from (1) Expectations that lead to (2) a Decision that is followed by (3) personal Experience in which the characteristics of the object of the decision are subject to (4) Perception which, in instances of sufficient (5) Response Involvement allows for a (6) Conf irmation/disconf irmation of the expectations and decision concerning the object. Note that the constructs are in no way labeled as specific to the consumer domain but can be operationalized in a variety of contexts. There are several features of the conceptualization worth noting here: (1) More constructs are proposed than in Day's (1977) model. Although the disconf irmation paradigm deals well with the cognitive dimension of satisfaction, more factors are necessary to understand the
105 total process. Each of the constructs is discussed in the following sections. (2) The six antecedent conditions are considered to be necessary and collectively sufficient for the arousal of the patterns of emotions characteristic of satisfaction. As such, the absence of any of these raises questions about the validity of any attempt to invoke the satisfaction construct. Thus, the conceptualization can be used as a template for the design of research to examine any type of satisfaction. (3) As with Day's model, the conceptualization is not meant to function as a causal model of the satisfaction process. The relationships designated in Figure 9 and the passage of time (denoted by movement to the right) are merely representative and do not encompass all of the influences that are possible. Derivation and validation of such causal theories will fellow demonstration of the necessity and collective sufficiency of the constructs. In the following sections, each antecedent condition is discussed in more detail. The discussions will generally be focused in the CS/D literature to allow clarity without loss of generality. Three of the constructs, expectations, perception and conf irmation/disconf irmation are well established in the dominant paradigm and are explicated in less detail than the three constructs that are being introduced. An evaluation of the role of this latter group (especially Decision and Response Involvement) constitutes the empirical portion of the dissertation.
106 Expectations A concept that has been central to the formation and use of the disconf irmation paradigm is the presence of prior expectations about the outcomes possible from a given situation. Comparison of the perception of actual outcomes against these expectations yields the disconf irmation judgment. Without expectations, satisfaction would be based solely on, and possibly would be indistinguishable from, perception. Within the CS/D literature, the source, nature and influence of expectations have enjoyed a prodigious amount of research attention. Cardozo's (1965) seminal paper assumed that expectations influence satisfaction and proposed an interaction between expectation and effort. Later findings suggested that expectations may even affect perception by masking reality or, at least, subjects' reports of their perceptions of reality (Anderson 1973; Anderson and Hair 1972; Deighton 1984; Olshavsky and Miller 1972; Olson and Dover 1978). Some of the results reported in Chapter 6, below, support this contention. This attention to the expectations construct has not waned in recent years. Broad models of the entire satisfaction process continue to center on the disconf irmation of expectations; and calls for better understanding of the construct are common (Day 1983; Woodruff, Cadotte and Jenkins 1983a). The problem with much of the research that attempts to address these issues is that it is not conducted within the context of these more complete theoretical networks, leading to results that may be bound by situation. Nevertheless, the concept of expectations has become axiomatic in theories of individual
107 satisfaction and is used to account for levels of satisfaction in macro, societal contexts (e.g., Summers and Granbois 1977). Less well agreed-upon is a specific definition of expectations. Among those offered have been 1) "the perceived likelihood that a product possesses a certain characteristic or attribute, or will lead to a particular event or outcome" (Olson and Dover 1976, p. 169) 2) "pre-decisional or pre-consumption beliefs about the overall performance of the product created by manufacturer's claims or test-report information" (LaTour and Peat 1979, p. 433) 3) prior attitude (in Fishbein's affect-belief sense) representing a homeostasis position prior to adaptation to a new experience (Oliver 1981) 4) expectations of satisfaction (Bahr 1982; Oliver 1981) 5) predicted performance of a brand (versus a performance norm within product class) (Woodruff, Cadotte and Jenkins 1983a) 6) "An expectation of product performance is simply a prediction of what performance will be expectations [and] norms are also measured for the social benefits of the consumption event and the total economic costs of the consumption event" (Day 1983, p. 113). A number of issues are unresolved in the treatment of expectations within the CS/D literature, as evidenced by the above definitions. Whereas earlier conceptualizations focused on attribute level
cognitions (e.g., Olson and Dover 1976), more recent theories have tended to operate at the overall perception level (Day 1983) and seem to approach the type of utility expectation discussed by Peter and Tarpey (1975). Furthermore, the limitation of expectations to simple predictions about the future is also not unanimously accepted. A recent review (Gilly, Cron and Barry 1983) of the research that focused on the expectations portion of the model pointed out that investigation in this area is still active. Of the four types of expectations suggested by Miller (1977), Expected, Deserved, Ideal and Minimum Tolerable, only two have received much support as discrete constructs. "Expected" expectation is essentially the probabilistic prediction most often referred to in satisfaction theories in both the consumer and job literatures (e.g., Filer 1952; LaTour and Peat 1978; Summers and Granbois 1977; Swan, Trawick and Carroll 1981; Wanous and Lawler 1972). Deserved expectations, on the other hand, are more of an equity phenomenon, based on the individual's "input" to the system (e.g., Sobel and McGuire 1977; Swan and Trawick 1980a). Although measures taken can discriminate between these types, no real difference in process is assumed. The other two types of expectations proposed by Miller, Ideal (the "wished for" level Â— similar to the aspirational level posited by Filer 1952) and Minimum Tolerable (absolute "bottom performance"), have not been discriminably measured, nor is it clear in the present conceptualization how they might affect satisfaction separately (Gilly, Crcn and Barry 1983; Sobel and McGuire 1977). It seems more likely
109 that all of these types of expectations combine to form an "Expected" expectation or simple prediction, as described by Day, that is reminiscent of a Bayesian probability judgment (Winkler 1972). Day (1983) and Woodruff, Cadotte and Jenkins (1983a, b; Cadotte, Woodruff and Jenkins 1983) treat expectations as a brand (or possibly item) specific construct. That is, the predictions made concern the specific target of the satisfaction feeling. At the more general (say product class) level, prior experience will have generated judgments about the normative perception for that situation (Cadotte, Woodruff and Jenkins 1987; Chase and Ericsson 1982; Chase and Simon 1973). These norms are presumed to have an influence separate from expectations on disconf irmation (Figure 3). This possibility is related to the complicating effects of expectations on relevant schemata (Alba and Hutchinson 1987) but the effect on CS/D constructs has not been demonstrated empirically, as yet. Finally, other "types" of expectations have been identified and discussed. One segment of the CS/D literature deals with Comparison Level theories of satisfaction. In essence, these propose that disconf irmation results from evaluating obtained perception in light of a Comparison level that is composed of predictions based on (1) one's prior personal experience (2) experiences of relevant referent persons and (3) unique characteristics of the present situation (LaTour and Peat 1978, 1979; Miller 1977). Although this is sometimes presented as a competitive explanation to disconf irmation (Barbeau and Quails 1984; Haines 1979; Swan and Martin 1981), the structure is consistent with the formation of a Bayesian probability judgment. In effect, the
1 10 so-called Comparison Level formulation is more a theory of the source of expectation than of its structure. Thus, this is better discussed with other theories of expectation formation, such as those dealing with advertising and interpersonal communication (Gilly, Cron and Barry 1983; LaTour and Peat 1980; Oliver 1979) than as an explication of separate constructs. As suggested by Gilly, Cron and Barry (1983),this attention to the source of expectation may reveal the process by which expectations change over time, thus creating different levels of satisfaction over time. The concept of expectation that is accepted for the dissertation, then, is essentially the adaptation level proposed by Oliver (1981), operationalized in a manner similar to Day (1983). In contrast to Day's treatment, however, the more general norms that he proposed be kept separate are believed to be encompassed within the prediction. This probability judgment results from a Bayesian-type process and can be generated for a target stimulus at any level of specificity (Table 2) that is within the person's experience. (Day and Woodruff et al. focus on brand or item satisf action--making it difficult to extrapolate beyond the concept of brand expectations and product class norms.) Within the present model, separate attribute level expectations are not thought to be relevant to the net effect of the satisfaction process. As will be discussed in the "Experience" section, salience of specific attributes may change through consumption or a discrepancy on an individual dimension may arouse a higher "Response Involvement" level. At this point, we are concerned only with an overall expectation that
11 1 results in a decision to commit to the stimulus that may become the target of a satisfaction feeling. Decision In the CS/D literature, satisfaction historically has been discussed as a "postpurchase" phenomenon. But rather than assume that this label merely designates a passage of time, it is argued that the purchase act itself is necessary for and can exert an influence upon the satisfaction process. At the more abstract level at which the conceptualization is presented, the required act is described as the decision to make a commitment to the satisfaction target. A great majority of the research in consumer behavior examines the factors which result in a consumer's choice or selection of a good. There has been comparatively little attention paid to the outcomes of such choices. The literature that arose relating cognitive dissonance (Carlsmith and Freedman 1963; Festinger 1957, 1964) to consumer behavior, however, did indeed focus on the choice and its effects on the individual. In their review of this literature, Cummings and Venkatesan (1976) argued that certain prerequisite conditions must be fulfilled in order that dissonance occur. Among these were cited "volition and irrevocable commitment to the decision (product choice)" (p. 304). These conditions circumscribe that which is being called decision in the conceptualization under discussion. 7 Some of the earliest papers that are cited often in the CS/D literature (e.g. Cardozo 1965; Cohen and Goldberg 1970) arose from this 7 Cummings and Venkatesan also argued that the decision must be important. This is treated separately under Response Involvement, below.
12 dissonance tradition. Cohen and Goldberg specifically manipulated decision (among other manipulations), as defined herein, to create "no dissonance" control groups. Unfortunately, no checks of this manipulation were reported, so the success of the strategy can not be easily assessed. Nevertheless, decision was believed to act to solidify cognitions and make them more resistant to change, thus arousing dissonance when the prior cognitions were disconfirmed (see Cialdini, Cacioppo, Bassett and Miller 1978 for a similar discussion). The dissonance literature is not without its limitations and criticisms. Even fairly early in the stream, Davidson and Kiesler (1964) raised concerns regarding observation of dissonance or its effects prior to or without a decision being made (also, Carlsmith and Freedman 1968). The authors cited Festinger's sharp distinction between "conflict" (a pre-decision construct) and "dissonance" (a post-decision construct) Janis (1959), on the other hand, thinking in terms of 'conflict resolution,' implies that there is little or no distinction between preand post-decision behavior and that systematic re-evaluation occurs both before and after the decision. (Davidson and Kiesler 1964, p. 10) A study conducted with thirteen to sixteen year-old girls supported Festinger's dichotomy, albeit not conclusively. Directional evidence is found for a decision effect on re-evaluation of job candidates. As Festinger points out, however, a counter-explanation merely requires an inference that delay between instruction and decision was sufficient to allow re-evaluation. It is also not clear whether the early dissonance experiments forced a decision under circumstances in which the subjects would not
1 13 ordinarily decide or try actively to avoid a decision (Braden and Walster 1964; Weick 1967). This forced decision may counter, or at least alter, the effect of the decision on the resulting cognitive structure. Freedom to choose is the immediate precursor of responsibility, for interpersonal life holds one accountable for what he chooses, not, ordinarily, for what he is forced to do. (Temerlin 1963, p. 39) This free, volitional decision is conceptually similar to the commitment construct elaborated by Kiesler and labeled "behavioral commitment" by Johnson (1973). Kiesler defined commitment as "the pledging or binding of the individual to behavioral acts" (1971, p. 30). This definition does not reflect all of the complexity of the construct, however. Kiesler gives more clues to the nature of commitment by proposing experimental manipulations that presumably would vary the level of commitment. These include varying 1) the explicitness of the act (of commitment) 2) the importance of the act for the subject 3) the degree of irrevocability of the act 4) the number of acts performed and 5) the degree of volition perceived by the person in performing the act. Similar manipulations of the decision construct proposed in the dissertation may also yield a complex structure. At this time, however, only a minimal structure is hypothesized. It is essential that an explicit, irrevocable, volitional choice be made (e.g., of a product, of behavior with respect to a product, etc.) in order that the
1 u satisfaction feeling, as conceptualized herein, result. It is argued that one act is sufficient to engage the satisfaction process; importance of the act is discussed below under Response Involvement. 8 Kiesler (1971; Kiesler and Pallak 1976) also argued that the result of "fixing" cognitions is central to his conceptualization of commitment (see also Aronson 1968; Festinger 1957; Gerard 1965, 1968; Lewin 1951, 1952). A similar case has been made in the marketing literature, Involvement results from the fact that important values or the person's self-image are engaged or made salient by a decision situation. This would lead to the arousal or drive activation that Mitchell (1979) has suggested. Commitment ([which is approximately equal to] position involvement) results when these values, self-images, important attitudes and so on become cognitively linked to a particular stand or choice alternative. (Crosby and Taylor 1983, p. 415) Tentative support for this proposition was derived from an analysis of data regarding attitudes before and after a vote on a referendum. A high commitment group of voters exhibited greater stability in voting preference over time than did a low commitment group. In addition, post-election perceptions of outcomes were more highly correlated with pre-election preferences for the high commitment than the low commitment group. In Kiesler' s view, the strength of resistance to change (or amount of dissonance generated) is likely to be related to the level of commitment which can be varied by manipulating the factors cited above or other aspects of the situation. Effort is one such variable that, 8 Another use of the term, commitment, with regard to consumer behavior mere closely approximates what is now called involvement which is discussed below Â— Robertson 1976.
1 15 as Kieslar stated, "should have a dramatic impact on behavioral commitment" (1971, p. 172) but its influence results from a process too complex to understand so early in the development of commitment theory. A similar argument might be raised in explaining a complex effect of effort on satisfaction (Cardozo 1965). For the purposes of the dissertation, various degrees of decisiveness and their impact on satisfaction are not discussed. At this point in the development of the conceptualization, it is simply argued that making a decision is a necessary antecedent to the initiation of the satisfaction process (see a related discussion by Weick 1967). In this regard, the definition of decision that has been adopted is consistent with the conceptualizations of early dissonance theorists who discussed separately the decision that was made and the importance of the situation in arousing dissonance (Brehm 1956, 1959; Brehm and Cohen 1962; Festinger 1957). It has been argued previously that the satisfaction process results in an emotional, motivational state which is similar to the motivation aroused by "dissonance inducing" processes (Kiesler and Pallak 1976). In a sense, the motivation in each case arises from the "ego involvement" (a concept separate from but related to Response Involvement, discussed later) that results from the necessity to make, and adhere to a decision (Filer 1952; Locke 1967). Others have suggested that this self evaluation phenomenon occurs and is motivating for the individual (see, for example, Calder 1973, p. 257). In the CS/D literature, one explanation that has been proposed for the function of expectations in determining satisfaction (versus a
1 16 Sxmple perceptual process) has centered on this justification of the consumption decision (Czepiel and Rosenberg 1977; Shapiro 1968). In essence, it is the decision that is disconf irmed, rather than the expectations (Carlsmith and Freedman 1968). This purchase concept is reflected in some models of the satisfaction process, but is more often ignored (Day 1983) or positioned off to the side, not directly impacting on the satisfaction result (Andreasen 1977a). The conceptualization of satisfaction being presented (Figure 9) models decision explicitly as a direct input to the disconf irmation judgment; expectations lead one to make such a commitment. Thus, one's ability to make the choice is relevant in determining the level of satisfaction which is ultimately experienced, a prediction that has some support from research findings dealing with the behavior of both individual (Folkes 1984a; Wes-tbrook 1980c) and collective (Lambert, Dornoff and Kernan 1977) consumers. Similar arguments can be raised from reports that individuals' self-concepts may be actively involved in consumer situations (Grubb and Grathwohl 1967; Sirgy 1982). That is, by committing to a product (or other "target"), the individual is risking his/her identity as a capable decision maker (Walton and Berkowitz 1985). Smith and Swinyard (1982), on the other hand, propose that trial rather than true commitment is the object of many purchase decisions. Their formulation seems flawed in that the decision appears to be between "buy versus don't buy" rather than one of "which alternative do I prefer?" It is this latter decision that is claimed herein to be the more common and, therefore, more central to the satisfaction process.
i 17 This proposal of decision as a crucial link in the satisfaction process is not new. Dickson and Wilkie (1978) raised specifically the issue that, without a decision, a postpurchase condition can not be present. Folkes (1984a, b), Russo (1979) and others have discussed the possibility of the decision to complain resulting from a process by which blame for a problem is attributed either to another (e.g., "He made me look bad, so I'll gripe to or about him") or to one's self (e.g., "I made a bad decision; I'll keep my mouth shut"). Admittedly, it is at times difficult to isolate the point at which a decision is made. Decision making by groups (e.g., families, industrial buying groups) may have emotional results for the individual members who feel the requisite volition was present. Less clear, however, is the case where the purchaser is different from the user. Many consumer behavior approaches to satisfaction presume that the consumer has voluntarily chosen the product in question. For example, the presence of free choice is a prerequisite to the arousal of cognitive dissonance. An interesting set of issues therefore arises when the consumer evaluates a product or service chosen by someone else. (Solomon 1986, p. 216) In circumstances such as this, the user would experience satisfaction if (1) there is sufficient volition in the choice to use the product, or (2) there was a decision made to accept the buyer's judgment (assuming the other conditions are also met). In the latter instance, the buyer acts as the user's agent and has had the (purchase) decision making authority delegated (Jensen and Heckling 1976; Ross 1973: Solomon 1986). The user still has a judgment at risk, sufficient to arouse emotion, but not the selection of the product. This can result in redefining the target of the feeling (e.g., from the product to the
1 18 buyer) and subsequent behaviors, but this issue is beyond the scope oi the dissertation. As defined so far, the pattern of emotions that comprises the satisfaction feeling does not include shame, guilt or other indications of personal responsibility (ego involvement). Although this would clarify the relationship with decision somewhat, it may not be necessary to incorporate these emotions. The scales used have, again, not been seeking a clear definition of satisfaction so that the omission of these internally directed emotions may be an artifact of the measurement procedure. It should also be noted that satisfaction, when acting as a precursor to ext ernally directed behaviors may evoke less strong internal responses. Thus, how decision affects and is reflected in satisfaction and how the selection of subsequent behaviors is influenced are empirical questions that can not be answered convincingly at this time. The foregoing discussion, however, does make clear that decision is necessary to a "purchase" situation. Satisfaction that is measured after a typical marketing experience pre-supposes this purchase event (cf. Crosby and Taylor 1982). When satisfaction is studied in the behavioral laboratory, however, this step can be and has been omitted at times. For example, in Cardozo's (1965) study, subjects were never given an option as to the task they were to perform, the "reward" they were to receive, or any other aspect of the experiment. Thus, the effort manipulation was not likely to have the expected impact (involvement may have been raised, but emotions were probably not aroused)
1 19 Similar to the Cardozo study, other early experiments dealt with factors influencing postpurchase evaluation and did not arouse satisfaction because there was no opportunity to make an explicit, volitional and irrevocable decision (Anderson 1973; Oliver 1977; Olshavsky and Miller 1972; despite their frequent citation as satisfaction studies. In some cases (e.g., LaTour and Peat 1980) it is suggested, but not clearly stated, that subjects may have chosen to participate in the study on the basis of a promised reward. The disconf irmation of such a choice may be adequate to drive the satisfaction process. More often, however, the necessity of demonstrating that purchase occurred prior to studying a "postpurchase" phenomenon is ignored. Even recent satisfaction research has neglected this issue (e.g., Churchill and Surprenant 1982; Wilton and Tse 1983). Some attempts to explain the effects of advertising messages that are counter to reality have invoked the disconf irmation paradigm and the satisfaction construct (Deighton 1984; Olson and Dover 1978). These, too, are predicated on the assumption that expectations are disconfirmed and ignore the opportunity to show that decisions were made based on the expectations In summary, this section has argued that decision is a necessary antecedent to the satisfaction process. Aspects of constructs labelled "decision" and "commitment" in consumer behavior and psychology literatures have been merged into the single construct presented here. It does not seem fruitful to attempt at this time to disentangle similar presentations of differently named constructs (Fendrich 1967;
120 Hartley 1967). Expectations doubtlessly influence the direction taken by this decision, but are insufficient to arouse disconf irmation. What enters into the disconf irmation, then, is the individual's self-concept of his/her decision making ability plus a summary expectation of the result of that decision. Moreover, the expectation is now more resistant to further changes because, in effect, more is at stake. This expectation is compared, within the disconf irmation process, against a perception of the "true" result of the decisionÂ—which only can be derived from personal experience. Experience The point was made earlier that in instances in which the purchaser and user of an item are different people, it is the user (if either) who will experience a satisfaction feeling. This follows from the assumption that, whereas decision or commitment making authority can be assigned, the experience of personal "consumption" can not. 9 In the general conceptualization, experience is defined to be sufficient interaction with the target (of the feeling) to allow evaluation of its characteristics (Fazio et al. 1982; Fazio et al. 1986). In the usual paradigm within the CS/D literature this simply amounts to consuming enough of a product to characterize its 9 The special case of satisfaction with gift-giving or donation is not addressed specifically here. Such situations can be treated under the same conceptual framework, but the problem of including expectations and perceptions with respect to the recipient's reaction is an additional complication which can be addressed better as a separate issue.
12i performance. 10 Rather than a hypothetical construct, experience constitutes an (often) observable activity or event that leads to the satisfaction process being initiated (given the other necessary conditions). In contrast to cognitive dissonance (which focuses on the decision event), satisfaction requires information additional to that available in making the decision. Without this added information, there could be no disconf irmation. (Satisfaction is sometimes referred to as a postconsumption, rather than merely postpurchase, concept-LaTour and Peat 1979. ) Surprenant and Churchill (1984) have raised the question of whether vicarious evaluation of product performance is adequate. In a comparison of groups that actually used a product versus others that were merely told of the product's performance, the latter had generally more compressed results (i.e., closer to mean responses). This finding corroborates prior arguments that subjects in role-playing experiments are likely to be less involved than those participating in more realistic situations (Sawyer 1977). The necessity of determining that personal experience has occurred prior to assessment of satisfaction had been noted in passing by some authors (e.g. Park and Bahr 1980); but in general, the issue is not raised. The importance of the problem can not be dismissed so lightly, however. The process by which an individual perceives an object, event or another individual and integrates that perception into knowledge is 13 Within the dissertation, "experience" is used to refer to this consumption. In other contexts, "prior experience" has been used to refer to one's history with the item, product class, etc. prior to the purchase decision (Alba and Hutchinson 1987).
I 22 not well understood. Fazio et al. (1982) examine the influence of experience on attitude accessibility, a conceptually similar but discrete concept as outlined above. The early studies of postpurchase phenomena were attempts to identify the influence of expectations on the perceptual process (Anderson 1973; Olson and Dover 1975). In essence, it was believed that the information that could be derived from perception is somehow altered by expectations (LaTour and Peat 1979). These theories depended on the assumption that the individual was free to draw conclusions from a given consumption situation. Occasionally, however, laboratory experiments or other studies that did not allow the opportunity for personal experience have been reported in the CS/D literature. Although it is not certain what the actual effects of such omissions are, there are suggestions from psychology as to how processing may be affected. For example, in a similar discussion of experience-based versus indirect (i.e., based on second-hand information) attitudes, Fazio and Zanna (1981) proposed that actual personal experience may not only make larger quantities of information available but may also change the focus of attention and attribution. Just as the previous section discussed the personal commitment required for satisfaction (i.e., risking one's decision making skill); experience challenges the individual's skill as a user ("Did I cause the good/bad performance?). Folkes reported that consumers do at times make such attributions concerning the cause of product failures (1984b; also, Krishnan and Valle 1979; Valie and Wallendorf 1977)
123 In addition to amounts of information available and types of attributions made, personal experience can influence how the information is processed further or what response is generated. Fazio and Zanna (1981) cite a study reporting similar attitudes but different behaviors found in samples of students who were either affected directly or only tangentially by a housing shortage. The literatures on decision making and judgment provide insights into which aspects of information affect its processing (Ofir 1982). Lyon and Slovic (1976), for example, reported that subjects prefer to base judgments on information that they perceive to be most specific to the judgment they are making and to ignore other available data that could result in a probabilistically better judgment. Essentially, they argued that specific information was more salient than more general, more abstract data. This process is demonstrated quite nicely in a consumer context in which a subject is seen wavering between objective information placed before her and information from prior personal experience (Smead, Wilcox and Wilkes 1980). Similarly, Kahneman and Tversky (1973) had reported that the relative representativeness of information affected its usefulness for individuals. In the current context, it might be argued that a consumer would find self-gathered information more representative of the specific item than information provided through other means. Bar-Hillel's (1980) discussion of relevance and specificity of information would seem to rule out even personal experience with a sample or exemplar that is not the specific target of the commitment (also, Kamins and Assael 1987).
24 In general, the process of gathering and evaluating information about a target object and generating a response is rather complex (cf. Zajonc 1968). The factors that influence what information is used and how it is used would seem to be sensitive to changes as gross as the omission of the actual object. Our knowledge is not yet so sophisticated that we can correct for this in experimental designs or analyses (Smith and Swinyard 1983). These arguments were suggested by Olson and Dover in an early report in which they concluded that personal experience, "may have a greater effect on the acquisition and/or change of cognitive structure elements than does information from sources such as advertising or word-of-mouth" (1976, p. 173). At the time, however those authors were making a distinction between the effects of expectations (based on the latter information sources) and disconf irmation on cognitive structure. They did not anticipate the potential use of indirect experience to generate disconf irmation. In fact, they reported that even, "usage experience with most consumer products may not create the extreme disconf irmation necessary to produce a contrast effect" (p. 173). The problem is not limited to the laboratory experiment. Locker and Dunt (1978), in their review of patient satisfaction in the United Kingdom, reported difficulty in determining whether surveys assessed satisfaction of consumers or only potential consumers. The surveying problem is further compounded by the as yet unknown effects of repeated purchases and/or experiences (Kennedy and Thirkell 1982; Lutz 1980) on measured satisfaction. With field surveys, such problems arise only when non-consumers attempt to respond or frequent consumers differ in
125 their timing of expectations and disconf irmation. If the sample is representative, then the extent of these problems should reflect the characteristics of the population. Justification of lack of personal experience in behavioral experiments is not so easy to provide. Sawyer (1977) argued convincingly that role-playing situations are not appropriate for the majority of experiments which purport to draw conclusions from the data (also, Orne 1962). Yet, consumers are frequently put in "What if?" situations and asked to provide rating scale responses (e.g., Churchill and Surprenant 1982; Jordan and Leigh 1984; Lambert, Dornoff and Kernan 1977; Mowen and Grove 1983; Sirgy 1984; Winter 1974). In other cases, consumers are asked to respond to the experiences reported by strangers (Deighton 1984) or to report their own satisfaction prior to use of the object (Sirgy 1984). Day (19-83) excluded specifically this latter "'anticipated satisfaction" from his general conceptualization. As with the other concepts in the current model, experience can be defined at various levels of specificity. If satisfaction with a retailer is of interest, for example, then experience with that retailer--rather than with a specific product--is the requirement. Most situations for which consumers can readily volunteer satisfaction reactions or report that they have engaged in satisfaction-driven behaviors will have a readily apparent experience episode enveloped in the description. This section has argued primarily that experimental designs must include not only an opportunity for decision, but also the opportunity for personal experience in order that a realistic satisfaction process
26 may obtain. The most basic cause of this requirement is the investigator's inability to pre-specify all of the applicable information in a form usable for the subject to derive his perception of "reality" Perception The CS/D literature has recognized since its inception that objectively defined products or product performance do not evoke the same response from all observers. A variety of factors can influence an individual's impression of reality. In essence, the following working definition holds, "Perception can be thought of as the individual's mental impression of a stimulus object" (Anderson and Hair, referring to product perception, 1972, p. 70). The studies that initiated interest in the postpurchase phase of consumer behavior (Anderson 1973; Anderson and Hair 1972; Cardozo 1965; Oliver 1977; Olshavsky and Miller 1972; Winter 1974) and are often cited as satisfaction studies (LaTour and Peat 1979) actually treated perception as the dependent variable. In effect, each investigation was more or less interested in the role that prior expectations played in altering one's impression of the performance of a product. A review of the literature dealing with the perception construct suggests that each of the preceding antecedent conditions, i.e., expectations, commitment and experience, may affect the individual's perception of the target (Kassarjian and Robertson 1973). Early CS/D research typically manipulated onlv two of these, expectations and/or experience (i.e., in this case, experience is analogous to the manipulations used to demonstrate good or bad product performance).
27 The conclusion that can be derived from these studies is that perception is largely the result of the performance that is obtained in actual experience, but can be significantly modified by expectations (especially if the expectations have been solidified by commitment). Recent theoretical treatments of satisfaction and the disconf irmation paradigm have focused on expectations and perception (Day 1983; Woodruff, Cadotte and Jenkins 1983b) although performance is still the most common operationalizat ion of the latter construct (e.g., Surprenant and Churchill 1984). In empirical research, perception dominates expectations as a cause of satisfaction (Churchill and Surprenant 1982; Sobel and McGuire 1977; Swan and Combs 1976) although each is usually confounded with disconf irmation (generally operationalized as the difference between expectations and performance/ perception). Perception is also thought to play a role in determining responsibility for a given situation occurring (i.e., self versus other perceived as a cause) that may influence complaining behavior (Kassarjian and Robertson 1973; Settle 1973). Further, perception of performance may impact directly on the definition of the individual's "evoked set" of alternatives (Narayana and Markin 1975), thus influencing repurchase without a satisfaction or attitudinal mechanism, per se. There are various explanations for why perceptions differ from objective reality. Choice of products and judgments of the relative preferability of attributes have been shown to be sensitive to the context in which the judgment is made, including the range of alternatives available (Assar and Chakravarti 1984; Chakravarri and
128 Lynch 1983; Olshavsky and Miller 1972) and there is some evidence that the presence of negative attributes or of negatively evaluated levels of attributes alters the manner in which items are evaluated (Lynch 1984b; Sobel and McGuire 1977). As Hoch and Ha (1986) demonstrate, objective reality itself is dependent on the ambiguity of product cues. Some products, e.g., paper towels which differ on manufacturing specifications, are easier to compare than others, e.g., shirts differing only on color (See also Bergh and Reid 1980.). Anderson (1973) described in some detail the more comprehensive theories of judgment that may have implications for perception. (1) Discrepant performance experiences are assimilated closer to positive expectations. In effect, being wrong is uncomfortable (here, it seems to be the decision being disconf irmed) and the individual reduces this discomfort by altering his sense of "reality" Â— essentially a dissonance explanation (Festinger 1957). (2) A somewhat similar conceptualization was presented as a generalized negativity theory. In this case, any inconsistency with expectations (good or bad) causes the discomfort that people seek to reduce. This is based on highly counterintuitive data (Aronson and Carlsmith 1962; Carlsmith and Aronson 1963) that have been difficult to replicate (Sampson and Sibley 1965). (3) The surprise resulting from experiencing a discrepancy between expectations and obtained performance causes the individual to magnify the disparity. This is derived from contrast theory, attributed to Hovland and his colleagues (see also Anderson and Hair 1972 for additional discussion of this). (4) Later adaptations of contrast theory suggest that both assimilation and contrast are processes available to the
129 individual; the choice of which is engaged is dependent on characteristics of the situation (e.g., the degree of disparity between expectations and reality--Sherif and Hovland 1961). There are undoubtedly influences on perception beyond those outlined above, but in most cases these will be situation specific. As Lynch (1984a) and others have argued, it may be true that many of the perception effects obtained in behavioral experiments are artifactual and do not influence behavior. There is a trade-off between using more concrete attributes that allow quantification of effects and using more ambiguous attributes that may be more realistic in the generation of those effects (Zeithaml 1981). For the most part, then, perception in satisfaction studies is likely to have been fairly close to performance, at least as manipulated, and not too seriously biased by other factors. Finally, Woodruff, Cadotte and Jenkins (1983a) have recently elaborated on the perception construct and how it may function in the process that results in a satisfaction feeling. They contend that performance perception is always salient after an experience, but that any discrepancy with expectations may or may not be above a threshold of detectability In effect, they argue that perception does not occur as a continuous mapping of performance attributes, but rather it results in a discontinuous series of summary evaluative "zones". The range that includes those levels of performance equivalent to that which was expected is labelled the "zone of indifference". Onlydiscrepancies beyond this range gain the attention of the individual. Thus, there may be some discrepancy threshold (or ambiguity threshold)
130 that determines whether the net perceptual process is to be data-driven or concept (expectation) -driven (Lachman, Lachman and Butterfield 1979; Neisser 1967) The conceptualization presented herein takes a slightly different view of the process. In the current model, not all perceptions of performance reach high enough levels of attention to gain awareness. Similar to the Woodruff, Cadotte and Jenkins' scheme, however, experiencing extreme performance can attract attention. This, in turn, can result in a comparison against expectations and an appraisal of the commitment. In general however, a minimal level of response involvement (interest, attention) is necessary to justify the effort of engaging in the disconf irmation process (this involvement may be aroused through perception of extraordinary performance or other aspects of the situation) Response Involvement As is true with many topics in marketing and consumer behavior, involvement is considered to be an important but not well-understood concept (that is, there is little agreement as to the treatment of its specific causes, effects and components). This has led to a number of inconsistent or contradictory theories, many of which are associated with some empirical validation attempts. By and large, conclusions drawn from this literature are no less equivocal than those based on the CS/D literature. Nevertheless, there is some consensus that the broad concept of involvement as a hypothetical construct is likely to influence cognitive processing (Burnkrant and Sawyer 1983). The involvement
131 concept has influenced thinking and research concerning the satisfaction construct and was useful in deriving the current conceptualization (Richins 1985; Richins and Bloch 1986, 1987). The definition and operationalization of the involvement construct that were chosen are somewhat more focused, but are not intended to be conclusive for all purposes in all fields. The general concept of involvement traces back to the earliest research in the CS/D literature. For instance, Cardozo's (1965) "effort" manipulation would be treated as involvement under some conceptualizations. Westbrook (1980a) measured effort and argued that it reflected different levels of involvement. This latter operationalization is closer to the manner in which involvement is conceptualized in the dissertation. That is, Cardozo's manipulation merely showed that subjects in each condition (high versus low search "effort") were equally willing to cooperate with the experimenter's definition of a necessary task. In the Westbrook study, greater effort could indicate greater interest in the product, greater attention to the decision or other such motivational states. 11 The studies just discussed were concerned with the effects on satisfaction of varying the level of involvement. This question is logically preceded by an issue that has had little specific attention in the CS/D literature. That is, the question of whether "true" 11 The possibility of these is acknowledged to suggest involvement. It is not argued that the construct was or was not operationalized in this particular investigation in a manner consistent with the perspective taken by the dissertation.
32 Table 6 "Triggers" of Response Involvement 1 The item and/or the purchase occasion has some special significance for the individual. 2. The social context in which the purchase is made and/or the item is consumed calls attention to the product and/or the purchaser. 3. The consumer has had previous experiences with the product or service which suggest caution. 4. The consumer has been advised to be careful in making the purchase by friends, consumer organizations, or consumer protection agencies 5. The consumer is inexperienced and poorly informed about the purchase and use of the product and is more conscious of all aspects of the situation than would normally be the case. 6. The consumer encounters some unexpected circumstance which suggests caution, receives information or advice from a salesperson which is in conflict with prior beliefs, or otherwise encounters a "surprise" in the purchase situation. 7. The consumer discovers after purchase that the product does not have all of the expected features, fails to perform as expected, or is defective or flawed in some way. 8. The consumer discovers after purchase that the product has desirable qualities or features which were not expected or otherwise performs at a higher level than expected. ;From Day 1977)
33 satisfaction can be evoked in situations of "true" low involvement is still outstanding (e.g., Zimmer and Deshpande 1984). Day's (1977) discussion of cues that may trigger the evaluative process that results in satisfaction provides the foundation for the present explication of involvement. The examples he provided of such cues (Table 6) parallel the representation of response involvement provided herein (Figure 9). That is, a high level of involvement may arise early in the purchase process as a result of prior knowledge (Day's examples 1-3), or very late, as a result of something perceived about the product (Day's examples 6-8), or at some point in between (Day's examples 4-5). The major contention, however, is that some "critical mass" of involvement must be attained in order to initiate the evaluation and evoke emotion (see also Olson and Dover 1976). The question of whether or not satisfaction always results from a consumption experience has been discussed at length. It has also been suggested that the more extreme behaviors thought to follow satisfaction require the higher motivation level that is generated by the emotion (compared to that derived from simple attitude). Involvement is conceived of as a kind of activation or arousal which enters the process and becomes the source of motivation (analogous to a conservation of energy principle) (Bloch and Richins 1983). Thus, to generate the higher motivation necessary for emotions, the involvement input would need to be greater than what might be required for attitude formation (or some other, less strong response) (see Locke 1969 for a similar model). The involvement literature itself generally deals with communication situations and correspondingly low involvement and
34 motivation levels, so it is not clear how a complete model of the interaction of these concepts may be developed. Involvement has been defined in a variety of ways. Rothschild's concept is close to that just explicated, "To me, involvement is a state of interest, motivation or arousal; in turn effort is a function of the level of involvement" (1984, p. 216). Ihis is more motivational than the result-oriented definition that he had at one time presented for response involvement "the complexity or extensiveness of cognitive and behavioral processes characterizing the overall consumer decision process" (Houston and Rothschild 1978b, p. 185; see also Stone 1984, for a similar discussion) Several authors have attempted to separate the issues of importance and involvement (Bloch and Richins 1983; Cohen 1983). Cohen (1982) also wrote of involvement as an "activation level" which may or may not be directed at a particular target. Operationalization of the construct "requires some measure of focused attention, ideally one which takes account of capacity available at the time for competing tasks" (p. 3; also, Greenwald and Leavitt 1984). All of these conceptualizations are generally consistent with a position that resulted from a special conference on involvement. "The proposed generic definition was that involvement is a state of motivation, arousal or interest. This state exists in a process. It is driven by current external variables (the situation; the product; the communication) and past internal variables (enduring; ego; central values). Its consequences are types of searching, processing and decision making" (Rothschild 1984, p. 217). Scales by which to measure
135 involvement have been developed (Belk 1984a; Bloch 1981; Sherrell and Shimp 1982) but care must be taken that related but undesired constructs are not confounded in the measurement (Cohen 1982). Most of the recently developed scales deal more with enduring than situational involvement (e.g., McQuarrie and Munson 1987, undated; Zaichkowsky 1985) In adopting the generic definition for the current conceptualization, some side issues arise. As discussed previously, the present use of the decision construct does not imply a high level of commitment (as is usually thought to be synonymous with arousal of involvement). Rather, the model posits some change in cognitive processes (including those relating to the self--hence, ego involvement) that results from making an irrevocable selection. There is some evidence from the CS/D literature (Muncy and Hunt 1984; Olson and Dover 1976; Winter 1974) and other satisfaction literature (Freedman 1964; Locke 1967) that decision effects are discriminable from the effects of variables which should function through involvement One such variable that is commonly included but not analyzed in consumer behavior studies is ownership (or personal relevance--Petty and Cacioppo 1986). A primary difference between most laboratory experiments and most field surveys is that ownership is seldom transferred in the laboratory. While several researchers have noted this discrepancy and included change of ownership as part of a study, ostensibly to increase involvement (Anderson and Hair 1972; LaTour and Feat 1980; Petty, Cacioppo and Schumann 1983; Winter 1974), they failed
36 to report whether or not the strategy succeeded. It may likely be the case that being given an inexpensive object is not sufficient, in itself, to evoke "motivation, arousal or interest". Thus, the question of whether ownership affects involvement and, in turn, arouses satisfaction has not been answered. Oliver and Bearden (1983) attempted to integrate involvement into satisfaction concepts and models. Although the issue they raised is clear, flaws in the methodology of the study render their results uninterpretable For example, the cover story used involved "free sampling" so no commitment was generated. Second, and more important for the present discussion, the special nature of the product (an amphetamine diet aid) suggests that all users were likely to be highly involved. Even the authors acknowledged that the major "manipulation" (actually, a discrimination based on a rating scale) may not have been effective Whereas Oliver and Bearden may not have measured low enough levels of involvement to discern the effects of various levels of the construct, other studies reported in the CS/D literature may not have generated sufficiently high levels to initiate the satisfaction process at all. The dissertation is concerned only with this latter issue, determining whether sufficient involvement is a prerequisite to evoking the satisfaction feeling. Later research might address the effects of intermediate levels of involvement. Of the studies for which the level of involvement reached is suspect, there are two types. In some studies, ownership is transferred but the product values are so low and nothing else about
n; the situation is remarkable (see Table 6) that at least some respondents were probably best described as "unemotional" (e.g., Anderson and Hair 1972; LaBarbera and Mazursky 1983). The clearer deficiency, of course, involves cases in which respondents were never led to believe they would own the "target" object (e.g., Olson and Dover 1976; Surprenant and Churchill 1984; Wilton and Tse 1983). In summary, the construct adopted within the current conceptualization is similar to the definition of involvement that was proposed by the conferees at New York University and the ope rationalization of involvement proposed by Houston and Rothschild (1978a, b). That is, response involvement may be aroused by enduring and/or situational factors; and ownership is likely to be a contributing but not sufficient condition for such arousal. Finally, the entire discussion of involvement is tied to Day's (1977) suggestion that satisfaction can only result when involvement was high enough to drive the evaluative process, conf irmation/disconf irmation. Conf irmat ion/Pi sconfirmat ion As depicted in Figure 9, each of the preceding factors contributes directly or indirectly to the conf irmation/disconf irmation process or judgment (for simplicity, this is referred to hereafter as disconf irmation) Although this concept lies at the heart of the disconf irmation paradigm, it has received comparatively little direct attention. In general, some type of valenced disconf irmation (Aronson and Carlsmith 1962) is included in a theory axiomatically and attention is focused on other constructs. However, this construct may represent either a comparison process that results in satisfaction or a judgment
138 that has been made en the basis of such a comparison (Oliver 1980a; Woodruff, Cadotte and Jenkins 1983a). Within the dissertation, disconf irmation is treated in the former sense, as a process. Although a judgment may also be made, it may not always be a conscious or salient result. Thus, the cognitive process is viewed as necessary for evoking the satisfaction feeling, but a specific measurable cognitive result of such a process is not required (more like Anderson's 1973 conceptualization). The argument here is similar to that raised by Zajonc (1984) that the emotion is more accessible than the cognition. At this point in the development of the present conceptualization, however, the details of this process are not proposed. Rather, the concept will be used at its more general level, consistent with most treatments As discussed previously, the disconf irmation paradigm, and therefore some form of the disconf irmation construct has been integral to the development of various satisfaction literatures. The early CS/D papers (Anderson 1973; Cardozo 1965; Olshavsky and Miller 1972), concerned with product evaluation, made a strong case for disconf irmation as an intervening process in quality assessment. These theories were based on perceptual theories that were discussed in a previous section. The use of product evaluation as the dependent variable suggests that the researchers believed that it occurred at the end of the process under investigation. This assumption requires a feedback process, however, inasmuch as disconf irmation cannot occur without some form of perception. These early theories thus required
39 that assimilation-contrast (Sherif and Hovland 1961) or other such processes both depend upon, and influence, perception (Filer 1952). As interest shifted to the more emotional satisfaction result, less was said about disconf irmation influences on perception. Although expectations are still hypothesized to have a potential effect on perception, disconf irmation is thought to result from the combined influence of these concepts. Furthermore, as mentioned in previous sections, the expectations are tempered by the decision that was made (Cialdini et al. 1978) and perception is affected by the experience in which it is gained (e.g., personal experience versus second-hand information) These formulations of the disconf irmation paradigm seem to assume implicitly or explicitly (Woodruff, Cadotte and Jenkins 1983a) that the cognitive process always occurs (Oliver 1980b,c). As discussed in the previous section it is not clear that this is true, especially when nothing has occurred to arouse higher levels of involvement (Day 1977). The difficulty in demonstrating this phenomenon arises from the tendency for measures of disconf irmation to trigger the process, much as attitude measures are believed to prompt attitude formation (Fazio and Zanna 1981). Even if the process does not involve as much "cognitive algebra" as some propose (Oliver 1980b), some motivation is likely to be required for the process to occur. Oliver is among the researchers who treat disconf irmation models at attribute levels. In contrast, Day (1983) and the present conceptualization view expectations, perception and disconf irmation at summary levels. The difficulty with the former viewpoint lies in
140 operationalizing salient and important attributes at all phases of the process, even in light of the "unforeseen" triggers of disconf irmation (Table 6). The problem of determining at what level the process occurs becomes magnified when integrated into the larger satisfaction models. For example, attempts to tie disconf irmation to repurchase intention (Oliver and Linda 1981) do not yield strong results when specific attributes or even summed attribute scales are used. Similarly, theorizing about the result of simple confirmation (Swan and Trawick 1979) depends on the salience and importance of the confirmation judgment. More recently, Oliver and Bearden (1985) report evidence that overall disconf irmation was a mediator between "inferred" disconf irmations and overall satisfaction. The problems in the definition and operationalization of the construct are also reflected in its measurement (Shuv-Ami 1984). Early research simply defined disconf irmation as the difference in value between expectations and perceptions. Later work demonstrated that rating scales (better/worse than expected) are improved methods (Moore and Shuptrine 1984; Prakash and Lounsbury 1983), but subject to the experimental reactivity problem discussed previously. Selection of a measurement technique, then, will depend on the purpose of the investigation and the sensitivity of any theory involved to possible contamination. In general, the recent use of the disconf irmation construct has been to assess its effect on satisfaction (e.g., Swan 1987). In such cases, rating scales may sensitize subjects to the extent of
141 disconf irmation. On the other hand, few investigators have demonstrated concern over initiating satisfaction processes by experimental situations or measurement instruments. Most such investigations have shown that disconf irmation does significantly affect satisfaction (e.g., Kennedy and Thirkell 1987; Moore and Shuptrine 1984). Moreover, this influence is directional; positive disconf irmation prompts satisfaction and negative prompts dissatisfaction, in contrast to the early models that posited dissatisfaction resulting from any disconf irmation (Aronson and Carlsmith 1962; Carlsmith and Aronson 1963; Sampson and Sibley 1965). Few of these models, however, have explained how satisfaction results from a cognitive process. In the discussion of emotions above, it was stated that emotions result from cognitive processes--even if such processes are not conscious. Thus, by defining satisfaction in emotional terms and defining disconf irmation as a cognitive process rather than a cognitive result, the present conceptualization integrates emotional and satisfaction theories in a manner that does not conflict with recent debates as to the precedence of cognitions and affect (Cohen 1981; Fazio, Powell and Herr 1983; Zajonc 1968, 1930). In summary, the conceptualization of satisfaction that is being proposed treats disconf irmation as a necessary result of the presence of the other antecedent conditions that have been discussed. Moreover, some level of satisfaction will result from this comparison processÂ— with no other construct currently hypothesized to intervene or moderate. The disconf irmation process operates at a summary or overall utility level, although it may be triggered by a particular salient
142 attribute. Satisfaction will not result in the absence of the disconf irmation process (nor, therefore, in the absence of any of the other antecedent conditions). Purpose of the Extended Conceptualization As presented, the current "six-antecedent" model may appear to be an unnecessary complication of the more commonly accepted "three-antecedent" formulations. It is useful, at this point, to reiterate and clarify the purpose of the conceptualization. Prior research has sought to understand levels of satisfaction; i.e., what causes a person to experience degrees of the feeling and how subsequent behavior is affected, in turn, by the feeling. The three commonly implicated constructs, expectations, perception and disconf irmation have been relatively consistent significant influences on measured satisfaction, albeit not always accounting for a great deal of variance. A potential cause of this loss of explanatory power has been uncertain definitions and operationalizations of the key dependent construct. That is, the situations contrived by researchers have not been demonstrated to replicate the conditions normally encountered in the marketplace. If interest in satisfaction arises from the desire to understand a market phenomenon, this would seem to be a serious flaw. The problem is exacerbated by the inability of current measurement techniques to assess the ability of an experimental situation to evoke satisfaction. Thus far, the dissertation has shown how the satisfaction construct relates to other purchase and post-purchase phenomena (e.g.,
'4 3 attitude, complaining) and has based a definition on this theoretical network. It has also proposed that the pattern of emotions that describes the dissatisfaction Â— satisfaction continuum can be measured with sufficient sensitivity to reveal the absence of the feeling. The model of the satisfaction process presented herein (Figure 9) takes an additional step. In order that an emotional response that can be called satisfaction will result, a situation must encompass all of the antecedent conditions presented in the model. This conceptualization, then, provides a template with which one may identify, describe, or create a satisfaction-evoking situation. The conceptualization might at first appear to comprise a rather restrictive set of criteria. It is important, however, to note that the antecedent conditions are not difficult to identify in the marketplace. In fact, the three concepts added by the dissertation, are the net result of a comparison of "typical" marketing conditions that result in satisfaction with conditions generated ostensibly to study the construct 2 For example, it has been argued that the decision construct can be represented by the purchase act. Satisfaction surveys and reports of complaining generally deal with items purchased by the respondent/complainer (or a decision making unit which included the respondent). These surveys appear to assume that a purchase encompasses the characteristics of volition, selection and irrevocability (in that the transaction has been consummated and 12 Again, the conceptualization generalizes to all satisfaction domains. The example of consumer behavior in the marketplace is used for consistency and clarity.
144 replacement with a different product entails some effort and "complaining") that were deemed necessary to constitute decision (e.g., Andreasen and Best 1977). Similarly, marketers interested in assessing levels of satisfaction with their products direct their surveys to those who have had personal experience with the item in question. A mail survey by a major domestic automobile manufacturer carries this instruction in bold print, "Please have the person who drives this car the most fill out this questionnaire". It is also common practice in the conduct of multi-product surveys (i.e., conducted in attempt to compare satisfaction levels across product classes) to not collect satisfaction ratings from respondents who report that they rarely or never use items in a particular category (e.g., Day and Landon 1976). Comparison studies such as these often identify more expensive products as the ones most responsible for consumer satisfaction and complaining (e.g., Shuptrine and Wenglorz 1930, 1981; TARP 1979). Although the increase in complaining activity may be partially attributable to the higher return expected from complaining about more costly items, this explanation does not hold for increased mention in satisfaction surveys. Quite simply, satisfaction responses to "high involvement" products (expensive, owned by the respondent) are evidently more salient and/or memorable. Other involvement arousing factors (total failure, physical danger) are also frequently stated in solicited and unsolicited accounts of dissatisfaction (e.g., Best 1981), although the activation level soon tapers off (Richins and Bloch 1986).
145 The discussion in this section is not meant to provide conclusive evidence of the usefulness of the expanded framework. Rather, it suggests that the specified necessary antecedent conditions for satisfaction to result are not particularly stringent, given experiences from non-experimental settings. 13 Adequacy of Prior Studies Given that the present conceptualization has been proposed as a template for the design of satisfaction research, it should also be useful in evaluating prior research efforts. This section of the chapter examines individual papers from the CS/D literature in light of the extended model and discusses the necessity of considering the entire situation in the design of satisfaction studies. Potential improvements to the studies that might have accrued from attention to the conceptualization are noted where appropriate. Table 7 summarizes this analysis (the structure is based on that used bv Wilkie and Pessemier 1973). The articles reviewed were chosen on several criteria: (1) articles considered to be "classic" in the CS/D literature (including those originally intended only to examine product perception), (2) articles that were noticeably deficient in one or more of the antecedent conditions, and (3) satisfaction articles that have appeared in a major journal ( Journal of Marketing ; Journal of Marketing Research; Journal of Consumer Research) or conference 13 The complete model, rather than just the individual concepts outlined above, is supported by as many as 129 out of 138 descriptions of "satisfying" or "dissatisfying" occasions collected in pilot studies (similar to the critical incidents described by Maddox 1981 and Swan and Combs 1 976)
146 Table 7 Prior CS/D Research and the Extended Conceptualization
147 Moore Richins 4 Robertson 4 WestShuptrine et al. Bloch Oliver brook 1984 1985 1986 1987 1987 s
148 Table 7--continued b) Experimental designs Citations Cohen & Olshavsky Lambert LaTour & Cardozo Goldberg & Miller Anderson Winter et al. Oliver Peat 1 965 1970 1972 1973 1974 1977 1977 1980 Satisfaction Concept
i.A.9 LaBarbera Mowen Wilton Surprenant Jordan Fisk Hoch Bolfing 4 4 & & 4 & & Cadotte 4 Mazursky GroveTse Churchill Leigh Folkes Young Ha et al. Woodruff Carsky 1983 1983 1983 1984 1984 1984b 1985 1986 1987 1987 1987' I
150 Notes: a: P0 = purchase occasion, U0 = use occasion, I = item, C = class, S = store or vendor, b: A = acute, E = enduring, c: C = cognitive definition, A = affective definition, none = no explicit concept, d: V = verbal, G = graphic, C = cognitive, E = emotional, 8= behavioral, = reliability test reported, e: P = present, A = absent, ? = cannot determine, M = manipulated, S = scaled as intermediate variable, C = causal relationship with satisfaction. f: D = dependent, S = scaled as intermediate variable, g: C = complaints/compliments, W = word-of-mouth, L = loyalty/ exit A = attitude, R = resultant from satisfaction, h: S = students, C = convenience, not strictly student, S = representative (random) sample attempted. i: Y = yes, study design conforms to criteria of expanded conceptualization, N = no, study design does not conform, j: S = support for conceptualization implied by results, ? = results ambiguous with respect to conceptualization.
151 proceedings (Advances in Consumer Research and CS/D&CB conference papers are examined only as selected by the other criteria) The table is divided into two parts to facilitate presentation and to reflect slightly different research goals and traditions. Reports of experimental investigations are summarized in part "a" of the table; surveys are listed in part "b". The discussion will focus on the experiments since correction of those flaws is a major purpose of the dissertation. The surveys are of interest because they reflect the constructs in realistic settings. In general, the surveys are not sensitive to conceptual differences between satisfaction and attitude, which opens some avenues to improvement via the extended conceptualization. The same issues are summarized for each set of studies. The rows of the table identify attention to, incorporation of, or a position taken on particular conceptual or methdological points. "Type", for example, refers to the satisfaction types suggested in Table 2. Satisfaction emotions can be aroused by targets at various levels of abstraction. Although the process is thought to be essentially the same, specific precursors and strength of emotions are likely to vary across the hierarchy. The next row, "Time", indicates the interval addressed by the investigation. The dissertation defines satisfaction to be emotional in nature; and quick to decay. This is more accurately reflected in the concept of acute rather than enduring satisfaction. The latter type, however, seems to have been of more interest to policymakers and other practitioners interested in long-run market reactions.
152 "Definition" attempts to identify the conceptual treatment of satisfaction. In some cases where no explicit definition was presented in the paper, it was possible to characterize the position taken based on either the context of the article or prior reports by the authorfs) (e.g., Oliver 1980a, 1987). The "Measures" used to assess satisfaction are characterized based on the categories in Figure 4. If a reliability test is reported, this is noted. Most crucial to the contribution of the dissertation is the examination of the critical "Conditions." In some cases, it is evident from the description of the methodology that a particular element of the process was (or was not) within the respondent's experience. Many of the constructs were explicitly manipulated or scaled as part of the investigation. In many instances, it is necessary to infer from the description of an experiment or from the qualifications of a survey sample whether or not the construct was present. If some, but not all, subjects experienced the enabling condition, it is generally identified as present. Finally, some studies sought causal relationships among variables. If evidence was found for a causal relationship with satisfaction, this is noted. Not all of the papers summarized in Table 7 sought to investigate satisfactionÂ—some never invoke the construct. Among these are papers that dealt with dissonance (Cohen and Goldberg 1970), involvement (Richins and Bloch 1986) and attitude (e.g., Winter 1974). Others scale satisfaction as an intermediate variable between pre-purchase and post-purchase constructs or outcomes (e.g., complaining, word-ofmouth) If a test of causality showed that a state or behavior
53 resulted from satisfaction, this is indicated. Inclusion criteria for such studies were broadly defined since it was felt that the proposed conceptualization makes recommendations valid for a variety of postpurchase investigations. Table 7 notes other methodological points that are often of interest in summarizing a body of literature, the subject pool, and target stimuli used. Student and other convenience samples are identified discretely from sampling plans that attempted some degree of representativeness--there were few truly random samples. For one study (Oliver 1987), a student sample was particularly appropriate for the issue under discussion and was categorized as representative. The "Target" identified is the stimulus used or manipulated. It is not always clear that subjects responded at this level, some seem to be willing to attribute item problems to store causes, or the reverse, and complicate measurement. The interpretation portion of Table 7 attempts to summarize the implications of each study in light of the proposed conceptualization and to evaluate the conceptualization against the empirical evidence in the study. The summary entries represent post hoc evaluations that allow considerable margin for bias. Thus, while none of the studies was judged to discount the conceptualization, it may be due to a broad concept of support. This notion of support, however, is secondary to the purpose of the table, which is to show how the research may have been better by attending to necessary factors in the designs.
154 Conclusions from the literature summary Table 7 is meant to be neither an exhaustive nor representative random list of CS/D studies. Nevertheless, it is instructive to look at patterns which may be found across the rows of the table. In particular, some of these patterns illustrate differences between experimental and survey research. Type For the most part, both series of studies have concentrated on item-specific satisfaction. This is not surprising and consistent with much of research in marketing and consumer behavior. Following attitude research tradition, CS/D researchers have found that individual items have more manageable attribute profiles than stores, product classes, and other more general consumer targets. On the other hand, reactions to experiences as specific as purchase or use occasions may be too varied to be predictable or interpretable That is, consumers may view one instance of unusual performance as a rare event" rather than as a disconf irming experience. Time The amount of evidence, and impact of disconf irmation, will vary across the time dimension. It is evident from Table 7 that survey and experimental designs tap different portions of the time continuum. Most experiments are conducted in a single session, allowing no time for satisfaction to erode. Surveys, on the other hand, usually measure response after relatively longer usage--owing to the difficulty of identifying and locating a sufficient number of respondents in an acute satisfaction stage. There are some exceptions in each case. Some of the experiments have utilized more longitudinal designs and have measured more delayed responses. Surveys have used critical incident
155 techniques to assess response at an earlier, more acute, time or have argued that very recent purchasers of highly involving products still experience stronger emotional response (e.g., Richins and Bloch 1986). This variation of time levels makes comparison of experiments and surveys untenable. It has been argued previously that satisfaction soon decays into attitude, which may be the construct measured by most of the surveys. If this is the case, then the findings of relatively low rates of satisfaction or dissatisfaction related behaviors in the marketplace are more plausible. Marketing practitioners should not expect complaining or word-of-mouth rates to correlate with levels of post-purchase attitude. Definition The conceptual distinction between satisfaction and attitude is best communicated by well-articulated definitions. This seldom has been the case in the studies reported. Many of the studies fail to provide an explicit definition of satisfaction, due either to a presumption that the concept is clear from prior research or to sheer neglect. Table 7 reports the types of definition provided (or, implicit in the treatment of the construct) or an indication that none existed. For those studies that did not purport to measure satisfaction, it is reasonable that the construct was not defined. What is more alarming is the frequency of investigations, often experiments, that neither define the construct nor take an identifiable position on its nature. In some cases, a report of the measurement scale(s) used for satisfaction suggest the philosophy adopted, but too often there are inconsistencies in the set of measures adopted or between measures and an espoused concept.
156 Measures Virtually all of the studies reported have gathered data using written questionnaires. The production mechanics involved have led to a dominance of verbal scales (i.e., it is much simpler to generate polished looking text than graphics). Within these, there has been a tendency to use scales that were more cognitive than emotional in nature. The experiments reported have been prone to this more so than the surveys. This measurement pattern across the literature creates an interesting discrepancy. The survey designs, which logically should be measuring something akin to attitude (enduring satisfaction), use more of the affective measures that should be sensitive to the emotions aroused by the particular item or incident (acute satisfaction). Thus some weaknesses in results may be reflecting these measurement errors (that can be cast in construct validity terms). Conditions Discussing and measuring satisfaction in the context of a given investigation requires confidence that the construct exists. The dissertation's claim that this has not always been so is reflected in this portion of the table. Even allowing for broad presumptions of presence, not all of the enabling conditions fare well. Since the earliest days of the disconf irmation paradigm, the concepts of expectations, performance perception, and confirmation or disconf irmation have dominated post-purchase research. Even when neither manipulated nor measured explicitly, each construct seems to be an implicit part of nearly all of the studies. For the sake of clarity, however, it would have been better to explain explicitly one's underlying model and the role of each construct in it. This is
157 particularly true in the case of the survey designs which often seem to result from an assumption that the respondent has performed a full decision, purchase, and post-purchase task. Surveys of satisfaction, unaccompanied by measures of expectations or absolute performance perception, tell an incomplete story about what underlies satisfaction levels. More critical are the patterns described for the new constructs. For those, the experimental design portion of Table 7 is the more illuminating. A majority of the experimental studies showed no evidence of the subjects' being forced--or even allowed--to make a decision. Thus, the expectations were never acted upon in anything resembling a purchase task. The absence of decision may have occurred either in a setting that included personal experience or in situations in which performance perception was available only as a description. It is unfortunate that these role-playing studies have come to be so common in CS/D research. Again, the problem is much more severe in the experimental rather than survey designs. The trade-off being made is, apparently, an attempt to generate involvement or enhance other manipulations at the cost of realism. The argument is that role-playing enables the use of expensive or difficult to manipulate products (Fisk and Young 1985, Surprenant and Churchill 1984), controls for past events, and prevents bias caused by deception (e.g., Fisk and Young 1985). While the concern over bias is admirable, the potential for bias seems much higher when both manipulations and measures are obtrusive (Sechrest 1979).
158 The desire to use expensive products as experimental stimuli often arises from a desire to create a more involving experimental situation. This would seem desirable given the conceptualization under discussion. Richins and Bloch (1986), among others, argue that this effort is misguided if the appropriate interaction of situation and enduring involvements is not taken into account. Indeed, the present chapter has argued that it is the personal relevance or "stake in the outcome" of the decision that is the important motivating force in the involvement mechanism. Thus, any benefit gained by using an expensive or interesting product is lost in the abstraction of role-playing. The experiments in Table 7 represent both entire role-playing situations Â— in which the manipulations are wholly hypothetical (e.g., Carsky 1987) Â— or partial role-playing Â— in which subjects are asked to imagine themselves to be buyers of a product that is physically present (e.g., Wilton and Tse 1983). In neither case, however, does the situation have any implications for the subject once the experimental session has concluded. These are noted as absent, or at best questionable, response involvement conditions. Satisfaction role and outcomes Because of the selection criteria used, it is not surprising that some form of satisfaction was a dependent variable in many of the studies summarized. Inclusion of satisfaction as a scaled intermediary between pre-purchase states and post-purchase behaviors was also common among both surveys and experiments. This provides some evidence in support of the grander models of post-purchase behavior.
159 The selection of behavior is not the same as would have been recommended by the present conceptualization. Both surveys and experiments emphasized "loyalty" behaviors--purchase of same or similar item Â— as outcomes of interest. While these are likely to be important from a managerial standpoint, it has been argued that such a response is more appropriate to attitude Â— following the Theory of Reasoned Action Â— than satisfaction processes. In the experiments attitude was often measured as a resultant construct, but with little causality testing. Such testing was most common in the broader based surveys which utilized regression techniques and a wider selection of dependent measures Methodology notes Across the studies, use of student subjects is widespread. Although this is desirable in one instance (Oliver 1987), as noted previously, the general pattern is bothersome. The combination of role-playing methodologies, lack of personal relevance and the use of student subjects has probably yielded a number of artifactual results. The products used as stimuli, while seldom inappropriate for a student sample, have not been targeted specifically to such limited samples. The variety of targets used has uncovered difficulties in making manipulations salient or generating sufficient involvement Interpretation Rather than attempt to globalize the interpretations at the bottom of the table, some comments can be made regarding selected papers as viewed from the vantage point of the extended conceptualization. The earliest papers generally thought to be satisfaction experiments (Anderson 1973, Cardozo 1965, Olshavsky and
160 Miller 1972) dealt more with product perception and consistently lacked any opportunity for the subject to make a decision regarding the product. This omission may help to explain why the researchers failed to find significant contrast effects: the expectations weren't "fixed" by an irrevocable commitment. Cardozo (1965) alluded to a contrast effect, but this study dealt more with equity of the experimental (shopping?) situation than with actual product involvement. That is, the expectation that was disconfirmed was the type and value of the gift received rather than its functionality. In general, these studies report findings consistent with simple perceptual or attitudinal processes On the other hand, an early investigation that did follow the conceptualization well (Cohen and Goldberg 1970) is often slighted in the satisfaction literature. While these authors did not claim to be engaged in satisfaction research, the conditions they set forth to generate dissonance plus post-experience reevaluation fit the criteria displayed in Table 7. Although none of their measures was adequate for satisfaction, the loyalty and attitude measures behaved as expected. Thus, while the study does not contribute specifically to knowledge about the satisfaction construct, it is a useful model for the design of the satisfaction investigations. Similarly, the study reported by Winter (1974) fits the criteria but was positioned as a study of attitudes. The arguments offered to distinguish a trial purchase from an adoption purchase offer weight to the necessity of the decision construct. Winter reports that "the purchase act, even for a low-involvement product such as scouring pads,
161 may be instrumental in altering consumer cognitions in the postpurchase period before actual consumption of the product takes place (1974, p. 170). The satisfaction process extends this reevaluation beyond the dissonance stage. While the importance of the decision construct has been highlighted, the lack of satisfaction measures hinders a more general endorsement of the conceptualization. The Lambert et al. (1977) paper is notable for its attempt to evaluate an industrial purchase process. While laudable for this goal, the lack of actual decision and the role playing structure of the situation hinder any interpretation of the findings. This lack of decision also reduces confidence in the findings of other papers that are discussed in more detail elsewhere (LaTour and Peat 1980, Oliver 1977). An investigation of "product trial" by Scott and Yalch (1980) is another study that could have been cast as a satisfaction study, but was not. As with the Cohen and Goldberg effort, this experiment fulfilled the criteria but included no measures of satisfaction nor discussion addressing directly the construct. Personal experience was found to affect evaluation of the target product as subjects gathered progressively more information about the product. A set of role playing experiments began the 1 980 s (Mowen and Grove 1983, Surprenant and Churchill 1984, Wilton and Tse 1983). The findings are consistent with cognitive operationalizations of satisfaction but can also be the result of experimental demands. Less obtrusive or emotional measures may have provided some divergence from intuition. Wilton and Tse (1983) reported some divergence between
162 attitude and satisfaction measures, colored somewhat by a surprisingly dispassionate assessment of a high-involving product. The Jordan and Leigh (1984) and Folkes (1984b) papers illustrate the dangers inherent in casual adoption of a paradigm. The former does not even match the "old" paradigm in that no disconf irmation is suggested. The authors merely assess post-purchase attitudes among children. Folkes, on the other hand, does not invoke the satisfaction construct or a satisfaction model but the examination of postconsumption attributions and behaviors certainly entails such emotions. Fisk and Young (1985) argue that role-playing was necessary for the selected product class (airline service), but never present a case for the appropriateness of the product class. It is impossible to infer the effects of the lack of decision, experience, and involvement on the criterion variables because no means were reported. The paper also did not discuss explicitly satisfaction results. In general, the paper reflects a casual interpretation of and approach to the disconf irmation paradigm. Among the most recent experiments reported, there is a distressing tendency to maintain the role-playing approach (Bolfing and Woodruff 1987, Carsky 1987, Hoch and Ha 1986). The concepts and definitions of satisfaction remain cognitive, although there is more emotion or affect reflected in some of the measures (e.g., How do you feel about ?). Nevertheless, the ability of such situations to arouse true emotions remains suspect. Bolfing and Woodruff (1987) claim to increase involvement, but in the absence of any personal relevance.
163 Cadotte, Woodruff, and Jenkins (1987) executed a design that fit the criteria by including an actual purchase in a field experiment. Their interest in satisfaction with the store (in this case, a restaurant) makes the results somewhat more difficult to compare with product satisfaction studies, but the old and extended disconf irmation models are generally supported. In addition, the study extends the concept of comparison standard beyond what had been considered simple expectations (although consistent with the dissertation's viewpoint). The implications of the survey reports are easier to summarize; most of these addressed actual purchase, consumption, post-purchase phenomena. Thus, it is possible to inferÂ— if not fully demonstrate Â— the presence of the enabling constructs in most survey accounts. Two studies sought to provide a new, Two-factor theory, approach to satisfaction (Maddox 1981, Swan and Combs 1976). The critical incident technique used in both of these studies is likely to result in descriptions of consumer experiences which fulfill the criteria, but those can only be interpreted on a case specific basis. Had the studies used satisfaction measures--particularly emotional ones--rather than simply classification as satisfying or dissatisfying experiences, the conceptualization may have garnered more support. The Oliver (1980a) survey does provide more emotional measures and attempts to distinguish satisfaction from attitude. The path analysis reflects a separation between the constructs such that satisfaction is found to be a significant cause of post-decision attitude. Although the subjects involved did undergo processes defined by the extended conceptualization, Oliver seems to force the satisfaction construct
164 upon them. That is, non-satisfaction is not allowed. This criticism is not unique in this study Â— few, if any, surveys disqualify respondents for absence of satisfaction and/or dissatisfaction. This is reflected in somewhat muddled treatments of confirmed expectations and their effects on consumers. Similar weaknesses make comparisons across studies difficult. The Anderson, Engledow and Becker (1979) study was more concerned with demonstrating LISREL than providing substantive knowledge in CS/D. Other recent surveys provide little control over time elapsed since the disconf irming experience and may be reporting attitude results (Bearden and Teel 1983, LaBarbera and Mazursky 1983, Moore and Shuptrine 1984). Westbrook's studies (1983, 1987) provide a decidedly emotional approach to satisfaction, but also assume that every consumer can be placed along the continuum at any time. It may better serve his needs to focus respondents' attention on critical incidents and measure emotional response to those. This would provide firmer evidence that a specific disconf irmation is involved and the target of the response. One could argue that the focus on a particular item (i.e., automobile) is sufficient; but the confirming versus disconf irming experiences maybe ambiguous. On a final note, the Richins and Bloch investigation (1986) was not a satisfaction study, nor did they collect satisfaction measures. It is included as a survey that explicitly mentions each of the enabling constructs and specifically investigates one of them, i.e., involvement. Of particular interest to CS/D and other consumer behavior researchers is the authors' caution regarding the difficulty
165 of manipulating situation involvement in the laboratory. One conclusion that might be drawn from their study is that involvement is inherently an individual characteristic and is better viewed as a blocking variable in a quasi-experiment rather than a manipulable factor. General conclusions Table 7 has a relatively simple overall story to tell. The main point is that the CS/D literature can be made more coherent by specification of criteria for satisfaction studies. Much effort is wasted on attempting to integrate results from inherently discrepant studies and paradigms. This prescription is particularly appropriate to experimental investigations. As discussed, surveys are not without fault. Although actual purchase and consumption are usually involved, the elapsed time suggests a confounding of satisfaction and attitude. More distinct concepts, measures and methodologies are needed if satisfaction is to be treated as a unique construct. The issues in Table 7 provide points of comparison among studies and research traditions. Perhaps a taxonomy can evolve that would enable better mapping from the laboratory experiment to the field (market) survey. Summary and Conclusions This chapter has integrated the literature in CS/D (reviewed in Chapter 2) with related literatures. The satisfaction construct was redefined as emotional in nature and generalizable to domains other than consumer behavior. A related concept, attitude was shown to be discriminable in its nature and effects on behavior. A general model
166 of the situation that results in the pattern of emotions defined to constitute satisfaction was presented and shown to be consistent with common consumer situations. The extant CS/D literature was then evaluated with reference to the model and it was demonstrated that a lack of adherence to the conceptualization may account for many of the weak and/or inconsistent results that have been reported. In the next chapter, the theory embodied in the model (Figure 9) is extended to specific hypotheses concerning the impact on satisfaction studies of omitting necessary conditions. A methodology for testing these hypotheses is then presented in Chapter 5.
CHAPTER 4 RESEARCH HYPOTHESES Introduction Preceding chapters have reviewed the evidence currently available regarding the antecedents and consequences of satisfaction. This review was augmented with discussions of literatures related to emotions and other operationalizations of satisfaction. The review culminated in the definition of satisfaction as a pattern of emotions which is evoked by a process slightly more complex than that assumed by the usual conceptualization. Three additional concepts were proposed as necessary for the arousal of satisfaction. Demonstration of the necessity of two of these concepts comprises the empirical portion of the dissertation. In this chapter, research hypotheses are derived from the model and theory that have been discussed. These general hypotheses are then interpreted in the context of a specific research methodology in Chapter 5. In the interest of clarity, the discussion will continue to focus on satisfaction within the consumer domain. Main Hypotheses Hypothesis 1a : When all necessary conditions are fulfilled (i.e., expectations, decision, experience, perception, high response involvement and disconf irmation are all present), a positive discrepancy (i.e. perception surpasses expectations) results in the pattern of emotions that constitutes satisfaction 167
168 Hypothesis 1b : When all necessary conditions are fulfilled, a negative discrepancy (i.e. perception falls short of expectations) results in the pattern of emotions that constitutes dissatisfaction (negative satisfaction) These hypotheses re-state the basic premise of the prevailing disconf irmation paradigm in terms that reflect the modified conceptualization being presented. In addition to the emotional measures, more standard measures of satisfaction (Figure 4) should also demonstrate standard results. When one or more of the necessary conditions is unfulfilled, however, there will be a divergence among the measures. As discussed in the previous chapter, the act of making a decision serves to solidify expectations and to make the object of the decision more personally relevant. Following the logic of hypotheses 1a and 1b, a lack of adequate emotional response in conditions in which decision was absent comprises support for a critical test of the necessity of decision in the satisfaction process. This test is founded on the distinction between satisfaction as a pattern of emotions and satisfaction as the response to a purported satisfaction scale. Chapter 3 described the motivational nature of emotions and argued that one source of the motivation necessary for experiencing the satisfaction emotion is response involvement. A logic similar to that described above suggests that it should be feasible to prevent satisfaction responses by creating a situation that entails low response involvement. At the present level of knowledge, prediction of how much response involvement is sufficient necessarily must be tautological (i.e., a sufficient level is associated with an emotional
169 response whereas an insufficient level does not evoke emotion). However, once again it is emphasized that standard satisfaction scales are not sensitive to the lack of emotion. Hypothesis 2 : Commonly used measures of satisfaction do not discriminate the absence of satisfaction or dissatisfaction in the presence of a perception-expectation discrepancy. That is, the usual measurement instruments are insensitive to the non-fulfillment of such necessary conditions as decision and sufficient response involvement. Hypothesis 3 : Measures of attitude do not discriminate the absence of satisfaction or dissatisfaction as results from the omission of a decision or adequate response involvement. The hypotheses in this group are predicated on the major argument presented in the dissertation: the use of standard satisfaction measures can yield responses that are better described as measurement artifacts than as the result of the consumption situation. In effect, omission of one or more of the necessary conditions allows for a detached, cognitive evaluation of the target, but will not arouse the emotions that constitute the satisfaction response. Attitude measures and the satisfaction scales from prior studies will be responsive only to the perceptual process that is driven by performance of the target object. These do not indicate that satisfaction responses will result (see Hypothesis 4, below). Secondary Hypotheses Hypothesis 4a : Satisfaction, as defined and identified by emotional response, is associated with a greater incidence of those behaviors (e.g., positive word-of-mouth, complimenting) usually thought to follow from the feeling. Hypothesis 4b : Dissatisfaction, as defined and identified by emotional response, is associated with a greater incidence of those behaviors (e.g., negative word-of-mouth, complaining) usually thought to follow from the feeling.
170 Hypothesis 4c : Attitude toward the brand is associated with a corresponding intent to repurchase. As discussed at length in Chapter 2, the original goal of the CS/D literature was to understand specific behaviors thought to result from satisfaction (Cardozo 1965). Chapter 3 presented a conceptualization that accounts for the weak relationships found frequently in empirical investigations. That is, although attitude will influence repurchase intention, it is not sufficiently motivating to initiate the other behaviors of interest. Hypothesis 5 : The likelihood of engaging in satisfaction-related behaviors increases with the strength of the satisfaction emotions, whether positive or negative. This hypothesis operationalizes the argument depicted in Figure 6, that the different behaviors require different levels of motivation. Chapter 2 pointed out that these motivational levels may be dependent on the strength of the satisfaction feeling. Previous attempts to establish the correlation between level of satisfaction and behavioral response have been handicapped by the lack of a sufficiently sensitive satisfaction measure (Richins 1983b). The present hypothesis assumes an adequately sensitive measure and that perceptions regarding the relative cost and benefit and the appropriate mechanism for completing each response are constant across subjects (Day 1984). Supplemental Hypotheses Hypothesis 6 : Undertaking a decision process directly influences perception of the outcome of the decision. When a decision has been made, variance from expectations is exaggerated. This hypothesis summarizes the basic contention of the dissonance literature that was adopted by the early consumer satisfaction
171 researchers. That is, Anderson, Cardozo and others were interested in the effects that purchase and consumption would have on one's ability to veridically perceive product performance. In this case, it is expected that any misperception that results would be consistent with assimilation-contrast theory (Sherif and Hovland 1961). In Chapter 3, it was argued that the personal risk aroused by decision is the most important cause of any shift in perception. Hypothesis 7 : Perception of a product (performance) has a direct influence on attitude toward that product. The CS/D literature has concentrated rather strongly on the relationships among perception, disconf irmation and satisfaction. Attitude, on the other hand, is viewed as affected by satisfaction alone (Figure 3). The current conceptualization reflects a different perspective; attitude can be influenced by either perception or feelings of satisfaction or both (Figure 8). This reflects Oliver's findings that suggest postusage satisfaction is influenced by the disconf irmation judgment which was based on perception gained from personal experience (1980a). Thus, the present model is more responsive to criticisms that objectively "good" or "poor" performance should not be totally counter-balanced by prior expectations. Hypothesis 8 : As response involvement increases, strength of emotional response increases. Again, there is a ceteris paribus condition implicit in the hypothesis. Hypotheses 5 and 8 provide a test of the mini-theory of "Conservation of Motivation" discussed earlier. The role of motivation in emotion research was elaborated in Chapter 3 and shown to be a critical distinguishing feature between satisfaction and attitude. The
172 present conceptualization explicitly links origins (response involvement) and outcomes (behavior) of motivation. Summary The dissertation presents an attempt to integrate a number of theories relevant to satisfaction into a coherent conceptualization. The theoretical network that is generated can not be evaluated in toto, but must be tested in parts. The hypotheses presented in this chapter constitute the initial phase of such testing. The main hypotheses are identified as such because the primary orientation of the dissertation is the definition of the construct and the presentation of a prototypical satisfaction situation. Thus, evaluation of the necessity of newly introduced constructs (in the present study, decision and response involvement) in the context of the definition adopted is imperative. The secondary hypotheses concern other aspects of the conceptualization which can be examined as part of the research that is envisaged, but which were not central to the initial design. As such, evidence gathered pertaining to these hypotheses should be interpreted as suggestive, in lieu of that which could be collected in studies specifically designed to investigate these issues. Finally, the supplemental hypotheses were drawn from arguments that have been raised in the CS/D and related literatures. The hypotheses, as stated, are consistent with both the present conceptualization and the prevailing wisdom of the CS/D literature. Contrary findings, however, would not be interpretable as damaging to the central model. As with the secondary hypotheses, the supplemental
173 hypotheses represent interesting, albeit tangentially important, portions of the overall theory. There are quite a number of issues raised in the literature review and conceptualization chapters that are not addressed by the empirical portion of the dissertation and are therefore not represented among the hypotheses. The investigation of what is essentially a network of theoretical constructs is necessarily an iterative process. A refining of theories and measures is implicit in any research stream, of which a study such as this is only the beginning. A plan for examining some of these issues is presented in more detail in Chapter 7.
CHAPTER 5 EXPERIMENTAL METHOD Introduction The concept of satisfaction has been defined as a general pattern of emotions that may arise in any of several domains. A general framework that relates the construct to its antecedents, outcomes and associated concepts has been presented in some detail. In the present chapter, the definition and conceptualization are operationalized and evaluated in light of specific experimental hypotheses derived from more general hypotheses presented in Chapter 4. The present chapter begins with a discussion of the development of the measures that will be used in the main study of the dissertation. The emotional measures that are derived constitute an operational definition of the construct that was tentatively defined in Chapter 3 (see also Figure 7b). The behavioral intention scales, as derived, are also considered to be measures of satisfaction, albeit indirect since they actually assess theoretical outcomes (Figure 9). Finally, the complete experimental methodology is presented with special emphasis on the manipulation of the proposed antecedent conditions. Operational versions of the hypotheses presented in Chapter 4 conclude the present chapter Pilot Study--Measure Development Westbrook's efforts to describe the patterns of emotions that arise during ownership (1983; Westbrook and Oliver 1984) have been 174
175 discussed previously and led to the proposed identification of the endpoints of the satisfaction continuum in terms of specific emotions. These studies were not intended for, nor were the measures taken adequate for, the derivation of such definitions. Therefore, a pilot studv was conducted to validate or revise the definitions that have been proposed. The studies reported by Westbrook provide no criteria by which to identify the emotional profiles that best describe satisfaction. The pilot study corrects for this in two ways. (1) The critical incident technique that has been introduced to the CS/D literature (Maddox 1981; Swan and Combs 1976) is used to concentrate responses at the ends (negative and positive) of the satisfaction continuum. (2) Measures of intent to engage in specific behaviors thought to result from satisfaction (as discussed in Chapter 2) are collected in addition to the more typical satisfaction scales used by Westbrook. A concise framework for the construction of psychometric scales has been presented in the marketing literature and can be adapted for use in the present study (Lundstrom and Lament 1976). The principles and philosophy implicit in the scale derivation are consistent with those of the author and those presented in other recent discussions of measure development (Churchill 1979; Nunnally 1978). The Lundstrom and Lamont framework presents five major considerations for the construction of a reliable and valid scale: (I) definition of the construct and selection of an item pool; (2) selection of a scaling procedure; (3) selection of scale items; (4) evaluation of the
176 reliability of the scale; and (5) evaluation of the validity of the scale. The pilot study examines two methods of measuring satisfaction, defined previously as "an emotional response conceptually distinct from behavioral responses" (Day 1933). One set of measures taps directly the emotional nature of the construct. The second set of measures discussed examines intent to behave in a particular manner. Although this latter set of scales, by definition, can measure the construct only indirectly, the model presented previouslv indicated the importance of behavioral response to the overall conceptualization. Obj ectives The principal goals of this study were to identify those emotional measures and those measures of behavioral intent that best discriminate between subjects responding to satisfying and dissatisfying occasions, respectively. Satisfaction scales and criteria for evaluation can then be constructed from these measures. M ethod Ninety-four students in an introductory Marketing class served as volunteer subjects for extra class credit. The study was conducted in various sized groups in a classroom setting and required about thirty minutes per session. Subjects were given a blank sheet of paper on which to first identify a recent consumer experience that was particularly dissatisfying and then, after all had completed this, identify a recent consumer experience that was particularly satisfying. (This order was
177 chosen because it was felt that a satisfying occasion would be more readily recalled if "primed" by an already salient dissatisfying occasion.) After sufficient time for all subjects to identify an occasion of each type, they were instructed to turn to and complete the first page of the questionnaire (Appendix A). On this page, they were to provide more details concerning either the satisfying (N=46) or dissatisfying (N=48) occasion. 1 In order to avoid hastily completed and unnecessarily brief descriptions of the specific occasions, subjects were instructed not to continue through the questionnaire until sufficient time had elapsed for a complete narrative (a pre-test indicated that 10 minutes was an adequate period). The remainder of the questionnaire was self-paced and consisted of emotion scales, satisfaction measures and behavioral scales (the entire questionnaire is contained in Appendix A). The narratives were required in order that subjects would revisit the emotional experience (Robinson 1976, 1980; Weiner, Russell and Lerman 1979). Measurement scales Westbrook employed an abbreviated version of Izard's Differential Emotions Scale, "owing to the length of the original scale (69 items)" (1983, p. 5). Yet, he chose not to use the abbreviated version that had already been supplied by Izard (1977). The shortened forms used by each of these authors have been discussed and are presented in Table 3. 1 In the course of analysis, it was determined that four subjects in the satisfying condition actually described, and completed the remainder of the study responding to, the dissatisfying occasion. This problem and isolated incidences of item non-response created some variations in sample size, as reported below.
Due to a lack of rational criteria by which to select between these and, in the face of Izard's estimate of 5 minutes as the time requirement for a 67-item edition of the scale (1972, p. 86), the full version was administered. In addition, adjectives were extracted from satisfaction scales to round out a 75-item list (added were: Dissatisfied, Satisfied, Pleased, Contented, Thrilled, Depressed, Wise, Foolish) A single randomized order was reproduced on all questionnaires, but individual subjects were assigned separate starting points and instructed to complete the entire list. Four satisfaction scales were selected from the assortment presented in Figure 4. The scales were chosen on the basis of reliability in prior testing (Oliver and Westbrook 1982; Westbrook and Oliver 1981) and production feasibility (i.e., verbal rather than graphic scales) The questionnaire concluded with a series of Likert type scales. A form of Oliver's (1980a) six-item satisfaction scale began the list and was followed by a series of behavioral intent scales (phrased as behavioral likelihood "I would ..." "Strongly Agree--Strongly Disagree"). This latter set was derived from a variety of sources including satisfaction surveys, guides for dissatisfied consumers and personal experience. The behavioral intent measures were intended to be exploratory in nature. Analysis The first of Lundstrom and Lamont s (1976) considerations has been satisfied; definition of the construct and identification of the item pool(s). The selection of Izard's DES as the basis for emotional
179 measures identified the scaling procedure as well as the item pool. Thus, a Likert-like scale will be used to measure emotions. For reasons similar to those elaborated by Lundstrom and Lamont (feasibility, simplicity), a Likert (1932) scaling procedure was also selected for the behavioral intent items. Comparisons of Likert and other procedures have generally concluded that the scales which result from the former are at least as reliable (e.g., Thurstone's technique, Edwards and Kenney 1946). The major task of the Pilot Study involved the selection of specific items as measures of the satisfaction construct. Several analyses of the emotion scales were conducted in order to identify those emotions which best characterize each end of the satisfaction continuum. The 75 items that constituted question 4 were subdivided four ways: items used by Westbrook (1983); items used by Izard (1977) in a shortened scale; those items that exhibited the highest factor scores in Izard's factor analysis; and the full scale. In general, the analyses conducted were of the following types: Factor Analysis of Emotions Â— A varimax principal components factor analysis was conducted for each of the four sets of emotion measures. These analyses were performed to ascertain whether the adjectives said to describe separate emotions would load appropriately. Also, given the restricted emotional patterns expected (i.e., only satisfaction and dissatisfaction), this analysis was expected to suggest which emotions might combine in each pattern. Discriminant, Cluster and Regression Analysis of Emotions-Scales were created by summing the items loading on each factor resulting from the prior analyses (Table 8). These intermediate scales were then employed in efforts to segment subjects. Discriminant analysis using assigned condition (satisfying versus dissatisfying occasion described), K-means cluster analysis looking for "good" two-cluster solutions and regression of more standard satisfaction measures on the emotional scales were expected to result in identification of
Table 8 Results of Factor Analysis of Emotion Items I. Using items from Izard (1977) Factors and loadings Emotion Item 1 2 3 4 5 6_ Joy Distress Disgust Anger Contempt Interest Attentive .8747 .7696 .7486 Fear Scared .9404 ,8326 ,9225 Surprise Surprised .8191 .8274 .7724 Shyness Sheepish .5707 Guilt Delighted
181 Table 8--continued Results from Westbrook's Items (29 items) Factor 1 Distress, Disgust, Anger, Contempt Factor 2 Interest, Joy Factor 3 Surprise Factor 4 Guilt, Shyness Factor 5 Fear Factor 6 Shame Results from items selected from Izard's Factor Analysis (49 items) Factor 1 Joy, Distress, Disgust, Anger Factor 2 Interest Factor 3 Fear Factor 4 Surprise Factor 5 Shyness Factor 6 (?') a Factor 7 Guilt Results from all items (75 items) Factor
182 the emotions most representative of satisfaction and dissatisfaction. Factor Analysis and Other Analyses of Behavioral Intent Measures Â— Rather than attempt to measure and scale actual behaviors in the Pilot Study, Likert scales of potential satisfaction related behaviors were used as a surrogate. As with the emotional scales a varimax principal component factor analysis was used to derive intermediate scales that were used in further analyses (Table 9) Discriminant analysis using assigned condition and mean comparisons across groups defined by emotional profiles were conducted to identify measures of behaviors most typical of satisfied and dissatisfied respondents. Results Given the exploratory nature of the study, the number and selection of subjects, the lack of blind judging and the intentional selection of extreme results, it would be misleading to conduct or report significance tests for this study. Where statistics are shown (e.g., Cronbach's alpha), these indicate criteria used in selecting items for further analysis. As a pilot study, the primary result is intended to be the actual measurement scales rather than statistics relating to their reliability and/or validity. Such estimates would have to be based on a sample separate from that used in derivation of the scales (Cureton 1950) Emotion scales In general, the factor analyses of emotional responses provided encouraging results. By and large, the adjectives that were predicted to describe a particular emotion did load on the same factor. Furthermore, the emotions that had been suggested in Chapter 3 as components of satisfaction and dissatisfaction also tended to load together (Table 8). As indicated above, the mean score on each scale listed in Table 8 was used in further analyses.
183 Table 9 Behavioral Intent Scales from Factor Analysis Item # from Factors and Factor loadings Appendix A Description 1 2 3 4 5 6 1 Satisfied with decision to purchase/ use* .8723 2 If do over, would feel differently 3 -.6750 3 Choice was wise a .8467 4 Feel bad about decision 3 -.7634 5 Did right thing a .7607 6 Not happy that I did what I did 3 -. 6378 7 Would buy other products of same brand .5547 14 Tell a friend good things .7530 20 Recommend product to a friend .7890 21 Recommend friend avoid product -.4811 .4720 22 Publicly endorse product .5987 12 Complain to higher authority .8511 15 Tell a friend bad things .7753 16 Write manufacturer and complain .8353 23 Publicly attack product .5387
184 Table 9 Â— continued Item // from loadings Appendix A Description Factors and Factor 3 4 5 6 18 Discard and replace with different brand 19 Discard and replace with different product 8 Buy other products from same seller 13 Compliment higher authority 17 Write manufacturer and compliment 10 Buy more of product as gift 11 Buy similar product as gift 9 Buy other products of same type 24 Fight to remove product from market Number of Items Coefficient a (based on Pilot) 11 4 .95 .83 7860
135 The emotional scales were used to predict whether the subject had been in the satisfied or dissatisfied condition of the experiment (discriminant analysis using PROC GLM from the Statistical Analysis System) and to predict the subject's responses on both the more typical satisfaction scales and the behavioral intent summed scales (regression and correlation in SAS) Two different cluster analysis programs (SAS and BMDP Â— K-means) were used to describe subsets of the sample. No single analysis led to firm conclusions. However, the bulk of the evidence suggested the following conclusions: 1. Izard's abbreviated DES (1977) was more reliable (Table 10) and yielded more consistent results than either Westbrook's version or scales developed within the present study. 2. The emotions that were most characteristic of persons in the "satisfied" condition and were most associated with highly positive satisfaction and behavioral intent responses were Interest, Joy and Surprise (Table 11). 3. The emotions that were most characteristic of persons in the "dissatisfied" condition and were most associated with highly negative satisfaction and behavioral intent responses were Anger, Disgust and Surprise (Table 11). These emotions, then, comprise working definitions of either end of the satisfaction continuum. The satisfaction end of the scale, at least, is consistent with the conclusion drawn by Westbrook, "joy and interest, once elicited in a consumption experience, are necessarily involved in behavior relevant to the stimulus the affect of surprise also is likely to be involved" (1987, p. 259). 2 Operational definitions were derived based on mean responses by subjects in each condition. For the main study of the dissertation, a satisfaction 2 Westbrook distinguishes the effect of surprise as an "amplifier" of other emotions. This suggests that some different algebraic representation of surprise versus other emotions--e g. multiplicative rather than additive, respectively, may be in order,
186 Table 10 Reliability Coefficients for Scales Used in Pilot Study Interest Joy Surprise Sadness Anger Disgust Scorn Fear Shame Guilt Number C Of Items a
137 Table 11 Adjectives Describing Ends of Satisfaction Continuum Satisfaction: Emotion Adjectives Interest Attentive, Concentrating, Alert Joy Delighted, Happy, Joyful Surprise Surprised, Amazed, Astonished Dissatisfaction: Emotion Adjectives Anger Enraged, Angry, Mad Disgust Feeling of Distaste, Disgusted, Feeling of Revulsion Surprise Surprised, Amazed, Astonished
pattern of emotions will consist of an average response equal to or greater than 3 (a response of "Moderately") on the items comprising the characteristic emotions. Similarly, dissatisfaction (negative satisfaction) is defined as an average response equal to or greater than 3 on items specific to that pattern of emotions (Table 11). When these definitions were compared with the responses of the 94 subjects, 36 met neither cut-off, 7 met both cut-offs, 50 were correctly classified into original condition and 1 subject from the dissatisfied condition was identified as satisfied. All of the "misclassif ied" subjects were from the dissatisfied condition of the study. Further examination showed that there was a tendency to report "dissatisfying" occasions that were resolved in the subject's favor and resulted in a net satisfaction. This explanation is corroborated by the responses on classical satisfaction scales. Those subjects who met neither cut-off (14 from the satisfied condition and 22 from dissatisfied) represent a different problem for the measurement technique. Either the definitions that have been constructed are not sufficiently sensitive, or these subjects were not satisfied (or dissatisfied) as defined by the conceptualization offered by the dissertation. Although not conclusive, some insight can be gained by comparison against the antecedents and resulting behavioral intentions Of the 36 respondents who met neither set of criteria, only three did not describe a situation which encompassed all six antecedents; nine did include some mention of the antecedents and 24 were counted as
most likely satisfying the antecedents. 3 These results indicate that caution is required in interpretation of the sufficiency of the antecedent conditions. That is, although an experience that fulfills the conditions should position a person along the satisfaction continuum, it may not be sufficient to arouse strong emotions. Thus, the 33 subjects who met the antecedent conditions but not the emotional criteria would be positioned in more central portions of the continuum. Description of this region is beyond the scope of the present study and not appropriate for the data collected. Future studies based on the conceptualization derived in the dissertation can make specific definitions and predictions regarding the middle range of the scale and at that time establish operational definitions based on emotional patterns Behavioral intent measures One advantage of the present conceptualization is the integration of specific results (behavioral, motivational) into the overall satisfaction theory. This enables the use of behavioral intent measures as well as satisfaction scales to compare subjects who have been classified by emotional response. Table 12 summarizes the mean responses to various dependent measures by experimental condition or subsequent emotional classification of subjects. There are a couple of points notable from the table: (1) The difference between survey groups (conditions) 3 The narratives provided by the subjects were scored by the author and required judicious interpretation, as with the early pilot findings reported. This procedure is adequate for the exploratory nature of these particular studies.
190 Table 12 Mean Responses by Condition and Classification Condition Classification Satisfied Dissatisfied Satisfied Dissatisfied Measure (n=42) (n=50) (n=29) (n=29) Dependent Measure 15 Scale U 4.35 2.53a 4.39 2.37a Scale 2 3.70 2.64a 3.73 2.52a Scale 3 4.27 2.96a 4.34 2.89a Scale 4 3.31 2.41a 3.46 2.19a Scale 5 2.96 2.30b 3.19 2.11* Scale 6 4.12 3.70^ 4.14 3.52Satisf iedDissatisf ieds 6.79 1.77a 6.62 1.69a Odds 9.29 2.48a 9.24 2.00a Percent Satisfied 93.10 20.96a 93.45 17.59a DeiightedTerrible 6.40 2.43a 6.59 2.31a Note: Significance of results is indicated by a = p< 0001 ; b = p< -01; c = p<.05; d = p< 10. Some scales have been reversed so that higher numbers are more positive f "Scales" are mean responses to behavioral intent scales derived from Table 9. Â§ Satisfaction measures are from question 5 of Appendix A, parts a through d.
191 resulted in significantly different responses across all measures. (2) As shown in Appendix A, the Likert type scales used to assess behavioral intention were centered at Neither Agree nor Disagree with a statement indicating what the subject might "normally do" under the circumstances described. The responses to the items comprising summary scales four, five, and six in Table 12 lie fairly close to this response, that had been assigned a value of three, and therefore are not likely to discriminate well between satisfied and dissatisfied (or indifferent) individuals. (3) The emotional classification procedure generated slightly more extreme means than those collected from the entire sample. While the summed scales are adequate to represent differential responses on the items, it is also of interest to determine which individual items may be useful in the main study. Table 13 summarizes the mean values for each of the dependent measures, by condition and by emotional classification. Those items that exhibit high absolute differences are candidates for inclusion as behavioral intent or behavioral measures in the main study of the dissertation. 4 Overall, behavioral intent, emotion and more standard measures or satisfaction have tended to converge on the distinction between groups. At this juncture, however, it would be inappropriate to attempt to compare the measures competitively. There is no reason to 'There is a level of abstraction problem raised by blind acceptance of these results. The situations described by the subjects in the pilot study involved a variety of products and services which may have aroused emotions focussed at any of the levels suggested by Table 2. For a given target of the emotion, behaviors specific to that target should be most affected, ceteris paribus.
192 Table 13 Mean Responses to Likert Items by Condition and Classification Condition Classification Satisfied Dissatisfied Satisfied Dissatisfied Measure*.* (n=42) (n=50) (n=28) (n=29) 2.08a 4.72 1.89a 2.68a 4.21 2.54a 2.32a 4.55 2.07a 3.10a 4.69 3.07a 2.86a 4.52 2.61* 3.26a 4.7 2 3.14* 2.76a 4.21 2.54a 2.92a 4.00 2.50a 3.60 3.86 3.36 2.16^ 3.24 1.96' 2.44 3.14 2.25= 2.46a 3.93 2.21a 2.14a 3.48 2.07^ 2.14a 4.24 2.04a 1.96a 3.31 1.86a 1
193 Table 13 Â— continued Condition Classification Satisfied Dissatisfied Satisfied Dissatisfied Measure^ f (n=42) (n=50) (n=28) (n=29) 16 Complain manufacturer 3.71 3.00^ 3.52 2.96 17 Compliment manufacturer 2.69 2.18= 2.90 2.00^ 18 Replace brand 4.24 2.90 a 4.31 2.82* 19 Replace product 4.31 3.02 a 4.38 2.96* 20 Recommend buy 4.29 2.20 a 4.38 2.11* 21 Recommend avoid 3.98 2.32 a 4.00 2.25 a 22 Endorse 3.55 1 94 a 3.66 1.71* 23 Attack 4.10 3 1 4 a 4.17 3.04^ 24 Ban 4.48 3.80 b 4.41 3.68^ Note: Significance of results is indicated by: a = p<.0001; b = p< .01 ; c p<-05; d = p< 10. e Some scales have been reversed so that higher numbers are more positive. f Measure numbers correspond to item numbers in Table 9.
194 expect divergence in any particular direction, nor was the studydesigned to provide such tests (a replication should indicate convergence among the measures). Moreover, the scales were developed iteratively using one another as criteria, so no true indicants of reliability are available. Likert type scales were derived with procedures similar to those usually prescribed. Of special note is that the situations were specifically chosen to create extreme responses on the various measures. Thus, the design of the study selected the extreme groups to use in measure selection, as is generally recommended (Edwards and Kenney 1946; Likert 1932; Shaw and Wright 1967) The behavioral intent measures may be useful in developing measures that are based on actual behavior. Behavioral measurement of attitude has been attempted, but involves considerable difficulty and cost to execute and, due to the other forces acting on behavior, may be of questionable reliability (Cook and Selltiz 1964; Fishbein 1966). it has been suggested that weighting behaviors may make for more reliable measures (Rosander 1937), but such weighting schemes tend to be unstable (Triandis and Triandis 1965). The best option would seem to be to collect both intent and actual behavior measures and look for convergence 5 5 An assumption implicit in the discussion of behavioral measures is that the experimental design will control such factors as cost and effort required for each behavior and, to an extent, expected result. That is, it is expected that perceived costs and benefits of various responses would be equivalent across subjects; only motivation would vary. This mitigates the arguments presented in the CS/D literature against complaints as a measure of satisfaction.
195 Summary of the pilot study The study used the critical incident technique that has been used successfully in the job satisfaction literature to revive the memory of particularly salient emotional events. Emotional definitions of the positive and negative endpoints of the satisfaction continuum were derived and shown to be feasible to ascertain with a simple questionnaire. The study also indicated behavioral intent measures that discriminated well between groups of subjects. The results from this study are used in deriving both measures and hypotheses for the major empirical portion of the dissertation. Research Conducted Purpose Having defined the satisfaction and dissatisfaction constructs as specific patterns of emotions, it is necessary to investigate the conditions that are hypothesized causes of the feelings. The study evaluated two of the three concepts that have been added to the conceptualization of satisfaction and whether each is necessary to the arousal of emotions. The measures derived in the Pilot Study just reported are used in conjunction with measures of behavioral intent and more traditional satisfaction measures to examine the appropriateness of the derived conceptualization. Design For the purpose of providing as clean a test of the hypotheses as possible, satisfaction with a single purchase/use occasion provided the focus of the study (Table 1). Also, the critical tests will focus on immediate response to the situation, before motivation begins to decay.
196 Six antecedent conditions have been proposed as necessary and collectively sufficient to arouse satisfaction emotions. The presence of two of the three introduced by the current conceptualization-decision and high response involvement Â— are manipulated factorially. The discrepancy between expectations and perception (perceived performance) is manipulated to either a strongly positive level (i.e., low expectations and good performance) or a strongly negative level (i.e., high expectations and poor performance). Thus, the overall design is a 2x2x2 factorial (the disconf irmation judgment is expected to occur when all of the other antecedent conditions are fulfilled) Target Product The design of the experiment made selection of the product to be used somewhat difficult. Several criteria were identified that needed to be met in order to conduct an effective and credible experiment. 1 ) The experimenter must be able to manipulate product performance in a simple manner that will yield a consistent and unambiguous perception. Yet, the manipulation of performance must be sufficiently subtle to avoid raising response involvement inadvertently. 2) The subject must be able to assess performance after a single trial which can be performed within an experimental setting. 3) Response involvement is dependent on enduring involvement which must therefore be fairly low in order to achieve the low involvement condition. On the other hand, the situation must be able to raise involvement credibly to a level sufficient to drive the satisfaction process 4) The product must be useful to the subjects in their ordinary lives so that preand post-trial choice measures are credible. 5) Similarly, prior consumption within the product class will help to establish more widely held norms
;97 for performance. Specific expectations, however, must still be subject to manipulation. 6) The overall procedure must be credible to the subject. That is, a condition requiring no decision and low response involvement would be suspect without a viable "cover story". As discussed, ground coffee (for automatic drip cof f eemakers) was chosen from a candidate list based in part on the products used in prior CS/D research (Table 7) and those discussed by respondents to pilot studies. Coffee was used in two early investigations (Cohen and Goldberg 1970; Olson and Dover 1976) and fits the criteria quite well. 1) Coffee has a limited number of salient attributes which suggests that performance perception can be manipulated via only a limited number of flavor dimensions such as strength of flavor and bitterness (Eastlack 1964; Olson and Dover 1976). The flavor quality of coffee can be lowered to a poor, but acceptable, level by a variety of techniques (e.g., use of a poor tasting additive) 2) In general, food products can be consumed and evaluated rather quickly. 3) Prior research suggests that a study which is focused on coffee selection can be relatively involving. On the other hand, coffee is not so intrinsically involving that a diversion would be unsuccessful 4) The subjects, whether students or a more general sample, can be screened to identify coffee drinkers 5) Coffee drinkers would be expected to have some expectations of normal flavor. Such subjects would be expected to have ample prior experience so that product class norms should dominate any comparison level judgments, rather than a specific alternative choice (Woodruff, Cadotte and Jenkins 1983a). Specific expectations regarding the experimental sample can be manipulated by use of unfamiliar brands or brands not yet available.
198 6) The study conducted by Cohen and Goldberg included many procedures that are adapted to the present study (1970). Specifically, they varied decision and non-decision, involvement (although they were concerned with ego involvement which the present conceptualization identifies more with the decision manipulation) and a taste manipulation. Rather than instant coffee as Cohen and Goldberg used, the study undertaken used coffee ground for automatic drip makers. This provided the ability to use a coffeemaker as a decoy product. The presence of a decoy enables the involvement manipulation to be presented to the subjects as a focus of attention or interest. Manipulations The basic manipulations are described in Table 14. Operationalization of these within the design necessitated slightly different procedures within each cell; therefore, there were eight different protocols (Appendix B) Discrepancy Central to most CS/D investigations is the discrepancy between expectations and perception. As mentioned previously, a subject participated in either a positive or a negative discrepancy condition (i.e., the present study does not examine the result of expectations being confirmed). Each discrepancy condition was defined by separate manipulations of expectations and product performance. The positive discrepancy condition required relatively low performance expectations be generated within the subjects. Thus, the information given about the brand had to be somewhat negative. Conversely, in order to maintain credibility regarding the purpose of the study and the product's existence as a competitor in the market, an
199 Table 14 Manipulation Summary Discrepancy 3 Positive: Low expectations were generated by advertising copy and subject instructions that focused on the inexpensive nature of the new blend. High performance perception was to be the result of using a premium grade--but not f lavored--ground roast coffee. Negative: High expectations were generated by advertising copy and subject instructions that focused on the premium quality of the new blend. Low performance perception was to be the result of using generic coffee blended with ground roast peanuts (80;20, respectively). Decision Present: Subjects were asked to select the coffee that would be prepared and sampled. Choice was to be made between the target brand and an unspecified brand of coffee of a quality appropriate to the experimental condition. Absent: Although the possibility of sampling either of two brands was presented, subjects were assigned to the target brand by the experimenter. Response Involvement High: Subjects were asked to participate in a test of a new coffee to be introduced soon in the local market. Each subject was told he/she would receive a "week's supply" of the evaluated brand. Low: Subjects were asked to participate in a test of a new coffeemaker to be introduced soon in the local market. Each subject was told he/she would receive a chance to win a coffeemaker in a lottery among study participants. Note: a Target product was a new ground roast coffee from an unnamed processor.
200 overall positive message had to be communicated. Therefore, the manipulation that was adopted presented the brand as very inexpensive but of "adequate" quality. This information was communicated to the subjects through multiple methods. At some point in the study, each subject in a positive discrepancy condition was told, "The coffee will be of adequate quality, but is primarily intended to be a cheap blend to help consumers cope with high prices brought about by the recent shortage. We are helping them to identify the minimal acceptable quality for the home market" (see Appendix B) In addition, subjects were asked to evaluate "mock-ups" of advertising proposed for the new brand (Figure 10). The advertisements were prepared by a commercial artist and were designed such that only the desired manipulations varied (i.e., she prepared a base advertisement on which copy blocks, et cetera could be substituted). 6 Subjects actually evaluated full-color photo-reproductions of handdrawn originals. Each subject evaluated two coffee advertisements as part of a booklet of four (also containing two coffeemaker advertisements, Figure 10a, 10b). In one advertisement the headline read, "And Now ... An Affordable Coffee From (Figure 10c). The body copy read, "A medium quality light coffee made from less costly coffee beans. Special roasting, grinding and blending of these beans keep processing costs at their lowest possible level. And there's less bitterness than you get with other economy brands. Look for coffee to be less costly than even your store's brand." 6 The advertisements were drawn by Maria T. Futules of Akron, Ohio.
201 The only can pictured on which labelling was readable identified the coffee as "Special Blend". The other advertisement featured differences only in the body copy and an accompanying cents-off coupon (Figure 10d). For the low expectations condition, the redemption value of the coupon was ten cents and the copy read, "originator of the automatic drip coffeemaker is proud to present an economical coffee, up to $1.00 less expensive per pound than even store brands. You still get acceptable quality in every cup. coffee provides delicate flavor and a light, clear brew. There's even less bitterness than other economy brands." As before, the graphic portion of the advertisement did not vary by condition. Having used these devices to create low expectations, it was necessary to find a product which would provide good performance and, therefore, a positive disconf irmation. The brand selected was "Uncle Charlie's Mountain Rich Coffee (see description in Figure 10e). 7 A preliminary taste test was accomplished by preparing and serving some of the coffee to faculty, staff and students in the College of Business Administration at the University of Akron. All agreed that the coffee was of good quality, at least equal to that normally prepared and sold in the college (although the normal brand varies). The negative discrepancy conditions required that high expectations be met with poor product performance. As before, expectations were generated through multiple communication modalities. Instructions read to the subjects were as follows, Uncle Charlie's Products Inc. Grand Rapids, MI 49508.
202 "The coffee will be very high in quality but only slightly more expensive than present top-of-theline coffees. We are helping the company identify the quality most preferred by consumers" (see Appendix B) Advertising copy was also written to generate high performance expectations. Replacing the copy in Figure 10c was, "A delicious full bodied coffee made from the finest beans available. Special roasting, grinding and blending of these beans bring out the finest aroma and taste of the coffee. And there's none of the bitterness of most brewed coffee. Look for coffee to be priced with your store's medium brand coffees." The label on the can was also changed to read "Premium Blend" (Figure 10f). The second advertisement was changed to read, originator of the automatic drip coffeemaker, is proud to present a premium coffee only about $1.00 more expensive per pound than other popular brands. This means you get premium quality in every cup. coffee provides full flavor and a dark, rich brew. There's no bitterness with this premium ground coffee." The face value of the coupon was seventy-five cents (Figure 10g). In order to create uniformly poor but credible product performance, the coffee was adulterated prior to conducting the study. For the negative discrepancy conditions, the base product was generic coffee purchased at a retail supermarket. 3 This coffee was mixed with roasted peanuts in an 30:20 ratio and re-ground at a regular grind 6 Kash N Karry Gainesville, FL.
203 In a Hurry? Coffee fir Ofte? The Quick-Brew from Brews Small Amounts Quickly and Easily Â• Up to 4 (5oz.)Cups. Â• Easy Fill. Â• Brews in Five Minutes. Â• Desighed -for Smaller Pdts No Need -for Coffee "to Sit in -the Pot -for Hours. Easy Clear\irvg, (a) Coffeemaker Advertisement Figure 10 Descriptive Stimuli
204 How Easy Is It To Use A Quick-Brew from ? \m Insert -Filter irfto basket ard waasum in your -favorite, brand erf drip drir\d Coefe. 3 After pourm^ (v\ tKe uJater, place, pot urwier tke -filter basket ardi 'pre&b tKe. automatic 3wftcVv i a the. *Or\" p osfti oa. Vjarfc 5" rAiiMcfces., a*vi. 2Fill tK& Calibrated po"t -to the desired T-up" level and pourinto -the water intake opervlng ENJOY (b) Coffeemaker Advertisement Figure 10--continued
205 And Now. An Affordable Coffee From A medium quality ligVtt coffee made Â•from less costly coffee, beans. Special roa-btiir^g; tfn'nding'and blerd'in^of "these, bean's, Keep prooessinc/ co^ts at "their lowest -posa Me (evet Atmd -there's lea, b'ctterness fhav\ you. get u>th other etorovAy bvdMs. Look for coffee, fjo be. (ess costly -than even stores bnav^di. (c) Positive Discrepancy Conditions Figure 10--continued
206 NEW FROM Our very own Coffee .Originator of -the, automatic drip coffeeirnake*'\s proud to pre^nt av\ ecorornicsl coffee. up-to % 1.00 less expensive, perpourd -than even store brands You. still sfefc acceptable, quality in every cup. ^^__^___^_ cot-fee provides delicate, -flavor ar>d a HaTht, clear brew. There's even less bitterness tWan other economy brands. (d) Positive Discrepancy Conditions Figure 10--continued
207 MOUNTAIN RICH DIFFERENCE Hand Picked Milds Coffee beans are identified by their characteristics, beginning at the bottom with the least expensive Robustas that are Mat on body and high on caffeine and acid They move up the quality Iddder through Semi-Roasts. Naturals Brazils and finally up to the gourmet quality beans callea Milds. Mountain Rich blends a higher percentage of Milds than any over-the-counter brand That's why it brews up more flavor with one-third the coffee Artful Blending We blend over 6 different types of beans from such diverse countries of origin ds Peru, Kenyd. Mexico dnd Colombia, in order to dchieve d superior fldvor thdt stays consistent in spite of seasonal changes Fast Roasted Because of the high percentage of Milds in our blend we can achieve a faster roast at a higher temperature ana thus get a deeper flavor without changing the character of the bean This results in an aromatic, full bodied cup of coffee Precision Ground The search for higher yield or getting more cups of coftee per pound has lead to d new generation of grinding methods Mountain Rich has been able to take advantage of this new technology to give you a finer, fluffier blena capable of delivering more than twice as much coffee per weight Nitrogen Packed The freshly grouna coffee is sealed in neutral nitrogen to prevent oxidation and to preserve the original aroma we have worked so hard to bring to its peak for your brewing pleasure Every way you compare it. Mountain Rich Coffee really is different But don't take our word for it Taste this difference for yourself SUPERB COFFEE AT HALF THE PRICE PER CUP! (e) Description of Product Figure 10--continued
208 How to make a perfect cup of coffee with your pack of MOUNTAIN RICH Open The Pack Take a moment to savor the aroma. Give yourself a preview of the full bodied flavor to come when you brew it m Put A Pinch In Your Hand Notice how uniformly fine it is ground Fine grinding is the secret to getting more cups per pound and Mountain Rich gives you three times more than the supermarket brands. Use Any Coffee Brewing Machine Pour the coffee evenly into the filter This 3 4 oz sample will make a standard 10 cup pot or even more depending on the strength you prefer Remember, this is a high yield blend that brews up to 3 times as much as your present brand. It you must perk, understand that you will be boiling away some of the delicdte flavor essence, so it will taste more bitter than a drip-brewed sample would Still, you should notice an improvement over your present brand SECRETS OF ENJOYMENT Â• Never make less than 3 4 of a pot or the water will pass through too fast to give you full flavor. Â• Never let coffee boil Ideal brewing temperature is 200 to 205 and holding temperature is 190 UCP251 Â• Clean your coffeemaker well after using A film of left-over coffee oils can make your next pot taste bitter without you knowing why. Take time to savor the aroma as well as the flavor Good coffee like fine wine, is a beverage to be experienced by all the senses (e)--continued Figure 10--continued
209 And Now. Premium Grade Coffee rrom A delicious -Full bodied coffee made, -from the finest beare available. Special roast'\*^ gnrelW^and blendir**" of these Deans bnVtf out tine finest aranra awdi tasfe. of the coffee. And there's none of Â•the bitterness of w\Oot brewed coffee. Look -for. (f) Negative Discrepancy Conditions Figure 10--continued
210 NEW FROM Our very own Coffee 75* 5 otto I mn S originator of "the, automatic drip coffeemaker is proud to pre&eyvt a premium coffee. -only about $1.00 more expensive per pound than otherpoplar brands. This means you aet premium quality in every cup, I coffee, provides full -flavor and a dark rich. Drew, There's no bitterness voitv-\ t.V\\& pvenvum, gTrourid coffee. Save 75* ON ANY Si Z-ff 09 T*iB< 75* (g) Negative Discrepancy Conditions Figure 10--continued
21 1 setting in a coffee grinder. 9 The ground product was checked for consistency and hand-packed into unlabelled #10 vegetable cans (with minimal headspace) The cans were sealed at room temperature and pressure. A series of taste tests culminating in trials by staff and graduate assistants in the Department of Food Science and Human Nutrition at the University of Florida identified the 80:20 blend as an appropriate quality for the study (from a range of 50:50 to 90:10). As described bv one secretary, "If I got this in a restaurant, I'd send it back. If I made it myself, I'd drink it." Additional (unannounced) trials with faculty, staff and students in the College of Business Administration at the University of Akron yielded similarly negative evaluations. Given these responses, the peanut-coffee beverage blend was retained as the poorly performing product. Decision Those subjects in conditions requiring that a decision be made were asked to choose the coffee they tasted. The decision was made more salient by having subjects indicate their choice on an appropriate questionnaire. The alternative in all cases was matched to the expected quality of coffee Â— albeit at a somewhat lower end of the range Â— in order to encourage volitional selection of the test brand (Scott and Yalch 1980) Specifically, subjects in the conditions in which a decision was to be made were told, Over the course of the study, we're comparing evaluations of several brands of coffee. Today we have both brand M and brand N. M was the brand 9 The peanuts were prepared as detailed in Table 15.
212 Table 15 Preparation of Peanuts for Coffee Blend 1 Peanuts pressed @ 40 tons pressure for approximately 30 minutes to remove oil. 2. Roasted @ 400F for approximately 3 hours to achieve a darkroasted hue similar to roasted coffee beans. 3. Skins removed by hand rubbing. 4. Mixed with coffee in 5-pound batch operation (i.e., four pounds of coffee plus 1 pound roasted peanuts).
213 described in the ads which you've just read. N is a locally available generic coffee [popular brand in this area]. We are, of course, most interested in your reaction to brand M, but since we're testing both M and N, you're free to choose either one. Please indicate the brand you have chosen in the space provided on the product evaluation form (Appendix B). The questionnaires used in these cells of the design provided spaces for indicating which was the "Brand Selected" and the "Brand Rejected" to make the freedom of decision more salient (Appendix C) Subjects in conditions where no opportunity for a decision was to be given were read a different set of instructions, Over the course of the study, we're comparing evaluations of several brands of coffee, including brand M and brand N. Today we'd like for you to evaluate brand M. You've just read some advertisements for this brand (Appendix B) In all cases, both brands were plainly visible in front of the subject. Ihe questionnaires used in the no decision portion of the experiment provided a space only for "Coffee Brand Evaluated" that the subject was instructed to fill by writing in an "M" Response involvement High response involvement was generated by emphasis on the role of the coffee product; the study is concerned with coffee and the subject received a gift of coffee for participating (Petty, Cacioppo and Schumann 1983; Sawyer 1975). Although all subjects ultimately chose and received a gift sample, only those in the high involvement conditions were apprised of this prior to the conclusion of the interview. Specifically, pre-tests of the experiment revealed that several devices were required to execute the response involvement manipulation convincingly. Set-up instructions for the high involvement conditions
214 specified that the gift packages of coffee (labelled paper bags purporting to be either brand "M" or brand "N") be in plain sight on the table throughout the study (Appendix B) Initial oral instructions included, Thank you for agreeing to participate in our study. Today we are asking consumers like yourself to evaluate coffee made by a major food processing company In return for your assistance and taking the time to contribute your honest opinions you will be given a week's supply of the coffee Before you actually try any product, we would like for you to look at some sample print advertisements. These ads are for the coffee and a related line of cof f eemakers Subjects read and evaluated advertisements for both the coffee (manipulating expectations as described above) and a coffeemaker. They were then either assigned to brand M or allowed to choose a brand to evaluate (depending on the "Decision" condition to which they were assigned) Following this they were told, "Remember, you will be given a week's supply of the brand you have evaluated." and presented with the appropriately labeled sample at this juncture. Following evaluation of the test coffee, subjects were given the opportunity to exchange the gift for the alternate brand if they so desired (as a measure of "re-purchase"). In contrast, the low response involvement conditions entailed diversion. That is, the study focused on the coffeemaker and the subjects were not apprised of the gift until all other dependent measures had been collected. Instead, a chance at a gift coffeemaker was the promised honorarium for participation. Blank cards for this
215 ow drawing were placed in plain view and one was completed during the brewing phase of the experiment (Appendix B) The explanation and instructions given to the subjects in the 1 involvement conditions were, Today we are asking consumers like yourself to evaluate a coffeemaker that may be introduced by a major food processing company. The company has modified an existing design for easier use. In return for your assistance and taking the time to contribute your honest opinions you will be given a chance to win a coffeemaker in a drawing that will be conducted at the end of this phase of the study ... Before you actually try any product, we would like for you to look at some sample print advertisements. These ads are for the coffeemaker and the company's own brand of coffee, also to be introduced soon As you use the coffeemaker, remember that you will have a chance to win one from the same company In addition to the chance at the coffeemaker that you were promised, I'm happy to be able to offer you a week's supply of coffee as a gift. The last of the above statements was the final instruction to subjects in low involvement conditions. Subjects The subjects for the study were 93 men and 110 women, ranging in age from 16 to 62 (when given). This reflects a convenience sample drawn in the student center on the campus of the University of Akron during the summer session and the summer/fall intersession period. This timing allowed the recruitment of parents who were on campus for orientation as well as the usual student patrons. Participation was voluntary and therefore self-selected. Assignment to experimental
216 condition, however, was predetermined randomly. All subjects claimed to drink regular (i.e., non-decaffeinated) coffee at least occasionally. Procedures In general, the study was designed to be as non-reactive as possible. It was argued previously that many of the CS/D findings can be explained as the result of subjects accommodating researchers by completion of satisfaction measures. The study used a cover story, a non-role playing experience and measures (e.g., revised DES) that do not suggest "obvious" responses in an effort to minimize obtrusiveness and any resulting demand artifacts (Sawyer 1975, 1977; Webb et al. 1966). The procedures described in Appendix B were constructed based on considerations of credibility of the procedure, given that the subjects were aware that they were participating in a Marketing study of some type (ostensibly, for a manufacturer). The protocols were constructed in such a way that the study could have been conducted in a shopping mall. On debriefing, no subject admitted to questioning the legitimacy of the research objective at the time of the experiment. The study was conducted by the principal investigator and three trained undergraduate assistants. 13 Each experimenter administered each condition of the design. The studies were conducted in a reserved room of the student center. One room was divided and two conditions were run concurrently, one positive and one negative discrepancy. This 10 I would like to acknowledge the contributions made by Nan Elrod, Chris Vandevort and the late Mike Albaugh in this regard.
217 corresponded to the need to use separate experimental supplies (cof feemaker, water supply, samples, et cetera) for the manipulation, as noted in the set-up instructions in each part of Appendix B. Background music was found to be effective at preventing inter-condition contamination Â— even from instructions read aloud. Having set-up for the appropriate condition (as determined by a list provided by the principal investigator), the experimenter attempted to locate a volunteer in public areas of the student center. Initial attempts to generate quota samples were frustrated by a relative scarcity of admitted coffee drinkers (Maxwell 1988 cites declining consumption--this was especially true among those of traditional college student age) and reluctance of prospects to commit for the requisite time. Ultimately, each loiterer, then each passerby, in turn was addressed, Excuse me, my name is and I'm conducting market tests for NEH research. Today we're interviewing people on campus regarding coffee products. Do you drink coffee, even if it's occasionally? [If no, the experimenters were instructed to simply thank the prospect anyway.] I'd appreciate it if you would take a short time to try our product and complete a questionnaire. The experimenters were instructed not to volunteer a time requirement but to respond that the test would take about 25 minutes, if asked. Some prospects asked what type of product was being tested. They were told either a ground coffee or coffeemaker (as appropriate to the involvement condition in which he/she would participate) but were given no further information. No mention of inducements was made as subjects were recruited. Volunteers were taken to the room where the experiments were being conducted.
218 On arrival, the subject was seated opposite the experimenter at a table containing the appropriate experimental materials. The same informed consent statement was read to each subject (Appendix C) None declined at this point or withdrew later due to risk. Since an adulterated food product was being used (i.e., the peanut-coffee blend), it was necessary to safeguard the subjects from unwitting exposure to potentially harmful contents (e.g., allergic reactions to the peanuts). To this end, a Food Sensitivity Questionnaire was developed (Appendix C). When presented to a subject, this was explained as "a standard form used whenever we have someone tasting food." In actual practice, the form was used to intercept anyone indicating a desire to avoid Peanuts/peanut products in conditions in which negative discrepancy was involved (Appendix C) Two prospective subjects were excused (but given honoraria) after indicating this category. One prospective subject, despite having been alerted to the use of "regular" coffee indicated an aversion to caffeine and was rewarded and excused. Finally, one subject wrote in a reference to an allergy to artificial creamers Â— then later used the artificial creamer that had been supplied. She completed the study and reported no serious ill-effects at de-briefing. Following administration of the Food Sensitivity Questionnaire, the subject was read the cover story and advertising evaluation instructions appropriate to his/her experimental condition. The advertisements were bound in a booklet that was color-coded to condition. The Advertisement Rating Form (Appendix C) was designed to force careful inspection of each advertisement. Manipulation checks of
219 the expectations generated by the advertisements are available more directly in the main questionnaire. The subjects then chose or were assigned a coffee to test. Each subject prepared his/her own coffee in the coffeemaker provided (in order to ensure credibility of the "coffeemaker test" portion of the design). The instruction cards, cof feemakers, and water bottles (distilled versus tap water) were all color-coded to ensure no contamination between discrepancy conditions. The amounts of coffee to be added were determined in informal taste tests and measured in graduated spoons that were provided (see Table 16 for instructions). The coffee brewed in approximately five minutes in Mr. Coffee Four Cup Coffeemakers (brand name and marks were obscured) During this interval, the subject was asked to complete the Classification Form (Appendix C). Although this provided useful demographic data, its primary purpose at this juncture was to prevent the boredom that pre-test subjects experienced while the coffee brewed. Experimenters were discouraged from making distracting conversation and told not to discuss the study or encourage conjectures by the subjects. On the completion of brewing, the subject was allowed to prepare the coffee as he/she would normally drink it using the artificial creamer, sugar, and sugar-substitute provided (the peanut flavor had been found to overwhelm cream and sugar). After initial sipping, the subject was given the main questionnaire (Appendix C) and allowed to complete it at his/her own pace. Continued drinking of the coffee was allowed (but neither encouraged nor discouraged) The final measure collected at this session was the selection of a gift sample of coffee,
220 Table 16 Coffee Preparation Instructions a. Positive discrepancy 1. Put filter in basket. 2. Add 4 teaspoons of coffee and slide basket into place. 3. Fill pot with water to bottom of metal band and pour into water intake opening. 4. Replace top on pot, place under filter basket and press switch into "on" position. Negative discrepancy 1. Put filter in basket. 2. Add 3 tablespoons of coffee and slide basket into place. 3. Fill pot with water to bottom of metal band and pour into water intake opening. 4. Replace top on pot, place under filter basket and press switch into "on" position.
221 that the experimenter recorded immediately on the questionnaire. At departure, subjects were given the gift sample, entered in the coffeemaker drawing (if this step had not been completed as part of a manipulation), and given a postage paid comment card that they were asked to send in at their leisure. Neither names nor identifying information were requested (some wrote this in anyway) but the cards were keyed to individual respondents by using the last four digits of the nine-digit zip code in the return address to record the appropriate subject number. One week subsequent to participation in the experiment, attempts were made to telephone each subject. The purpose of this was to administer a follow-up interview and debriefing statement (Appendix C) These attempts were by the principal investigator, with some assistance from one of the student experimenters. Any subject not contacted was mailed a written debriefing notice to ensure that the entire sample was disabused (Appendix C) These subjects could not be questioned regarding suspicions about the study. No subject mailed a reply card beyond the pre-set one week limit. 11 Measures Table 17 summarizes the measures that were collected in the study (the full text and actual format of the questionnaire are in Appendix C) The questions were derived from a variety of sources including the review of satisfaction measures that was depicted in Figure 4 as well as questions used in the Pilot Study reported above. 1 1 Those subjects who were sent written notice could conceivably have had time to respond before being told it was unnecessary. This did not occur.
222 The table is organized in a hierarchical fashion. That is, each successive grouping of major measures is considered to be more strongly representative of satisfaction as generally treated in the CS/D literature and as defined by the conceptualization presented in Chapter 3. The first two groups of measures for example, product evaluation and brand attitude measures, assess what are usually considered to be separate constructs. The measures in group 3, satisfaction scales, enjoy some degree of consensus among researchers but none would be a unanimous choice. Emotional measures, group 4, are presented herein as direct measures of the satisfaction construct Â— but this is an advancement not yet accepted by CS/D researchers. Finally, the measures in group 5 represent the behaviors in which CS/D researchers claim to have an interest (Cardozo 1965). The origin of the measures in each group is outlined below. Product evaluations. Brewed ground coffee was selected as the test product in part because of the paucity of salient attributes. As shown in Table 17 and Appendix C, subjects were asked to evaluate the prepared sample as to bitterness, strength of flavor, overall quality, darkness of color and overall taste. These scales were derived from those used by Olson and Dover (1976) and by Cohen and Goldberg (1970). 12 Although the scales purport to measure different dimensions of the coffee, it was expected that on analysis, these responses would 2 Mention should also be made of the semantic differential scale developed by Eastlack 1964. This scale was specifically designed for the evaluation of ground coffee but was unknown to the investigator at the time of the present study Â— and, apparently, was also unknown to Olson and Dover.
223 Table 17 Measure Summary Group 1 : Product Evaluations (measured on seven-point semantic differential styled scales) Bitter/Not Bitter Strong in Flavor/Weak in Flavor Poor Quality/Good QualityLight in Color/Dark in Color Good Tasting/Bad Tasting Group 2: Brand Attitude (measured on seven-point semantic differential styled scales) Good/Bad Unsatisfactory/Satisfactory Unfavorable /Favorable Worthless /Valuable Like/Dislike Disapprove /Approve Group 3: Satisfaction (adapted from Figure 4) "How satisfied were you with the coffee?": Very Dissatisfied, Somewhat Dissatisfied, Slightly Dissatisfied, Slightly Satisfied, Somewhat Satisfied, Very Satisfied (off-scale option: Not Satisfied nor Dissatisfied) "Knowing what you know now, what are the chances in ten (10) that you would choose to use the coffee again?": (No Chance) ... 10 (Certain). "Which word or phrase most closely reflects your satisfaction with the coffee?": Delighted, Pleased, Mostly Satisfied, Mixed, Mostly Dissatisfied, Unhappy, Terrible (off-scale option: Neutral, Never Thought About It) "Which face comes closest to expressing how you feel about the coffee?" (followed by Kunin's 1955 five equal interval faces) Oliver's (1980a) Likert-type scales: "I am satisfied about my choice of coffee. If I had it to do all over again, I would feel differently about choosing the coffee. My choice of coffee was a wise one. I feel bad about my decision regarding the coffee. I think that I did the right thing when I chose the coffee. I am not happy that I chose the coffee that I did.
224 Table 17 Â— continued Group 4: Emotion (instructions adapted from Izard 1977) "Sampling the coffee makes me feel:". Response categories were Not at all or Very Slightly, Slightly, Moderately, Considerably, Very Strongly for each of the 35 items. Group 5: Behavior & Behavioral Intent Likert type scales: "The next time I need coffee, I will consider buying the brand I tried (if available). The next time I serve coffee to guests, I would use the brand I tried." Actual behaviors: Brand of gift selected, word-of-mouth reported, usage rate in first week, service to other members of household, service to guests, postcard return with comments. Behavior intent: "If you had the opportunity, would you consider repurchasing the coffee? Have you tried to repurchase the coffee?" Group 6: Manipulation Check Discrepancy: "Overall, my expectations about the coffee were: Too high; It was worse than I thought Too low; It was better than I thought." (seven-point scale); The coffee is worse than I expected Better than I expected (seven-point semantic differential styled scales). Decision: "I had the option to choose the brand of coffee to evaluate." (seven-point Likert type scale). Response Involvement: "Evaluation of the coffee: Was interesting to me/Was not interesting to me, Was important to me/was not important to me, Was not difficult to evaluate/was difficult to evaluate, Took a lot of thought/Did not take a lot of thought, Task was very involving/Task was not very involving, I put a lot of thought into evaluation/I put no thought into evaluation." (seven-point semantic differential styled scales) Expectations: "Based on the information given to you at the beginning of the study, what did you expect about the coffee? Did you expect it to be: Very Bitter Verv not Bitter, Very Strong in Flavor Very Weak in Flavor, Very Poor Quality VeryGood Quality, Very Light in Color Very Dark in Color, Very Good Tasting Very Bad Tasting, Very Inexpensive Very Expensive." Perceptions: See Product Evaluation in Group 1.
225 reveal only one evaluative factor. That is, the relatively accelerated personal experience should enhance any naturally occurring halo effect. Brand attitude scales Attitude scales are at the same time more established yet more controversial than either perceptual evaluation or satisfaction scales. Having adopted a conceptual definition of attitude in Chapter 2, it remains necessary only to select measures consistent with that definition. Semantic differential scales that have been used in past research to assess overall attitude were chosen with endpoints labelled good-bad, worthless-valuable (e.g., Lutz 1975), satisfactory-unsatisfactory (e.g., Moore, Hausknecht and Thamodaran 1986), favorable-unfavorable, like-dislike, approvedisapprove (e.g., Fishbein and Ajzen 1975). Satisfaction scales A battery of satisfaction scales was assembled based on the results of the Pilot Study and other competitive tests of common satisfaction measures (Westbrook and Oliver 1981). Selected for inclusion were (1) a semantic differential ranging from Very Dissatisfied to Very Satisfied (Figure 4a, #7); (2) a scale based on chances in ten of re-using the product (Figure 4c, #31); (3) a categorical verbal scale from Delighted to Terrible (Figure 4b, #23); (4) a smiling/frowning face scale using Kunin's (1955) characterized faces selecting those numbered 1, 4, 6, 8, 10 to achieve approximately equal intervals (Kunin also tested circle faces similar to those depicted in Figure 4 b, #28); and (5) Oliver's (1980a) Likert scale adapted for the coffee choice (Figure 4a, #18). Emotion scales Since the emotional conceptualization of satisfaction is a major contribution of the dissertation, the selection
226 of emotion scales was the subject of much of the foregoing discussion. Izard's abbreviated DES (1977) was adapted by insertion of references to "the coffee you have now sampled." All subjects were given the same randomized order of adjectives for their response. The Pilot Study indicated that random starting positions create too much confusion for a self-administered questionnaire. Behavioral intention and behavior measures Likert scales referring to intent to consider buying or serving the coffee were found to be useful in the Pilot Study and were therefore used in the main study as well. Those respondents who were contacted by telephone were also asked if they would consider repurchasing the coffee (only a dichotomous yes/no response was recorded) A variety of actual behaviors (or self-reports of actual behaviors) were recorded: (1) gift brand chosen at the conclusion of the initial session (although not usually a dependent variable in CS/D studies, product selection is a common measure in consumer behavior studies; e.g., Cohen and Goldberg 1970; Gorn 1982; Wilson, Mathews and Harvey 1975); (2) self-reported usage (number of pots of coffee made with gift sample and whether served to others in or beyond the household); (3) self-reported wordof -mouth; (4) self-reported attempts to re-purchase; and (5) return ot a completed reply card (i.e., an actual compliment/complaint behavior). Manipulation check measures In order to claim confidence in the research design as a test of the conceptualization, it is useful to demonstrate that the appropriate constructs were manipulated as intended. The present study was concerned with three major manipulated constructs: Discrepancy, Decision and Response Involvement. As
227 discussed above, discrepancy was manipulated by separate operationalizations of Expectations and Perception; this is reflected in the manipulation checks for the construct. There were two direct assessments of perceived discrepancy: (1) a semantic differential scale describing the coffee as "Worse than I expected Better than I expected;" and (2) a semantic differential scale describing overall expectations about the coffee as "Too high; it was worse than I thought Too low; it was better than I thought." In addition, discrepancy can be derived from a comparison of each individual's expectations versus perception of the product. These are available as semantic differential scales evaluating the coffee's bitterness, strength of flavor, overall quality, darkness of color and overall taste. A scale measuring expectations of, "Inexpensive Expensive,' was also included as a check of that manipulation (since the advertisements referred to price differentials). The decision construct was defined as a behavior or activity. As designed in the present study, no real manipulation check is required since the subject either performed the task or did not. One Likert scale, however, assessed agreement with the statement, "I had the option to choose the brand of coffee to evaluate," as a check on the salience of the manipulation. 13 Response involvement is another construct that cannot be measured in a manner that would be endorsed by all researchers. One argument is 13 In the same section of the questionnaire were questions about reactions to the study itself Â— as checks on the informed consent instructions and the overall task experience. These are reported only briefly in Chapter 6.
228 that the presentation of the gift early in the session defined the focus of involvement and, as with decision, no check is required (see Petty, Cacioppo and Schumann 1 983--although in their study involvement was reflected in the expectation of a gift rather than determined by the presentation of an actual gift as in the present study) The viewpoint that response involvement is a measurable construct is represented by semantic differential scales regarding the evaluation of the coffee and whether that task was: interesting, important, difficult, requiring of thought, involving, generating of actual thought (Sherrell and Shimp 1982). It may be necessary to compare, within subject, involvement with evaluation of the coffee versus involvement with evaluation of the cof f eemaker 1 u Analysis The analysis of the experiment is based on the research hypotheses that were introduced in Chapter 4. Presented below are strategies for using the measures defined above to test each hypothesis or set of hypotheses Hypotheses 1a and 1b predict a three-way interaction among the manipulations, as they affect subjects' emotions. Specifically, in conditions in which response involvement is high and a decision is present, critical emotions are maximized (Figure 11a). When discrepancy is positive, this pattern results for measures of interest, u The majority of the measures just discussed can be matched to analogous scales relating to the coffeemaker. The purpose for including these was twofold: 1) the measures were required to maintain the mundane reality of evaluating both products; and 2) null results from the coffeemaker measures serve as "no manipulation" controls for most conditions.
229 joy and surprise. When discrepancy is negative, the emotions affected are anger, disgust and surprise. Empirically, the hypotheses are tested through analysis of variance using mean responses on the appropriate patterns of emotions as criteria, along with follow-up tests as required. An absence of response to the decision and response involvement manipulations, as measured by standard satisfaction and attitude scales, is predicted by hypotheses 2 and 3. As shown in Figure 11b, the only factor expected to affect these criteria is discrepancy. Thus, an analysis of variance is expected to show no significant interaction, only a main effect of discrepancy. 15 Behavior and behavioral intention effects are predicted by hypotheses 4 and 5. Hypotheses 4a and 4b predict that criterion behaviors reflect an interaction among all manipulated variables: discrepancy, decision and response involvement (Figure 11c). These behaviors were measured either directly or by self-report of actual behavior, as discussed above, and the hypotheses are examined by analysis of variance with follow-ups. Hypothesis 4c predicts a significant simple correlation between attitude toward the product (coffee) and intention to re-purchase (Figure 11d). This hypothesis is tested by Pearson correlation. Also tested by Pearson correlation (point-biserial) are the relationships reflected in hypothesis 5; that 15 The hypotheses do not suggest an equivalency of absolute scale response between satisfaction and attitude scales. Rather, it is a parallel in the pattern of responses showing insensitivity to the manipulations which is predicted.
230 Critical Emotion Strength / S Low involvement High involvement No decision (a) Predicted Emotional Responses Decision Satisfaction or Attitude Scales No decision Decision No decision Decision Positive discrepancy Negative discrepancy (b) Predicted Satisfaction and Attitude Scale Responses Figure 11 Hypothesized Relationships
Postcards returned or Persons spoken to 231 High involvement Low involvement No decision Decision (c) Predicted Behavioral Responses Repurchase intent Attitude (d) Predicted Correlation Between Attitude and Repurchase Figure ll--continued
232 Criterion Behavior Emotion Strength (e) Predicted Correlation Between Emotion Strength and Behavior Product Perception Scales Positive discrepancy Negative discrepancy No Decision Decision (f) Predicted Relationship Between Decision and Product Perception Figure ll--cont inued
233 Attitude Product Perception (g) Predicted Correlation Between Product Perception and Attitude Mean Emotion Strength Response Involvement (h) Predicted Correlation Between Response Involvement and Emotion Strength Figure 1 l--cont inued
234 is, that word-of-mouth or other critical behaviors are more likely as emotions are more intense (Figure 1 1 e ) Hypothesis 6 predicts that decision adds contrast to a perceptual discrepancy (Figure 11f). This relationship is evaluated through analysis of variance in product perception as determined by product quality (discrepancy manipulation) and decision; no interaction or main effect of response involvement is predicted. Follow-up tests of significant relationships are conducted. Perception, in turn, determines attitude toward the product as predicted in hypothesis 7. This relationship, depicted in Figure 11g, is tested through Pearson correlation using the scales derived for these constructs. Finally, hypothesis 8 predicts a correlation between achieved response involvement and resulting emotional strength. This is testable through either Pearson correlation between the response involvement scale and average response to emotion scales (Figure 11h) or a (bi-serial) correlation between manipulated response involvement and emotion. Summary This chapter has used the available literature, reviewed in Chapter 2, to derive and present an experiment to evaluate and demonstrate the conceptualization extended in Chapter 3. Specifically, analyses of data that address the hypotheses presented in Chapter 4 are proposed. Chapter 6 presents the results of these analyses as well as the outcomes from pretests of the experiment.
CHAPTER 6 STUDY RESULTS Introduction To this point, an extended conceptualization of satisfaction has been presented and an experiment to evaluate its structure has been discussed. In this chapter, the results of analysis of data from that experiment are reported for the hypotheses and analysis plan in preceding chapters. Presented first is a discussion of the pretests performed in order to fine tune the experimental design that was elaborated in Chapter 5. These were necessary to determine the procedures appropriate for achieving the desired manipulations while maintaining a high level of experimental reality. The main study results are divided into two sections: manipulation check results, and results of hypothesis tests. The manipulation check measures indicated that caution is appropriate in further interpretation of main study results. The chapter concludes with an integrative discussion of the findings. Following this, Chapter 7 presents implications for and modifications to the conceptualization. Pretest Findings The previous chapter alluded to pretests of the experiment. A variety of these were conducted as the manipulations and protocols were developed and modified. The goal was a design that effectively 235
236 manipulated the constructs of interest in a format that could be operationalized in a field setting as well as maintain credibility with the subjects. The basic logic that was followed required revisions with each administration until subject debriefing revealed acceptable procedures Initial drafts of advertising copy and the questionnaire were evaluated for clarity of meaning and logical flow by faculty and staff of the Marketing Department of the University of Akron. Individuals underwent the experimental procedure in order to identify problems with the logic, instructions, timing, or products. Revisions were based on expert and/or naive subject judgments made at each administration. On the basis of these tests, final advertising copy and approximate instructions were approved, administration time for the study was estimated at thirty minutes per subject, and the questionnaire was repeatedly revised. Participants in the pre-test indicated a decided preference for being allowed to add creamer and/or sweetener to the coffee before attempting to evaluate it. This was usually expressed as a desire to give a fair judgment against their normal brand. Since prior testing had suggested that such additions were not masking the peanut of f-f lavor 1 --and "normal" preparation had the potential to enhance evaluation of the unadulterated coffee--it was decided to allow this practice but not encourage it. 1 This was determined in conjunction with the choice of the 80:20 coffee:peanut ratio, discussed in Chapter 5. Addition of cream/ sweetener was allowed during the University of Florida taste tests in the event that such procedure proved necessary during the actual study.
237 The final major adjustment made as a result of pretesting was made to alleviate subject boredom. One major advantage of the design was that each subject was able to unseal and prepare his/her own coffee. However, even using the small coffeemaker, brewing required approximately five minutes. This time passed slowly for the participants who had been given nothing to do while the brewing took place. For this reason, the Classification Information form (Appendix E) was developed. The form served dual purposes as a method of collecting useful background data and as a study-related task to occupy subjects' time. The open-ended questions on the form that evolved provided adequate cushion to keep subjects interested throughout the brewing period without creating demands for additional time. Having designed the basics of the study, it was next necessary to evaluate the procedures in a setting that more nearly approximated the conditions which would be found in the actual field administration. For this purpose, students in a Marketing Research class were recruited to participate in field trials. 2 The students were naive to the purpose of the study and were told only that it was being conducted by another marketing instructor. On arrival, each was given the approximate cover storv (as appropriate) reportedin Appendix B. Since the purpose of this pretest was to evaluate the basic manipulations, only two conditions were utilized no decision, low involvement, and positive discrepancy versus decision, high involvement, and negative discrepancy. Those volunteers who were not 2 I would like to acknowledge the cooperation of Dr. J.E. Wilkinson in providing volunteers for this purpose.
238 coffee-drinkers served only to evaluate the finished advertisements. The study was conducted in a classroom (the only suitable location available) with each student participating individually (n = 16 at various times during one week). Responses to the questionnaire as well as individual debriefing were used to evaluate the study. As expected, the decision manipulation was the most straightforward to implement and to evaluate. Subjects in the decision condition perceived a freedom of choice in coffee whereas those in the no decision condition did not (means of 7.0 versus 2.5, respectively, on the manipulation check item on the questionnaire). As further reinforcement of the perceived freedom of decision, it should be noted that one subject in the decision condition chose the alternate brand (in this case, the store brand) to evaluate. Manipulation of disconf irmat ion required successful manipulation of both product perception and prior expectancies. Based on prior evaluations, final forms of the stimuli were used in this iteration. The advertisements designed to create positive expectations were viewed more positively than those designed to create negative expectations (means of 5.9 versus 4.4, respectively on averaged expectations scales). Measurement of expectations occurred either immediately after exposure to the advertisement or after some delay. The delay did not appear to affect expectations (means of 5.3 versus 5.6, respectively). Debriefing of some subjects suggested that the "negative advertisements were too negative for the instructions that were given. The instructions were modified (as they appear now in Appendix 3) to better coincide with the tone of the advertisements.
239 This pretest was the first opportunity for naive respondents to assess product perception (i.e., of the coffee) based on preparation with the coffeemakers that were purchased for the main study. After some variation in formulation, the recipes shown in Table 14 were accepted as providing suitably distinct quality levels. In particular, the positive discrepancy preparation required only slightly more coffee than recommended by the coffee label (in order to overcome a perception of weakness). The negative discrepancy preparation was not perceived to be stronger but was made so that both bitterness and peanut flavor were evident. Since subjects were relatively familiar with drip coffeemakers but not with coffeemakers of this size and capacity (i.e., four cup), they found the instructions to be fairly simple to follow and the coffee amounts used unremarkable. This pretest also provided an opportunity to assess and modify the involvement manipulation. On debriefing, subjects were asked about the experiment's focus and their interest in the products being evaluated. Initial results suggested an inadequate divergence between conditions. Subsequently, the protocols were adjusted to more closely approximate the involvement manipulation used by Petty, Cacioppo, and Schumann (1983). For example, the presentation of a gift (i.e., the coffee sample or the coffeemaker drawing card) was moved to an earlier point in the experiment to make the presentation a salient part of the manipulation. These adjustments led to procedures described previously (and detailed in Appendix B) that were found to be adequate to direct attention to the product meant to engender higher involvement (i.e., either the coffee or the coffeemaker).
240 The student subjects in this pretest were also asked to evaluate the full experimental situation for realism and any suspicion-arousing or off-putting aspects. Overall, they found the cover story to be credible and accepted the claim that this was a test of a product evaluation situation for a food-product company. None admitted to concern about the identity of the particular manufacturer. When asked, none felt that the coffeemaker (i.e., the machine itself) was responsible for the flavor of the coffee that was prepared. Some were intrigued by the "novelty" of the smaller machine but agreed that it was standard in appearance and operation. The final pretest was to serve as a check of the experiment under field operating conditions. Thus, this was conducted at the location selected for the experiment just prior to the days reserved for main study. Subjects (n = 7) were similar to those who would be available for the experiment and were entirely naive to the nature of the pretest. The pretest was conducted by the principal investigator and the student experimenter charged with supervising the study. J The major purpose of this pretest was verification rather than modification. Debriefing of subjects showed that the stated industrypurpose was believed--although it was necessary to claim that the experimenters were working to raise funds for a student group. The involvement manipulation (i.e., presenting the gift earlier) was 3 I would like to acknowledge the effort and suggestions by the late Mike Albaugh. In particular, he was successful at recruiting and motivating subjects. Mike also provided a portable sound system which was used to play background music during the study. This strategy, combined with the use of a room divider enabled us to conduct two concurrent experiments--! e one positive and one negative disconf irmation condition--in the same location.
241 successful at directing attention, although it was necessary to slightly reword the measures that were to be used to check the manipulation in the main study (i.e., in Appendix E the questions regarding evaluation of the cof fee-cof f eemaker originally referred to choice Â— Sherrell and Shimp 1982). Quality perceptions of the coffee also appeared consistent with the design. Subjects admitted to no suspicions regarding the manufacturers or any purposes of the study other than that explained. Some were curious about the purpose of the emotional measurement page--Izard s Differential Emotions Scale Â— but were content with experimenters' claimed ignorance. Finally, this pretest suggested that recruiting coffee-drinkers was to be more difficult than anticipated (as described in Chapter 5, this eventually resulted in a less stringent selection procedure). Main Study Results Manipulation Check Measures Decision The check on the decision manipulation was straightforward. There was a large and significant difference between decision conditions in agreement with a Likert-type item stating, "I had the option to choose the brand of coffee to evaluate." Mean agreement for subjects allowed to make a decision was 6.6 out of 7; those not making a decision responded with a mean agreement of 2.6. The manipulation explained 63% of the variance in the responses to the item, using Hays' omega-squared formulation (Keppel 1982). A statistically significant effect of the response involvement manipulation (for low versus high involvement conditions, means were 4.9 versus 4.3, respectively) explained less than one percent of the
242 variance on the manipulation check measure and is not taken as damaging to the manipulation (Table 18). Other measures provide reaffirmation of the success of the decision manipulation. A similar Likert-type item that dealt with the freedom to choose the coffeemaker to evaluate showed no significant effects of manipulations (F 7|13 6 < I, Table 19) and a mean rating of 2.2 out of 7 (choice of coffeemaker did not, in fact, vary among conditions). Also reassuring is the fact that, out of a total of 108 subjects who were offered a choice, five subjects chose to evaluate the brand proffered as the alternative to the company's "own" test brand (i.e., the one described verbally and in the advertisements). Data from these subjects were omitted from analyses of post-purchase and post-trial responses. The fact that these choices occurred, however, strongly argues that choice of the test product was in no way forced by the design. Discrepancy Two scales directly assessing the perception of a discrepancy between expectations and achieved quality were discussed in Chapter 5. For the evaluation of the discrepancy in information received regarding the target coffee product, the two seven-point scales were averaged ("Coffee was worse/better than I expected" and "My expectations were too high/low; It was worse/better than I thought, Cronbach's alpha = .79, Table 20). Analogous measures regarding perception of the coffeemaker were collected as a further check. These did not correlate as well and were analyzed separately (Cronbach's alpha = .57, Table 21). Using the combined scale as a criterion for the coffee perceptual discrepancy, the manipulation was predicted to generate lower scores
243 Table 18 ANOVA Summary of Decision
244 Table 19 ANOVA Summary of Coffeemaker Decision Source df Sums of Squares F omegasquared Model 7 11.49 0.53 Discrepancy (DI) 1 0.53 0.17 Decision (DE) 1 3.25 1 .04 Involvement (I) 1 0.59 0.19 DE x DI 1 0.72 0.23 I x DI 1 1 .69 0.54 I x DE 1 3.23 1 .04 I x DE x DI 1 1 .30 0.42 Residual 186 581.20 Cell Means (variance) [n] 1. Low Involvement, No Decision, Positive Discrepancy 2.56 (3.17)  2. High Involvement, No Decision, Positive Discrepancy 1.84 (2.06)  3. Low Involvement, No Decision, Negative Discrepancy 2.43 (4.35)  4. High Involvement, No Decision, Negative Discrepancy 2.42 (3.04)  5. Low Involvement, Decision, Positive Discrepancy 2.00 (2.75)  6. High Involvement, Decision, Positive Discrepancy 2.13 (2.55)  7. Low Involvement, Decision, Negative Discrepancy 1.96 (3.37)  8. High Involvement, Decision, Negative Discrepancy 2.13 (3.85)  Main Effect Means Low Involvement 2.23 No Decision 2.31 Positive Discrepancy 2.13 High Involvement 2.13 Decision 2.05 Negative Discrepancy 2.23 Note: Based on responses to a Likert-type scale on which 1 = Strongly Disagree and 7 = Strongly Agree with the statement, "I had the option to choose the model of coffeemaker to evaluate." Significance of results is indicated by: a = p<.0001; b = p< .01 ; c = p<.05; d = p< 1
245 Table 20 Selection of Coffee Discrepancy Scale a) Factor Analysis Results 3 Items Factor 1 Factor 2 1 Coffee is not bitter .69 2 Coffee is strong -87 3 Coffee is quality .78 4 Coffee is dark .32 5 Coffee is good .81 6 Coffee is better .86 7 Expectations too low .71 variance explained 41% 19% / 6C b) Cronbach's Alpha Computation 13 Perceived Discrepancy = mean of (item 6, item 7). Pearson Correlation Coefficient: Item 6, item 7 .65 a = .79 Note : a Principal component analysis using varimax rotation b Using formulation in Carmines and Zeller 1979
246 Table 21 Selection of Coffeemaker Combined Discrepancy Scale a) Factor Analysis Results 3 Items Factor 1 Factor 2 1 Coffeemaker is simple to use -75 2 Coffeemaker is fast .74 3 Coffeemaker is simple to fill .70 4 Coffeemaker is attractive .70 5 Coffeemaker is better .71 5 Expectations too low .56 variance explained 31% 23%/ 54% b) Cronbach's Alpha Computation' Combined Discrepancy = mean of (item 5, item 6). Pearson Correlation Coefficient: Item 5, item 6 = .402 a = .57 Note: a Principal component analysis using vanmax rotation b Using formulation in Carmines and Zeller 1979
247 in the negative discrepancy conditions than in the positive discrepancy conditions. Mean responses on the coffee scale of 4.0 for negative discrepancy and 4.7 for positive discrepancy were significantly different in the appropriate direction. However, the manipulation explained only about 5% of the variance (Table 22). Analyses of the same scales, as applied to the coffeemaker evaluation, revealed no significant omnibus F test for either the combined scale (Table 23) or the two separate scales (Tables 24, 25). Significant differences associated with the response involvement manipulation were found on the combined scale and apparently resulted from differences on the expectations scale. These differences were not expected and, in light of the non-significant omnibus test and the low variance explained, are not examined further (Tables 23, 24). The processes underlying these results on the discrepancy measures may be better examined by separate inspections of expectations about and perceptions of the products. Expectations about the coffee were measured using six 7-point semantic differential scales using the important coffee attributes derived previously: bitter/not bitter, strong/weak in flavor, poor/good quality, light/dark in color, good/bad tasting, and inexpensive/expensive (Olson and Dover 1976). A principal components factor analysis, with varimax rotation, yielded only two dimensions for these scales (Table 26). The bitterness expectation scale showed no effect of the manipulations (F184 < 1, Table 27). The mean response to this scale was 4.56, or "Slightly Not Bitter." Since subjects in all conditions were promised coffee that would be of at least "acceptable"
Table 22 ANOVA Summary of Coffee Discrepancy 248 Source df Sums of Squares omegasquared Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual 31 .10
249 Table 23 ANOVA Summary of Coffeemaker Combined Discrepancy Source df Sums of Squares Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual 5.83 0.10 0.19 4.55 0.00 0.18 0.79 0.01 236.34 Cell Means (variance) [n] 0.66 0.08 15 6 2d 0.00 0.14 0.63 0.01 omegasquared 01 1. Low Involvement, No Decision, Positive Discrepancy 4.66 (1.49)  2. High Involvement, No Decision, Positive Discrepancy 4.18 (1.60)  3. Low Involvement, No Decision, Negative Discrepancy 4.57 (1.12)  4. High Involvement, No Decision, 5. Low Involvement, Decision, 6. High Involvement, Decision, 7. Low Involvement, Decision, 3. High Involvement, Decision, Negative Discrepancy 4.18 (0.46) [ 2 5 j Positive Discrepancy 4.48 (1.64)  Positive Discrepancy 4.23 (1.09)  Negative Discrepancy 4.37 (1.45)  Negative Discrepancy 4.26 (1.18)  Main Effect Means Low Involvement 4.52 High Involvement 4.21 No Decision 4.39 Positive Discrepancy 4.39 Decision 4.34 Negative Discrepancy 4.34 Note: Analysis of mean responses on combined seven-point scales describing the coffeemaker as 1 = "Worse than I expected" to 7 = "Better than I expected"; expectations about the coffeemaker as 1 = "Too high; It was worse than I thought" to 7 = "Too low; it was better than I thought." Results are for a combined scale representing the average response, Cronbach's alpha = .57. Significance of results is indicated by: a = p< .0001 ; b = p< 1 ; c = p<.05; d = p< 1
Table 24 ANOVA Summary of Coffeemaker Discrepancy 250 Source df Sums of Squares omegasquared Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual 1 1 I86 14.83
Table 25 ANOVA Summary of Coffeemaker Expectancy Disconf irmation 251 Source Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual df Sums of Squares 185 5.08 0.89 1 .46 0.35 0.03 0.21 0.73 1 .34 289.00 omegasquared 0.46 0.57 0.94 0.22 0.02 0.13 0.47 0.86 Cell Means (variance) [n] 1. Low Involvement, No Decision, Positive Discrepancy 5.04 2. High Involvement, No Decision, Positive Discrepancy 4.60 3. Low Involvement, No Decision, Negative Discrepancy 4.70 4. High Involvement, No Decision, Negative Discrepancy 4.72 5. Low Involvement, Decision, Positive Discrepancy 4.60 6. High Involvement, Decision, Positive Discrepancy 4.74 7. Low Involvement, Decision, Negative Discrepancy 4.54 3. High Involvement, Decision, Negative Discrepancy 4.48 Main Effect Means ,62) ,25) ,13) ,21) 42) ,84)       39)  62)  Low Involvement 4.72 High Involvement 4.64 No Decision 4.77 Positive Discrepancy 4.74 Decision 4.59 Negative Discrepancy 4.61 Note: Based on responses to a semantic differential scale describing the coffeemaker as 1 = "worst than I expected" to 7 = "Better than I expected". Significance of results is indicated by: a = p< 0001 ; b = p< 01 ; c = p<.05; d = p< 1
252 Table 26 Selection of Coffee Expectations Scale a) Factor Analysis Results 3 Items Factor 1 Factor 2 1 Coffee is not bitter .93 2 Coffee is strong .76 3 Coffee is quality .68 4 Coffee is dark 77 5 Coffee is good tasting .67 6 Coffee is expensive .54 variance explained 39% 23% / 62% b) Cronbach's Alpha Computation Coffee expectations = mean of (items 1,2,3,4,5,6). Pearson Correlation Coefficients: Item 2 Item 3 Item 4 Item 5 .324 .422 .553 .398 .487 .466 .119 .396 .343 .210 = .73 Note: a Principal component analysis using varimax rotation b Subjects were asked what they expected the coffee to be. c Using formulation in Carmines and Zeller 1979
Table 27 ANOVA Summary of Coffee Bitterness Expectations 253 Source df Sums of Squares omegasquared Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DF x DI Residual 17.68 3.13 0.57 184 1 0, 5, 6, 0. 509. ,96 ,00 .64 38 72 69 0.91 1 .13 0.21 0.71 0.00 2.04 2.30 26 Cell Means (variance) [n] Low Involvement, No Dec is High Involvement, No Decis Low Involvement, ?Jo Decis High Involvement, No Decis Low Involvement, Decision High Involvement, Decision Low Involvement, Decision High Involvement, Decision on, on, on, on, Positive Discrepancy 4.54 (3.30)  Positive Discrepancy 4.44 (2.671  Negative Discrepancy 5.27 (2.49; [22; Negative Discrepancy 4.24 (2.36) l 2j] Positive Discrepancy 4.20 (2.92)  Positive Discrepancy 4.58 (3.56)  Negative Discrepancy 4.67 (2.23)  Negative Discrepancy 4.61 (2.61)  Main Effect Means Low Involvement 4.65 High Involvement 4.46 No Decision 4.60 Positive Discrepancy 4.44 Decision 4.51 Negative Discrepancy 4.68 Note : Analysis of seven-point scale describing expectations about the coffee as 1 = Very Bitter to 7 = Very Not Bitter. Significance of results is indicated by: a = p<.0001; b = p< 01 ; c p<.05; d = p< 1
254 quality, it seems reasonable that none would expect bitterness. Apparently those subjects in the negative discrepancy (high expectations) conditions did not attempt to discriminate degrees of "not bitter." A composite measure of the remaining coffee expectation scales was used as the major indicant of expectations (Cronbach's alpha = .73, Table 26). The advertisements and product information given verbally to the subjects was essentially unidimensional with respect to product attributes. That is, no complex relationships among good and bad values were suggested. Thus, the fact that a negatively valenced attribute, bitterness, was the only one perceived differently can be taken as reassuring evidence that the subjects were attending closely to the task. Measures of expectations regarding the coffeemaker were generated and presented in a similar fashion. All four scales--dif f icult/simple to use, fast/slow, difficult/simple to fill, unattractive/attractive-ioaded on one factor in a principal component analysis and were treated as measures of one gestalt expectation (Cronbach's alpha = .69, Table 28). As with the preliminary information regarding the coffee, the advertisements and instructions regarding the coffeemaker did not provide any discrepant attribute information. It should be noted that the type and quantity of information provided to the subjects regarding each product were consistent with what might occur in the marketplace. The responses to the measures discussed above suggest that a single expectation level is generated
255 Table 28 Selection of Coff eemaker Expectations Scale a) Factor Analysis Results 3 Items' 3 Factor 1 1 Coffeemaker is simple to use 70 2 Coffeemaker is fast .80 3 Coffeemaker is simple to fill .77 4 Coffeemaker is attractive .60 variance explained 52% b) Cronbach's Alpha Computation Coffeemaker Expectations = mean of (items 1-4) Pearson Correlation Coefficients: Item 1 Item 2 Item 3 Item 2 .368 Item 3 .322 .589 Item 4 .jo5 .261 .236 a = .69 Note: a Unrotated principal component analysis
256 from which more specific attribute expectations can be derived. Thus, absent any discriminating or discrepant information, the subjects in the present study reported highly correlated expectations. The only exception to this pattern was a scale that may have been created with an inadvertently restricted range. Substantive analyses of the composite scales for coffee and coffeemaker expectations produced results consistent with the desired manipulations. Subjects in negative discrepancy conditions reported much higher expectations regarding the coffee, mean score = 5.57, than did those in positive discrepancy conditions, mean score = 4.10 (p< 0001 Table 29). This difference explained approximately onethird of the overall variance in responses to the composite scale, using Hays' omega-squared criterion. An interaction between discrepancy manipulation and decision availability accounted for less than one percent of the variance and does not qualify the main effect. The corresponding analysis of the coffeemaker expectation scale was predicted to show no effect of experimental condition. The nonsignificant omnibus F test is adequate evidence of this F 7)1 36 = 1.30, p< 25 Table 30). As discussed in detail in Chapter 5, only positive information was given for the coffeemaker, so the mean rating of its expected quality of 5.80 on the 7-point scale lends credence to the validity of the procedures and measures. A statistically significant difference in mean expectations between the no-decision (5.99) and decision (5.61) conditions explained only about 3% of the variance. Since this was not predicted and is associated with a non-significant omnibus test, it is best considered spurious in this study.
257 Table 29 ANOVA Summary of Composite Coffee Expectations Source Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual Low Involvement, High Involvement, Low Involvement, High Involvement, Low Involvement, High Involvement, Low Involvement, High Involvement, df Sums of Squares F omegasquared 86 109.32 104.89 0.10 0.02 3.23 0.23 1 .01 0.07 203.59 14.27^ 95. 83* 0.09 0.02 2.95^ 0.21 0.93 0.07 331 .007 Cell Means (variance) [n. No Decis No Decis No Decis No Decis Decision Decision Decision Decision Positive Discrepancy 4.27 (1.60)  Positive Discrepancy 4.24 (1.44)  Negative Discrepancy 5.49 (0.66)  Negative Discrepancy 5.13 (0.70)  Positive Discrepancy 3.95 (1.17)  Positive Discrepancy 4.17 (0.78)  Negative Discrepancy 5.49 (0.73)  Negative Discrepancy 5.59 (0.71)  Main Effect Means Low Involvement 4.82 High Involvement 4.83 No Decision Decision 4.80 4.85 Positive Discrepancy 4.10 Negative Discrepancy 5.57 Note: Analysis of mean responses on combined seven-point scales describing the expected quality of the coffee as: weak/strong in flavor, poor/good quality, light/dark in^ color, bad/good tasting, inexpensive/expensive, and assigned values of one/seven respectively. Results are for a combined scale representing mean response across the scales, Cronbach's alpha = .73. Significance of results is indicated by: a = p<.0001; b = p< 1 ; c = p<.05; d = p<.10.
253 Table 30 ANOVA Summary of Coffeemaker Expectations Source df Sums of Squares omegasquared Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual 186 Low Involvement, High Involvement, Low Involvement, High Involvement, Low Involvement, High Involvement, Low Involvement, High Involvement, 9.91 0.85 7.23 0.40 0.17 0.01 0.60 0.79 1.30 0.78 6.65* 0.37 0.16 0.01 0.56 0.72 ,03 202.16 Cell Means (variance) [n] No Decision, Positive Discrepancy 5.99 (0.97)  No Decision, Positive Discrepancy 5.93 (1.23)  No Decision, Negative Discrepancy 6.21 (0.49)  No Decision, Negative Discrepancy 5.86 (0.98)  Decision, Positive Discrepancy 5.56 (0.90)  Decision, Positive Discrepancy 5.47 (1.99)  Decision, Negative Discrepancy 5.64 (1.22)  Decision, Negative Discrepancy 5.77 (0.87)  Main Effect Means Low Involvement 5.84 High Involvement 5.76 No Decision Decision 5.99 5.61 Positive Discrepancy 5.74 Negative Discrepancy 5.86 Note : Analysis of mean responses on combined seven-point scales describing the expected quality of the coffeemaker as very difficult/very simple to use, very slow/very fast, very difficult/very simple to fill and unattractive/attractive. Results are reported for a combined scale representing the mean of the four items, Cronbach's alpha = .69. Significance of results is indicated by: a = p<.0001; b = p<.01; c = p<.05; d = p<. 10.
259 The apparent weakness in the discrepancy manipulation rests in the subjects' perception of product quality (possibly due to individual differences in taste discrimination Â— Buchanan, Givon and Goldman 1987). Measures of perception did not reveal the simple main effect that was predicted based on objective quality evaluations of the target products. Detailed analyses of product perception results suggested that individual differences in flavor perception sensitivity were responsible for the deviation from expectations. 4 The questionnaire was constructed with the assumption that the five coffee perception scales could be averaged to serve as the sole measure of coffee perception. 5 Similarly, the coffeemaker perception was expected to be unidimensional ; enabling the use of means across four scales. Principal component factor analyses, using a varimax rotation, did not return these patterns. Rather, the coffee scales seemed to describe three factors that were labeled as follows for convenience: Flavor (mean of bitter/not bitter and strong/weak in flavor scales); Quality (mean of poor/good quality and bad/good tasting scales); and Color (light/dark in color scale) (Table 31). The coffeemaker perceptions also appeared to have been the result of separable evaluations: Speed (slow/fast scale); Fill (difficult/simple 4 Analyses of satisfaction and attitude scales, reported below, do show a similarly weak but significant effect of the discrepancy manipulation. These provide some assurance that the desired effect is present, albeit either not as strong or as sensitively measured as would have been preferred. 5 No companion measure to the inexpensive/expensive expectation scale was attempted since no additional information was made available to the subjects regarding this attribute. The "worse/better than I expected" item which appeared in the same section as the perception scales, Appendix C, was discussed above.
260 Table 31 Selection of Coffee Perception Scales a) Factor Analysis Results 3 Items Factor 1 Factor 2 1 Coffee is not bitter .59 2 Coffee has weak flavor -87 3 Coffee is good quality .78 4 Coffee is dark color -.32 5 Coffee is good tasting .81 6 Coffee is better than b expected .86 7 Expectations too low b .78 variance explained 41% 1 9%/ 60% b) Cronbach's Alpha Computation Flavor = mean of (item 1, item 2) Quality = mean of (item 3, item 5) Color d = item 4 Pearson Correlation Coefficients: Item 1 item 2 = .322; a = .49 Item 3, item 5 = .500; a = .67 Note: a Principal component analysis using varimax rotation b These items weren't designed to be part of the perception scale, included as a check of confounding between perception and disconf irmation. c Using formulation in Carmines and Zeller 1979. d This item correlated too weakly to be included in either scale.
261 to fill scale); Simple (difficult/simple to use scale); and Attractiveness (unattractive/attractive scale) (Table 32). Among the three coffee perception scales, two did not appear responsive to experimental manipulations. Neither Quality (F 7)188 = 1.55) nor Color (F 7jl84 = 0.93) demonstrated effects that were significantly explained by the ANOVA model (Tables 33 and 34, respectively) Effects within each model (a response involvement by discrepancy interaction for coffee quality and a response involvement main effect for coffee color) did reach statistical significance, but were relatively weak and should not be given a great deal of credence in light of non-significant omnibus F-tests. 6 The Flavor scale did exhibit significant effects of the manipulations. Specifically, the coffee in the positive discrepancy conditions was perceived to be somewhat weaker than that in the negative discrepancy conditions. This main effect is qualified by a significant interaction between decision making and discrepancy. When the subject did not decide on the coffee to sample, the difference in evaluation of flavor strength (M = 4.13 for positive discrepancy, M = 3.95 for negative discrepancy) is not significant. Having decided on the brand of coffee to sample, subjects rated the coffee in the 5 It is interesting to note the apparent effects for future investigation. The interaction seemed to indicate differential discrimination in perception depending on focus of involvement. Under "high" involvement, the coffee was perceived to be of higher quality ir the positive discrepancy conditions, M = 5.3 than in the negative discrepancy conditions, M = 4.6. On the other hand, under "low" involvement subjects in the positive discrepancy conditions perceived the coffee to be of no better quality M = 4.92 than did subjects in the negative discrepancy conditions M = 5.24, the latter means are not significantly different at a = .05 using Tukey's Honestly Significant Difference test.
262 Table 32 Correlations Among Coffeemaker Perception Scales Items 1 Simple to use 2 Fast 3 Simple to fill 4 Attractive Pearson Correlation coefficients: Item 1 Item 2 Item 3 Item 2 .053 Item 3 .278 .358 Item 4 .163 .039 .073 Note: Each scale was analyzed and is reported separately due to low correlations
Table 33 ANOVA Summary of Coffee Quality Perception 263 Source df Sums of Squares Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual Cell Means (variance) [n. 18
Table 34 ANOVA Summary of Coffee Color Perception 264 Source df Sums of Squares omegajquared Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual 184 1 .87
265 positive discrepancy conditions as weaker in flavor, M = 4.32, than did subjects rating the negative discrepancy product, M = 3.38 (using Tukey's Honestly Significant Difference test a = .05). Note, however, that this interaction accounts for only about 2% of the variance in the results (Table 35). Because the correlation of the two measures used for the Flavor scale was so low (Cronbach's a = .49), the individual scales were also examined separately. Alone, the bitterness scale did not respond to the manipulations (F 7)186 = 1.69, p< 1 1 Table 36). The apparent perception of the peanut blend as more bitter, mean = 4.29, than the packaged coffee, mean = 4.86 was appropriate Â— even if the omnibus test did not reach a conventional significance level. The perception of flavor strength was more complex; the three-way interaction among the manipulations is depicted in Figure 12. The interaction explains only about one percent of the variance, but qualifies the interaction between decision and discrepancy as well as the discrepancy main effect (Table 37). For the positive discrepancy conditions, only the low involvement, no decision group perceived the coffee to be stronger, mean 2.96, than did the other subjects (based on Tukey's Honestly Significant Difference text, a = .05). In the negative discrepancy conditions, the low involvement, no decision group was again the outlier. In this case, however, the blend was rated less strong, mean = 3.64, by those in the low involvement, no decision condition (the high involvement, no decision mean, 3.2, was not significantly different from either 3.64 or its high involv.ement, decision counterpart, 2.78).
Coffee strength Very 7.0 r Weak 266 6.0 5.0 4.0 3.0 Very 2.0 Strong 1.0 0.0 Low involvement High involvement J L No decision Decision No decision Decision Positive discrepancy Negative discrepancy Figure 12 Perceptions of Coffee Strength Interaction Among Experimental Conditions
Table 35 ANOVA Summary of Coffee Flavor Perception 257 Source df Sums of Squares omegasquared Model Discrepancy (DI) Decision (DE) Involvement DE x DI I x DI I x DE I x DE x Residual (I) DI 32.56
Table 36 ANOVA Summary of Coffee Bitterness Scale 2b8 Source df Sums of Squares omegasquared Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual 7 1 1 1 1 1 1 1 18b 34.27 15.13 49 ,02 33 4.92 0.47 3.33 539.07 1 .69 5.22= 1 .20 1 .73 0.82 1 .70 0.16 1.15 ,02 Cell Means (variance) [n] Low Involvement, No Decision, Positive Discrepancy 4.80 (3.33)  High Involvement, No Decision, Positive Discrepancy 4.96 (3.12)  Low Involvement, No Decision, Negative Discrepancy 5.04 (2.86)  High Involvement, No Decision, Negative Discrepancy 4.04 (2.48)  Low Involvement, Decision, Positive Discrepancy 4.92 (2.95)  High Involvement, Decision, Positive Discrepancy 4.75 (2.72)  Low Involvement, Decision, Negative Discrepancy 4.19 (2.64)  High Involvement, Decision, Negative Discrepancy 3.91 (3.08)  Lew Involvement 4.72 High Involvement 4.43 Main Effect Means No Decision 4.71 Positive Discrepancy 4.86 Decision 4.44 Negative Discrepancy 4.29 Note : Analysis of a scale describing the coffee as 1 = Very Bitter to 7 = Very Not Bitter. Significance of results is indicated by: a = p<.0001; b = p< 01 ; c = p<.05; d = p< 1
Table 37 ANOVA Summary of Coffee Flavor (single) Scale 269 Source df Sums of Squares omegasquared Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual 187 44.54 14.65 0.85 0.49 16.76 2.68 0.50 7.91 426.46 Cell Means (variance) [n] 2.79b 6.42 = 0.37 0.21 7.35^ 1 18 0.22 3.47^ .026 .031 .012 Low Involvement, No Decision, Positive Discrepancy 2.96 (2.21)  High Involvement, No Decision, Positive Discrepancy 3.80 (3.33)  Low Involvement, No Decision, Negative Discrepancy 3.64 (1.86)  High Involvement, No Decision, Negative Discrepancy 3.20 (1.67)  Low Involvement, Decision, Positive Discrepancy 3.92 (2.74)  High Involvement, Decision, Positive Discrepancy 3.75 (2.11)  Low Involvement, Decision, Negative Discrepancy 2.62 (1.45)  High Involvement, Decision, Negative Discrepancy 2.78 (2.91)  Main Effect Means Low Involvement 3.27 High Involvement 3.39 No Decision Decision 3.39 Positive Discrepancy 3.61 3.27 Negative Discrepancy 3.04 Note : Analysis of a scale describing the coffee as 1 = Very Strong in Flavor to 7 Very Weak in Flavor. Significance of results is indicated by: a = p< 0001 ; b = p< 01 ; c = p<.05; d = p<. 10.
270 It is interesting that the group that was assigned a "noncritical" test product perceived it differently in each case. This could be due to a number of processes: a stronger assimilation effect; greater attribution to the coffeemaker, and the like. While it is tempting to speculate on such causes, the fact that the pattern emerges for one measure and with relatively small effect sizes elicits caution. It is sufficient to note that the general pattern of flavor strength ratings was consistent with the product formulation manipulation. Among the coffeemaker perception scales (no correlation was high enough to justify combining items, Table 32), only the one showing evaluations of speed revealed significant effects explainable by the experimental manipulations (Tables 38, 39, 40, 41). Although the variance explained was small, each manipulation demonstrated a significant main effect: discrepancy, 2% of variance, decision, 2% of variance, and involvement 1% of variance. Speed was perceived to be higher (coffeemaker was faster) in the positive discrepancy, M = 5.37, no decision, M = 5.36, and high response involvement (with coffee), M = 5.31 conditions than in the negative discrepancy, M = 4.75, decision present, M = 4.78, and low involvement, M = 4.83, conditions. No manipulation effects were hypothesized, but speculative interpretations of these data may be borne out in future studies. The involvement effect might signal some effectiveness of that manipulation Â— subjects were more critical when the coffeemaker was the target product. The difference in beverage formulation may have affected the actual speed and therefore perception of speed between discrepancy conditions--
Table 38 ANOVA Summary of Coffeemaker Speed Perception 271 Source df Sums of Squares omegasquared Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual 49, 18. 15, 10.47 0.37 2.91 0.75 0.00 >32 35 2.11<= 0.02 c 67 = 11 d 11 87 22 0.00 ,02 .02 ,01 Cell Means (variance) [n] Low Involvement, No Decision, Positive Discrepancy 5.28 (4.13)  High Involvement, No Decision, Positive Discrepancy 6.12 (1.36)  Low Involvement, No Decision, Negative Discrepancy 4.83 (4.51)  High Involvement, No Decision, Negative Discrepancy 5.16 (3.47)  Low Involvement, Decision, Positive Discrepancy 4.76 (3.52)  High Involvement, Decision, Positive Discrepancy 5.33 (2.41)  Low Involvement, Decision, Negative Discrepancy 4.46 (3.46)  High Involvement, Decision, Negative Discrepancy 4.57 (4.17)  Main Effect Means Low Involvement 4.83 High Involvement 5.31 No Decision 5.36 Positive Discrepancy 5.37 Decision 4.78 Negative Discrepancy 4.75 Note : Analysis of responses to a seven-point scale describing the coffeemaker as 1 = "Very Slow" to 7 = "Very Fast". Significance of results is indicated by: a = p< 000 1 ; b = p< 01 ; c = p<.05; d = p< 10.
Table 39 ANOVA Summary of Coffeemaker Filling Perception 272 Source df Sums of Squares omegasquared Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual ,89 ,58 ,11 ,65 ,73 09 ,13 187 0.98 542.03 0.57 1 .68 0.25 0.14 0.83 0.91 0.03 0.22 Cell Means (variance) [n] 1. Low Involvement, No Decision, Positive Discrepancy 4.76 (5.94)  2. High Involvement, No Decision, Positive Discrepancy 5.36 (5.41)  3. Low Involvement, No Decision, Negative Discrepancy 5.86 (4.31)  4. High Involvement, No Decision, Negative Discrepancy 5.60 (3.42)  5. Low Involvement, Decision, Positive Discrepancy 5.08 (4.91)  6. High Involvement, Decision, Positive Discrepancy 5.29 (4.22)  7. Low Involvement, Decision, Negative Discrepancy 5.35 (3.36)  8. High Involvement, Decision, Negative Discrepancy 5.26 (4.47)  Main Effect Means Low Involvement 5.24 High Involvement 5.38 No Decision 5.38 Positive Discrepancy 5.12 Decision 5.24 Negative Discrepancy 5.51 Note: Analysis of responses to a seven-point scale describing the coffeemaker as 1 = "Very Difficult to Fill" to 7 = "Very Simple to Fill". Significance of results is indicated by: i = p<.0001; b = p< 01 ; c = p<.05; d = p< 1
273 Table 40 ANOVA Summary of Coffeemaker Use Perception Source df Sums of Squares F omegasquared Model 7 8.90 0.65 Discrepancy (DI) 1 0.21 0.11 Decision (DE) 1 3.80 1 .94 Involvement (I) 1 0.02 0.01 DE x DI 1 1.93 0.99 I x DI 1 0.02 0.01 I x DE 1 0.06 0.03 I x DE x DI 1 2.91 1 .49 Residual 188 367.77 Cell Means (variance) [n] Low Involvement, No Decision, Positive Discrepancy 6.12 (2.44) [25 High Involvement, No Decision, Positive Discrepancy 6.36 (1.41) [25. Low Involvement, No Decision, Negative Discrepancy 6.61 (1.61) [23 High Involvement, No Decision, Negative Discrepancy 6.40 (1.50) [25. Low Involvement, Decision, Positive Discrepancy 6.32 (0.98) [25. High Involvement, Decision, Positive Discrepancy 6.00 (3.39) [24 Low Involvement, Decision, Negative Discrepancy 5.92 (2.23) [26 High Involvement, Decision, Negative Discrepancy 6.13 (2.12) [23 Main Effect Means Low Involvement 6.23 No Decision 6.36 Positive Discrepancy 6.20 High Involvement 6.23 Decision 6.09 Negative Discrepancy 6.26 Note : Analysis of responses to a seven-point scale describing the coffeemaker as 1 = "Very Difficult to Use" to 7 = "Very Simple to Use". Significance of results is indicated by: a p<.0001; b = p< .01 ; c = p<.05; d = p< 1
Table 41 ANOVA Summary of Coffeemaker Appearance Perception Source df Sums of Squares omegasquared Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual 187 21.37
275 although this was not noticeable to the experimenter. 7 Finally, having made a decision to try a particular product, subjects may have been more anxious to assess their decision Â— and therefore less patient with the coffeemaker Â— than those who did not choose the beverage product. Other post hoc interpretations are possible; these account for the results (which may be found to be spurious on replication). The data relating to the discrepancy manipulation reveal more complexity than originally intended. The advertisements and instructions communicated differential expectations quite well. One trial of the actual coffee beverage appeared to have been less convincing in the face of the prior testimony. The ratings that resulted, however, were in appropriate directions and generally significant (albeit weakly so). As appropriate, there were few such differences for analogous measures dealing with the "decoy" product, the coffeemaker. In both cases, however, the subjects appeared to have carefully weighed the information newly acquired in product trial against that which was presented "and the greater dimensionality" to them. This was evidenced by the difference in discrimination among individual scales. Involvement Since the response involvement manipulation was based on the one used by Petty et al. (1983), it would seem natural to replicate the manipulation check measure as well. However, "asking subjects what gift they had been told to expect" (p. 140) reduces to a 7 Also, it should be recalled that identical Â— but different Â— coffeemakers were used to avoid contamination. There may have been slight differences in operating characteristics between discrepancyconditions
276 extraneous measure when Â— as is the case in the present study Â— the subjects actually have a gift in hand and are unaware that they may be offered the opportunity to exchange it after completing the questionnaire. Thus a position could be taken that response involvement was successfully manipulated whenever procedures were followed properly. This is similar to the argument raised for the decision manipulation and fully consistent with the implementation of the construct by Petty et al. 8 In an effort to avoid adopting such a dogmatic stance, manipulation check measures adapted from Sherrell and Shimp (1982) were used (Appendix C) At best, this selection was ill-advised. However, these were the only measures located that claimed any relevance to response involvement (note that the Petty et al. measure does not directly tap the construct) Analyses of the scales which rated "evaluation of the cof f ee/cof f eemaker" as: interesting/not interesting to me, important/not important to me, difficult/not difficult to evaluate, took/did not take a lot of thought, task was/was not very involving, and I put a lot/no thought into evaluation, were conducted on individual as well as aggregate scales. Once again, the expected unidimensionality did not obtain (Tables 42, 43) and sub-scales were created. 8 As originally designed, the procedure for the present study mirrored the Petty et al. formulation more closely. During the pretests, it became evident that it was necessary to award the "gift" prior to product trial in order to convince subjects of the veracity of the offer. This alteration in the protocol rendered the Petty et al. check measure redundant at best. No analogous measure could be derived that would not arouse subject suspicion.
Table 42 Selection of Coffee Involvement Scale Combinations 277 a) Factor Analysis Results 3
27i Table 43 Selection of Coffeemaker Involvement Scale Combinations a) Factor Analysis Results 3 Factor 1 Factor 2 .84 .74 -.65 .82 .81 .46
279 Analyses of variance performed on the scales, both individually and in sets, yielded few significant results detectable by omnibus F statistics (most F's < 1, Table 44). The lack of any discernible patterns among or within the models that did reach a conventional level of significance suggests that these may have been due to chance and might therefore be considered spurious. At this point, it is appropriate to consider deeming the measures inadequate rather than to conclude that the manipulation was nonexistent, albeit recognizing that this creates a tenuous philosophical position. However, such a conclusion can be defended on several grounds. Most to the point is the fact that the scale's developers had similar difficulty detecting a within product difference, "manipulation did generate different degrees of cognitive activity in the highand low-involvement groups, but that the two verbal methods were simply incapable of detecting this" (Sherrell and. Shimp 1982, p. 107). The authors further opined that the lack of sensitivity of the verbal measures may have been due to decay during the interval between manipulation and measurement. A chronometric measure (decision time) did yield the results expected for the manipulation, providing some confirmation of this conjecture. Others have despaired the lack of adequate measures, "the reactive nature of research procedures can confound the measurement of response involvement. More [sic] participation in research can heighten the involvement level of a subject; artificially high levels of response involvement may result" (Houston and Rothschild 1978b, p. 186). While heightened involvement did not seem to be a particular
o u Â— > o> c
281 problem in the present study (means were generally around the middle range Â— 3 to 5 Â— of the 7-point scales), a restriction of range due to a ceiling effect would be one explanation for the lack of variance among conditions In summary, the measures of the response involvement measure appear to have been flawed. The manipulation itself, however, is wellgrounded in both prior research and self-reports of subjects in the pretests reported herein. This can be taken as presumptive evidence that the manipulation was executed as defined. This suggests that a conservative approach to data analysis is appropriate, without losing completely the response involvement factor. As argued by Sternthal, Tybout and Calder, "manipulation checks deserve no special status in rigorous tests of theory (1987, p. 119). That is, negative or ambiguous responses on these scales should not be weighted more heavily than responses on any other scales. This issue is discussed further in Chapter 7. A conservative approach to the data is in order. The response involvement factor and its interactions remain in the analyses reported below. Discussion of its effects Â— especially where specific hypotheses were made--is constrained in recognition of the problem with demonstrating conclusively the presence of a strong manipulation. Tests of Hypotheses The manipulations defined an experiment that was designed to evaluate the utility of the extended conceptualization via application of the measures of emotions. That is, the satisfaction measures derived from the DES should show effects attributable to the
282 manipulated additional constructs, decision and response involvement. Other measures Â— perception, attitude, common satisfaction Â— should be consistent with other studies conducted under the disconf irmation paradigm. Attitude measures Among the most basic responses to a product evaluation is thought to be the development or modification of an attitude. Hypothesis 3 asserted that attitude measures would be insensitive to the decision opportunity and response involvement manipulations. The pattern of means anticipated is depicted in Figure 11b. As noted in the previous chapter, a six-item scale was devised to assess attitude toward the coffee. A principal components factor analysis demonstrated only the one dimension (obviating the need for rotation) explaining 87% of variance, which enabled the use of the mean across the scales as an attitude index (Cronbach's a = .97, Table 45). In this case, the model did explain a significant, albeit not large, proportion of the variance in the criterion (F 7;181 = 1.89, p < .10, Table 46). The majority of this explanatory power lies in the discrepancy manipulation (F 1>181 = 7.96, p < .01, omega-squared = .04). Subjects in positive discrepancy conditions had a more favorable attitude, M = 5.30, than did subjects in negative discrepancy conditions, M = 4.67. This finding is particularly interesting given the somewhat discouraging results revealed by the discrepancy manipulation checks.
283 Table 45 Reliability of Coffee Attitude Scale a) Factor Analysis Results 3 Factor 1 .94 .96 .91 .85 .95 .97
Table 46 ANOVA Summary of Coffee Attitude Ratings 284 Source df Sums of Squares omegasquared Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual 181 32.18 19.37 1 .75 0.59 8.71 0.83 0.02 0.89 440.29 Cell Means (variance) [n_ 1 .89^ 7.96b 0.72 0.24 3.58^ 0.34 0.01 0.37 Low Involvement, No Decision, High Involvement, No Decision, Low Involvement, No Decision, High Involvement, No Decision, Low Involvement, Decision, High Involvement, Decision, Low Involvement, Decision, High Involvement, Decision, Positive Discrepancy 5.55 (2.51) Positive Discrepancy 5.68 (1.25) Negative Discrepancy 4.47 (3.04) Negative Discrepancy 4.61 (2.31) Positive Discrepancy 4.81 (2.34) Positive Discrepancy 5.17 (2.48) Negative Discrepancy 4.87 (2.35) Negative Discrepancy 4.69 (3.47) Main Effect Means 04 01         Low Involvement 4.96 High Involvement 5.04 No Decision 5.13 Positive Discrepancy 5.30 Decision 4.88 Negative Discrepancy 4.67 Note : Analysis of mean responses to combined seven-point semantic differential scales describing the respondent's feelings about the coffee as bad/good, unsatisfactory/satisfactory, unfavorable/favorable, worthless/valuable, dislike/like, disapprove/approve; scored one/seven respectively. Results are for a combined scale representing the average response, Cronbach's alpha = .97. Significance of results is indicated by: a = p< 0001 ; b = p< 1 ; c = p<.05; d = p< 1
285 It may be that the subjects couldn't express, or weren't given an opportunity to express, themselves prior to this stage of the questionnaire. That is, it was believed that the perception and disconf irmation scales measured important, salient attributes. Either subjects could not identify the specific cause of displeasure or they were not provided with a measure appropriate to their need. Thus, the manipulation check measures may, again, have been weaker than the manipulation. Also, it should be noted that even in the negative discrepancy conditions, the mean attitude was slightly favorable. Further elaboration of these problems, and implications for the conceptualization, are presented in the general discussion in Chapter 7. The only other effect that was significant was a decision opportunity interaction with discrepancy (explaining about 1% of the variance); with the remaining F statistics each less than one. The best understanding of the data can be gained from a graphic representation (Figure 13). The overall pattern matches closely that which was hypothesized. The pattern for the (non-significant) interaction between involvement and discrepancy is reassuring. The unexpectedly significant interaction between decision and discrepancy once again suggests that making a decision makes subjects more critical of the product in their initial evaluation. (Hypothesis 6 proposed that a contrast effect would be evidenced, but in the same direction as the discrepancy.) This was true in the positive discrepancy conditions in which the unadulterated coffee was rated 5.62 by those subjects who had no decision opportunity but only 5.00 by those who decided on the
286 Combined Attitude 7.0 p 6.0 5.0 4.0 3.0 No decision Â— Â— Â• Low involvement High involvement Decision No decision Decision Positive discrepancy Negative discrepancy Figure 13 Attitude Toward Coffee Mean Response by Experimental Condition
287 product. In the negative discrepancy conditions, the ratings among subjects not offered a decision, M = 4.55, and the ratings among those who did decide to try the coffee, M = 4.78, did not differ (a = .05, Tukey's HSD). Once again, there is mixed support for the overall conceptualization. While Hypothesis 3 does receive some statistical support (i.e., the non-significance of involvement or decision main effects), the resultant means indicate that there is a response to the manipulations. This suggests a confounding between attitude (Hypothesis 3) and satisfaction (Hypothesis 4) as currently measured in CS/D research. In light of this, it is not disquieting to find that involvement and decision did have some influence on the direction of effects. The six-item scale measuring attitude toward the coffeemaker also proved to be unidimensional A principal components factor analysis revealed one factor, explaining 85% of the variance, which enabled use of a composite scale (Cronbach's a = .96, Table 47). As before, a lack of significant effects, as demonstrated by omnibus and effect Fstatistics, was to be expected for the coffeemaker attitude measures. The results indicate this to be true, with subjects overall granting the coffeemakers a favorable, M = 5.30, rating (Table 48). Perception measures One controversy in the CS/D literature has centered on the influence of perception on satisfaction. Because the dissertation distinguishes satisfaction from attitude, the effects of perception on each construct are of interest.
288 Table 47 Reliability of Coffeemaker Attitude Scale a) Factor Analysis Results 3 Factor 1 .91 .95 .96 .87 .90 .94
Table 48 ANOVA Summary of Coffeemaker Attitude Ratings 289 Source df Sums of Squares omegasquared Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual 181 8.57 2.29 3.49 1 .22 0.46 1 .33 1 .71 0.27 387.81 Cell Means (variance) [n] 0.57 1 .07 1 .63 0.57 0.22 0.62 0.80 0.13 Low Involvement, No Decision, Positive Discrepancy 5.35 (2.50) [25 High Involvement, No Decision, Positive Discrepancy 5.25 (3.26) [25 Low Involvement, No Decision, Negative Discrepancy 5.92 (0.70) [17 High Involvement, No Decision, Negative Discrepancy 5.32 (1.28) [25 Low Involvement, Decision, Positive Discrepancy 5.07 (2.71) [25 High Involvement, Decision, Positive Discrepancy 5.19 (2.54) [24 Low Involvement, Decision, Negative Discrepancy 5.28 (1.65) [25 High Involvement, Decision, Negative Discrepancy 5.22 (2.03) [23 Low Involvement 5.36 High Involvement 5.24 Main Effect Means No Decision 5.42 Positive Discrepancy 5.21 Decision 5.19 Negative Discrepancy 5.40 Note: Analysis of mean responses to combined seven-point semantic differential scales describing the respondent's feelings about the coffeemaker as bad/good, unsatisfactory/ satisfactory, unfavorable/favorable, worthless/valuable, dislike/like, disapprove/approve; scored one/seven respectively. Results are for a combined scale representing the average response, Cronbach's alpha = .97. Significance of results is indicated by: a = p< 000 1 ; b = p< .01; c = p<.05; d = p< 1
290 Hypothesis 7 proposed a direct relationship between perception of and attitude toward an object (i.e., a product such as the coffee). Such a simple relationship might be found in cases for which there is little other information on which to build, such as the trial of a new product. In the present study, subjects evaluated both a food product (coffee) and a mechanical product (cof feemaker) then rendered attitude judgments of each. For both types of product, correlations between perception measures and attitude measures were fairly strong (Table 49). While reassuring at first, there are some problems revealed in the table. The measures that showed no significant correlations related to the coffee's flavor strength and color. The other measures were phrased such that they could be interpreted as having evaluative dimensions analogous to attitude (although no true measures of evaluative dimension of attitude were collected). Indeed, regression analyses showed that two of the variables Â— poor/good quality, bad/good tasting Â— explained about 60% of the variance in attitude (no further interpretation of the regression analyses is warranted) So although the hypothesis seems to be supported by the significant correlations, the findings can be questioned on the basis of confounding of perception and attitude. The criticism is blunted somewhat by the analysis of the coffeemaker data. Here, too, the correlations are significant; but the perception measures are less likely to be interpretable as direct measures of attitude. While some evaluative expression is present in the coffeemaker perception scales, the scales are more clearly attribute-specific.
a u-i m -u ^o ij ^o N m n N CI Â— I C"> CN Â— ( co
292 In summary, better perception measures would have been desirable (e.g., those of Eastlack 1964, rather than Olson and Dover 1976 and Cohen and Goldberg 1970) and some measurement technology that would have avoided common method bias would have been preferred. Nevertheless, the hypothesis receives support, albeit from debatable evidence. Perception, itself, was expected to be affected by the extended conceptualization. That is, the early CS/D studies described in Table 7 actually addressed performance perception. Not all, however, reflected the complete prepurchase, purchase, postpurchase sequence that is postulated to be necessary in assessing generalizable behavior. Hypothesis 6 stated that undertaking a decision process influences directly the perception of the outcome of the decision. So, in addition to the effect of the presence or absence of decision on satisfaction, there should be an effect on measures of discrepancy from expectations The direction of this discrepancy was manipulated as part of the overall design, so there should have been an interaction between the effects of manipulated discrepancy and decision opportunity on perceived discrepancy. This problem of directionality can be alleviated by a number of different transformations; reported here are both raw and squared discrepancy means (Table 50). It was expected that the transformation would show a simple main effect of decision opportunity, so one-way Analyses of Variance (essentially, t-test-s) were performed and are reported in Table 50 for
293 both sets of dependent measures. The full ANOVA model was also applied to both sets of data when no significant simple effects were uncovered. As shown in the table, the only statistically significant influence on the dependent measures was the manipulated discrepancy. The simplest explanation for the failure to uncover a contrast effect is that it was not present. That is, either the discrepancies were perceived veridically or assimilation effects were present. This is consistent with reports of difficulty finding such an effect that date to the earliest days of CS/D research (e.g., Cardozo 1965, Olshavsky and Miller 1972). It is evident that quality differences were detected, so the assimilation was not as great as reported in some prior studies (e.g., Olson and Dover 1976, 1978). In summary, however, the analyses did not show that contrast did or could have occurred in this study. Satisfaction measures If perception (and expectations) was manipulated in the desired direction, this should be reflected in the measures that are commonly used in CS/D studies to assess satisfaction. This is consistent with treatment under both the old and the extended conceptualizations. Hypothesis 2 states that commonly used measures of satisfaction respond to discrepancy between expectations and product performance even in the absence of a decision or adequate response involvement. One interpretation of this hypothesis is that only a main effect of discrepancy is predicted, with the positive discrepancy conditions netting a more positive satisfaction response. As depicted in Figure 11b, however, an additional effect, an interaction between involvement
294 Table 50 Significance Tests For Perceived Discrepancy Measures Squared Variables Measure e mean F, 194 ^7,137 mean F-j a 1 94 F 7j1 3 7 1 "Expectations too high/low" 4.1 0.02 2.26= 2. "Coffee worse/better than expected" 4.6 2.65 1.45 3.
295 and discrepancy, may obtain due to closer scrutiny of the product under high involvement conditions. The mean responses are drawn to illustrate a slight, albeit not statistically significant, such effect. 9 Although allegedly measuring the same construct, the satisfaction scales are analyzed separately. This reflects the philosophical stand that the scales are being evaluated competitively (that is, as described in Chapter 3, no one scale is universally accepted among CS/D researchers). It is possible that there will be divergence in results among the scales. 10 For this reason, the resultant means are displayed in Figure 14 for each scale (in the interest of simplicity, the scales are referred to by labels consistent with Westbrook's 1983 usage: 14a Satisfaction; 14b Odds; 14c D-T; 1 4d Faces and 14e Oliver). Although not perfect images of one another, the scales do exhibit a fairly high degree of similarity among themselves. The patterns also suggest some correspondence with that hypothesized in Figure 11b. 9 It should be acknowledged at this point that possible weaknesses in the involvement and discrepancy manipulations render interpretation of results arguable. While an effect of discrepancy is expected, failure to find any effect of involvement could be dismissed as due to weakness in the manipulation. This problem has been discussed with the involvement manipulation check and further caveats are issued in the general discussion, Chapter 7. 10 Obviously "consistency with hypotheses" is not an adequate criterion for selecting that satisfaction measure which best represents "truth". Control of measurement error and sensitivity are likely to be evidenced by F-tests for the effects within each measure.
1 296 Satisfied 7.0 r 6.0 5.0 4.0 3.0 I L Low involvement High involvement No decision Decision No decision Decision Positive discrepancy Negative discrepancy Figure 14 Mean Responses to Standard Satisfaction Scales by Experimental Condition a) Satisfied Scale
297 Odds 8.0 7.0 6.0 5.0 4.0 3.0 Low involvement High involvement No decision Decision No decision Decision Positive discrepancy Negative discrepancy Figure 14--continued b) Odds Scale
298 D-T 7.0 r 6.0 5.0 4.0 3.0 Low involvement High involvement No decision Decision No decision Decision Positive discrepancy Negative discrepancy Figure 14--cont inued c) Delighted-Terrible Scale
299 Faces 7.0 r 6.0 5.0 4.0 Low involvement High involvement 3.0 No decision Decision No decision Decision Positive discrepancy Negative discrepancy Figure 14--continued d) Faces Scale
n 300 Oliver 7.0 r 6.0 5.0 4.0 3.0 Low involvement High involvement No decision Decision No decision Decision Positive discrepancy Negative discrepancy Figure 14--continued e) Oliver's Likert-type Scale
30" Analyses of variance indicate that significance of the results is less uniform, and therefore more ambiguous to interpret (Tables 51, 52, 53, 54, and 55). Specifically, only the Odds measure demonstrated variability that was explained by the model at a conventional significance level (p < .10, Table 52). Total variance (sensitivity) for this measure is greater than that for the others due to the larger number of potential responses. Of the remaining scales, Faces (F 7]183 = 1.59, p < .14, Table 54) and Oliver (F 7)186 = 1.52, p < .16, Table 55) demonstrated effects that were better explained by the ANOVA model. Examining the results of analysis of each measure reveals interesting information. Using the Odds scale as the criterion, the expected discrepancy main effect was evidenced (accounting for approximately 3% of variance) and the interaction between discrepancy and involvement was in an appropriate pattern albeit not significant (p < .25). In addition to these effects was a significant (p < .10) response to the decision manipulation that indicated that the product was viewed more negatively in the conditions in which a decision was made (M = 5.45) than in those that allowed no decision (M = 6.15). Examination of the individually plotted means suggests that this was true for the positive discrepancy conditions but not the negative discrepancy conditions (using Tukey's HSD test, albeit in the face of a nonsignificant, p < .32, decision by discrepancy interaction). Overall, the data from the Odds scale seem to suggest that the manipulations did have their desired effects. Moreover, the Odds scale itself is more sensitive than predicted under Hypothesis 2.
Table 51 ANOVA Summary of Coffee Satisfaction Ratings 302 Source df Sums of Squares Model Discrepancy (DI) Decision (DE) Involvement (I) BE x DI I x DI I x DE I x DE x DI Residual 1 Low Involvement, 2. High Involvement, 3. Low Involvement, 4. High Involvement, 5. Low Involvement, 6. High Involvement, 7. Low Involvement, 8. High Involvement, 18.70 0.74 183 0.05 554.69 0.88 1 .03 2.04 0.86 0.24 1 .91 0.20 0.02 Cell Means (variance) [n omegasquared No Decis No Decis No Decis No Decis Decision Decision Decision Decision on, on, on, on, Positive Discrepancy 5.50 (2.96)  Positive Discrepancy 5.76 (2.36)  Negative Discrepancy 5.50 (1.98)  Negative Discrepancy 5.00 (2.61)  Positive Discrepancy 5.16 (3.14)  Positive Discrepancy 5.13 (3.30)  Negative Discrepancy 5.35 (3.52)  Negative Discrepancy 4.68 (4.42)  Main Effect Means Low Involvement 5.37 High Involvement 5.16 No Decision Decision 5.44 Positive Discrepancy 5.39 5.09 Negative Discrepancy 5.14 Note: Analysis of seven-point scale which asked, "How satisfied were you with the coffee?" Responses ranged from 1 = "Very Dissatisfied" to 7 = "Very Satisfied", including 4 = "Mixed" and = "Not Satisfied Nor Dissatisfied". Significance of results is indicated by: a = p<.0001; b = p<.01; c = p<.05; < = p<. 10.
Table 52 ANOVA Summary of Coffee Odds Scale 303 Source df Sums of Squares omegasauared Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual 102.14 57.00 23.64 1 .14 7.95 10.77 0.83 0.61 1485.10 Cell Means (variance) [n] 85d 22b ,99d 14 01 36 10 0.08 .03 .01 Low Involvement, No Decision, High Involvement, No Decision, Low Involvement, No Decision, High Involvement, No Decision, Low Involvement, Decision, High Involvement, Decision, Low Involvement, Decision, High Involvement, Decision, Positive Discrepancy 6.56 (7.34)  Positive Discrepancy 7.20 (4.00)  Negative Discrepancy 5.43 (6.17)  Negative Discrepancy 5.36 (7.07)  Positive Discrepancy 5.48 (9.09)  Positive Discrepancy 6.08 (6.08)  Negative Discrepancy 5.38 (11.69) Negative Discrepancy 4.83 (11. 70 ) [ 23 ] Main Effect Means Low Involvement 5.7 2 High Involvement 5.89 No Decision Decision 6.15 5.45 Positive Discrepancy 6.33 Negative Discrepancy 5.26 Note : Analysis of eleven point scale rating the chances in ten the respondent "would choose to use the coffee again". Responses ranged from = "No Chance" to 10 = "Certain". See Appendix C for question wording and structure. Significance of results is indicated by: a = p< 0001 ; b = p< 01 ; c = p<.05; d = p< 1
Table 53 ANOVA Summary of Coffee Delighted-Terrible Scale 304 Source df Sums of Squares Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual 13.18 4.96 0.75 0.02 0.24 1 .70 0.39 5.50 334.64 Cell Means (variance) [n 1 .06 2.79^ 0.42 0.01 0. 14 0.96 0.22 3.09^ omegasquared .01 Low Involvement, No Decision, High Involvement, No Decision, Low Involvement, No Decision, High Involvement, No Decision, Low Involvement, Decision, High Involvement, Decision, Low Involvement, Decision, High Involvement, Decision, Positive Discrepancy 5.32 (1.31)  Positive Discrepancy 5.24 (0.69)  Negative Discrepancy 4.78 (1.45)  Negative Discrepancy 5.00 (2.83)  Positive Discrepancy 4.88 (1.94)  Positive Discrepancy 5.29 (1.95)  Negative Discrepancy 5.15 (2.06)  Negative Discrepancy 4.52 (1.99)  Main Effect Means Low Involvement 5.04 High Involvement 5.02 No Decision Decision 5.09 4.97 Positive Discrepancy 5.1. Negative Discrepancy 4.8. Note : Analysis of seven-point scale describing satisfaction with the coffee. Responses ranged from 1 = "Terrible" to 7 = "Delighted" and included 4 = "Mixed" and = "Neutral, Never Thought About It". See Appendix C for question and response wording. Significance of results is indicated by: a = p< 0001 ; b = p< 01 ; c = p<.05; d = p< 1
305 Table 54 ANOVA Summary of Coffee Faces Scale Source df Squares
Table 55 ANOVA Summary of Oliver's Six Item Scale 305 Source df Sums of Squares omegasquared Model Discrepancy (DI) Decision (DE) Involvement DE x DI I x DI I x DE I x DE x DI Residual (I) 186 15.25 1.52 8.77 0.25 0.07 3.75 0.95 0.01 266.93 Cell Means (variance) [n] 1 .52 1.06 6.11 = 0.18 0.05 2.61 0.66 0.00 High Involvement, No Decision, Low Involvement, Decision, High Involvement, Decision, Low Involvement, Decision, High Involvement, Decision, Main Effect Means Low Involvement High Involvement 4.77 4.84 No Decision Decision 4.59 5.02 03 Low Involvement, No Decision, Positive Discrepancy 4.56 (0.72)  High Involvement, No Decision, Positive Discrepancy 4.76 (0.67)  Low Involvement, No Decision, Negative Discrepancy 4.69 (1.18)  Negative Discrepancy 4.35 (0.96)  Positive Discrepancy 4.87 (1.78)  Positive Discrepancy 5.38 (1.13)  Negative Discrepancy 4.95 (2.39)  Negative Discrepancy 4.87 (2.69)  Positive Discrepancy 4.89 Negative Discrepancy 4.72 Note : Analysis of mean responses to six seven-point Likert type items patterned after Oliver (1980): "I am satisfied about my choice of coffee; If I had it to do all over again, I would feel differently about choosing the coffee; My choice of coffee was a wise one; I feel bad about my decision regarding the coffee; I think that I did the right thing when I chose the coffee; I am not happy that I chose the coffee that I did. Mean responses are reported after reversing the First, Third, and Fifth scales so that 7 = Strongly Agree (with positive statements) and 1 = Strongly Disagree (Cronbach's alpha = .85). Significance of results is indicated by: a = p< 0001 ; b = p < 1 ; c = p<.05; d = p< 1
307 The next most interesting pattern of responses were those to the D-T scale (Table 53). Since the diagrams in Figures 11 and 14 effectively depict three-way breakdown of means, without a significant interaction hypothesized, it is useful to examine the one significant three-way interaction that did obtain (accounting for, however, only 1% of variance). Pairwise comparisons (Tukey's HSD test) indicate that the means that were obtained for this measure, as shown in Figure 14c, are adequate representations of the pattern hypothesized and depicted in Figure 11b, with one additional complication. Apparently, the act of making a decision caused the divergence in effect of the involvement manipulation within the range of responses generated by the discrepancy manipulation. A significant main effect of discrepancy is also present indicating a more favorable evaluation of the product under positive discrepancy conditions, M = 5.18, than under negative discrepancy conditions, M = 4.88, omega-squared = .01. It is interesting to note that the obtained pattern is very much in accord with that predicted for the emotional measures of satisfaction and dissatisfaction (Figure 11a). Therefore, although Hypothesis 2 is not well supported by the DT results, Hypotheses 1a and 1b receive unexpected support. Of the three remaining scales, little needs to be said at this juncture. The analysis of variance in responses to the satisfaction scale revealed no significant effects (Table 51) and the main effects found to be significant for the Faces scale and Oliver's six-item scale (a = .85, Table 56) were consistent with the patterns already discussed. Mean responses to the Faces scale showed a more negative
3( response to the negative discrepancy conditions, M = 3.43, than to the positive, M = 3.77 (accounting for approximately 3% of the variance in response to this 5-point scale). Oliver's scale seems to indicate that decision enhances satisfaction (M = 5.02) compared with no decision (M = 4.59) although this is questionable in light of no other effects (particularly, effects of discrepancy and its interactions). All in all, the patterns of results in Figure 14 display more sensitivity to the experimental manipulations than originally predicted. The first implication that can be drawn is that the manipulations, discrepancy and decision opportunity at least, were more effective than the check measures suggest. Secondly, some of the satisfaction measures may be more sensitive than may have been indicated by the global judgment offered earlier by the dissertation. This increases confidence regarding the manipulations and other analyses of the data. Further confidence is gained by examination of analogous scales for the coffeemaker product. In essence, only involvement (as the counterpoint to coffee involvement) was manipulated. Examination of the ANOVA tables for Satisfaction, Odds, D-T, and Faces scales evaluating the coffeemakers reveals no significant omnibus F-test (in fact, only one F of the four exceeded a value of one, Tables 57, 58, 59, 60). In light of strong expectations of no effect, these results are sufficient to judge the isolated statistically significant effects to be trivial (Tables 58, 60). It should be noted that satisfaction with the coffeemakers was generally good, but not outstanding (on the Satisfaction and D-T 7-point scales, means were 5.45 and 5.13, respectively; on the 11 -point Odds scale,
H Table 56 Reliability of Oliver's Six-item Scale a) Factor Analysis Results 3 Itemsb Factor 1 1 Satisfied with choice .74 2 Feel differently again (-) .59 3 Wise choice .77 4 Feel bad (-) .77 5 Did right thing .85 6 Not happy (-) .83 variance explained 58% b) Cronbach's Alpha Computation Pearson Correlation Coefficients: Item 3 Item 4 Item 5 .435 .674 .534 .499 .742 .628
Table 57 ANOVA Summary of Coffeemaker Satisfaction Ratings 310 Source df Sums of Squares omegasquared Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual 18.24 1.52 6.70 6.48 0.60 0.03 1 .30 2.42 586.35 Cell Means (variance) [n] 0.84 0.49 2.15 2.08 0.19 0.01 0.42 0.77 Low Involvement, No Decision, Positive Discrepancy 5.64 (3.16) [25. High Involvement, No Decision, Positive Discrepancy 5.36 (3.41) [25. Low Involvement, No Decision, Negative Discrepancy 6.17 (0.97) [23. High Involvement, No Decision, Negative Discrepancy 5.40 (3.58) [25. Low Involvement, Decision, Positive Discrepancy 5.44 (3.01) [25. High Involvement, Decision, Positive Discrepancy 5.04 (4.13) [24 Low Involvement, Decision, Negative Discrepancy 5.31 (2.94) [26. High Involvement, Decision, Negative Discrepancy 5.30 (3.68) [23 Main Effect Means Low Involvement 5.63 High Involvement 5.28 No Decision 5.63 Positive Discrepancy 5.37 Decision 5.28 Negative Discrepancy 5.54 Note : Analysis of seven-point scale which asked "How satisfied were you with the coffeemaker?" Responses ranged from 1 = "VeryDissatisfied" to 7 = "Very Satisfied", including 4 = "Mixed" and = "Not Satisfied Nor Dissatisfied". Significance of results is indicated by: a = p< 0001 ; b = p< 01 ; c = p<.05; d = p< 10.
Table 58 ANOVA Summary of Coffeemaker Odds Scale 311 Source df Sums of Squares omegasquared Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual 71
312 mean response was 6.14 and on the 5-point Faces scale, mean response was 3.64). This is perfectly consistent with the discrepancy scale results reported earlier. Behavior measures In addition to the need to debrief experimental participants, one purpose for the callback procedure was to collect self-reported behavioral data. This type of data collection has not been reported previously in the CS/D literature, but was thought to be a potentially rich source of information. Unfortunately, it proved to be more difficult to contact the subjects than had been anticipated. Of the 196 subjects who provided data used in analyzing the primary hypotheses, 83 were actually contacted one week after participating in the evaluation portion of the experiment. Since these contacts were approximately evenly distributed across conditions, the results are interpreted as though from a random sample, but with caution appropriate to the 42% response (Table 61). Hypotheses 4 and 5 had been identified as secondary to the major contribution of the dissertation and the behavioral measures (self-reports) were collected, in part, to evaluate the feasibility of the methodology. Hypothesis 4 stated that satisfaction and dissatisfaction are each associated with a greater incidence of their respective theoretical outcomes Â— positive versus negative word-of-mouth, complimenting versus complaining--and that attitude is associated with intent to repurchase. Hypothesis 5 stated that the likelihood of behavior increases with strength of satisfaction or dissatisfaction emotion. Very few subjects reached criterion levels of satisfaction emotions (12) or dissatisfaction emotions (3), which rules out a test or discussion of
Table 59 ANOVA Summary of Coffeemaker Delighted-Terrible Scale 313 Source df Sums of Squares omegasquared Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual 187 9.52 1.58 2.76 1 .82 1 .93 0.27 0.01 1 .50 329.02 0.77 0.90 .57 .04 .10 .15 0.01 0.85 Cell Means (variance) [n_ Low Involvement, No Decision, Positive High Involvement, No Decision, Positive Low Involvement, No Decision, Negative High Involvement, No Decision, Negative Low Involvement, Decision, Positive High Involvement, Decision, Positive Low Involvement, Decision, Negative High Involvement, Decision, Negative Discrepancy 5.20 (1 .50)  Discrepancy 4.92 (1 .74)  Discrepancy 5.48 (0.53)  Discrepancy 5.40 (3.50)  Discrepancy 5.00 (1 .83)  Discrepancy 5.04 (1.78)  Discrepancy 5.23 (1.62)  Discrepancy 4.77 (1.42)  Low Involvement 5.22 High Involvement 5.04 Main Effect Means No Decision 5.24 Positive Discrepancy 5.04 Decision 5.02 Negative Discrepancy 5.23 Note : Analysis of seven-point scale describing satisfaction with the coffeemaker. Responses ranged from 1 = "Terrible" to 7 = "Delighted" and included 4 = "Mixed" and = "Neutral, Never Thought About It." See Appendix C for question and response wording. Significance of results is indicated by: a = p< .0001 ; b = p<.01; c = p<.05; d = p< 1
Table 60 ANOVA Summary of Coffeemaker Faces Scale 314 Source df Sums of Squares omegasquared Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual 136 5.38 0.03 0.28 1 .31 0.83 0.17 2.63 0.18 173.36 Cell Means (variance) [n. 0.82 0.04 0.30 1 .41 0.90 0.19 2.82* 0.20 ,01 1. Low Involvement, No Decision, Positive Discrepancy 3.80 (1.00)  2. High Involvement, No Decision, Positive Discrepancy 3.40 (0.83)  3. Low Involvement, No Decision, Negative Discrepancy 3.96 (0.50)  4. High Involvement, No Decision, Negative Discrepancy 3.56 (1.09)  5. Low Involvement, Decision, Positive Discrepancy 3.56 (1.26)  6. High Involvement, Decision, Positive Discrepancy 3.75 (0.98)  7. Low Involvement, Decision, Negative Discrepancy 3.58 (0.89)  8. High Involvement, Decision, Negative Discrepancy 3.52 (0.86)  Main Effect Means Low Involvement 3.72 High Involvement 3.56 No Decision 3.67 Positive Discrepancy 3.63 Decision 3.60 Negative Discrepancy 3.65 Note: Analysis of five point graphic scale rating "how you feel about the coffeemaker". Responses were re-scaled such that 1 = most negative face and 5 = most positive face. See Appendix C for wording and appearance of question. Significance of results is indicated by: a = p<.0001; b = p< 01 ; c = p<.05; d = p< 1
Table 61 Number of Subjects Contacted by Telephone at One Week Delay 315 Experimental Condition Number Number Not Contacted Contacted Total Low Involvement, No Decision, Positive Discrepancy High Involvement, No Decision, Positive Discrepancy Low Involvement, No Decision, Negative Discrepancy High Involvement, No Decision, Negative Discrepancy Low Involvement, Decision, Positive Discrepancy High Involvement, Decision, Positive Discrepancy Low Involvement, Decision, Negative Discrepancy High Involvement, Decision, Negative Discrepancy 13 11 1 1 11 12 83 12 14 12 18 14 14 14 15 113 25 25 23 25 25 24 26 23
316 hypothesis 4 in its strictest interpretation (Table 66). Removing the restriction imposed by the criterion levels, and allowing satisfaction and dissatisfaction to vary along continua, collapses the essential points of hypotheses 4 and 5. Therefore, these are discussed together. The most straightforward assessment of the degree of association between variables is by correlation analysis. Measured levels of satisfaction, dissatisfaction and attitude toward the product were correlated (product-moment) with self-reported behavior or behavioral intention. 11 The results of these correlation analyses are reported here for the entire sample and for sub-samples blocked by discrepancy condition. The separate sample procedure allows satisfaction and dissatisfaction to follow different continua and generate--or respond to--separate psychological processes. Of greatest interest was the association between the emotional measures of satisfaction and dissatisfaction and their respective behavioral results. The satisfaction emotions shared no significant correlation with most of the behaviors for which effects were hypothesized Â— positive word of mouth, return of a complimenting postcard, consumption of the product, serving the product to others in the household or to guests Â— when examined either across the sample or just for those in the positive discrepancy conditions (Table 62 and Table 63, respectively). The only exception to this was a small positive relationship with a dummy-coded variable indicating whether or not positive word-of-mouth was reported (r=.22, p<.05). No such 1 1 The selection of the gift product was a behavior observed by the experimenter.
317 relationship was found with the measure of the extent of word-of-mouth (i.e., the number of people spoken to positively). The more conventional measures of satisfaction fared a little better in the full sample, with significant correlations with the incidence and frequency of positive or negative word-of-mouth. These relationships appear to have been much stronger for those subjects in the negative discrepancy conditions (Table 64) than in the positive discrepancy conditions (Table 63), although the interpretations must be tempered in light of the low recontact rates. Similar discouraging results were encountered for the emotional measures of dissatisfaction and the reports of negative behaviors (Table 62). The dissatisfaction measure exhibited a slight relationship with the return of a postcard with complaints (r=.12, p< .10) as well as inverse relationships with the return of a complimenting card (r=-.13, p< 10) and serving the coffee to guests (r=-.29, p<.05). Contrary to theory, however, was a direct relationship with engaging in positive word-of-mouth (r=.21, p< 10) Again, the results were mirrored in the within discrepancy analyses, with the positive discrepancy conditions apparently accounting for the correlation between the positive word-of-mouth and dissatisfaction emotion measures (Tables 63 and 64). 12 12 Similar correlation analyses were conducted within other subsets of the data. Dividing the data set at either the midpoint of each satisfaction scale or at the median response to each scale did not affect markedly the interpretability of the significant correlations. Cross-tabulations and/or tests of differences between means were similarly unenlightening. Consequently, none of these analyses are reported nor discussed here.
Â•1-1 318 .* 6 co co co CN i-l Ov Xi CX5 r- en iÂ—i Â£ 5 5 CD I Â— H 00 r> i-i Â—i to
319 CO < | um > c J 5* o ^ 2 CJ "* w 4-1 Â™ W 0)
ti u 05 < 1 to 4J S 3 <" z a h x n n ri 320 S u > o < Â£ Â— i O T3 "0 5 C u-, O 4J o O -i 41 O Kl 41 05 en co .c -a h -a c 4> 05 l~ 4> w Â— I O cfl Â§ Ia a. o B 4) (0O^5CU3O>C> 003 E c/5 4-> a. t-i-H u co x: Â— oj
321 It is interesting to note that the satisfaction measures adapted from prior research did, overall, correlate more highly with the target behaviors. However, the measure used for attitude toward the coffee also varied with the behaviors. This indicates that Â— at least at the time of measurement Â— the constructs are highly confounded, using such similar assessment instruments. Further evidence of the time-sensitive nature of the processes can be seen in the relative magnitudes of the correlations. The strongest correlations were among the satisfaction and attitude measures with the behavioral intent measures that were collected at the same time, as well as the same affective or cognitive measures correlated with gift selection behavior. Other correlations, collected after a delay and possible reconsideration of the product, were less strong. Also, consistent with the Theory of Reasoned Action, the behavioral intention measures were generally more highly related to attitude Â— and the satisfaction measures Â— than were the reports of actual behaviors. In summary, the conclusions that can be drawn from the behavioral assessment procedures are tentative at best. The rate of re-contact was too low to inspire confidence and netted very small sample sizes in some blocked analyses. Further, those analyses that depended on the occurrence of relatively extreme behaviors (e.g., return of a selfaddressed, stamped postcard) would require much larger overall samples, as these behaviors proved to be rare events. While hypotheses regarding such events are testable comparing against appropriate statistical distributions, the results of the present study do not display effects that warrant such treatment.
322 The logic behind the hypotheses and analyses of behavioral results, however, remains appealing. The CS/D literature is rooted in an interest in postpurchase behaviors. As discussed previously, one difficulty in finding behavioral effects in CS/D research arises from the interference of constraints on behavior. The preceding analysis attempted to control these extraneous factors at a constant level. Another approach would be to evaluate a construct, such as motivation to behave, which acts earlier in the process. Hypothesis 8 makes explicit the "conservation of motivation" concept elaborated earlier. It states that strength of emotional response varies with involvement. Again, correlation analysis is appropriate to the question. Since response involvement was both manipulated and measured, it was possible to analyze the effect on emotions (particularly those hypothesized to be characteristic of satisfaction feelings) of assigned condition and perceived involvement. The former was evaluated by means tests and the latter by correlation. There were no significant differences in mean emotion strength (individual emotions) detectable by either t-test or ANOVA (details of which are not reported here. There were, however, some modest correlations between measures of involvement and emotions for the overall sample as well as subsets separated by discrepancy condition (Table 65). None of the correlations was very large, however, suggesting very limited support of the hypothesis. As noted in the manipulation check discussion, the response involvement manipulation was the most equivocal and the measures of
323 Table 65 Correlations Between Involvement and Emotion Measures a) Full data set (n=188) Involvement e Emotions f Interesting Important Difficult Involving Evaluation Satisfaction Dissatisfaction
Note to Table 65: : Reported are Pearson correlations; a: p<.0001, b p< .01 c: p<.05, d: p< 1 e Involvement measures as defined previously. f Emotion measures as defined previously. g Reported are: first positive n = (99) then negative (n = 90) discrepancy within each pair. 324
52 5 involvement were thought to be unreliable. This may reflect a floor effect in which the chosen situation was unable to adequately distract a subject from an inherently uninvolving product. Again, at best only tenuous support can be derived for the hypothesis. Measures based on the Differential Emotions Scale The most direct assessment of the influence of the constructs added by the conceptualization on satisfaction, as defined, is by analysis of the DES patterns of emotions. Prediction of effects measurable by this technique was the focus of the hypotheses central to the contribution of the extended conceptualization. Hypotheses 1a and 1b stated that, in the presence of all enabling conditions, a positive discrepancy from expectations to performance perception results in a pattern of emotions that constitutes satisfaction and a negative discrepancy results in a pattern of emotions that constitutes dissatisfaction. The pilot study reported in Chapter 5 concluded that appropriate definitions of these patterns of emotions required average ratings of 3 on the Differential Emotions Scale items which described one's feelings of interest, joy, and surprise (satisfaction) or anger, disgust and surprise (dissatisfaction) This measurement device was quite disappointing for the present study. An examination of responses by condition experienced reveals no mean equal to nor exceeding 3.0 (Tables 67 and 68). An inspection of the data at the level of individual subject response shows that few subjects ever reached the criterion level of satisfaction or dissatisfaction (Table 66).
1 326 Table 66 Number of Subjects Reaching Criterion Emotion Levels Experimental Treatments Number (%) At or Above Criterion Satisfaction Dissatisfaction Low Involvement, No Decision, Positive Discrepancy High Involvement, No Decision, Positive DiscrepancyLow Involvement, No Decision, Negative Discrepancy High Involvement, No Decision, Negative Discrepancy Low Involvement, Decision, Positive Discrepancy High Involvement, Decision, Positive Discrepancy Low Involvement, Decision, Negative Discrepancy High Involvement, Decision, Negative Discrepancy 3 (12)
Table 67 ANOVA Summary of Satisfaction Emotions 327 df Sums of Squares omegasquared Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual 7 1 1 1 1 1 1 1 187 4.70 0.18 3.13 0.37 0.00 0.15 0.29 0.58 94.66 1.33 0.35 6.18= 0.74 0.01 0.30 0.57 1 .14 Cell Means (variance) [n] .03 Low Involvement, No Decision, Positive Discrepancy 1.88 (0.71)  High Involvement, No Decision, Positive Discrepancy 1.70 (0.52)  Low Involvement, No Decision, Negative Discrepancy 1.64 (0.35)  High Involvement, No Decision, Negative Discrepancy 1.80 (0.39)  Low Involvement, Decision, Positive Discrepancy 2.09 (0.71)  High Involvement, Decision, Positive Discrepancy 1.98 (0.45)  Low Involvement, Decision, Negative Discrepancy 2.09 (0.44)  High Involvement, Decision, Negative Discrepancy 1.87 (0.45)  Low Involvement 1.93 High Involvement 1.83 Main Effect Means No Decision 1.76 Positive Discrepancy 1.91 Decision 2.01 Negative Discrepancy 1.86 Note : Analysis of mean responses to nine Differential Emotions Scale items which were defined as comprising the satisfaction pattern of emotions: attentive, concentrating, alert, delighted, happy, joyful, surprised, amazed, astonished; Cronbach's alpha = .85. Significance of results is indicated by: a = p < 6 1 ; b = p< 1 ; c = p<.05; d = p< 1
Table 68 ANOVA Summary of Dissatisfaction Emotions 328 Source df Sums of Squares Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual 1 Low Involvement, 2. High Involvement, 3. Low Involvement, 4. High Involvement, 5. Low Involvement, 6. High Involvement, 7. Low Involvement, 8. High Involvement, 1 .88 0.01 1 .01 0.21 0.09 0.23 0.11 0.20 38.16 Cell Means (variance) [n] No Decision, No Decision, No Decision, No Decision, Decision, Decision, Decision, Decision, Positive Positive Negative Negative Positive Positive Negative Negative Discrepancy Discrepancy Discrepancy Discrepancy Discrepancy Discrepancy Discrepancy Discrepancy F omegasquared 1 .32 0.03 4.98= 1 .02 0.44 1 .12 0.53 1 .01 ,02 1 .28
329 Analyses of variance were conducted on each composite scale. 13 Neither scale showed the expected interaction among the three manipulated factors. The only significant effect on either measure was a main effect of the decision manipulation (Tables 67, 68). In each case, making a decision appears to have enhanced emotional response. For the satisfaction measure, mean responses were 1.76 for the no decision conditions versus 2.01 for the decision conditions (accounting for about 3% of the variance). Mean responses to the dissatisfaction measure were 1.22 in the no decision conditions and 1.37 in the decision conditions (accounting for about 2% of the variance). In light of the non-significant omnibus F-tests and the relatively slight explanatory power, even these results must be viewed with caution. One diagnostic strategy of interest is the analysis of each component emotion. Each emotion was measured using three items from the DES (see Table 69 for calculation of Cronbach alphas) and could also be expected to show response to the manipulations. Of the five emotions, none exhibited a significant omnibus F indicating statistically significant variation, although some effects Â— interpreted as a priori Â— appeared significant for each emotion except interest (Table 70). The effects witnessed on the satisfaction and dissatisfaction scales are reflections of responses to the joy and surprise scales (satisfaction) and the anger and surprise scales (dissatisfaction), all 13 As discussed in Chapter 5, the component emotions are not expected to correlate directly so an internal reliability coefficientsuch as Cronbach' s alpha--is less meaningful for these data. Nevertheless, values of .85 and .90 were calculated for satisfaction and dissatisfaction, respectively, and are reported.
330 Table 69 Reliability of Emotion Measures a) Factor Analysis Results 3 Basic Measurement Factor Factor Factor Item Item 1 2 3 Interest Attentive (E1) .81 Concentrating (E2) .72 Alert (E3) .74 Joy Delighted (E4) .77 Happy (E5) .77 Joyful (E6) .63 Surprise Surprised (E7) .61 Amazed (E8) .67 Astonished (E9) .74 Disgust Feeling of Distaste (E10) .84 Disgusted (E1 1 ) .77 Feeling of Revulsion (E12) .70 Anger Enraged (E13) -80 Angry (E14) .28 Mad (E15) .56 variance explained 24.6% 19.5% 1 3 8%/ 5 7
331 Table 69 Â— continued b) Cronbach's Alpha Computations For Each Emotion 13 Pearson Correlation Coefficients attentive concentrating concentrating .54 a = .80 (Interest) a = .80 (Joy) alert
1 332 Table 69 Â— continued c) Cronbach's Alpha Computations for Satisfaction/Dissatisfaction Patterns Pearson Correlation Coefficients Â— Satisfaction E1 E2 E3 E4 E5 E6 E7 E8 E2
Table 70 ANOVA Summary of Interest 333 Source df Sums of Squares omegasquared Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual 187 4.76 0.00 1 .65 0.00 0.04 0.27 1 .26 1 .63 194.56 0.65 0.00 1 .59 0.00 0.04 0.26 1 .21 1 .57 Cell Means (variance) [n. Low Involvement, No Decision, Positive Discrepancy 2.20 (1.51)  High Involvement, No Decision, Positive Discrepancy 2.10 (1.17)  Low Involvement, No Decision, Negative Discrepancy 1.92 (0.69)  High Involvement, No Decision, Negative Discrepancy 2.33 (1.02)  Low Involvement, Decision, Positive Discrepancy 2.33 (0.87)  High Involvement, Decision, Positive Discrepancy 2.28 (0.94)  Low Involvement, Decision, Negative Discrepancy 2.47 (0.91)  High Involvement, Decision, Negative Discrepancy 2.20 (1.18)  Main Effect Means Low Involvement 2.24 High Involvement 2.23 No Decision 2.14 Positive Discrepancy 2.23 Decision 2.33 Negative Discrepancy 2.24 Note : Analysis of mean responses to three Differential Emotions Scale items which were defined as comprising the interest emotion: attentive, concentrating, alert; Cronbach's alpha = .80. Significance of results is indicated by: a = p<.0001; b = p< 01 ; c = p<.05; d = p< 1
Table 71 ANOVA Summary of Joy 334 Source df Sums of Squares F omegasquared Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual Low Involvement, High Involvement, Low Involvement, High Involvement, Low Involvement, High Involvement, Low Involvement, High Involvement, 7 1 1 1 1 1 1 1 187 38 69 15 25 ,63 ,31 ,03b .71 .03 0.01 0.00 0.01 0.17 136.97 0.01 0.00 0.02 0.23 Cell Means (variance) [n] No Decision, Positive Discrepancy 1.91 (0.88) [25. No Decision, Positive Discrepancy 1.68 (0.74) [25 No Decision, Negative Discrepancy 1.66 (0.63) [22 No Decision, Negative Discrepancy 1.53 (0.53) [25. Decision, Positive Discrepancy 2.15 (1.23) [25 Decision, Positive Discrepancy 2.07 (0.73) [24 Decision, Negative Discrepancy 2.04 (0.42) [26 Decision, Negative Discrepancy 1.83 (0.71) [23. Main Effect Means Low Involvement 1.95 High Involvement 1.77 No Decision Decision 1 .69 2.02 Positive Discrepancy 1 Negative Discrepancy 1 .95 ,77 Note: Analysis of mean responses to three Differential Emotions Scale items which were defined as comprising the joy emotion: delighted, happy, joyful; Cronbach's alpha = .80. Significance of results is indicated by: a = p<.0001; b = p< 01 ; c = p<.05; d = p<. 10.
Table 72 ANOVA Summary of Surprise 335 Source df Sums of Squares F omegasquared Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual 4.82 0.00 3.07 187
Table 73 ANOVA Summary of Anger 336 Source df Sums of Squares omegasquared Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual 1 Low Involvement, 2. High Involvement, 3. Low Involvement, 4. High Involvement, 5. Low Involvement, 6. High Involvement, 7. Low Involvement, 8. High Involvement, 2.01 0.07 0.92 0.43 0.14 0.05 0.13 0.23 33.50 Cell Means (variance) [n] No Decision, No Decision, No Decision, No Decision, Decision, Decision, Decision, Decision, Positive Positive Negative Negative Positive Positive Negative Negative Disc Disc Disc Disc Disc Disc Disc Disc repancy repancy repancy repancy repancy repancy repancy repancy Main Effect Means 1.61 0.40 5.15= 2.44 0.81 0.28 0.72 1.31 .09 .01 .07 .07 .27 .22 .28 .03 (0.11) (0.00) (0.06) (0.06) (0.3C) (0.45) (0.41) (0.02) Low Involvement 1.18 High Involvement 1.08 No Decision Decision 1 .06 1.20 02        T23] Positive Discrepancy 1.15 Negative Discrepancy 1.12 Note : Analysis of mean responses to three Differential Emotions Scale items which were defined as comprising the anger emotion: enraged, angry, mad; Cronbach's alpha = .46. Significance of results is indicated by: a = p< 000 1 ; b = p< 01 ; c = p<.05; d = p< 1
Table 74 ANOVA Summary of Disgust 337 Source Model Discrepancy (DI) Decision (DE) Involvement (I) DE x DI I x DI I x DE I x DE x DI Residual df Sums of Squares 7 1 1 1 1 1 1 1 187 1 19 0.00 0.07 0.00 0.21 0.82 0.00 0.06 41 .38 Cell Means (variance) [n] omegasquared 0.77 0.02 0.31 0.01 0.96 3.71^ 0.00 0.27 ,01 1 Low Involvement, No Decision, Positive 2. High Involvement, No Decision, Positive 3. Low Involvement, No Decision, Negative 4. High Involvement, No Decision, Negative 5. Low Involvement, Decision, Positive 6. High Involvement, Decision, Positive 7. Low Involvement, Decision, Negative 8. High Involvement, Decision, Negative Discrepancy 1 .23 (0.25) [25." Discrepancy 1.05 (0.04)  Discrepancy 1 .14 (0.08) [22; Discrepancy 1 .29 (0.45) [25; Discrepancy 1 .29 (0.22) [25; Discrepancy 1.19 (0.26) [24; Discrepancy 1 .14 (0.12) [26; Discrepancy 1 .23 (0.36) [23; Main Effect Means Low Involvement 1.20 High Involvement 1.19 No Decision Decision 1 .18 1 .21 Positive Discrepancy 1.19 Negative Discrepancy 1.20 Note: Analysis of mean responses to three Differential Emotions Scale items which were defined as comprising the disgust emotion: Feeling of distaste, feeling of revulsion, disgusted; Cronbach's alpha = .60. Significance of results is indicated by: a = p< 0001 ; b = p< 01 ; c = p<.05; d = p< 1
338 of which returned a decision main effect (Tables 71, 72, 73). In each case, the mean response was higher for those conditions in which subjects made a decision than for those in which no decision was allowed, although none approached the previously discussed 3.0 level. Specifically, mean responses to the Joy items were 1.69 and 2.02; to Surprise 1.42 and 1.68; to Anger 1.06 and 1.20 for the no decision and decision conditions, respectively (variance explained ranged from 2 to 3%). Mean responses to disgust revealed yet another pattern that was significant but accounted for a very small amount of variance (1%, Table 74). The effect that seemed to obtain was that under high involvement, the negative discrepancy conditions (i.e., the peanut coffee) created a significantly higher level of disgust (M = 1.26) than did the positive discrepancy conditions (M = 1.12). However, under low involvement, the reverse was true and disgust was higher in the positive discrepancy conditions (M = 1.26) than in the negative discrepancy conditions (M = 1.14), both pairwise differences are significant at a = .05 using Tukey's HSD test). A distraction explanation may well account for these results; but they are probably best considered trivial, given the very small proportion of variance in the measure that was explained. Overall, hypotheses 1a and 1b were not supported by these analyses. Only the decision manipulation demonstrated any consistent effect on the emotion scales as measured. Nearly identical results obtained from analyses of emotion scales responses normalized for each subject's mean emotional rating. Either the DES is not sensitive
339 enough for this task, or the manipulations were not strong enough to drive the process (or both). Implications for the conceptualization presented in the dissertation are best discussed in light of all results and are reserved for the general discussion and Chapter 7. Summary and Discussion Although the integrative interpretations are reserved for chapter 7, some general conclusions are in order. The study suffered from some operational flaws resulting from the ambitious design. The effort to include three factors led to operationalization decisions (e.g., choice of target and distractor products) that may have weakened some of the tests. In particular, involvement is difficult to manipulate reliably and should probably be measured more accurately and used as a blocking variable or covariate. Overall, the hypotheses received mixed support. Hypotheses 1a and 1b, predicting emotional patterns resulting from the manipulations, received little support from the DES measures--but the more common satisfaction scales indicated that some effect, at least, was present. The lack of such effects for the coffeemaker product validated further the contentions of hypotheses 1a and 1b, but the sensitivity of these measures was contrary to the prediction of hypothesis 2. Similarly, in evaluating hypothesis 3, greater effects on attitude measures than expected were found. The unexpectedly strong performance of the common satisfaction and attitude measures was repeated in correlation analyses with behaviors, hypotheses 4 and 5. Although emotion strengths did not correlate well
340 with behavior, these alternate measures did indicate some relationship. The analyses were hampered by low response rates. Among the supplemental hypotheses, the prediction of hypothesis 6 that decision would enhance a contrast effect on product quality perception received no support. The contrast effect remains elusive and the decision opportunity manipulation had no apparent effect. The direct relationship between perception and attitude that was predicted by hypothesis 7 did receive strong support. Evidence for hypothesis 8, postulating a direct relationship between involvement and emotion strength, was weak. The measures of emotions used to detect effects on consumers' emotions did not show the responses that were predicted. There are at least two reasons for this: (1) the measures are not sufficiently sensitive to detect emotions evoked by consumption, and (2) no emotions were evoked. Both of these explanations reflect the fact that the measures were derived using a critical incident technique. It is, perhaps, true that the conditions evoked under those circumstances are more prominent than those felt in response to the typical consumer packaged good (although some descriptions did involve such products). In other words, true satisfaction or dissatisfaction may be relatively rare emotions and more difficult to arouse than so far argued. On a more positive note, the manipulation of discrepancy between expectations and perception seemed to mirror reasonably the results of prior investigations. While subjects were reluctant to admit to "error" in evaluating the poor quality product, their gift selections showed a marked trend away from the adulterated goods. The only
341 manipulation with a significant effect on gift selection was discrepancy. Eighty-four percent (84%) of the subjects in positive discrepancy conditions chose the test brand M (good quality coffee) as an honorarium for participation. In contrast, only 58% of the subjects in negative discrepancy conditions chose the brand that they had tested (poor quality Â— peanut coffee) as an honorarium (chi-square = 18.05, 1 df, p<.0001; overall, chi-square = 22.66, 1 df, p< 01 ) Finally, the major contribution of the analyses, as a body, arises from the revealed effects of decision opportunity. While neither strong nor omnipresent, these effects were sufficient to demonstrate the importance of attending to decision as part of the consumption experience in designing a study. Including a purchase-like decision can help to eliminate a demand artifact (experimental situation effect) that need not exist in a consumer study. Chapter 7 evaluates these findings in light of the literature reviewed and the theory espoused. It concludes by offering proposals for improvements to the present design as well as directions for further investigation.
CHAPTER 7 CONCLUSION Introduction Based on an extensive review of the CS/D literature and related literatures, revisions to the dominant conceptualization of satisfaction (and dissatisfaction) were proposed. The essence of the new conceptualization was that the satisfaction/dissatisfaction continuum is the emotion-laden product of a specific psychological process. The intended contributions of the dissertation were the generation of measures of the construct and the delineation of elements necessary to the process. The research that was described in Chapter 5 and the results reported in Chapter 6 suggest additional modifications to the conceptualization. These results and their conceptual implications are discussed, improvements in the research design are proposed, and additional research that extends application of the conceptualization is suggested. Summary of Findings Pilot Study The original pilot study was conducted as a test of the critical incident technique and to identify the emotional profile of satisfied consumers and dissatisfied consumers. The critical incident methodology was successful at causing subjects to focus on memorable, emotion-laden events. Based on responses to the Differential Emotions 342
343 Scale (DES), criterion levels of responses to emotion scales were established to define the endpoints of a satisfaction/dissatisfaction continuum. By its nature, the critical incident technique evokes memories of rare (critical) events. The motivation (involvement) level is thereby raised and emotions are likely to be strong relative to any aroused by less remarkable events. This fact should alert researchers to the difficulty of arousing satisfaction or dissatisfaction--indeed, any true emotion Â— in an experimental setting. This is not to say that such a manipulation is impossible to execute. Rather, the situation argues for either stronger manipulations that more reliably generate the desired effect or for more attention to the causes and effects of differences in responses by individuals. In addition, the study of satisfaction (especially, but not limited to CS/D) should probably take greater strides toward development of methodologies like critical incident which allow the individual respondent to reach higher arousal levels As documented in the dissertation, the emotions evoked in the experimental setting were not strong. The strategies implemented to examine the effects of several factors and at the same time avoid experimental demand artifacts necessitated more subtle manipulations. The effects that obtained suggest that changes did occur in hypothesized, or at least understandable, patterns. This finding argues for a conclusion that broadens the dogmatic stand taken in Chapters 2 and 3. That is, rather than absolute thresholds defining at a universal level the presence or absence of satisfaction, the patterns
344 of emotions can be seen to vary through longer continua. Response to the strength of these emotions will vary by individual and by situation. Moreover, the strength of satisfaction (or dissatisfaction) reflects the influence of the manipulated factors. Manipulations Thus, although the measurement criteria are open to debate, the proposed conceptualization, "Satisfaction is the feeling which results from expectations that lead to a decision which is followed by personal experience in which the characteristics of the object of the decision are subject to perception which, in instances of sufficient response involvement allows for a conf irmat ion/ d is confirmation of the expectations and decision concerning the object," has not been rejected. Two of the three added constructs, decision and response involvement, have been found to affect some criterion measures of satisfaction The strongest manipulation, determined by manipulation check measures as well as bv the effect on criterion variables, was decision opportunity. The two basic conditions, choice versus assignment of test product, accurately mirrored the methodology described for studies discussed in the literature review. Subjects were able to differentiate accurately between decision and non-decision conditions as well as identify the lack of decision opportunity regarding the cof feemaker There was considerably more variation in the discrepancy manipulation. While the average results were directionally correct, individual-level data were not as encouraging. The manipulation of
345 discrepancy was able to account only for a small amount, less than 5%, of the variance in a scale designed to measure the perception of discrepancy between expectations and achieved performance. Measured expectations tracked well the intended manipulations. As determined in Chapter 6, the fluctuations centered on the perception of the coffee product rather than on expectations generated by advertisements and study instructions. This is likely to have resulted from an assimilation effect exaggerated by the limited personal experience with the product. This finding partially replicates the report of inaccurate product perception in Olson and Dover (1978). Note that while the results reflect limited experience with the product, the ability to personally perceive and evaluate the product was present. As chapter 3 argued, this is an antecedent to the satisfaction process. The perception of evidence as disconf irming may vary depending on the individual's relative confidence in the prior versus posterior information (Ofir 1982). The effect of personal experience versus second-hand information on the outcome of emotional processes is unknown. In order that such strong emotional processes may occur, sufficient motivation (involvement) must be present or aroused by the situation. The intended manipulation of involvement may not have occurred in the main study reported. The effect of the manipulation can be characterized as a variation in personal relevance that had the capacity to stimulate a limited variation in response involvement. The manipulation check measures adopted for this factor did not indicate
346 any reliable effect, although this may have been due to measure insensitivity The problem that arises for satisfaction researchers is that there are no generally-accepted measures Â— nor, indeed, conceptualizations of involvement. The manipulation that was adopted was based on one by Petty et al. (1983) which was later associated with issue involvement, and not response involvement (Petty and Cacioppo 1986). Although the execution reflected the enduring involvement plus situation involvement definition derived by the dissertation, treatment by Petty and Cacioppo conflicts with this. Recent advances in measurement have focused en enduring rather than response involvement (e.g., McQuarrie and Munsor. 1987, Zaichkowsky 1985, 1987). It may be to the advantage of CS/D researchers to focus on enduring involvement and its measures. Rather than seek to manipulate strong motivations and feelings, we might do better to measure them and attempt to explain them at the individual level. This argues for more attention to product, situation and individual interactions than has been paid previously. Individual differences in enduring involvement seem to account for more variance in response than manipulated situational involvement. The better strategy would be, perhaps, to utilize sampling plans resembling self-selected quasi-experimental designs (Cook and Campbell 1979) or make better use of critical incident techniques (Robinson 1976, 1980). The counterargument, of course, is that covariation, blocking, or other explicit recognition or enduring aspects of the subjects is a more tenuous experimental strategy than direct manipulation.
347 The problems experienced with the manipulation check measures recall the discussion presented cogently by Sternthal et al. (1987). Manipulation check measures can be interpreted as one indication of the status of the theoretical construct in the operationalization of an experiment (Perdue and Summers 1986). However, these are neither the only nor the most important indicants of the effectiveness of a manipulation. Measures in CS/D, as in consumer behavior in general, are not sufficiently well-developed that consensus exists regarding even the correct instruments to use. Thus, a comparative approach seems to be well in order. Tests of Hypotheses The main study sought to evaluate specific major and secondary hypotheses that had been derived from the revised conceptualization. As noted, these hypotheses garnered various levels of support resulting in mixed interpretations of the conceptualization. The first hypothesis, dealing with the criterion levels of the satisfaction emotions, found little evidence in its favor. It is possible to explain this in a number of ways: the conceptualization is simply wrong; problems with the DES indicate a lack of sufficient sensitivity to detect differences in consumer behavior; differences among individuals led to unreliability in manipulations and measures; and/or the situation did not arouse sufficiently strong emotions. Of these, the last is most likely and a potential underlying cause to two others. A weak manipulation of involvement could account for emotions too weak to be accurately measured or overwhelmed by individual traits.
348 The first explanation will not be entertained in its fullest at this time What should be recognized and acknowledged at this juncture is the impact of involvement. The dissertation discussed the notion of a "conservation of motivation" as analogous to the laws of thermodynamics. Motivation may be stored and released but is neither created nor lost. The emotional expression of motivation represents an intermediate stage in the progression from involvement to behavior. Whereas the conceptualization, as presented, indicated a minimum threshold of involvement to enable the emotion-generating process (i.e., "sufficient" response involvement), a revised viewpoint would allow a more direct process. That is, strength of emotion: 1) varies with involvement (arousal); and 2) influences likelihood of behavior. From this standpoint, the negative findings on the first hypothesis can be interpreted as arousal of emotions too weak to be characterized by the individuals or determine their behavior (even such simple behaviors as completing the DES). The findings in tests of the second hypothesis bolster the foregoing contention. That is, the measures of satisfaction that were derived from the literature were not supposed to be able to detect variations in decision or response involvement. The fact that they did show a response to these manipulations, as well as to the discrepancy manipulation, is evidence that some consumer behavior process was affected. The lack of such influence on the coffeemaker satisfaction measures argues further for the process having been influenced by the experimental manipulations.
349 This pattern argues for modification of the stance taken by the dissertation. It had been argued that measures of attitude would not be distinguishable from the measures of satisfaction which were adapted from the CS/D literature. The emotional measures of satisfaction were expected to unconfound the constructs, operationalizing the conceptual distinction developed in Chapter 3. This dissertation posed and was attempting to help resolve the question of separability of the constructs. Similarity in behavior of the standard satisfaction measures and the attitude measures was expected for both the coffee and the coffeemaker; the emotional measures were expected to diverge from these. The sensitivity of both former sets of measures to the manipulation of decision opportunity in the case of coffee suggests great similarity in the processes influencing and influenced by each construct. This conclusion is consistent with Oliver's (1981) notion that satisfaction either decays into attitude or decays leaving attitude (analogous to the loss of energy through heat, friction, etc.). What is missing is evidence that the concepts are discrete at all. As emphasized from the introductory chapter forward, satisfaction is thought to comprise the more strongly motivational construct. Thus, there should be distinction in measures of emotional strength (the DES was not sensitive to this in this study) as well as in behavioral outcomes. Hypotheses 4 and 5 examined behaviors. As above, there were no distinct patterns that would differentiate satisfaction effects from attitude effects in this study. One pattern that did emerge simply confirmed that similarity in time and method (e.g., both written
350 responses) enhances correlation between attitude and behavior measures. Two satisfaction measures, Odds and D-T, showed slightly stronger correlations with behavior than did the other measures. This could be because they are anchored in more emotional terms and therefore more consistent with the conceptualization, or simply because there were more responses available to subjects and the improved correlations resulted from less restrictive measurement ranges. The hypothesized effect of decision on perceived discrepancy (hypothesis 6) simply did not occur; in fact the means suggested a slight assimilation rather than contrast effect. Because the decision opportunity manipulation was fairly strong elsewhere, failure to tmd the expected effect was not likely a result of the failure of the manipulation (see LaTour and Peat 1979 for a discussion of problems with contrast effects). This tentative hypothesis was identified as supplementary because of its exploratory nature. Thus, it is better to interpret the results as evidence against the particular process than as failure of either the manipulation or the measures. The measures used in evaluating hypothesis 7 were also questioned subsequent to the analysis. Some of the measures of perception could be construed as partially confounded with attitude due to the evaluative aspects of anchor points. These were the measures that correlated well with the defined attitude scales. Both products, the coffee and the coffeemaker, were chosen on the basis of their having relatively few features. This enabled quick, confident evaluation by the subjects and simpler manipulation of product quality in order to achieve the discrepancy manipulation. The cost of this strategy was
351 paid in not having sufficient perceptual features to provide a rigorous test of this supplemental hypothesis. Although the supporting correlations were high, the lack of correlation on some dimensions raises caution in accepting the simple hypothesis. Finally, the previously cited weakness in the measurement of response involvement can be implicated in the failure to find a correlation between involvement and emotion strength. In addition, the DES may share responsibility for lack of sensitivity in the present situation. If a more liberal interpretation of the conceptualization is accepted, then low response involvement may have generated low levels of emotions and both ranges were too restricted to show correlation. The harsher interpretation is that the involvement level was so low that no emotions were aroused. Implications of Findings The implications of the research can be summarized rather succinctly. The major contribution is the' identification of decision as an important factor in satisfaction processes. Most of the dependent measures were affected by this manipulation, either directly or through an interaction. Omission of decision in a research design that purports to evaluate satisfaction, and probably any post-purchase phenomenon, creates an experience different from what was intended. The importance of the decision construct in satisfaction research, or postpurchase research in general is highlighted in two very recent reports. Tse and Wilton (1988), in a study utilizing a role-playing task, report that product performance outperformed disconf irmation as a cause or predictor of satisfaction. This is not surprising, given the
352 research design that was reported. Subjects' expectations were manipulated by presentation of written product evaluations. Each subject was then assigned a product to evaluate, and allowed to examine and operate the item. As predictable by the conceptualization, performance perceptionÂ—made concrete by personal experienceÂ—dominated the influence on satisfaction of abstract expectations or disconf irmation Â— unsecured by a decision. The importance of decision opportunity is echoed in a discussion by Oliver and DeSarbo (1988). They used decision opportunity, albeit in a role-playing context, to manipulate attribution of locus of control. Although the effect was weak, they concluded, "the locus of attribution effect displays the basic notion that persons are more satisfied when they feel responsible for the decision generating that satisfaction, ceteris paribus (p. 504)." The weakness of the effect that obtained may have been due, in part, to the use of a role-playing task. The lack of personal relevance likely led to low response involvement and, therefore, constrained satisfaction responses. Attempts to address further these issues relating to the effects of decision in the present research were largely unsuccessful for several reasons. The DES proved to be not sufficiently sensitive to detect low levels of emotions, if indeed these were present. The manipulation of response involvement through manipulation of personal relevance was not powerful enough to arouse emotions which would lead to reliable measures or predictable behaviors. Some success was achieved in implementing behavioral measures. While such behaviors as complaining or engaging in word-of-mouth are
353 denigrated in the CS/D literature as measures of satisfaction levels (due primarily to the restricted range which results from low incidence of such behaviors), the study reported here indicates that observation of behavior can be a useful measure when the perceived cost:benefit is approximately equal across individuals. An unexpected contribution of the dissertation arose from the analyses conducted in checking the discrepancy manipulation. After receiving the information in the advertisements and instructions, subjects claimed relatively simple, unidimensional expectations about both products, coffee and coffeemaker. After only a brief experience with each product, the subjects were prepared to make considerably more discriminating judgments. This finding has practical applications in both research and managerial settings, as it suggests storage or availability of more complex semantic information regarding the item. Research in Marketing frequently depends on single exposures to novel stimuli. The level of detail in the advertisements used in this study was not remarkably different from that in much of consumer research. Thus, it can be argued that the simple expectations that resulted were also typical of what might occur under laboratory conditions. The elaboration of the evaluative framework after a briet experience with the product suggests that efforts to compare expectations and perceptions directly may be misleading, at least when dealing with novel stimuli (Cadotte et al. 1983, Scott and Yalch 1930). For CS/D researchers interested in theory development, this is an interesting process to map and explain. For others, it is important to note that there is likely to be a considerably different evaluation
354 generated between a naive consumer and one with even minor personal experience This difference has market implications as well. Advertisement pretesting frequently omits trial or disconf irmation as part of the design. Whereas an advertisement may elicit a generally positive response, failure or weakness on one or more attributes may create a decidedly more negative impression than once thought. Even if the post-purchase evaluation is compensatory, discriminable attribute scores allow for potentially unfavorable weightings. This argues for closer attention to consumer experience in the initial evaluation of products and communications. Future Outlook Revisions to Present Study If even a partial replication of the research reported here were to be undertaken, attention to and correction of some of the limitations would be necessary. The most crucial change would focus on response involvement. Alternatives to the strategy pursued in the present study, manipulating involvement within product, would involve manipulating involvement by using different products (although defining equivalence on all other dimensions would be nearly impossible) or by treating involvement as an individual difference variable in a blocking or covariate design. Any of these approaches requires an ability to measure the construct more accurately than is presently feasible. Measures of enduring involvement do not weigh the instantaneous arousal implied by the "conservation of motivation" formulation (e.g., Apsler and Sears 1968, Rhine and Severance 1970).
155 The DES needs additional development to be fully appropriate to marketing and consumer behavior situations. An expansion of the response categories may be adequate to increase the sensitivity of the instrument. Further development work could produce a research stream of its own. Similarly, the development of behavioral measurement technology for CS/D would be a fruitful line of inquiry to pursue. In the confines of improving the present study, however, it is sufficient to enhance the sensitivity of these measures as well. This can be accomplished by adopting a callback plan that generates a higher response rate, more strongly encouraging behaviors that do not require contact by the researcher (e.g., give each respondent a postcard that can be returned as a "vote" in favor of the product and one that votes against the product), or even better handling of unstructured responses extensions Additional research beyond this basic design can also extend the findings or correct some of the limitations. The major finding was that decision opportunity affected many of the dependent variables of interest. The decision manipulation was strong, but simple. It would be of interest to examine the influence of variations of the decision opportunity along the lines proposed by Kiesler for the commitment construct (1971 Â— see Chapter 3). A more recent investigation, not yet fully analyzed nor documented, used the DES to investigate the satisfaction levels of recent homebuyers (recent was defined as within two years prior to the data collection). In this study, out of 162
356 respondents who met the ownership tenure criterion, 76 met or exceeded the 3.0 satisfaction criterion and only seven met the 3.0 dissatisfaction criterion (interestingly, four met or exceeded both criteria) Additional efforts to document use of the instrument in field settings would be helpful in improving the measures for general use. The influence of the apparent changes in cognitive structure should be examined more carefully. Why/how often does this occur? How soon does structure stabilize? Is this an example of schema formation? Does the satisfaction reaction generalize across the satisfaction types proposed in Table 2? On a similar note, is advertising perceived differently by novice versus experienced consumers? How much experience is enough to stabilize processing? Finally, what other factors are important in the satisfaction process? Do these models truly generalize across satisfaction domains (e.g., employee satisfaction, marital satisfaction, life satisfaction, etc.)? At what point do satisfaction and attitude merge? At what point do they separate? Are merge and separate adequate characterizations? In summary, there is still much to be learned and understood about satisfaction. What is needed is a fuller elaboration of a consensus paradigm that can provide ideas and strategies for research.
APPENDIX A PILOT STUDY QUESTIONNAIRE Name (optional Sex Age CONSUMER RESEARCH PROJECT Please think about an occasion on which you, as a consumer, felt particularly dissatisfied with a product. Please answer the following questions with reference to this one occasion 1 What type of product was involved? How long ago did this occur (approximately) Briefly describe the situation; be sure to mention anything you mav have done afterward. PLEASE DON'T GO ON UNTIL TOLD TO DO SO. 357
358 Appendix A--continued This portion of the study consists of a number of words that describe different emotions or feelings. Please indicate the extent to which each word describes the way you felt on the occasion you described in question 3. Record your answers by circling the appropriate number on the five-point scale following each word. Presented below is the response scale for indicating the degree to which each word describes the way you felt. 12 3 4 5 Very Slightly Slightly Moderately Considerably Very or not at all Strongly In deciding on your answer to a given item or word, consider the feeling connoted or defined by that word. For example, if during the occasion you described in question 3 you felt very slightly or not at all "interested" you would circle the number 1 on the scale if you felt "interested" to a moderate degree you would circle 3; if you felt very strongly interested you would circle 5, and so forth. Remember, you should respond based on how you felt during the occasion you described in question 3. Please do not consider events that occurred afterwards, (e.g., a complaint that was resolved). Work at a good pace. It is not necessary to ponder; the first answer you decide on for a given word is probably the most valid. You should be able to finish this section in about 5 minutes Fiease turn to the next page and begin with the word marked by an arrow (-->). Choose a response for each word, in turn, until you reach the end of the list.
359 Appendix A--continued 1 Very Slightly or not at all
360 Appendix A Â— continued 1 Very Slightly or not at all
36' Appendix A--continued a. How satisfied were you with the product? Very Somewhat Slightly Mixed Slightly Dissatisfied Dissatisfied Dissatisfied Satisfied Somewhat Very Neither Don't Know/ Satisfied Satisfied Don't Care i. Knowing what you know now if a similar situation arose, what ari the chances in ten (10) that you would purchase (use) the product again? 0123456789 10 No Chance Certain c. Overall, were you satisfied or dissatisfied with the product? 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% .100% Completely Half & Completely Dissatisfied Half Satisfied Which word or phrase most closely reflects your satisfaction with the product? Delighted Pleased Mostly Mixed Mostly Unhappy Terrible Satisfied Dissatisfied Neutral Never Thought About It
36: Appendix A--continued Please circle the word or phrase that most clearly reflects your agreement with each statement. I am satisfied with my decision to purchase (use) the product. Strongly Agree Agree Neither Disagree Strongly Disagree 2 If I had it to do all over again, I would feel differently about the product. Strongly Agree Agree Neither Disagree Strongly Disagree 3 My choice of the product was a wise one. Strongly Agree Agree Neither Disagree Strongly Disagree I feel bad about my decision concerning the product. Strongly Agree Agree Neither Disagree Strongly Disagree 5 I think that I did the right thing when I decided to get (use) the product. Strongly Agree Agree Neither Disagree Strongly Disagree I am not happy that I did what I did about the product. Strongly Agree Agree Neither Disagree Strongly Disagree f. What would you normally do under the circumstances of the occasion you described in question 3? Please circle the word or phrase that most closely reflects your agreement with each statement no matter whether or no t you actually responded that way on this particular occasion 7 I would buy other products of the same brand. Strongly Agree Agree Neither Disagree Strongly Disagree Â£ I would buy other products from the same seller. Strongly Agree Agree Neither Disagree Strongly Disagree
363 Appendix A Â— continued 9 I would buy other products of the same type. Strongly Agree Agree Neither Disagree Strongly Disagree 10 I would buy more of the same product as a gift. Strongly Agree Agree Neither Disagree Strongly Disagree 11 I would buy similar products as gifts. Strongly Agree Agree Neither Disagree Strongly Disagree 1 2 I would complain to a "higher authority" (owner, manager) Strongly Agree Agree Neither Disagree Strongly Disagree 13 I would extend a compliment to a "higher authority" (owner, manager) Strongly Agree Agree Neither Disagree Strongly Disagree 14 I would tell a friend the good things about my experience. Strongly Agree Agree Neither Disagree Strongly Disagree 15 I would tell a friend the bad things about my experience. Strongly Agree Agree Neither Disagree Strongly Disagree 15 I would write directly to the manufacturer (original supplier) and complain. Strongly Agree Agree Neither Disagree Strongly Disagree 17 I would write directly to the manufacturer (original supplier) to compliment them. Strongly Agree Agree Neither Disagree Strongly Disagree
364 Appendix A Â— continued I would discard the product and replace it with a different brand as soon as possible. Strongly Agree Agree Neither Disagree Strongly Disagree 9 I would discard the product and replace it with a different product as soon as possible. Strongly Agree Agree Neither Disagree Strongly Disagree 2 3 I would recommend the product to a friend. Strongly Agree Agree Neither Disagree Strongly Disagree 21 I would recommend a friend avoid the product. Strongly Agree Agree Neither Disagree Strongly Disagree 22 I would publicly endorse the product (e.g., in advertising). Strongly Agree Agree Neither Disagree Strongly Disagree 23 I would publicly attack the product (e.g., in advertising). Stronglv Agree Agree Neither Disagree Strongly Disagree 1L I would fight to have the product removed from market. Strongly Agree Agree Neither Disagree Strongly Disagree
APPENDIX B INSTRUCTIONS FOR ADMINISTRATION OF EACH EXPERIMENTAL CONDITION A-H Set-Up Instructions: Cell 1 Common Set-up: paper towels clean cup and spoon sugar, creamer, sweetener pencil coffee filters measuring spoons Special Set-up: gift packages: out of view drawing cards: visible coffee maker: red water bottle: red coffee instructions: red brand M: Uncle Charlie's brand N: generic ad booklet: white scissors 365
366 Appendix B Â— continued SPECIFIC INSTRUCTIONS CELL 1 CONDITION 001 LOW INVOLVEMENT, NO DECISION, POSITIVE DISCREPANCY On arrival, Subject is read informed consent: The study in which you are participating concerns your evaluations of certain common products. You will be asked to sample products and complete a questionnaire as part of the study. Your responses will be kept confidential. You are free to withdraw from the study at any time. There is no monetary compensation for participation. I will try to answer any questions you may have about the procedures. The Subject completes the "Food Sensitivity Questionnaire". Study terminates if sensitivity to caffeine is indicated. The cover story for the study is read to the subject. Thank you for agreeing to participate in our study. Today we are asking consumers like yourself to evaluate a coffeemaker that may be introduced by a major food processing company. The company has modified an existing design for easier use. In return for your assistance and taking the time to contribute your honest opinions you will be given a chance to win a coffee maker in a drawing that will be conducted at the end of this phase of the study Â— that is, in about 4 weeks.
367 Appendix B Â— continued 4. Read the instruction, Before you actually try any product, we would like for you to look at some sample print advertisements. These ads are for the coffee maker and the company's own brand of coffee, also to be introduced soon. The coffee will be of adequate quality, but is primarily intended to be a cheap blend to help consumers cope with high prices brought about by the recent shortage. We are helping the company identify the minimal acceptable quality for the home market. On the sheet provided, please write down what you think is the main product feature that each ad is trying to communicate. Then rate how convincing you find the ad to be. Do this for each of the ads. Give the Subject the book of "negative" ads and the rating form. The Subject is shown the test coffee. Over the course of the study, we're comparing evaluations of several brands of coffee, including brand M and brand N. Today we'd like for you to evaluate brand M. You've just read some advertisements for this brand. As you use the coffee maker, remember that you will have a chance to win one from the same company. 6. Show Subject coffeemaker instructions and have Subject make a full (4-cup) pot. As the coffee brews, give the Subject the "Classification" form to comDlete 7. The Subject tastes the coffee which he/she has prepared. (The use of cream and/or sugar should be allowed, but not encouraged.)
Appendix B--continued 8. The Subject is asked to complete the questionnaire containing the major measures. Offer the Subject a choice of gift coffee. In addition to the chance at the coffee maker that you were promised, I'm happy to be able to offer you a week's supply of coffee as a gift. You may choose either brand M (which you've just tried) or brand N, a locally available generic brand. Mark the brand selected on the last page of the questionnaire at column 62. 9. The subject is thanked for participation, entered into the coffee maker drawing, given gifts and dismissed. Interviewer initials the back of the questionnaire. 10. After one week, delay measures are collected by telephone: self-report of Word-of-Mouth, self-report of the amount of use, self-report of purchase or attempt to purchase the product, Also, comment cards will be provided. Those returned within one week will provide a measure of complaining/ complimenting. 11. After collection of the delay measures, the Subject will be questioned about suspicions regarding the hypotheses. The subject is then debriefed.
369 Appendix B Â— continued Set-Up Instructions: Cell 2 Common Set-up: paper towels clean cup and spoon sugar, creamer, sweetener pencil coffee filters measuring spoons Special Set-up: gift packages: visible drawing cards: out of sight coffee maker: red water bottle: red coffee instructions: red brand M: Uncle Charlie's brand N: generic ad booklet: white scissors
370 Appendix B--continued SPECIFIC INSTRUCTIONS CELL 2 CONDITION 101 HIGH INVOLVEMENT, NO DECISION, POSITIVE DISCREPANCY On arrival, Subject is read informed consent: The study in which you are participating concerns your evaluations of certain common products. You will be asked to sample products and complete a questionnaire as part of the study. Your responses will be kept confidential. You are free to withdraw from the study at any time. There is no monetary compensation for participation. I will try to answer any questions you may have about the procedures. 2. The Subject completes the "Food Sensitivity Questionnaire' Study terminates if sensitivitv to caffeine is indicated. 3. The cover story for the study is read to the Subject. Thank you for agreeing to participate in our study. Today we are asking consumers like yourself to evaluate coffee made by a major food processing company. The coffee will be of adequate quality, but is primarily intended to be a cheap blend to help consumers cope with high prices brought about by the recent shortage. We are helping them to identify the minimal acceptable quality for the home market. In return for your assistance and taking the time to contribute your honest opinions you will be given a week's supply of the coffee.
37' Appendix B Â— continued 4. Read the instruction, Before you actually try any product, we would like for you to look at some sample print advertisements. These ads are for the coffee and a related line of coffee makers. On the sheet provided, please write down what you think is the main product feature that each ad is trying to communicate. Then rate how convincing you find the ad to be. Do this for each of the ads Give the Subject the book of "negative" ads and the rating form. The Subject is shown the test coffee. Over the course of the study, we're comparing evaluations of several brands of coffee, including brand M and brand N. Today we'd like for you to evaluate brand M. You've just read some advertisements for this brand. Remember, you will be given a week's supply of the brand you have evaluated. Give the Subject the gift bag at this point. 6. Show Subject coffeemaker instructions and have Subject make a full (4-cup) pot. As the coffee brews, give the Subject the "Classification" form to complete. 7. The Subject tastes the coffee which he/she has prepared. (The use of cream and/or sugar should be allowed, but not encouraged.)
372 Appendix B Â— continued The Subject is asked to complete the questionnaire containing the major measures. Offer the Subject a choice of gift coffee. At this time, I'd like for you to choose your gift of coffee. You may either keep brand M (which you've just tried) or choose brand N, a locally available generic brand. You will also receive a chance to win a coffee maker similar to this one in a drawing to be held at the end of this phase of the study, in about six weeks. Mark the selection on the last page of the questionnaire at column 62. 9. The subject is thanked for participation, entered into the coffee maker drawing, given gifts and dismissed. Interviewer initials the back of the questionnaire. 10. After one week, delay measures are collected by telephone: self-report of Word-of-Mouth, self-report of the amount of use, self-report of purchase or attempt to purchase the product. Also, comment cards will be provided. Those returned within om week will provide a measure of complaining/complimenting. 11. After collection of the delay measures, the Subject will be questioned about suspicions regarding the hypotheses. The subject is then debriefed.
373 Appendix B Â— continued Set-Up Instructions: Cell 3 Common Set-up: paper towels clean cup and spoon sugar, creamer, sweetener pencil coffee filters measuring spoons Special Set-up: gift packages: out of sight drawing cards: visible coffee maker: black water bottle: black coffee instructions: black brand M: peanut brand N: Food Club ad booklet: green can opener
374 Appendix B Â— continued SPECIFIC INSTRUCTIONS CELL 3 CONDITION 002 LOW INVOLVEMENT, NO DECISION, NEGATIVE DISCREPANCY 1. On arrival, Subject is read informed consent: The study in which you are participating concerns your evaluations of certain common products. You will be asked to sample products and complete a questionnaire as part of the study. Your responses will be kept confidential. You are free to withdraw from the study at any time. There is no monetary compensation for participation. I will try to answer any questions you may have about the procedures. The Subject completes the "Food Sensitivity Questionnaire". Study terminates if sensitivity to caffeine or peanuts/peanut products is indicated. 3. The cover story for the study is read to the subject. Thank you for agreeing to participate in our study. Today we are asking consumers like yourself to evaluate a coffeemaker that may be introduced by a major food processing company. The company has modified an existing design for easier use. In return for your assistance and taking the time to contribute your honest opinions you will be given a chance to win a coffee maker in a drawing that will be conducted at the end of this phase of the study Â— that is, in about 4 weeks. 4. Read the instruction, Before you actually try any product, we would like for you to look at some sample print advertisements. These ads are for the coffee maker and the company's own brand of coffee, also to be introduced soon.
375 Appendix B--continued The coffee will be very high in quality but only slightly more expensive than present top-of-theline coffees. We are helping the company identify the quality most preferred by consumers. On the sheet provided, please write down what you think is the main product feature that each ad is trying to communicate. Then rate how convincing you find the ad to be. Do this for each of the ads Give the Subject the book of "positive" ads and the rating form. 5. The Subject is shown the test coffee. Over the course of the study, we're comparing evaluations of several brands of coffee. Today we'd like for you to evaluate brand M. You've just read some advertisements for this brand. As you use the coffee maker, remember that you will have a chance to win one from the same company. 6. Show Subject coffeemaker instructions and have Subject make a full (4-cup) pot. As the coffee brews, give the Subject the "Classification" form to cormolete 7. The Subject tastes the coffee which he/she has prepared. (The use of cream and/or sugar should be allowed, but not encouraged.) The Subject is asked to complete the questionnaire containing the major measures. Offer the Subject a choice of gift coffee. In addition to the chance at the coffeemaker that you were promised, I'm happy to be able to offer you a week's supply of coffee as a gift. You may choose either brand M (which you've just tried) or brand N, a popular brand in this area. Mark selection on the last page of the questionnaire at column 62.
376 Appendix B--continued 9. The Subject is thanked for participation, entered into the coffeemaker drawing, given gifts, and dismissed. Interviewer initials the back of the questionnaire. 10. After one week, delay measures are collected by telephone: self-report of Word-of-Mouth, self-report of the amount of use, self-report of purchase or attempt to purchase the product. Also, comment cards will be provided. Those returned within one week will provide a measure of complaining/ complimenting 11. After collection of the delay measures, the Subject will be questioned about suspicions regarding the hypotheses. The subject is then debriefed.
377 Appendix B--continued Set-Up Instructions: Cell 4 Common Set-up: paper towels clean cup and spoon sugar, creamer, sweetener pencil coffee filters measuring spoons Special Set-up: gift packages: visible drawing cards: out of sight coffeemaker: black water bottle: black coffee instructions: black brand M: peanut brand N: Food Club ad booklet: green can opener
378 Appendix B--continued SPECIFIC INSTRUCTIONS CELL 4 CONDITION 102 HIGH INVOLVEMENT, NO DECISION, NEGATIVE DISCREPANCY 1. On arrival, Subject is read informed consent: The study in which you are participating concerns your evaluations of certain common products. You will be asked to sample products and complete a questionnaire as part of the study. Your responses will be kept confidential. You are free to withdraw from the study at any time. There is no monetary compensation for participation. I will try to answer any questions you may have about the procedures. The Subject completes the "Food Sensitivity Questionnaire. Study terminates if sensitivity to caffeine or peanut/ peanut products is indicated. 3. The cover story for the study is read to the subject. Thank you for agreeing to participate in our study. Today we are asking consumers like yourself to evaluate coffee made by a major food processing company. The company plans to introduce a very high quality coffee that is only slightly more expensive than present top-of-the-Iine coffees We are helping them identify the quality most preferred by consumers. In return for your assistance and taking the time to contribute your honest opinions you will be given a week's supply of the coffee. 4. Read the instruction, Before you actually try any product, we would like for you to look at some sample print advertisements. These ads are for the coffee and a related line of cof f eemakers
379 Appendix B Â— continued On the sheet provided, please write down what you think is the main product feature that each ad is trying to communicate. Then rate how convincing you find the ad to be. Do this for each of the ads Give the Subject the book of "positive" ads and the rating form. 5. The Subject is shown the test coffee. Over the course of the study, we're comparing evaluations of several brands of coffee, including brand M and brand N. Today we'd like for you to evaluate brand M. You've just read some advertisements for this brand. Remember, you will be given a week's supply of the brand you have evaluated. Give the Subject the gift coffee at this point. 6. Show Subject coffeemaker instructions and have Subject make a full (4-cup) pot. As the coffee brews, give the Subject the "Classification" form to comDlete. 7. The Subject tastes the coffee which he/she has prepared. (The use of cream and/or sugar should be allowed, but not encouraged.) The Subject is asked to complete the questionnaire containing the major measures. Offer the Subject a choice of gift coffee. At this time, I'd like for you to choose your gift of coffee. You may either keep brand M (which you've just tried) or choose brand N, a popular brand in this area. You will also receive a chance to win a coffeemaker similar to this one in a drawing to be held at the end of this phase of the study, in about six weeks. Mark the brand selected on the last page of the questionnaire at column 62.
380 Appendix B Â— continued 9. The subject is thanked for participation, entered into the coffeemaker drawing, given gifts and dismissed. Interviewer initials the back of the questionnaire. 10. After one week, delay measures are collected by telephone: self-report of Word-of-Mouth, self-report of the amount of use, self-report of purchase or attempt to purchase the product. Also, comment cards will be provided. Those returned within one week will provide a measure of complaining/ complimenting. 11. After collection of the delay measures, the Subject will be questioned about suspicions regarding the hypotheses. The subject is then debriefed.
381 Appendix B--continued Set-Up Instructions: Cell 5 Common Set-up: paper towels clean cup and spoon sugar, creamer, sweetener pencil coffee filters measuring spoons Special Set-up: gift packages: out of sight drawing cards: visible coffeemaker: red water bottle: red coffee instructions: red brand M: Uncle Charlie brand N: generic ad booklet: white scissors
382 Appendix B Â— continued SPECIFIC INSTRUCTIONS CELL 5 CONDITION 01 1 LOW INVOLVEMENT, DECISION, POSITIVE DISCREPANCY 1. On arrival, Subject is read informed consent: The study in which you are participating concerns your evaluations of certain common products. You will be asked to sample products and complete a questionnaire as part of the study. Your responses will be kept confidential. You are free to withdraw from the study at any time. There is no monetary compensation for participation. I will try to answer any questions you may have about the procedures. The Subject completes the "Food Sensitivity Questionnaire. Study terminates if sensitivity to caffeine is indicated. 3. The cover story for the study is read to the subject. Thank you for agreeing to participate in our study. Today we are asking consumers like yourself to evaluate a coffeemaker made by a major food processing company. The company has modified an existing design for easier use. In return for your assistance and taking the time to contribute your honest opinions you will be given a chance to win a coffeemaker in a drawing that will be conducted at the end of this phase of the study Â— that is, in about 4 weeks. 4. Read the instruction, Before you actually try any product, we would like for you to look at some sample print advertisements. These ads are for the coffeemaker and the company's own brand of coffee, also to be introduced soon. The coffee will be of adequate quality, but is primarily intended to be a cheap blend to help consumers cope with high prices brought about by the recent shortage. We are helping the company
383 Appendix B Â— continued identify the minimal acceptable quality for the home market. On the sheet provided, please write down what you think is the main product feature that each ad is trying to communicate. Then rate how convincing you find the ad to be. Do this for each of the ads. Give the Subject the book of "negative" ads and the rating form. 5. The Subject is shown the test coffee. Over the course of the study, we're comparing evaluations of several brands of coffee, including brand M and brand N. M was the brand described in the ads which you've just read. N is a locally available generic coffee. We are, of course, most interested in your reaction to brand M, but since we're testing both M and N, you're free to choose either one. Please indicate the brand you have chosen in the space provided on the product evaluation form. Subject is given questionnaire and allowed to indicate choice of "Brand Selected" and "Brand Rejected". If Brand M is not selected, continue interview but do not count toward quota for this cell. As you use the coffeemaker, remember that you will have a chance to win one from the same company. 6. Show Subject coffeemaker instructions and have Subject make a full (4-cup) pot. As the coffee brews, give the Subject the "Classification" form to complete. 7. The Subject tastes the coffee which he/she has prepared. (The use of cream and/or sugar should be allowed, but not encouraged.)
384 Appendix B Â— continued The Subject is asked to complete the questionnaire containing the major measures. Offer the Subject a choice of gift coffee. In addition to the chance at the coffeemaker that you were promised, I'm happy to be able to offer you a week's supply of coffee as a gift. You may choose either brand M or brand N. Mark the brand selected on the last page of the questionnaire at column 62. 9. The subject is thanked for participation, entered into the coffeemaker drawing, given gifts and dismissed. Interviewer initials the back of the questionnaire. 10. After one week, delay measures are collected by telephone: self-report of Word-of-Mouth, self-report of the amount of use, self-report of purchase or attempt to purchase the product. Also, comment cards will be provided. Those returned within one week will provide a measure of complaining/ complimenting 11. After collection of the delay measures, the Subject will be questioned about suspicions regarding the hypotheses. The subject is then debriefed.
385 Appendix B--continued Set-Up Instructions: Cell 6 Common Set-up: paper towels clean cup and spoon sugar, creamer, sweetener pencil coffee filters measuring spoons Special Set-up: gift packages: visible drawing cards: out of sight coffeemaker: red water bottle: red coffee instructions: red brand M: Uncle Charlie's brand N: generic ad booklet: white scissors
386 Appendix B Â— continued SPECIFIC INSTRUCTIONS CELL 6 CONDITION 1 1 1 HIGH INVOLVEMENT, DECISION, POSITIVE DISCREPANCY 1. On arrival, Subject is read informed consent: The study in which you are participating concerns your evaluations of certain common products. You will be asked to sample products and complete a questionnaire as part of the study. Your responses will be kept confidential. You are free to withdraw from the study at any time. There is no monetary compensation for participation. I will try to answer any questions you may have about the procedures. 2. The Subject completes the "Food Sensitivity Questionnaire. Study terminates if sensitivity to caffeine is indicated. 3. The cover story for the study is read to the subject. Thank you for agreeing to participate in our study. Today we are asking consumers like yourself to evaluate coffee made by a major food processing company. The coffee will be of adequate quality, but is primarily intended to be a cheap blend to help consumers cope with high prices brought about by the recent shortage. We are helping them to identify the minimal acceptable quality for the home market. In return for your assistance and taking the time to contribute your honest opinions you will be given a week's supply of the coffee you choose. Read the instruction, Before you actually try any product, we would like for you to look at some sample print advertisements. These ads are for the coffee and a related line of cof feemakers
387 Appendix B Â— continued On the sheet provided, please write down what you think is the main-product feature that each ad is trying to communicate. Then rate how convincing you find the ad to be. Do this for each of the ads Give the Subject the book of "negative" ads and the rating form. 5. The Subject is shown the coffee choices. Over the course of the study, we're comparing evaluations of several brands of coffee. Today we have both brand M and brand N. M was the brand described in the ads which you've just read. N is a locally available generic coffee. We are, of course, most interested in your reaction to brand M, but since we're testing both M and N, you're free to choose either one. Remember, you will be given a week's supply of the brand which you have chosen. Please indicate the brand you have chosen in the space provided on the product evaluation form. Subject is given questionnaire and allowed to indicate choice of "Brand Selected" and "Brand Rejected". If Brand M is not selected, continue interview but do not count toward quota for this ceil. Give the Subject the gift which was selected. 6. Show Subject coffeemaker instructions and have Subject make a full (4-cup) pot. As the coffee brews, give the Subject the "Classification" form to complete. 7. The Subject tastes the coffee which he/she has prepared. (The us of cream and/or sugar should be allowed, but not encouraged.)
388 Appendix B Â— continued 8. The Subject is asked to complete the questionnaire containing the major measures. Offer the Subject a choice of gift coffee. At this time, I'd like for you to confirm your choice of gift. You may either keep the brand you chose earlier or exchange it for the other brand. You will also receive a chance to win a coffeemaker similar to this one in a drawing to be held at the end of this phase of the study, in about six weeks. Mark the brand selected on the last page of the questionnaire at column 62. 9. The subject is thanked for participation, entered into the coffeemaker drawing, given gifts and dismissed. Interviewer initials the back of the questionnaire. 10. After one week, delay measures are collected by telephone: self-report of Word-of-Mouth, self-report of the amount of use, self-report of purchase or attempt to purchase the product. Also, comment cards will be provided. Those returned within one week will provide a measure of complaining/ complimenting 11. After collection of the delay measures, the Subject will be questioned about suspicions regarding the hypotheses. The subject is then debriefed.
389 Appendix B Â— continued Set-Up Instructions: Cell 7 Common Set-up: paper towels clean cup and spoon sugar, creamer, sweetener pencil coffee filters measuring spoons Special Set-up: gift packages: out of sight drawing cards: visible coffeemaker: black water bottle: black coffee instructions: black brand M: peanut brand N: Food Club ad booklet: green can opener
390 Appendix B Â— continued SPECIFIC INSTRUCTIONS CELL 7 CONDITION 012 LOW INVOLVEMENT, DECISION, NEGATIVE DISCREPANCY 1. On arrival, Subject is read informed consent: The study in which you are participating concerns your evaluations of certain common products. You will be asked to sample products and complete a questionnaire as part of the study. Your responses will be kept confidential. You are free to withdraw from the study at any time. There is no monetary compensation for participation. I will try to answer any questions you may have about the procedures. 2. The Subject completes the "Food Sensitivity Questionnaire, Study terminates if sensitivity to caffeine or peanuts/ peanut products is indicated. The cover story for the study is read to the subject. Thank you for agreeing to participate in our study. Today we are asking consumers like yourself to evaluate coffeemaker that may be introduced by a major food processing company. The company has modified an existing design for easier use. In return for your assistance and taking the time to contribute your honest opinions you will be given a chance to win a coffeemaker in a drawing that will be conducted at the end of this phase of the study--that is, in about 4 weeks. Read the instruction, Before you actually try any product, we would like for you to look at some sample print advertisements. These ads are for the coffeemaker and the company's own brand of coffee, also to be introduced soon.
391 Appendix B--continued The coffee will be very high in quality, but only slightly more expensive than present top-of-theline coffees. We are helping the company identify the quality most preferred by consumers. On the sheet provided, please write down what you think is the main product feature that each ad is trying to communicate. Then rate how convincing you find the ad to be. Do this for each of the ads Give the Subject the book of "positive" ads and the rating form. 5. The Subject is shown the coffee choices. Over the course of the study, we're comparing evaluations of several brands of coffee. Today we have both brand M and brand N. M was the brand described in the ads which you've just read. N is a popular brand in this area. We are, of course, most interested in your reaction to brand M, but since we're testing both M and N, you're free to choose either one. Please indicate the brand you have chosen in the space provided on the product evaluation form. Subject is given questionnaire and allowed to indicate choice of "Brand Selected" and "Brand Rejected". If Brand M is not selected, continue interview but do not count toward quota for this cell. As you use the coffeemaker, remember that you will have a chance to win one from the same company. 6. Show Subject coffeemaker instructions and have Subject make a full (4-cup) pot. As the coffee brews, give the Subject the "Classification" form to complete. 7. The Subject tastes the coffee which he/she has prepared. (The use of cream and/or sugar should be allowed, but not encouraged.)
392 Appendix B Â— continued The Subject is asked to complete the questionnaire containing the major measures. Offer the Subject a choice of gift coffee. In addition to the chance at the coffeemaker that you were promised, I'm happy to be able to offer you a week's supply of coffee as a gift. You may choose either brand M or brand N. Mark the brand selected on the last page of the questionnaire at column 62. 9. The subject is thanked for participation, entered into the coffeemaker drawing, given gifts and dismissed. Interviewer initials the back of the questionnaire. 10. After one week, delay measures are collected by telephone: self-report of Word-of-Mouth, self-report of the amount of use, self-report of purchase or attempt to purchase the product. Also, comment cards will be provided. Those returned within one week will provide a measure of complaining/ complimenting. 11. After collection of the delay measures, the Subject will be questioned about suspicions regarding the hypotheses. The subject is then debriefed.
393 Appendix B Â— continued Set-Up Instructions: Cell 8 Common Set-up: paper towels clean cup and spoon sugar, creamer, sweetener pencil coffee filters measuring spoons Special Set-up: gift packages: visible drawing cards: out of sight coffeemaker: black water bottle: black coffee instructions: black brand M: peanut brand N: Food Club ad booklet: green can opener
394 Appendix B Â— continued SPECIFIC INSTRUCTIONS CELL 8 CONDITION 112 HIGH INVOLVEMENT, DECISION, NEGATIVE DISCREPANCY 1. On arrival, Subject is read informed consent: The study in which you are participating concerns your evaluations of certain common products. You will be asked to sample products and complete a questionnaire as part of the study. Your responses will be kept confidential. You are free to withdraw from the study at any time. There is no monetary compensation for participation. I will try to answer any questions you may have about the procedures. The Subject completes the "Food Sensitivity Questionnaire. Study terminates if sensitivity to caffeine or peanut/ peanut products is indicated. 3. The cover story for the study is read to the subject. Thank you for agreeing to participate in our study. Today we are asking consumers like yourself to evaluate coffee made by a major food processing company. The company plans to introduce a very high quality coffee that is only slightly more expensive than present top-of-the-line coffees We are helping them identify the quality most preferred by consumers. In return for your assistance and taking the time to contribute your honest opinions you will be given a week's supply of the coffee you choose. Read the instruction, Before you actually try any product, we would like for you to look at some sample print advertisements. These ads are for the coffee and a related line of cof f eemakers
395 Appendix B Â— continued On the sheet provided, please write down what you think is the main product feature that each ad is trying to communicate. Then rate how convincing you find the ad to be. Do this for each of the ads Give the Subject the book of "positive" ads and the rating form. 5. The Subject is shown the test coffee. Over the course of the study, we're comparing evaluations of several brands of coffee. Today we have both brand M and brand N. M was the brand described in the ads which you've just read. N is a popular brand in this area. We are, of course, most interested in our reaction to brand M, but since we're testing both M and N, you're free to choose either one. Remember, you will be given a week's supply of the brand which you have chosen. Please indicate the brand you have chosen in the space provided on the product evaluation form. Subject is given questionnaire and allowed to indicate choice of "Brand Selected" and "Brand Rejected". If Brand M is not selected, continue interview but do not count toward quota for this cell. Give the Subject the gift which was selected. 6. Show Subject coffeemaker instructions and have Subject make a full (4-cup) pot. As the coffee brews, give the Subject the "Classification" form to complete. 7. The Subject tastes the coffee which he/she has prepared. (The use of cream and/or sugar should be allowed, but not encouraged.) The Subject is asked to complete the questionnaire containing the maior measures.
396 Appendix B Â— continued Offer the Subject a choice of gift coffee. At this time, I'd like for you to confirm your choice of a gift. You may either keep the brand you chose earlier or exchange it for the other brand. You will also receive a chance to win a coffeemaker similar to this one at the end of this phase of the study, in about six weeks. Mark the brand selected on the last page of the questionnaire at column 62. 9. The subject is thanked for participation, entered into the coffeemaker drawing, given gifts and dismissed. Interviewer initials the back of the questionnaire. 10. After one week, delay measures are collected by telephone: self-report of Word-of-Mouth, self-report of the amount of use, self-report of purchase or attempt to purchase the product. Also, comment cards will be provided. Those returned within one week will provide a measure of complaining/ complimenting 11. After collection of the delay measures, the Subject will be questioned about suspicions regarding the hypotheses. The subject is then debriefed.
APPENDIX C QUESTIONNAIRE AND SUBJECT FORMS FROM MAIN STUDY University of Akron Division of Consumer Testing Food Sensitivity Questionnaire Many of the product studies which we administer include an opportunity to sample various foods. When this is the case, we ask participants to notify us of any known allergies or other dietary restrictions. You will not be asked to try any products containing those ingredients you have told us about. PLEASE PLACE A CHECK (vO NEXT TO ANY OF THE FOLLOWING WHICH YOU CANNOT OR PREFER NOT TO EAT FOR ANY REASON SIMPLY MARK EACH SUCH ITEM. YOU WILL NOT BE ASKED TO EXPLAIN ANY SELECTION. Sugar Salt Caffeine Chocolate Saccharin Aspartame (NutraSweet) MSG (monosodium glutamate) Other artificial sweeteners Other artificial preservatives Fish Beef /beef derivatives Pork/pork derivatives Poultry/poultry derivatives Liver/other organ meats Milk/dairy products Calcium Eggs/egg derivatives Peanuts/peanut products Other nuts. 'nut products Vegetat les ( specify) Fruits (specify) Soybean/ Soybean derivatives Please note here any known dietary restriction not covered above. Signature Date 397
393 Appendix C Â— continued Advertisement Rating Form Please evaluate each advertisement individually Ad #1 What product feature(s) is this ad trying to communicated Place a check (Â•") in the space that indicates how convincing you find this advertisement to be. Not at all Somewhat Moderately Extremely Convincing Convincing Convincing Convincing Ad #2 What produce feature(s) is this ad trying to communicate? Place a check (Â•) in the space that indicates how convincing you find this advertisement to be. Not at all Somewhat Moderately Extremelv Convincing Convincing Convincing Convincing Ad #3 What product feature(s) is this ad trying to communicate: Place a check (Â•) in the space that indicates how convincing you find this advertisement to be. Not at all Somewhat Moderately Extremely Convincing Convincing Convincing Convincing Ad #4 What product feature(s) is this ad trying to communicate? Place a check (vO in the space that indicates how convincing you find this advertisement to be. Not at all Somewhat Moderately Extremely Convincing Convincing Convincing Convincing
399 Appendix C Â— continued Classification Information Name : Phone : Mailing address: Sex: M F Age: Employed: Yes No Average number of cups of coffee (all kinds) you drink each day: Do you prepare coffee yourself: at home? always sometimes never at work? always sometimes never What brand(s) of coffeemaker do you/have you used? What brand(s) of coffee do you/have you used' Please list every brand of coffee you can think of: Please list every brand of coffeemaker vou can think of: Please check the spaces that describe what you would consider to be an "ideal" coffee. For me, an ideal coffee is: ModerModerVery ately Slightly Neither Slightly ately Very Bitter Not Bitter Strong in Weak in flavor in flavor Poor Good Quality ~~ Quality Light in Dark in Color ~" Color Good Bad Tasting Tasting
400 Appendix C Â— continued BRAND SELECTED BRAND REJECTED Questionnaire Thank you for agreeing to participate in our market test. At this point, we would like to know your reactions to the product you have tried and to other aspects of the market test. Part I Directions : In this section, please place a check (/) in the space that best represents your opinion or reaction to parts of the study. For example, if you feel that today's weather is Very Sunny but Moderately Cold you would mark the appropriate scales as follows: Today's weather is Very Moderately Slightly Neither Slightly Moderately Very Cloudy \/_ Sunny Hot _v/ Cold Please mark only one space on each line, but be certain not to miss any line. Remember, base your answers on today's study. It you have any questions, ask the interviewer: The Coffeemaker is: ModerModerVery ately Slightly Neither Slightly atelv Very Difficult _' Simple to to Use L 'se Fast s l w Simple to ~ ~~ "_ Difficult to Fill '" to Flll_ Unattractive Attractive Better __ ~~ "_ Worse than than I expected ~" I expected The Coffee is: ModerModerVery ately Slightly Neither Slightly ately Very Bitter Not Bitter Strong in ~ Weak in Flavor "~ Flavor Poor Good Quality ~ Quality Light in Dark in Color ~" Color Good Bad Tasting Tasting Worse than Worse than I expected I expected
401 Appendix C--continued COFFEE BRAND EVALUATED Questionnaire Thank you for agreeing to participate in our market test. At this point, we would like to know your reactions to the product you have tried and to other aspects of the market test. Part I Directions In this section, please place a check (Â•/ ) in the space that best represents your opinion or reaction to parts of the study. For example, if you feel that today's weather is Very Sunny but Moderately Cold you would mark the appropriate scales as follows: Today's weather is Very Moderately Slightly Neither Slightly Moderately Very Cloudy
402 Appendix C--continued Part II Directions This portion of the study consists of a number of words that describe emotions or feelings. Please indicate the extent to which each word describes the way you feel about the coffee that you have now sampled. Record your answer by circling the appropriate number on the five point scale following each word. Presented below is the response scale for indicating the degree to which each word describes the way you feel about the coffee. 12 3 4 5 Not at all or VeryVery Slightly Slightly Moderately Considerably Strongly In deciding on your answer to a given item or word consider the feeling connoted or defined by that word. Then, if sampling the coffee makes you feel that way not at all or very slightly you would circle the number 1 on the scale; if you feel that way to a moderate degree, you would circle 3; if you feel that way very strongly you would circle 5, and so forth. Remember, you are requested to make your responses on the basis of the way you feel about the coffee you sampled. Work at a good pace. It is not necessary to ponder; the first answer you decide on for a given word is probably the most valid. You should be able to finish this section in about 3 minutes. Sampling the coffee makes me feel: Not at all or
403 Appendix E Â— continued Not at all or
404 Appendix C Â— continued How satisfied were you with the coffeemaker? Very Somewhat Slightly SlightlyDissatisfied Dissatisfied Dissatisfied Mixed Satisfied Somewhat Very Not Satisfied Satisfied Satisfied Nor Dissatisfied Knowing what you know now, what are the chances in ten (10) that you would choose to use the coffee again? 0123456789 10 No Certain Chance Knowing what you know not, what are the chances in ten (10) you would choose the coffeemaker again? 1 2 3 4 5 6 7 8 9 10_ n; Certain Chance Which word or phrase most closely reflects your satisfaction with the coffee? Mostly Mostly Delighted Pleased Satisfied Mixed Dissatisfied Unhappy Terrible Neutral Never Thought About It Which word or phrase most closely reflects your satisfaction with the coffeemaker? Mostly Mostly Delighted Pleased Satisfied Mixed Dissatisfied Unhappy Terrible Neutral Never Thought About It
405 Appendix C Â— continued Which face comes closest to expressing how you feel about the coffee? Which face comes closest to expressing how you feel about the cof feemaker? Part IV Directions This section consists of words and phrases that describe the study and the products. Please place a check (Â• ) on the portion of the scale that best reflects your feelings. Evaluation of the Coffee Was interesting to me Was not interesting to me Was important to me Was not important to me Was not difficult to Was difficult to evaluate evaluate Took a lot of Did not take a lot thought of thought Task was very ^ Task was not very involving ~ involving I put a lot of I P ut no thought thought into into evaluation evaluation
406 Appendix C Â— continued Evaluation of the Coffeemaker Was interesting to me Was not interesting to me Was important to me Was not important to me Was not difficult to Was difficult to evaluate evaluate Took a lot of Did not take a lot thought "~ of thought Task was very Task was not very involving involving I put a lot of I put no thought thought into into evaluation evaluation Coffee Good Bad Unsatisfactory """ Satisfactory Unfavorable Favorable Worthless ~ "" Valuable Like Dislike Disapprove '_ Approve Coffeemaker Good Bad Unsatisfactory Satisfactory Unfavorable ~~ "" Favorable Worthless Valuable Like Dislike Disapprove Approve
407 Appendix C Â— continued Part V Directions The questions in this section refer to what you expected before trying the coffeemaker and coffee. Please place a check (>/) on the portion of each scale that best reflects your feelings. Based on the information given to you in the advertisements at the beginning of the study, what did you expect about the coffeemaker? Did you expect it to be ModerModerVery ately Slightly Neither Slightly ately Very Difficult Simple to to Use Use Fast slow SimDle to Difficult to Fill to Fill Unattractive __ Attractive Better Worse than than I "~ : expected expected The Coffee is: ModerModerVery ately Slightly Neither Slightly ately Very Bitter 1 Not Bitter Strong in "_ ~~ Weak in Flavor """ Flavor Poor lood Quality Quality Light in Ddrk in Color ~ Color Good Bad Tasting Tasting Inexpensive expensive Overall, my expectations about the coffeemaker were: Too high; To lov; '> It was worse Accurate It was better Than I thought Than l thought Overall, my expectations about the coffee were: Too high; Too low; It was worse Accurate It was better Than I thought Tha x thought
408 Appendix C--continued Part VI Directions In this last section, please circle the word or phrase that most closely reflects your agreement with each statement. The next time I need coffee, I will consider buying the brand I tried (if available) Strongly Moderately Slightly Slightly Moderately Strongly Agree Agree Agree Neither Disagree Disagree Disagree The next time I need a coffeemaker for myself, I will consider buying the model I tried (if available). Strongly Moderately Slightly Slightly Moderately Strongly Agree Agree Agree Neither Disagree Disagree Disagree The next time I serve coffee to guests, I would use the brand I tried. Strongly Moderately Slightly Slightly Moderately Strongly Agree Agree Agree Neither Disagree Disagree Disagree If I wanted to give a coffeemaker as a gift, I would consider giving the model I tried. Strongly Moderately Slightly Slightly Moderately Strongly Agree Agree Agree Neither Disagree Disagree Disagree Overall, my experience in the study was enjoyable. Strongly Moderately Slightly Slightly Moderately Strongly Agree Agree Agree Neither Disagree Disagree Disagree I had the option to leave the study at any time. Strongly Moderately Slightly Slightly Moderately Strongly Agree Agree Agree Neither Disagree Disagree Disagree I had the option to choose the brand of coffee to evaluate. Strongly Moderately Slightly Slightly Moderately Strongly Agree Agree Agree Neither Disagree Disagree Disagree
409 Appendix E Â— continued I had to option to choose the model of coffeemaker to evaluate. Strongly Moderately Slightly Slightly Moderately Strongly Agree Agree Agree Neither Disagree Disagree Disagree Given the opportunity, I would participate in another product study. Strongly Moderately Slightly Slightly Moderately Strongly Agree Agree Agree Neither Disagree Disagree Disagree I am satisfied about my choice of coffee. Strongly Moderately Slightly Slightly Moderately Strongly Agree Agree Agree Neither Disagree Disagree Disagree If I had it to do all over again, I would feel differently about choosing the coffee. Strongly Moderately Slightly Slightly Moderately Strongly Agree Agree Agree Neither Disagree Disagree Disagree My choice of coffee was a wise one. Strongly Moderately Slightly Slightly Moderately Strongly Agree Agree Agree Neither Disagree Disagree Disagree I feel bad about my decision regarding the coffee. Strongly Moderately Slightly Slightly Moderately Strongly Agree Agree Agree Neither Disagree Disagree Disagree I think that I did the right thing when I chose the coffee. Strongly Moderately Slightly Slightly Moderately Strongly Agree Agree Agree Neither Disagree Disagree Disagree I am not happy that I chose the coffee that I did. Strongly Moderately Slightly Slightly Moderately Strongly Agree Agree Agree Neither Disagree Disagree Disagree
410 Appendix C Â— continued Sirs : I recently had the opportunity to try your new coffee. I wanted to make the following comments about it: Market
41 1 Appendix C Â— continued Follow-Up Interview Subject number: Date of first interview: Name : Phone : Call back date: 1. time: result: 2. time: result: 3. time: result: 4. time: result: 5. time: result: Get the participant on the line, then: Good (morning, afternoon, evening), I'm calling for NEH research. One week ago, you participated in a study which we are conducting. I'd like to ask you 6 brief questions about the coffee you tried at that time 1 How many times have you told someone about the coffee you tried a week ago? (number of people) 2. What did you tell them? circle: positive negative 3. How manv times have you used the gift sample you were given (# pots): 4. Have you served it to: others in household? yes no guests? yes no 5. If you had the opportunity, would you consider repurchasing the coffee? yes no 6. Have you tried to repurchase the coffee? yes no At this time, I want to inform you that this is not a test market study being conducted by a food processing company. Rather, this is part of an academic study of customer satisfaction being conducted at the University of Akron. The advertisements and products which you evaluated were prepared for this study. The coffee sample you were given in the paper bag was not modified and may be used per the label instructions. A drawing for a coffeemaker will be held as soon as the study is completed. If you have any questions, I can provide you with the name and phone number of the professor conducting the study. If ycu have not yet mailed the postcard that you were given, please discard it. We are only interested in your response prior to this call. Thank you again for your assistance. (Doug Hausknecht 375-6303)
412 Appendix C Â— continued Marketing Department University of Akron Akron, OH 44325 (stamp) (participant's address) Dear We have attempted to contact you by telephone in regard to the coff ee/cof feemaker study in which you participated. In addition to a few follow-up questions, we wished to inform you that the study was conducted as part of academic research on consumer satisfaction/ dissatisfaction and was not at the request of any manufacturer. If you have not yet returned your "comment card", please discard it at this time. The coffee you received as a gift (foil pouch in the paper bag) was not altered and may be used as indicated in the directions. A drawing for a coffeemaker will be held when all the data are collected. If you have any questions about the study, please call Professor Hausknecht at the University of Akron (216) 375-6303. Thank you for vour assistance.
REFERENCES Adler, Roy D. and Larry M. Robinson (1980), "A Consumer Satisfaction Model Based on Job Satisfaction Theory," in Refining Concepts and Measures of Consumer Satisfaction and Complaining Behavior eds. H. Keith Hunt and Ralph L. Day, Bloomington, IN: Indiana University, 19-22. Aiello, Albert Jr. and John A. Czepiel (1979), "Customer Satisfaction in a Catalog Type Retail Outlet: Exploring the Effect of Product, Price and Attributes," in New Dimensions of Consumer Satisfaction and Complaining Behavior eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana University, 129-135. John A. Czepiel, and Larry J. Rosenberg (1977), "Scaling the Heights of Consumer Satisfaction: An Evaluation of Alternative Measures," in Consumer Satisfaction, Dissatisfaction and Complaining Behavior ed. Ralph L. Day, Bloomington, IN: Indiana University, 43-50. Alba, Joseph W. and J. Wesley Hutchinson (1987), "Dimensions of Consumer Expertise," Journal of Consumer Research 13 (March), 411454. Alderson, Wroe (1965), Dynamic Marketing Behavior: A Functionalist Theory of Marketing Homewood, IL: Richard D. Irwin, Inc. Allen, Chris T., Karen A. Machleit, and Susan S. Marine (1987), "On Assessing the Emotionality of Advertising Via Izard's Differential Emotions Scale," paper presented at Association for Consumer Research Conference, Cambridge, MA: Oct. 8-11. Anderson, Rolph E. (1973), "Consumer Dissatisfaction: The Effect of Disconfirmed Expectancy on Perceived Product Performance," Journal of Marketing Research 10 (February), 38-44. and Joseph F. Hair, Jr. (1972), "Consumerism, Consumer Expectations and Perceived Product Performance," in Association for Consumer Research, Annual Proceedings #3 ed. M. Venkatesan, Ann Arbor, MI: Association for Consumer Research, 67-79. Anderson, Ronald D., Jack L. Engledow and Helmut Becker (1979), "Evaluating the Relationships Among Attitude Toward Business, Product Satisfaction, Experience and Search Effort," Journal of Marketing Research 16 (August), 394-400. 413
414 Andreasen, Alan R. (1977a), "A Taxonomy of Consumer Satisfaction/ Dissatisfaction Measures," Journal of Consumer Affairs 11 (Winter), 11-24. (1977b), "A Taxonomy of Consumer Satisfaction/ Dissatisfaction Measures," in Conceptualization and Measurement of Consumer Satisfaction and Dissatisfaction ed. H. Keith Hunt, Cambridge, MA: Marketing Science Institute, 11-35. and Arthur Best (1977), "Consumers Complain Does Business Respond?" Harvard Business Review 55 (July-August), 93-101 Andrews, Frank M. and Stephen B. Witney (1976), Social Indicators of Well-Being New York, NY: Plenum Press. Apsler, Robert and David 0. Sears (1968), "Warning, Personal Involvement, and Attitude Change," Journal of Personality and Social Psychology 9 (Number 2), 16 2-166. Arndt, Johan (1967), "Role of Product-Related Conversations in the Diffusion of a New Product," Journal of Marketing Research 4 (August), 291-295. Aronson, Elliot (1968a), "Discussion: Expectancy vs. Other Motives," in Theories of Cognitive Consistency: A Sourcebook eds. Robert P. Abelson et al., Chicago, IL: Rand McNally and Company, 491-493. and J. Merrill Carlsmith (1962), "Performance Expectancy as a Determinant of Actual Performance," Journal of Abnormal and Social Psychology 65 (Number 3), 178-182. Assar, Amardeep and Dipankar Chakravarti (1 984), "Attribute Range Knowledge: Effects on Consumers' Evaluation of Brand-Attribute Information and Search Patterns in Choice," in American Marketing Association 1984 Educators' Conference Proceedings ed. Russell W. Belk, Chicago, IL: American Marketing Association, 62-67. Atkinson, John W. (1957), "Motivational Determinants of Risk-Taking Behavior," reprinted in Research in Consumer Behavior eds. David T. Kollat, Roger D. Blackwell, and James F. Engel, New York, N.Y.: Holt, Rinehart and Winston, Inc., 31-47. and David Birch (1978), An Introduction to Motivation 2nd edition, N<;w York. NY: Van Nostrand. Averill, James R. (1980), "A Constructivist View of Emotion," in Emotion: Theory, Research, and Experience Volume 1, eds. Robert Plutchik and Henry Ke Herman, New York, NY: Academic Press, 305-339.
415 Bahr, William J. (1982), "Consumer Dissatisfaction or Disappointment: The Critical Difference," in New Findings on Consumer Satisfaction and Complaining eds Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana University, 33-39. Bar-Hillel, Maya (1980), "The Base-Rate Fallacy in Probability Judgments," Acta Psychologica 44, 211-233. 3arbeau, J. Bradley and William J. Quails (1984), "Consumers' Perceptions of Attributes and Behavioral Intentions: An Extended Comparison-Level Model," in Advances in Consumer Research: Volume 1_1_, ed. Thomas C. Kinnear, Provo, UT: Association for Consumer Research, 143-147. Barnes, James G. and Karen R. Kelloway (1980), "Consumerists : Complaining Behavior and Attitudes Toward Social and Consumer Issues," in Advances in Consumer Research: Volume 7 ed. Jerry C. Olson, Ann Arbor, MI: Association for Consumer Research, 329-334. Bearden, William 0. (1983), "Profiling Consumers Who Register Complaints Against Auto Repair Services," Journal of Consumer Affairs 17 (Winter), 315-335. and J. Barry Mason (1984), "An Investigation of Influences on Consumer Complaint Reports," Advances in Consumer ^ Research: Volume 1 1 ed. Thomas C. Kinnear, Provo, UT: Association for Consumer Research, 490-495. and Richard L. Oliver (1985), "The Role of Public and Private Complaining in Satisfaction with Problem Resolution," Journal of Consumer Affairs 19 (Winter), 222-240. and Jesse E. Teel (1983), "Selected Determinants of Consumer Satisfaction and Complaint Reports," Journal of Marketing Research 20 (February), 21-28. Becherer, Richard C, Fred W. Morgan, and Lawrence M. Richard (1982), "The Job Characteristics of Industrial Salespersons: Relationship to Motivation and Satisfaction," Journal of Marketing 46 (Fall), 125-135. Belk, Russell W. (1984a), "Three Scales to Measure Constructs Related to Materialism: Reliability, Validity and Relationships to Measures of Happiness," in Advances in Consumer Research: Volume 1_1_, ed. Thomas C. Kinnear, Provo, UT: Association for Consumer Research, 291-297. (1984b), "Manifesto for a Consumer Behavior of Consumer Behavior," in Proceedings of 1984 Winter Educators' Conference: Scientific Method in Marketing eds. Paul F. Anderson and Michael J. Ryan, Chicago, IL: .American Marketing Association, 163-167.
416 (1984c), "Against Thinking," in Proceedings of 1984 Winter Educators' Conference: Scientific Method in Marketing eds. Paul F. Anderson and Michael J. Ryan, Chicago, IL: American Marketing Association, 57-60. (1986), "Art Versus Science as Ways of Generating Knowledge About Materialism," in Perspectives on Methodology in Consumer Research eds. David Brinberg and Richard J. Lutz, New York, NY: Springer-Verlag, 3-36. John F. Sherry, Jr. and Melanie Wallendorf (1988), "A Naturalistic Inquiry into Buyer and Seller Behavior at a Swap Meet," Journal of Consumer Research 14 (March), 449-470. Melanie Wallendorf, John Sherry, Morris Holbrook, and Scott Roberts (1988), "Collectors and Collecting," in Advances in Consumer Research Volume 15, ed. Michael J. Houston, Provo, UT: Association for Consumer Research, 548-553. Bergh, Bruce G. Vanden and Leonard N. Reid (1980), "The Roleof Newspaper Advertisements in the Confirmation or Disconf irmation of Consumer Expectations in a Product Trial Situation," in Advertising 1980: Voice of a Nation at Work ed. James E. Haefner, Urbana, IL: University of Illinois, 61-64. Bernacchi, M. D., Ken Kono, and G. L. Willette (1980), "An Analysis of Automobile Warranty Service Dissatisfaction," in Refining Concepts and Measures of Consumer Satisfaction and Complaining Behavior eds. H. Keith Hunt and Ralph L. Day, Bloomington, IX: Indiana University, 141-143. Bernardo, John J. and Lawrence X. Tarpey, Sr. (1980), "Brand Loyalty as a Theoretical Construct," in Lamb and Dunne Theoretical Developments in Marketing eds. Charles W. Lamb, Jr. and Patrick M. Dunne, Chicago, IL: American Marketing Association, 198-201. Best, Arthur (1981), When Consumers Complain Xew York, NY: Columbia University Press. Biehal, Gabriel J. (1983), "Consumer's Prior Experiences and Perceptions in Auto Repair Choice," Journal of Marketing 47 (Summer), 82-91 Bjorklund, Gail and Richard Bjorklund (1979), "An Exploratory Study of Toddler's Satisfaction With Their Toy Environments," in Advances in Consumer Research: Volume 6 ed. William L. Wilkie, Ann Arbor. MI: Association for Consumer Research, 400-406. Bloch, Peter H. (1981), "An Exploration into the Scaling of Consumers' Involvement With a Product Class," in Advances in Consumer Research: Volume 8 ed. Kent B. Monroe, Ann Arbor, MI: Association for Consumer Research, 61-65.
417 and G. D. Bruce (1984), "The Leisure Experience and Consumer Products: An Investigation of Underlying Satisfactions," Journal of Leisure Research 16, 74-88. and Marsha L. Richins (1983), "A Theoretical Model for the Study of Product Importance Perceptions," Journal of Marketing 47 (Summer), 69-81 [loom, Paul N. (1982a), "Research on Consumerism: Opportunities and Challenges," in Advances in Consumer Research: Volume 9 ed. Andrew Mitchell, Ann Arbor, MI: Association for Consumer Research, 520-522. (1982b), Consumerism and Beyond: Research Perspectives on the Future Social Environment Cambridge, MA: Marketing Science Institute and Stephen A. Greyser (1981), Exploring the Future of Consumerism Cambridge, MA: Marketing Science Institute. Bohl, Don Lee (ed.) (1987), Close to the Customer: An American Management Association Research Report on Consumer Affairs New York, NY: American Management Association. Bolfing, Claire P., and Robert B. Woodruff (1987), "Effects of Situational Involvement on Consumers Use of Standards in Satisfaction/Dissatisfaction Processes," paper presented at 10th Consumer Satisfaction/Dissatisfaction and Complaining Behavior Conference, Montreal, Quebec, August. Braden, Marcia and Elaine Walster (1964), "The Effect of Anticipated Dissonance on Pre-Decision Behavior," in Leon Festinger Conflict, Decision and Dissonance Stanford, CA: Stanford University Press, 145-151 Brehm, Jack W. (1956), "Postdecision Changes in the Desirability of Alternatives," Journal of Abnormal and Social Psychology 52 (May) 384-389. (1959), "Increasing Cognitive Dissonance by a Fait Accompli," Journal of Abnormal and Social Psychology 58 (May), 379-382. and Arthur R. Cohen (1962), Explorations in Cognitive Dissonance New York, NY: John Wiley & Sons Inc. Srewer, Marilynn B. (1977), "An Information Processing Approach to Attribution of Responsibility," Journal of Experimental Social Psychology 13, 53-69.
418 Brown, Stephen W. and Teresa A. Swartz (1984), "Consumer Medical Complaint Behavior: Determinants of and Alternatives to Malpractice Litigation," Journal of Public Policy and Marketing 3, 85-98. Buchanan, Bruce, Moshe Givon and Arieh Goldman (1987), "Measurement of Discrimination Ability in Taste Tests: An Empirical Investigation," Journal of Marketing Research 24 (May), 154-163. 3urke, Marian, W. David Conn, and Richard J. Lutz ( 1 978a) ,_ "Using Psychographic Variables to Investigate Product Disposition Behaviors," Working Paper #64, Center for Marketing Studies, UCLA. W. David Conn and Richard J. Lutz (1978b), "Using Psychographic Variables to Investigate Product Disposition Behaviors," in Research Frontiers in Marketing: Dialogues and Directions 1978 Educator's Proceedings Series #43, ed. Subhash C. Jain, Chicago, IL: AiMA, 321-326. Burnkrant, Robert E. and Alain Cousineau (1975), "Informational and Normative Social Influence in Buyer Behavior," Journal of Consumer Research 2 (December), 206-215. and Alan G. Sawyer (1983), "Effects of Involvement and Message Content on Information Processing Intensity, in Information Processing Research in Advertising ed. Richard J. Harris, Hillsdale, NJ: Lawrence Erlbaum Assoc, 43-64. Cadotte, Ernest R. Robert B. Woodruff, and Roger L. Jenkins ( 1 987) "Expectations and Norms in Models of Consumer Satisfaction," Journal of Marketing Research 24 (August), 305-314. Robert B. Woodruff, and Roger L. Jenkins (1983), "Norms and Expectation Predictions: How Different Are the Measures?" in International Fare in Consumer Satisfaction and Complaining Behavio r' eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana University, 49-56. Calder, Bobby J. (1973), "Cognitive Consistency and Consumer Behavior," in Perspectives in Consumer Behavior revised edition, eds. Harold H. Kassarjian and Thomas S. Robertson, Glenview, IL: Scott, Foresman and Co., 247-263. Cardozo, Richard N. (1965), "An Experimental Study of Customer Effort, Expectation and Satisfaction," Journal of Marketing Research 2 (August), 244-249. Carey, J. Ronald, Steven H. Clicque, Barbara A. Leighton, and Frank^ Milton (1976), "A Test of Positive Reinforcement of Customers," Journal of Marketing 40 (October), 98-100.
419 Carlsmith, J. Merrill and Elliot Aronson (1963), "Some Hedonic Consequences of the Confirmation and Disconf irmation of Expectancies," Journal of Abnormal and Social Psychology 66 (Number 2), 151-156. and Jonathan L. Freedman (1968), "Bad Decisions and Dissonance: Nobody's Perfect," in Theories of Cognitive Consistency: A Sourcebook eds Robert P. Abelson et al., Chicago, IL: Rand McNally and Company, 485-490. Carmines, Edward G. and Richard Z. Zeller (1979), Reliability and Validity Assessment Sage University Paper series on Quantitative Analysis in the Social Sciences, 07-017, Beverly Hills, CA: Sage Publications Carroll, Stephen J. and Henry L. Tosi (1977), Organizational Behavior Chicago, IL: St. Clair Press. Carsky, Mary L. (1987), "Negative Information at the Point-ofSale: Can It Moderate Satisfaction with New Automobiles?" Paper presented at 10th Consumer Satisfaction/ Dissatisfaction and Complaining Behavior Conference, Montreal, Quebec, August. Chakravarti, Dipankar and John G. Lynch (1983), "A Framework for Exploring Context Effects on Consumer Judgment and Choice," in Advances in Consumer Research: Volume 10 eds. Richard P. Bagozzi and Alice M. Tybout, Ann Arbor, MI: Association for Consumer Research, 289-297. Charell, Ralph (1973), How I Turn Ordinary Complaints into Thousands of Dollars New York, NY: Dell Publishing Co. Chase, William G. and K. Anders Ericsson (1982), "Skill and Working Memory," in The Psychology of Learning and Motivation: Volume 16 ed. Gordon H. Bower, New York, NY: Academic Press, 1-58. and Herbert A. Simon (1973), "Perception in Chess," Cognitive Psychology 4 (January), 55-81. Churchill, Gilbert A. Jr. (1979), "A Paradigm for Developing Better Measures of Marketing Constructs," Journal of Marketing Research 16 (February), 64-73. Churchill, Gilbert A. Jr. and Anthony Pecotich (1982), "A Structural Equation Investigation of the Pay Satisfaction Valence Relationship Among Salespeople," Journal of Marketing 46 (Fall), 114-124. and Carol Surprenant (1932), "An Investigation into the Determinants of Customer Satisfaction," Journal of Marketing Research 19 (November), 491-504.
420 Cialdini, Robert B., John T. Cacioppo, Rodney Bassett, and John A. Miller (1978), "Low-Ball Procedure for Producing Compliance: Commitment Then Cost," Journal of Personality and Social Psychology 36 (Number 5), 463-476. Clarke, Keith and Russell W. Belk (1979), "The Effects of Product Involvement and Task Definition on Anticipated Consumer Effort," in Advances in Consumer Research: Volume 6 ed. William L. Wilkie, Ann Arbor, MI: Association for Consumer Research, 313-318. Cobb, Cathy J., Gary C. Walgren, and Mary Hollowed (1987), "Differences in Organizational Responses to Consumer Letters of Satisfaction and Dissatisfaction," in Advances in Consumer Research: Volume 14 eds. Melanie Wallendorf and Paul Anderson, Provo, UT: Association for Consumer Research, 227-231 Coca-Cola (1982), "Measuring the Grapevine: Consumer Response and Word-of-Mouth," study conducted by Technical Assistance Research Programs, Inc. Cohen, Joel B. (1981), "The Role of Affect in Categorization: Toward a Reconsideration of the Concept of Attitude," in Advances in Consumer Research: Volume 9 ed. Andrew Mitchell, Ann Arbor, MI: Association for Consumer Research, 94-100. (1982), "Involvement: Separating the State from Its Causes and Effects," Working Paper #33, Center for Consumer Research, University of Florida. (1983), "Involvement and You: 1000 Great Ideas," Advances in Consumer Research: Volume 10 eds. Richard P. Bagozzi and Alice M. Tybout, Ann Arbor, MI: Association for Consumer Research, 324-327. and Marvin E. Goldberg (1970), "The Dissonance Model in Post-Decision Product Evaluation," Journal of Marketing Research 7 (August), 315-321 and Michael J. Houston (1972), "Cognitive Consequences of Brand Loyalty," Journal of Marketing Research 9 (February), 97-99. Converse, Philip E. (1970), "Attitudes and Non-attitudes: Continuation of a Dialogue," in The Quant i tative Analysis of Social Problems ed. Edward R. Tufte, Reading, MA: Addison-Wesley 168-189. (1974), "Comment: The Status of Nonatt itudes .American Political Science Review 68, 650-666. 3ook, Stuart W. and Claire Selltiz (1964), "A Multiple Indicator Approach to Attitude Measurement," reprinted in Readings in Attitude Theory and Measurement ed. Martin Fishbein, New York, NY: John Wiley & Sons, 1967, 220-235.
421 Cook, Thomas D. and Donald T. Campbell (1979), Quasi -Experiment at ion: Design and Analysis Issues for Field Settings Boston, MA: Houghton Miffflin Company. Costley, Carolyn L. (1988), "Meta Analysis of Involvement Research," in Advances in Consumer Research Volume 15, ed. Michael J. Houston, Provo, UT: Association for Consumer Research, 554-562. Crosby, Lawrence A. and James R. Taylor (1983), "Psychological Commitment and Its Effects on Post-Decision Evaluation and Preference Stability Among Voters," Journal of Consumer Research 9 (March), 413-431 and James R. Taylor (1982), "Consumer Satisfaction with Michigan's Container Deposit Law An Ecological Perspective," Journal of Marketing 46 (Winter), 47-60. Cummings, William H. and M. Venkatesan (1976), "Cognitive Dissonance and Consumer Behavior: A Review of the Evidence," Journal of Marketing Research 13 (August), 303-308. Cureton, Edward E. (1950), "Validity, Reliability and Baloney," reprinted in Principles of Educational and Psychological Measurement eds. William A. Mehrens and Robert L. Ebel, Chicago, IL: Rand McNally, 1967, 271-273. Czepiel, John A. (1980). Managing Customer Satisfaction in Consumer Service Businesses Cambridge, MA: Marketing Science Institute. and Larry J. Rosenberg (1977), "The Study of Consumer Satisfaction: Addressing the 'So What' Question, in Conceptualization and Measurement of Consumer Satisfaction and Dissatisfaction ed. H. Keith Hunt, Cambridge, MA: Marketing Science Institute, 92-119. Davidson, Jon R. and Sara B. Kiesler (1964), "Cognitive Behavior Before and After Decisions," in Leon Festinger, Conflict, Decision and Dissonance Stanford, CA: Stanford University Press, 10-21. Day, Ralph L. (1977), "Extending the Concept of Consumer Satisfaction," in Advances in Consumer Research: Volume 4 ed. William D. Perreault, Jr., Atlanta: Association for Consumer Research, 1 49-1 54 (1980a), "How Satisfactory Is Research on Consumer Satisfaction?" in Advances in Consumer Research: Volume 7 ed. Jerrv C. Olson, Ann Arbor, MI: Association for Consumer Researcl 593-597.
422 (1980b), "Research Perspectives on Consumer Complaining Behavior," in Theoretical Developments in Marketing eds. Charles W. Lamb, Jr. and Patrick M. Dunne, Chicago, IL: American Marketing Association, 211-215. (1983), "The Next Step: Commonly Accepted Constructs for Satisfaction Research," in International Fare in Consumer Satisfaction and Complaining Behavior eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana University, 113-117. (1984), "Modeling Choices Among Alternative Responses to Dissatisfaction," Advances in Consumer Research: Volume 11 ed Thomas C. Kinnear, Provo, UT: Association for Consumer Research, 496-499. and Stephen B. Ash (1979), "Consumer Response to Dissatisfaction With Durable Products," in Advances in Consumer Research: Volume 6 ed. William L. Wilkie, Ann Arbor, MI: Association for Consumer Research, 438-444. and Muzaffer Bodur (1978), "Consumer Response to Dissatisfaction With Services and Intangibles," in Advances in Consumer Research: Volume 5 ed. H. Keith Hunt, Ann Arbor, MI: Association for Consumer Research, 263-272. and Muzaffer Bodur (1979), "Analysis of Average Satisfaction Scores of Individuals Over Product Categories, in New Dimensions of Consumer Satisfaction and Complaining Behavior eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana University, 184-189. and E. Laird Landon, Jr. (1976), "Collecting Comprehensive Consumer Complaint Data by Survey Research," in Advances in Consumer Research: Volume 3 ed. Beverlee B. Anderson, Ann Arbor, MI: Association for Consumer Research, 263-268. DeBell, Margaret and Rachel Dardis (1979), "Extending Product Life: Technology Isn't the Only Issue," in Advances in Consumer Research: Volume 6 ed. William L. Wilkie, Ann Arbor, MI: Association for Consumer Research, 381-385. Deighton, John (1984), "The Interaction of Advertising and Evidence," Journal of Consumer Research 11 (December), 763-770. Dickson, Peter R. and William L. Wilkie (1978), The Consumption of Household Durables: A Behavioral Review Cambridge, MA: Marketing Science Institute.
423 Diener, Betty J. (1980), "Selective Perception and Consumer Dissatisfaction (Or... The Case of the Rabid Rabbits)," in Refining Concepts and Measures of Consumer Satisfaction and Complaining Behavior eds H. Keith Hunt and Ralph L. Day, Bloomington, IN: Indiana University, 168-172. (1977), "The Fulfillment of Consumer Promotions as a Source~of Consumer Dissatisfaction," in Conceptualization and Measurement of Consumer Satisfaction and Dissatisfaction ed. H. Keith Hunt, Cambridge, MA: Marketing Science Institute, 333-354. Donnelly, James H. Jr., and John M. Ivancevich (1970), "Post-Purchase Reinforcement and Back-out Behavior," Journal of Marketing Research 7 (August), 399-400. Eastlack, J.O. Jr. (1964), "Consumer Flavor Preference Factors in Food Product Design," Journal of Marketing Research 1 (February), 38-42. Edwards, Allen L. and Kathryn Claire Kenney (1946), "A Comparison of the Thurstone and Likert Techniques of Attitude Scale Construction," reprinted in Readings in Attitude Theory and Measurement ed. Martin L. Fishbein (1967), New York, NY: John Wiley & Sons, 249-256. Eisenberger, Kenneth (1977), The Expert Consumer: A Complete Handbook Englewood Cliffs, NJ : Prentice-Hall. Engel, James F. Robert J. Kegerreis, and Roger D. Blackwell (1969), "Word-of-Mouth Communication by the Innovator," Journal of Marketing 33 (July), 15-19. Evans, Richard H. (1983), "Diagnosing Customer Satisfaction/ Dissatisfaction and Behavioral Intention: Expectancy Value Theory Versus Equity Theory," in International Fare in Consumer Satisfaction and Complaining Be havior eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana University, 64-68. Fazio, Russell H. Jeaw-mei Chen, Elizabeth C. McDonel, and Steven J. Sherman (1982), "Attitude Accessibility, Attitude-Behavior Consistency, and the Strength of the Object-Evaluation Association," Journal of Experimental Social Psychology 13, 339-357. Martha C. Powell, and Paul M. Herr (1983), "Toward a Process Model of the Attitude-Behavior Relation: Accessing One's Attitude upon Mere Observation of the Attitude Object," Journal of Personality and Social Psychology 44 (Number 4), 723-735.
424 David M. Sanbonmatsu, Martha C. Powell, and Frank R. Kardes (1986), "On the Automatic Activation of Attitudes," Journal of Personality and Social Psychology 50 (Number 2), 229-238. and Mark P. Zanna (1981), "Direct Experience and Attitude-Behavior Consistency," in Advances in Experimental Social Psychology: Volume 14 ed. Leonard Berkowitz, New York, NY: Academic Press, 161-202. Feick, Lawrence F. (1984), "Do Consumers Complain Sequentially?" in Proceedings of 1984 Winter Educator's Conference: Scientific Method in Marketing eds. Paul F. Anderson and Michael J. Ryan, Chicago, IL: American Marketing Association, 89-92. (1987), "Latent Class Models for the Analysis of Behavioral Hierarchies," Journal of Marketing Research 24 (May), 174-186. Feinman, Jeffrey (1979), The Purple Pages New York, NY: Hawthorn Books, Inc. Feldman, Daniel C. and Hugh J. Arnold (1983), Managing Individual and Group Behavior in Organizations New York, NY: McGraw-Hill Book Co. Fendrich, James M. (1967), "A Study of the Association Among Verbal Attitudes, Commitment and Overt Behavior in Different Experimental Situations," reprinted in The Consistency Controversy ed. Allen E. Liska, New York, NY: John Wiley & Sons, 1972, 95-108. Festinger, Leon (1957), A Theory of Cognitive Dissonance Evanstcn, IL: Row, Peterson and Co. (1964), Conflict, Decision and Dissonance, Stanford, CA: Stanford University Press. Filer, Robert J. (1952), "Frustration, Satisfaction, and Other Factors Affecting the Attractiveness of Goal Objects," Journal of Abnormal and Social Psychology 47, 203-212. Fishbein, Martin (1972), "The Search for Attitudinal-Behavioral Consistency," reprinted in Perspectives in Consumer Behavior Revised Edition eds. Harold H. Kassarjian and Thomas S. Robertson, Glenview, IL: Scott, Foresman and Co, 1973, 210-219. (1966), "The Relationship Between Beliefs, Attitudes and Behaviors," reprinted in Research in Consumer Behavior eds. David T. Kollat, Roger D. Blackwell and James F. Engel, New York, NY: Holt, Rinehart and Winston, 1970, 216-235.
425 and leek Ajzen (1975), Belief, Attitude, Intention and Behavior: An Introduction to Theory and Research Reading, MA: Add i son-Wesley. Fisk, Raymond P. and Clifford E. Young (1985), "Disconf irmation of Equity Expectations: Effects on Consumer Satisfaction With Services," in Advances in Consumer Research, Volume 12 eds.^ Elizabeth Hirschman and Morris B. Holbrook, Provo, UT: Association for Consumer Research, 340-345. Folkes, Valerie S. (1984a), "An Attributional Approach to Postpurchase Conflict Between Buyers and Sellers," Advances in Consumer Research: Volume 11 ed. Thomas C. Kinnear, Provo, UT: Association for Consumer Research, 500-503. (1984b), "Consumer Reactions to Product Failure: An Attributional Approach," Journal of Consumer Research 10 (March), 398-409. Susan Koletsky, and John L. Graham (1987), "A Field Study of Causal Inferences and Consumer Reaction: The View from the Airport," Journal of Consumer Research 13 (March), 534-539. and Barbara Kotsos (1986), "Buyers' and Sellers' Explanations for Product Failure: Who Done It?" Journal of Marketing 50 (April), 74-80. Fornell, Claes and Nicholas Didow (1980), "Economic Constraints on Consumer Complaining Behavior," in Advances in Consumer Research: Volume 7 ed. Jerry C. Olson, Ann Arbor, MI: Association for Consumer Research, 318-323. and Birger Wernerfelt (1987), "Defensive Marketing Strategy by Customer Complaint Management: A Theoretical Analysis,"' Journal of Marketing Research 24 (November) 337-346. and Robert A. Westbrook (1979), "An Exploratory Study ^ of Assertiveness, Aggressiveness, and Consumer Complaining Behavior," Advances in Consumer Research: Volume 6 ed. William L. Wilkie, Ann Arbor, MI: Association for Consumer Research, 105-110. and Robert A. Westbrook (1984), "The Vicious Circle^of Consumer Complaints," Journal of Marketing 48 (Summer), 68-78. Freedman, Jonathan L. (1964), "Involvement, Discrepancy and Change," Journal of Abnormal and Social Psychology 69 (September), 290-295. Futrell, Charles M. and A. Parasuraman (1984), "The Relationship of Satisfaction and Performance to Salesforce Turnover," Journal of Marketing 48 (Fall), 33-40.
426 Gerard, Harold B. (1965), "Deviation, Conformity, and Commitment," in Current Studies in Social Psychology eds. Ivan D. Steiner and Martin Fishbein, New York, NY: Holt Rinehart and Winston, Inc., 263-277. (1968), "Basic Features of Commitments," in Theories of C ognitive Consistency: A Sourcebook (eds. Robert P. Abelson et al., Chicago, IL: Rand McNally and Company, 456-463. Gibson, Lawrence D. (1981), "The Psychology of Food: Why We Eat What We Eat When We Eat It," Food Technology (February), 54-56. Gilly, Mary C, William L. Cron, and Thomas E. Barry (1983), "The Expectations-Performance Comparison Process: An Investigation of Expectation Types," in International Fare in Consumer Satisfaction and Complaining Behavior eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana University, 10-16. and Betsy D. Gelb (1982), "Post-Purchase Consumer Processes and the Complaining Consumer," Journal of Consumer Research 9 (December), 323-328. Goering, Patricia A. (1985), "Effects of Product Trial on Consumer Expectations, Demand and Prices," Journal of Consumer Research 12 (June) 7 4-82. Goodwin, Stephen A., Vijay Mahajan, and Bhal J. Bhatt (1979), "On Consumer Dissatisfaction: Consumer Arbitration as an Alternative Dispute Resolution Mechanism," Advances in Consumer Research: Volume 6 ed. William L. Wilkie, Ann Arbor, MI: Association for Consumer Research, 1979, 460-465. Gorn, Gerald J. (1982), "The Effects of Music in Advertising on Choice Behavior: A Classical Conditioning Approach," Journal of Marketing 46 (Winter), 94-101. and Clark Leavitt (1984), "Audience Involvement in Advertising: Four Levels," Journal of Consumer Research 11 (June), 581-592. Greenwald, Anthony G. and Clark Leavitt (1984), "Audience Involvement in Advertising: Four Levels," Journal of Consumer Research 11 (June), 581-592. Gronhaug, Kjell (1977), "Exploring Consumer Complaining Behavior: A Model and Some Empirical Results," Advances in Consumer Research: Volume 4 ed. William D. Perreault, Jr., Atlanta: Association for Consumer Research, 159-165.
427 and Johan Arndt (1980), "Consumer Dissatisfaction and Complaining Behavior as Feedback: A Comparative Analysis of Public and Private Delivery Systems," Advances in Consumer Research: Volume 7 ed. Jerry C. Olson, Ann Arbor, MI: Association for Consumer Research, 324-328. and Gerald Zaltman (1981), "Complainers and Noncomplainers Revisited: Another Look at the Data," Advances in Consumer Research: Volume 8 ed. Kent B. Monroe, Ann Arbor, MI: Association for Consumer Research, 83-87. Grubb, Edward L. and Harrison L. Grathwohl (1967), "Consumer Self-Concept, Symbolism and Market Behavior: A Theoretical Approach," Journal of Marketing 31 (October), 22-27. Gurwitz, Sharon B. and Kenneth A. Dodge (1977), "Effects of Confirmations and Disconf irmations on Stereotype-Based Attributions," Journal of Personality and Social Psychology 35, 495-500. Guttman, Louis and Edward A. Suchman (1947), "Intensity and a Zero Point for Attitude Analysis," reprinted in Readings in Attitude Theory and Measurement ed. Martin Fishbein, New York, NY: John Wiley & Sons, 1967, 267-276. Haines, George H. Jr. (1979), "Three Papers on Consumer Satisfaction/Dissatisfaction: A Comment," in Advances in Consumer Research: Volume 6 ed. William L. Wilkie, Ann Arbor, MI: Association for Consumer Research, 450-452. Hartley, Eugene L. (1967), "Attitude Research and the Jangle Fallacy," in Attitude, Ego-Involvement and Change eds Carolyn W. Sherif and Muzafer Sherif, New York, NY: John Wiley & Sons, 88-104. Havlena, William J. and Morris B. Holbrook (1986), "The Varieties of Consumption Experience: Comparing Two Typologies of Emotion in Consumer Behavior," Journal of Consumer Research 13 (December), 394-404. Hawes, Douglass K. Roger D. Blackwell, and W. Wayne Talarzyk (1976), "Consumer Decisions to Reduce or Stop Using Products and Services: Preliminary Results of a Nationwide Study," Advances in Consumer Research: Volume 3 ed. Beverlee B. Anderson, Ann Arbor, MI; Association for Consumer Research, 102-109. Heberlein, Thomas A., Daniel Linz, and Bonnie P. Ortiz (1982), "Satisfaction, Commitment and Knowledge of Customers on a Mandatory Participation Time-of-Day Electricity Pricing Experiment," Journal of Consumer Research 9 (June), 106-114.
428 Heller, Harry E. (1982), "If You Want to Know What is Important, Don't Ask," in Straight Talk About Attitude Research ed. Joseph Chasm, Chicago, IL: American Marketing Association, 1-11. Helson, Harry (1948), "Adaptation-Level as a Basis for a Quantitative Theory of Frames of Reference," Psychological Review 55 (November), 297-313. Herndon, Booton (1972), Satisfaction Guaranteed: An Unconventional Report to Today's Consumers New York, New York: McGraw-Hill. Herzberg, Frederick, Bernard Mausner, and Barbara B. Snyderman (1959), The Motivation to Work Second edition, New York, NY: John Wiley & Sons, Inc. Hoch, Stephen J. and Young-Won Ha (1986), "Consumer Learning: Advertising and the Ambiguity of Product Experience," Journal of Consumer Research 13 (September), 221-233. Horowitz, David (1979), Fight Back, and Don't Get Ripped Off San Francisco, CA: Harper & Row. Houston, Michael J. and Michael L. Rothschild (1978a), "AParadigm for Research on Consumer Involvement," Working Paper, University of Wisconsin-Madison. (1978b), "Conceptual and Methodological Perspectives on Involvement," in Research Frontiers in Marketing: Dialogues and Direction s, ed. Subhash C. Jain, Chicago, IL: American Marketing Association, 184-187. Hover, Wayne D. (1984), "An Examination of Consumer Decision Making for a Common Repeat Purchase Product," Journal of Consumer Research 11 (December), 822-829. Hughes, Donald A. (1977), "An Investigation of the Relation of Selected Factors to Consumer Satisfaction," in Conceptualization and Measure ment of Consumer Satisfaction and Dissatisfaction ed. H. Keith Hunt, Cambridge, MA: Marketing Science Institute, 300-332. Hunt, H. Keith (1976), "CS/D: The Program Planning and Evaluation Perspective," in Advances in Consumer Research: Volume 3 ed. Beverlee 3. Anderson, Ann Arbor, MI: Association for Consumer Research, 259-260. ed. (1977a), Conceptualization and Measurement of Consumer Satisfaction and Dissatisfaction "Introduction" and "CS/D Â— Overview and Future Research Directions," Cambridge, MA: Marketing Science Institute.
429 (1977b), "CS/D: Bits and Pieces," in Consumer Satisfaction, Dissatisfaction and Complaining Behavior ed. Ralph L. Day, Bloomington, IN: Indiana University Conference April 1977, '38-41. (1982), "Consumer Satisfaction/Dissatisfaction: Of Continuing Interest? To Whom? Why?" in Consumerism and Beyond: Research Persectives on the Future Social Environment ed. Paul N. Bloom, Cambridge, MA: Marketing Science Institute. (1983), "A CS/D & CB BibliographyÂ— 1 982, in "" International Fare in Consumer Satisfaction and Complaining Behavior eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana University, 132-154. Hunt, Shelby D. (1976), Marketing Theory: Conceptual Foundations of Research in Marketing Columbus, OH: Grid, Inc. Huppertz, John W. (1979), "Measuring Components of Equity in the Marketplace: Perceptions of Inputs and Outcomes by Satisfied and Dissatisfied Consumers," in New Dimensions of Consumer Satisfaction and Complaining Behavior eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana University, 140-143. Sidney J. Arenson, and Richard H. Evans (1978), "An Application of Equity Theory to Buyer-Seller Exchange Situations," Journal of Marketing Research 15 (May), 250-260. Isen, Alice M. Margaret Clark, and Mark F. Schwartz (1976), "Duration of the Effect of Good Mood on Helping: 'Footprints on the Sands of Time'," Journal of Personality and Social Psychology 34 (Number 3), 385-393. Izard, Carroll E. (1972), Patterns of Emotions: A New Analysis of Anxiety and Depression New York, NY: Academic Press. (1977), Human Emotions, New York, NY: Plenum Press. Jacoby, Jacob and Robert W. Chestnut (1978), Brand Loyalty: Measurement and Management New York, NY: John Wiley & Sons. and James J. Jaccard (1981), "The Sources, Meaning and Validity of Consumer Complaint Behavior: A Psychological Analysis," Journal of Retailing 57 (Fall), 4-24. James, William (1890), The Principles of Psychology Volumes 1 and 2, reprint editions, New York, NY: Dover Publications, Inc. Jensen, Michael C. and William H. Meckling (1976), "Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure," Journal of Financial Economics 3 (October), 305-360.
430 Johnson, Michael P. (1973), "Commitment: A Conceptual Structure and Empirical Application," The Sociological Quarterly 14 (Summer), 395-406. Jordan, Kathi A. and James H. Leigh (1984), "Race and Sex Differences of Children in Satisfaction and Other Factors Associated With Video Games," in Advances in Consumer Research, Volume 11 ed. Thomas Kinnear, Ann Arbor, MI: Association for Consumer Research, 94-99. Kahneman, Daniel and Amos Tversky (1973), "On the Psychology of Prediction," Psychological Review 80 (Number 4), 237-251. Kamins, Michael A. and Henry Assael (1987), "Two-Sided Versus One-Sided Appeals: A Cognitive Perspective on Argumentation, Source Derogation and the Effect of Disconf irming Trial on Belief Change," Journal of Marketing Research 24 (February), 29-39. Kassarjian, Harold H. and Thomas S. Robertson, eds. (1973), Perspectives in Consumer Behavior Revised Edition, Glenview, IL: Scott, Foresman and Co. Kennedy, John R. and Peter C. Thirkell (1987), "An Extended Perspective on the Antecedents of Satisfaction," paper presented at 10th Consumer Satisfaction/Dissatisfaction and Complaining Behavior Conference, Montreal, Quebec, August. and Peter C. Thirkell (1982), "The Effect of Automobile Product Experience on Attribute Disconf irmation and Importance," in New Findings on Consumer Satisfaction and Complaining eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana University, 2-10. Keppel, Geoffrey (1982), Design and Analysis: A Researcher's Handbook Englewood Cliffs, NJ: Prentice-Hall. Kiesler, Charles A. (1971), The Psychology of Commitment: Experiments Linking Behavior to Belief New York, NY: Academic Press. and Michael S. Pallak (1976), "Arousal Properties of Dissonance Manipulations," Psychological Bulletin 83 (Number 6), 1014-1025. Klemke, Lloyd W. (1982), "Exploring Juvenile Shoplifting," Sociology and Social Research 67, 59-75. Knauer, Virginia H. (1983), Consumer's Resource Handbook Pueblo, CO: Consumer Information Center. Koepp, Stephen (1985), "All Wrapped Up in Company Logos," Time (December 2) 70
431 Kourilsky, Marilyn and Trudy Murray (1981), "The Use of Economic Reasoning to Increase Satisfaction with Family Decision Making," Journal of Consumer Research 8 (September), 183-188. Krishnan, S. and Valerie A. Valle (1979), "Dissatisfaction Attributions and Consumer Complaint Behavior," Advances in Consumer Research: Volume 6 ed. William L. Wilkie, Ann Arbor, MI: Association for Consumer Research, 445-449. Kunin, Theodore (1955), "The Construction of a New Type of Attitude Measure," Personnel Psychology 8 (Spring), 65-77. LaBarbera, Priscilla A. and David Mazursky (1983), "A Longitudinal Assessment of Consumer Satisfaction/ Dissatisfaction: The Dynamic Aspect of the Cognitive Process," Journal of Marketing Research 20 (November), 393-404. Lachman, Roy, Janet L. Lachman and Earl C. Butterfield (1979), Cognitive Psychology and Information Processing: An Introduction Hillsdale, NJ : Lawrence Erlbaum Associates. Laird, James D. (1974), "Self-Attribution of Emotion: The Effects of Expressive Behavior on the Quality of Emotional Experience," Journal of Personality and Social Psychology 29 (Number 4), 475-486. Lambert, David R. Ronald J. Dornoff, and Jerome B. Kernan (1977), "The Industrial Buyer and the Postchoice Evaluation Process," Journal of Marketing Research 14 (May), 246-251. Lmgmeyer, Lynn and George Miaoulis (1981), "The Consumer and the Health Care Proces," in Advances in Consumer Research: Volume 7, ed. Jerrv C. Olson, Ann Arbor, MI: Association for Consumer Research', 612-615. LaPiere, Richard T. (1934), "Attitudes Vs. Actions," Social Forces 13 (Number 2), 230-237. Lastovicka, John L. and E. H. Bonfield (1982), "Do Consumers Have Brand Attitudes?" Journal of Economic Psychology 2, 57-75. LaTour, Stephen A. (1978), "Determinants of Participant and Observer Satisfaction With Adversary and Inquisitorial Modes of Adjudication," Journal of Personality and Social Psychology 36 (Number 1 2 ) 1 53 1 1 545 and Nancy C. Peat (1978), "Effects of Prior Experience and Quality of the Best Alternative on Consumer Satisfaction, Complaints and Purchase Termination Responses to Quality Decline," Working Paper #5, University of Florida Center for Consumer Research
432 and Nancy C. Peat (1979), "Conceptual and Methodological Issues in Consumer Satisfaction Research," Advances in Consumer Research: Volume 6 ed. William L. Wilkie, Ann Arbor, MI: Association for Consumer Research, 431-437. and Nancy C. Peat (1980), "The Role of Situationally Produced Expectations, Other's Experiences and Prior Experience in Determining Consumer Satisfaction," Advances in Consumer Research: Volume 7 ed. Jerry C. Olson, Ann Arbor, MI: Association for Consumer Research, 588-592. Leavitt, Clark (1976), "Consumer Satisfaction as the Ultimate Life Force," in Advances in Consumer Research: Volume 3 ed. Beverlee B. Anderson, Ann Arbor, MI: Association for Consumer Research, 252-258. (1977), "Consumer Satisfaction and Dissatisfaction: Bi-Polar or Independent?" in Conceptualization and Measurement of Consumer Satisfaction and Dissatisfaction ed. H. Keith Hunt, Cambridge, MA: Marketing Science Institute, 132-149. Leigh, James H. (1984), "What Buyers Prefer the Next Time Around," Tierra Grande (Number 24), 12-13. Levy, Danna and Carol Surprenant (1982), "A Comparison of Responses co Dissatisfaction with Products and Services," in Conceptual and Empirical Contributions to Con sumer Satisfaction and Complainin g Behavior eds. H. Keith Hunt and Ralph L Day, Bloomington, IN: Indiana University, 43-49. Lewin, Kurt (1951), Field Theory in Social Science New York, NY: Harper. (1952), "Group Decision and Social Change," in Readings in Social Psychology eds. G. Swanson, T. Newcomb and E. Hartley, New York, NY: Holt. Lewis, Robert C. (1983), "Consumers Complain What Happens When Business Responds?" in International Fare in Consumer Satisfaction and Complaining Behavior eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana University, 88-94. Liefeld, John (1980), "Urban/Rural Consumer Expectations and Evaluations of Their Consumer Realities," in Refining Concepts and Measures of Consumer Satisfaction and Complaining Behavior eds. H. Keith Hunt and Ralph L. Day, Bloomington, IN: Indiana University, 95-101.
433 Likert, Rensis (1932), "The Method of Constructing an Attitude Scale," reprinted in Readings in Attitude Theory and Measurement ed. Martin Fishbein, New York, NY: John Wiley & Sons, 1967, 90-95. Lippert, Alice and Maurice MacDonald (1981), "Evaluating User Satisfaction With a Consumer Protection Hotline," Journal of Consumer Affairs 15 (Summer), 136-145. Locke, Edwin A. (1967), "Relationship of Success and Expectation to Affect on Goal-Seeking Tasks," Journal of Personality and Social Psychology 7 (Number 2), 125-134. (1969), "What Is Job Satisfaction?," Organizational Behavior and Human Performance 4 (November), 309-336. Locker, David and David Dunt (1978), "Theoretical and Methodological Issues in Sociological Studies of Consumer Satisfaction With Medical Care," Social Science and Medicine 12 (July), 283-292. Lundstrom, William J. and Laurence M. Lamont (1976), "The Development of a Scale to Measure Consumer Discontent," Journal of Marketing Research 13 (November), 373-381. Lutz, Richard J. (1980), "On Getting Situated: The Role of Situational Factors in Consumer Research," Advances in Consumer Research: Volume 3 ed. Beverlee B. Anderson, Ann Arbor, MI: Association for Consumer Research, 659-663. (1975), "Changing Brand Attitudes Through Modification of Cognitive Structure," Journal of Consumer Research 1 (March), 4959. Lynch, John G. Jr. (1984a), "Comments on 'Intentions and Behaviors'," Advances in Consumer Research: Volume 11 ed. Thomas C. Kinnear, Provo, UT: Association for Consumer Research, 156-158. (1984b), "Adventures in Paramorphic Modeling: Models of Consumers' Processing of Negative Information," Working Paper University of Florida Center for Consumer Research. Lyon, Don and Paul Slovic (1976), "Dominance of Accuracy Information and Neglect of Base Rates in Probability Estimation," Acta Psychologica 40, 287-298. MacCorquodale, Kenneth and Paul E. Meehl (1948), "On a Distinction Between Hypothetical Constructs and Intervening Variables," Psychological Review 55 (Number 1), 95-107. MacDougall, Curtis D. (1958), Hoaxes New York, NY: Dover Publications.
434 Maddox, R. Neil (1981), "Two-Factor Theory and Consumer Satisfaction: Replication and Extension," Journal of Consumer Research 8 (June), 97-102. Marks, Lawrence J., Susan Higgins, and Michael A. Kamins (1988), "Investigating the Experiential Dimensions of Product Evaluation," Advances in Consumer Research: Volume 15 ed. Michael J. Houston, Provo, UT: Association for Consumer Research, 114-121. Maslow, A. H. (1943), "A Theory of Human Motivation," Psychological Review 50, 370-396. Maxwell, John C, Jr. (1988), "Coffee Consumption Continues to Cool Off," Advertising Age (June 13), 14. McNeal, James U. (1969), "Consumer Satisfaction: The Measure of Marketing Effectiveness," MSU Business Topics (Summer), 31-35. McQuarrie, Edward F. and J. Michael Munson (1987), "The Zaichkowsky Personal Involvement Inventory. Modification and Extension," in Advances in Consumer Research: Volume 14 Melanie Wallendorf and Paul Anderson (eds.), Provo, UT, Association for Consumer Research, 36-40. and J. Michael Munson (undated), "A Multidimensional, Arousal-Based Measure of Consumer Involvement," Unpublished working paper, Department of Marketing, Santa Clara University, Santa Clara, CA 95053. Mehrabian, Albert and James A. Russell (1974), An Approach to Environmental Psychology Cambridge, MA: MIT Press. Miller, John A. (1977), "Studying Satisfaction, Modifying Models, Eliciting Expectations, Posing Problems, and Making Meaningful Measurements," in Conceptualization and Measurement of Consumer Satisfaction and Dissatisfaction ed. H. Keith Hunt, Cambridge, MA: Marketing Science Institute, 72-91. Mills, Michael K. (1981), "Deviance and Dissatisfaction: An Exploratory Study," Advances in Consumer Research: Volume 8 ed. Kent B. Monroe, Ann Arbor, MI: Association for Consumer Research, 682-686. Mitchell, Andrew A. (1979), "Involvement: A Potentially Important Mediator of Consumer Behavior," in Advances in Consumer Research: Volume 6 ed William L. Wilkie, Ann Arbor, MI: Association for Consumer Research, 191-196.
435 Mittal, Bonwari and Myung-Sou Lee (1988), "Separating Brand-Choice Involvement from Product Involvement Via Consumer Involvement Profiles," in Advances in Consumer Research Volume 15, ed. Michael J. Houston, Provo, UT : Association for Consumer Research, 43-49. Mizerski, Richard W. Jon B. Freiden and Robert C. Greene, Jr. (1983), "The Effect of the 'Don't Know' Option on TV Ad Claim Recognition Tests," in Advances in Consumer Research: Volume 10 eds. Richard P. Bagozzi and Alice M. Tybout, Ann Arbor, MI: Association for Consumer Research, 283-287. Mook, Douglas G. (1987), Motivation: The Organization of Action New York, NY: W.W. Norton and Company. Moore, Danny L. Douglas Hausknecht, and Kanchana Thamodaran (1986), "Time Compression Response Opportunity and Persuasion," Journal of Consumer Research 13 (June), 85-99. Moore, Ellen M. and F. Kelly Shuptrine (1984), "Disconf irmation Effects on Consumer Satisfaction and Decision Making Processes," Advances in Consumer Research: Volume 11 ed. Thomas C. Kinnear, Provo, UT: Association for Consumer Research, 299-304. Morgan, Fred W. and Dana I. Avrunin (1982), "Consumer Conduct in Product Liability Litigation," Journal of Consumer Research 9 (June), 47-55. Morganoskv, Michelle Ann and Hilda Mayer Buckley (1987), "Complaint Behavior: Analvsis by Demographics, Lifestyle and Consumer Values," Advances in Consumer Research: Volume 14 eds. Melanie Wallendorf and Paul Anderson, Provo, UT: Association for Consumer Research, 223-226. Morris, Earl W. (1977), "A Normative Deficit Approach to Consumer Satisfaction," in Conceptualization and Measurement of Consumer Satisfaction and Dissatisfaction ed. H. Keith Hunt, Cambridge, MA: Marketing Science Institute, 240-274. Mary Winter, and Ivan F. Beutler (1976), "A Normative Deficit Model of Consumer Behavior," in Advances in Consumer Research : Volume 3 ed. Beverlee B. Anderson, Ann Arbor, MI: Association for Consumer Research, 161-165. Moschis, George P. (1976), "Social Comparison and Informal Group Influence," Journal of Marketing Research 13 (August), 237-24^ Dena Saliagas Cox, and James J. Kellaris (1987), "An Exploratory Study of Adolescent Shoplifting Behavior," in Advances in Consumer Research: Volume 14 eds. Melanie Wallendorf and Paul Anderson, Provo, UT: Association for Consumer Research, 526-530.
436 Mowen, John C. and Stephen J. Grove (1983), "Search Behavior, Price Paid, and the 'Comparison Other': An Equity Theory Analysis of Post Purchase Satisfaction," in International Fare in Consumer Satisfaction and Complaining Behavior eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana University, 57-63. Muncy, James A. and Shelby D. Hunt (1984), "Consumer Involvement: Definitional Issues and Research Directions," in Advances in Consumer Research: Volume 11 ed. Thomas C. Kinnear, Provo, UT: Association for Consumer Research, 193-196. Narayana, Chem L. and Rom J. Markin (1975), "Consumer Behavior and Product Performance: An Alternative Conceptualization," Journal of Marketing 39 (October), 1-6. Neisser, Ulric (1967), Cognitive Psychology New York, NY: Meredith Publishing. Norpoth, Helmut and Milton Lodge (1985), "The Difference between Attitudes and Nonattitudes in the Mass Public: Just Measurements?" American Journal of Political Science 29, 291-307. Nunnally, Jum C. (1978), Psychometric Theory second edition, New York, NY: McGraw-Hill. Ofir, Chezy (1982), "Judgment Under Uncertainty: The Integration of Singular and Distributional Information," Working Paper Number 32, University of Florida Center for Consumer Research. Olander, Folke (1980), "Recent Developments in European Consumer Policy Research," in Advances in Consumer Research: Volume 7 ed. Jerry C. Olson, Ann Arbor, MI: Association for Consumer Research, 56-65. (1977), "Consumer Satisfaction Â— A Skeptic's View," in Conceptualization and Measurement of Consumer Satisfaction and Dissatisfaction ed. H. Keith Hunt, Cambridge, MA: Marketing Science Institute, 409-452. Oliva, Terence A. and Alvin C. Burns (1978), "Catastrophe Theory as a Model for Describing Consumer Behavior," Advances in Consume r Research: Volume 5 ed. H. Keith Hunt, Ann Arbor, MI: Association for Consumer Research, 273-276. Oliver, Richard L. (1977), "Effect of Expectation and Disccnf irmat ion on Postexposure Product Evaluations: An Alternative Interpretation," Journal of Applied Psychology 62 (Number 4), 480-486. (1979), "An Interpretation of the Attitudinal and Behavioral Effects of Puffery," Journal of Consumer Affairs 13 (Summer) 8-27
4 37 (1980a), "A Cognitive Model of the Antecedents and Consequences of Satisfaction Decisions," Journal of Marketing Research 17 (November), 460-469. (1980b), "Conceptualization and Measurement of in Disconf irmation Perceptions in the Prediction of Consumer Satisfaction," in Refining Concepts and Measures of Consumer Satisfaction and Complaining Behavior eds. H. Keith Hunt and Ralph L. Day, Bloomington, IN: Indiana University, 2-6. (1980c), "Theoretical Bases of Consumer Satisfaction Research: Review, Critique and Future Direction," in Charles W. Lamb, Jr. and Patrick M. Dunne (eds.) Theoretical Developments Marketing Chicago, IL: American Marketing Association, 206-210. (1981), "Measurement and Evaluation of Satisfaction Processes in Retail Settings," Journal of Retailing 57 (Fall), 25-48. (1987), "An Investigation of the Interrelationship Between Consumer (Dis) Satisfaction and Complaint Reports," Advances in Consumer Research: Volume 14 eds. Melanie Wallendorf and Paul Anderson, Provo, UT: Association for Consumer Research, 218-222. and William 0. Bearden (1983), "The Role of Involvement in Satisfaction Processes," Advances in Consumer Research: Volume 10 eds. Richard P. Bagozzi and Alice M. Tybout, Ann Arbor, MI: Association for Consumer Research, 250-255. and William 0. Bearden (1985), "Disconf irmation Processes and Consumer Evaluations in Product Usage," Journal of Business Research 13 (June), 235-246. and Wayne S. DeSarbo (1988), "Response Determinants in Satisfaction Judgments," Journal of Consumer Research 14 (March), 495-507. and Gerald Linda (1981), "Effect of Satisfaction and^Its Antecedents on Consumer Preference and Intention," Advances in Consumer Research: Volume 8 ed. Kent B. Monroe, Ann Arbor, MI: Association for Consumer Research, 88-93. and Robert A. Westbrook (1982), "The Factor Structure of Satisfaction and Related Postpurchase Behavior," in New Findings on Consumer Satisfaction and Complaining eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana University, 11-14. Olshavsky, R. W. and John A. Miller (1972), "Consumer Expectations, Product Performance and Perceived Quality," Journal of Marketing Research 9 ( February), 19-21.
438 Olson, Jerry C. and Philip A. Dover (1976), "Effects of Expectation Creation and Disconf irmation on Belief Elements of Cognitive Structure," Advances in Consumer Research: Volume 3 ed. Beverlee B. Anderson, Ann Arbor, MI: Association for Consumer Research, 168-175. and Philip A. Dover (1978), "Cognitive Effects of Deceptive Advertising," Journal of Marketing Research 15 (February) 29-38. """" Orne, Martin T. (1962), "On the Social Psychology of the Psychological Experiment: With Particular Reference to Demand Characteristics and Their Implications," American Psychologist 17, 776-783. Ortinau, David J. (1982), "Testing the Influence of Time/Usage Experience on Consumers' Perceived, Expected and Actual Satisfaction Attitudes Toward Automobiles," in New Findings on Consumer Satisfaction and Complaining eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana University, 23-28. Oshikawa, Sadaomi (1969), "Can Cognitive Dissonance Theory Explain Consumer Behavior?" Journal of Marketing 33 (October), 44-49. Park, C. Whan and William J. Bahr (1980), "A Situational Analysis of Communication Effect: A New Product Purchase by Mail Order," in Advances in Consumer Research: Volume 7 ed. Jerry C. Olson, Ann Arbor, MI: Association for Consumer Research, 650-654. and V. Parker Lessig (1977), "Students and Housewives: Differences in Susceptibility to Reference Group Influence," Journal of Consumer Research 4 (September), 102-110. Pe-due, Barbara C. and John 0. Summers (1986), "Checking the Success of Manipulations in Marketing Experiments," Journal of Marketing Research 23 (November), 317-325. Peter, J. Paul (1984), "What a MESS: Comments on Materialism, Ethnocentrism and Symbolism Session," in Advances in Consumer Research: Volume 1 1 ed. Thomas C. Kinnear, Provo, UT: Association for Consumer Research, 298. and Lawrence Tarpey (1975), "A Comparative Analysis of Three Consumer Decision Strategies," Journal of Consumer Resear ch 2 (June), 29-37. Peterson, Robert A., Wayne D. Hoyer and William R. Wilson (1986), The Role of Affect in Consumer Behavior Lexington, MA: D.C. Heath and Company. Petty, Richard E. and John T. Cacioppo (1986), Communication and Persuasion: Central and Peripheral Routes to Attitude Change New York, NY: Springer-Verlag
439 John T. Cacioppo and David Schumann (1983), "Central and Peripheral Routes to Advertising Effectiveness: The Moderating Role of Involvement," Journal of Consumer Research 10 (September), 135-146. Pfaff, Martin (1977), "The Index of Consumer Satisfaction: Measurement Problems and Opportunities," in Conceptualization and Measurement of Consumer Satisfaction and Dissatisfaction ed. H. Keith Hunt, Cambridge, MA: Marketing Science Institute, 36-72. Pierce, John C. and Douglas D. Rose (1974), "Nonattitudes and American Public Opinion: The Examination of a Thesis," American Political Science Review 68, 626-649. Prakash, Ved and John W. Lounsbury (1983), "A Reliability Problem in the Measurement of Disconf irmation of Expectations," in Advances in Consumer Research: Volume 10 eds. Richard P. Bagozzi and Alice M. Tybout, Ann Arbor, MI: Association for Consumer Research, 244-249. Pucely, Marya J., Richard Mizerski, and Pamela Perrewe (1988), "A Comparison of Involvement Measures for the Purchase and Consumption of Pre-Recorded Music," in Advances in Consumer Research Volume 15, ed. Michael J. Houston, Provo, UT: Association for Consumer Research, 37-42. Reitz, H. Joseph (1981), Behavior in Organizations revised edition, Homewood, IL: Richard D. Irwin, Inc. Resnik, Alan, Brian Gnauck and Rodney Aldrich (1977), "Corporate Responsiveness to Consumer Complaints," in Consumer Satisfaction, Dissatisfaction and Complaining Behavior ed. Ralph L. Day, Bloomington, IN: Indiana University, 148-152. and Robert R. Harmon (1983), "Consumer Complaints and Managerial Response: A Holistic Approach," Journal of Marketing 47 (Winter), 86-97. Reymert, Martin L. ed. (1928), Feelings and Emotions: The Wittenberg Symposium New York, NY: Arno Press. Rhine, Ramon J. and Laurence J. Severance (1970), "Ego-Involvement, Discrepancy, Source Credibility, and Attitude Change," Journal of Personality and Social Psychology 16 (No. 2), 175-190. Richins, Marsha L. (1987), "A Multivariate Analysis of Responses to Dissatisfaction," Journal of the Academy of Marketing Science 15 (Fall), 24-31.
440 (1985), "Factors Affecting the Level of Consumer Initiated Complaints to Marketing Organizations," in Consumer Satisfaction, Dissatisfaction and Complaining Behavior eds. H. Keith Hunt and Ralph L. Day, Bloomington, IN: Indiana University, 82-85. (1983a), "Negative Word-of-Mouth by Dissatisfied Consumers: A Pilot Study," Journal of Marketing 47 (Winter), 63-78. (1983b), "Word-of-Mouth as an Expression of Product Dissatisfaction," in International Fare in Consumer Satisfaction and Complaining Behavior eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana University, 100-104. (1982), "An Investigation of Consumers' Attitudes Toward Complaining," Advances in Consumer Research: Volume 9 ed. Andrew Mitchell, Ann Arbor, MI: Association for Consumer Research, 502-506. and Peter H. Bloch (1987), "The Role of Situational and Enduring Involvement in Post-Purchase Product Evaluation," Paper presented at 10th Consumer Satisfaction/Dissatisfaction and Complaining Behavior Conference, Montreal, Quebec, August. and Peter H. Bloch (1986), "After the New Wears Off: The Temporal Context of Product Involvement," Journal of Consumer Research 13 (September), 280-285. and Teri Root-Shaffer (1988), "The Role of Involvement and Opinion Leadership in Consumer Word-of-Mouth: An Implicit Model Made Explicit," in Advances in Consumer Research Volume 15. ed. Michael J. Houston, Provo, UT: Association for Consumer Research, 32-36. Risley, George (1972), Modern Industrial Marketing New York, NY: McGraw-Hill Robertson, Thomas S. (1976), "Low-Commitment Consumer Behavior," Journal of Advertising Research 16 (April), 19-24. John R. Rossiter, and Scott Ward (1985), "Consumer Satisfaction Among Children," in Advances in Consumer Research, Volume 12 eds. Elizabeth C. Hirschman and Morris B. Holbrook, Provo, UT: Association for Consumer Research, 279-284. Robinson, John A. (1980), "Affect and Retrieval of Personal Memories,' Motivation and Emotion 4 (Number 2), 149-174. (1976), "Sampling Autobiographical Memory," Cognitive Psychology 8, 578-595.
441 Robinson, Larry M. and Robert L. Berl (1980), "What About Compliments: A Follow-Up Study on Customer Complaints and Compliments," in Refining Concepts and Measures of Consumer Satisfaction and Complaining Behavior eds. H. Keith Hunt and Ralph L. Day, Bloomington, IN: Indiana University, 144-148. George G. Trebbi, and Roy D. Adler (1983), "Taking It to the Top: A Profile of Customers Who Complain to the CEO, in International Fare in Consumer Satisfaction and Complaining Behavior eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana University, 95-99. Rodgers, Alice L. and Daniel J. Sweeney (1980), "Satisfaction With Retail Stores as Reflected in Consumers' Opinions," in Refining Concepts and Measures of Consumer Satisfaction and Complaining Behavior eds. H. Keith Hunt and Ralph L. Day, Bloomington, IN: Indiana University, 153-158. Rokeach, Milton (1970), Beliefs, Attitudes and Values San Francisco, CA: Jossey-Bass, Inc. Rosander, A. C. (1937), "An Attitude Scale Based Upon Behavior Situations," Journal of Social Psychology 8, 3-15. Rosenthal, Robert (1982), "Valid Interpretation of Quantitative Research Results," in New Directions for Methodology of Social and Behavioral Science: Forms of Validity in Research eds. David Brinberg and L. Kidder, San Francisco, CA: Jossey-Bass, 59-75. Ross, Ivan (1985), "Consumer Initiated Communication Audit," in Consumer Satisfaction, Dissatisfaction, and Complaining Behavior eds. H. Keith Hunt and Ralph L. Day, Bloomington, IN: Indiana University, 73-81 and Kathy Gardner (1985), "The Use of Consumer Initiated Communication as Marketing Research Data: The View of Consumer Affairs Professionals," unpublished manuscript, University of Minnesota. Ross, Stephen A. (1973), "The Economic Theory of Agency: The Principal's Problem," American Economic Review 63 (May), 134-139. Rothschild, Michael L. (1984), "Perspectives on Involvement: Current Problems and Future Directions," Advances in Consumer Research: Volume 11 ed. Thomas C. Kinnear, Provo, UT: Association for Consumer Research, 216-217. Ruekert, Robert W. and Gilbert A. Churchill, Jr. (1984), "Reliability and Validity of Alternative Measures of Channel Member Satisfaction," Journal of Marketing Research 21 (May), 226-233.
442 Rusbult, Caryl E. (1983), "A Longitudinal Test of the Investment Model: The Development (and Deterioration) of Satisfaction and Commitment in Heterosexual Involvements," Journal of Personality and Social Psychology 45 (Number 1), 101-117. and Isabella Zembrodt (1983), "Responses to Dissatisfaction in Romantic Involvements: A Multidimensional Scaling Analysis," Journal of Experimental Social Psychology 19, 274-293. Russo, J. Edward (1979), "Consumer Satisfaction/Dissatisfaction: An Outsider's View," in Advances in Consumer Research: Volume 6 ed. William L. Wilkie, Ann Arbor, MI: Association for Consumer Research, 453-455. Sampson, Edward A. and Linda B. Sibley (1965), "A Further Examination of the Confirmation or Nonconf irmation of Expectancies and Desires," Journal of Personality and Social Psychology 2 (Number 1), 133-137. Sawyer, Alan G. (1975), "Demand Artifacts in Laboratory Experiments in Consumer Research," Journal of Consumer Research 1 (March), 20-30. (1977), "The Role of Role Playing in Experiments About Consumer Behavior," in Contemporary Marketing Thought eds. Barnett A. Greenberg and Danny N. Bellenger, Chicago, IL: American Marketing Association, 191-194. Parker M. Worthing, and Paul E. Sendak (1979), "The Role of Laboratory Experiments to Test Marketing Strategies," Journal of Marketing 43 (Summer), 60-67. Scammon, Debra L. and Mary Jane Sheffet (1984), "Regulation in the 80' s: What is the Role of Consumer Researchers?" in Advances in Consumer Research: Volume 11 ed. Thomas C. Kinnear, Provo, UT: Association for Consumer Research, 463-465. Schachter, Stanley (1964), "The Interaction of Cognitive and Physiological Determinants of Emotional State," in Advances in Experimental Social Psychology ed. Leonard Berkowitz, New York, NY: Academic Press, 49-80. Scherf, Gerhard W.H. (1974), "Consumer Dissatisfaction as a Function ci Dissatisfaction with Interpersonal Relationships," Journal of Applied Psychology 59 (Number 4), 465-471. Schmalensee, Richard (1978), "A Model of Advertising and Product Quality," Journal of Political Economy 86 (Number 3), 485-503. Scott, Carol A. (1976), "The Effects of Trial and Incentives on Repeat .Purchase Behavior," Journal of Marketing Research 13 (August), 263-269.
443 and Richard F. Yalch (1980), "Consumer Response to Initial Product Trial: A Bayesian Analysis," Journal of Consumer Research 7 (June), 32-41 Sechrest, Lee (1979), Unobtrusive Measurement Today San Francisco, CA: Jossey-Bass, Inc. Serafin, Raymond (1987), "Auto Makers Stress Consumer Satisfaction," Advertising Age February 23, S-12. Settle, Robert B. (1973), "Attribution Theory and Consumer Behavior," in Perspectives in Consumer Behavior revised edition eds. Harold H. Kassarjian and Thomas S. Robertson, Glenview, IL: Scott, Foresman and Co., 64-74. Shapiro, Benson P. (1968), "The Psychology of Pricing," Harvard Business Review 46 (July-August), 14, 25, 160. Shaw, Marvin E. and Philip R. Costanzo (1982), Theories of Social Psychology second edition, New York, NY: McGraw-Hill. and Jack M. Wright (1967), Scales for the Measurement of Attitudes New York, NY: McGraw-Hill. Shelly, Maynard W. ed. (1972) Analyses of Satisfaction Volumes I, II, III, New York, NY: MSS Information Corporation. Sherif, Muzafer and Carl I. Hovland (1961), Social Judgment: Assimilation and Contrast Effects in Communication and Attitude Change New Haven, CT: Yale University Press. Sherman, Elaine and Leon G. Schiffman (1984), "Applying Age-Gender Theory from Social Gerontology to Understand the Consumer Well-Being of the Elderly," in Advances in Consumer Research: Volume 11 ed. Thomas C. Kinnear, Provo, UT: Association for Consumer Research, 569-573. Sherrell, Daniel L. and Terence A. Shimp (1982), "Consumer Involvement in a Laboratory Setting," in An Assessment of Marketing Thought and Practice ed. Bruce J. Walker, William 0. Bearden, William R. Darden, Patrick E. Murphy, John R. Nevin, Jerry C. Olson, Barton A. Weitz, Chicago, IL: American Marketing Association, 104-108. Sheth, Jagdish N. (1971), "Word-of-Mouth in Low-Risk Innovations," Journal of Advertising Research 11 (June), 15-18. Shuptrine, F. Kelly and Gerhard Wenglorz (1981), "Comprehensive Identification of Consumers' Marketplace Problems and What They Do About Them," in Advances in Consumer Research, Volume VIII ed. Kent Monroe, Ann Arbor, MI: Association for Consumer Research, 687-692.
444 and Gerhard Wenglorz (1980), "Product Satisfaction/ Dissatisfaction: How Content Are Consumers with Products They Buy and Use?" in Refining Concepts and Measures of Consumer Satisfaction and Comp laining Beha vior eds H. Keith Hunt and Ralph L. Day, Bloomington, IN: Indiana University, 149-152. Shuv-Ami, Avichai (1984), "Measures of Disconf irmation: An Integration," in Proceedings of 1984 AMA Winter Educators' Conference: Scientific Method in Marketing eds. Paul F. Anderson and Michael J. Ryan, Chicago, IL: American Marketing Association, 233-237. Silverman, Irwin (1968), "Expectancy Disconf irmation and the Choice of Negative Alternatives: Dissonance Avoidance or Situational Demands?" in Theories of Cognitive Consistency: A Sourcebook (eds. Robert P. Abelson et al., Chicago, IL: Rand McNally and Company, 479-484. Singh, Jagdip (1988), "Consumer Complaint Intentions and Behavior: Definitional and Taxonomical Issues," Journal of Marketing 52 (January), 93-107. Sirgy, M. Joseph (1982), "Self-Concept in Consumer Behavior: A Critical Review," Journal of Consumer Research 9 (December), 287-300. (1984), "A Social Cognition Model of Consumer Satisfaction/Dissatisfaction," Psychology and Marketing 1 (Number 2, Summer) 27-44 Smead, Raymond J., James B. Wilcox, and Robert E. Wilkes (1980), "An Illustration and Evaluation of a Joint Process Tracing Methodology: Eye Movement and Protocols," Advances in Consumer Research: Volume 7 ed. Jerry C, Olson, Ann Arbor, MI: Association for Consumer Research, 507-512. Smith, Darlene B. and Paul N. Bloom (1984), "Is Consumerism Dead or Alive? Some New Evidence," Advances in Consumer Research: Volu me JJ_> ed. Thomas C. Kinnear, Provo, UT: Association for Consumer Research, 369-373. Smith, Patricia C, Lome M. Kendall, and Charles L. Hulin (1969), The Measurement of Satisfaction in Work and Retirement: A Strategy for the Study of Attitudes Chicago, IL: Rand McNally and Co. Smith, Robert E. and William R. Swinyard (1982), "Information Response Models: An Integrated Approach," Journal of Marketing 46 (Winter), 81-93.
445 and William R. Swinyard (1983), "Attitude-Behavior Consistency: The Impact of Product Trial Versus Advertising," Journal of Marketing Research 20 (August), 257-262. Sobel, Robert S. and Howard McGuire (1977), "Subtractive Models of Satisfaction: Cognitive Processes or Measurement Artifact?" Journal of Psychology 96 (July), 251-258. Solomon, Michael R. (1986), "The Missing Link: Surrogate Consumers in the Marketing Chain," Journal of Marketing 50 (October), 208-218. Solomon, Richard L. and John D. Corbit (1974), "An Opponent-Process Theory of Motivation: I. Temporal Dynamics of Affects," Psychological Review 81 (March), 119-145. Steers, Richard M. and Lyman W. Porter (1975), Motivation and Work Behavior New York, NY: McGraw-Hill. Stein, Karen F. (1980), "Explaining Ghetto Consumer Behavior: Hypotheses from Urban Sociology," Journal of Consumer Affairs 14 (Summer), 232-241 Sternthal, Brian, Alice M. Tybout and Bobby J. Calder (1987), "Confirmatory Versus Comparative Approaches to Judging Theory Tests," Journal of Consumer Research 14 (June), 114-125. Stone, Robert N. (1984), "The Marketing Characteristics of Involvement," in Advances in Consumer Research: Volume 11 ed. Thomas C. Kinnear, Provo, UT: Association for Consumer Research, 210-215. Strumpel, Burkhard, ed. (1974), Subjective Elements of Weil-Being Paris: Organisation for Economic Co-operation and Development. Summers, John 0. and Donald H. Granbois (1977), "Predictive and Normative Expectations in Consumer Dissatisfaction and Complaining Behavior," in Advances in Consumer Research: Volume 4 ed. William D. Perreault, Jr. Atlanta: Association for Consumer Research, 155-158. Surprenant, Carol and Gilbert A. Churchill, Jr. (1984), "Can Role^ Playing be Substituted for Actual Consumption?" in Advances in Consumer Research: Volume 11 ed. Thomas C. Kinnear, Provo, UT: Association for Consumer Research, 122-126. Swan, John E. (1983), "Consumer Satisfaction Research and Theory: Current Status and Future Direction," in International Fare in Consumer Satisfaction and Complaining Behavior eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana University, 124-129.
446 (1987), "Consumer Satisfaction Related to Disconfirmation of Expectations and Product Performance," Paper presented at 10th Consumer Satisfaction/ Dissatisfaction and Complaining Behavior Conference, Montreal, Quebec, August 1987. and Maxwell G. Carroll (1980), "Patient Satisfaction: An Overview of Research 1965-1978," in Refining Concepts and Measures of Consumer Satisfaction and Complaining Behavior eds. H. Keith Hunt and Ralph L. Day, Bloomington, IN: Indiana University, 112-118. and Linda Jones Combs (1976), "Product Performance and Consumer Satisfaction: A New Concept," Journal of Marketing 40 (April), 25-33. and Warren S. Martin (1981), "Testing Comparison Level and Predictive Expectations Models of Satisfaction," Advances in Consumer Research: Volume 8 ed. Kent B. Monroe, Ann Arbor, MI: Association for Consumer Research, 77-82. and I. Frederick Trawick (1979), "Testing an Extended Concept of Consumer Satisfaction," in New Dimensions of Consumer Satisfaction and Complaining Behavior eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana University, 56-61 and I. Frederick Trawick (1983), "Satisfaction With and Support of a Continuously Utilized Service: A Public Radio Station," in International Fare in Consumer Satisfaction and Complaining Behavior eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana University, 75-80. and I. Frederick Trawick (1980a), "Satisfaction Related Predictive Vs. Desired Expectations," in Refining Concepts and Measures of Consumer Satisfaction and Complaining Behavior eds. H. Keith Hunt and Ralph L. Day, Bloomington, IN: Indiana University, 7-12. and I. Frederick Trawick (1980b), "Triggering Cues and the Evaluation of Products as Satisfactory or Dissatisfactory," in Proceedings of the 1979 Educators' Conference Chicago, IL: American Marketing Association, 231-234. I. Frederick Trawick, and Maxwell G. Carroll (1981), "Effect of Participation in Marketing Research on Consumer Attitudes Toward Research and Satisfaction with a Service," Journal of Marketing Research 18 (August), 356-363. Tapp, Shelley R. (1984), "Brunswik's Lens Model: A Review and Extension to Consumer Research," in Advances in Consumer Research: Volume 1 1 ed. Thomas C. Kinnear, Provo, UT: Association for Consumer Research, 103-108.
447 Technical Assistance Research Programs, Inc. (1979), Consumer Complaint Handling in America: Final Report Washington, DC: TARP. Temerlin, Maurice Kahn (1963), "On Choice and Responsibility in a Humanistic Psychotherapy," Journal of Humanistic Psychology 3, 35-48. Triandis, Harry C. and Leigh M. Triandis (1965), "Some Studies of Social Distance," reprinted in Readings in Attitude Theory and Measurement ed. Martin Fishbein, New York, NY: John Wiley & Sons, 1967, 199-207. Trombetta, William L. and Timothy L. Wilson (1975), "Foreseeability of Misuse and Abnormal Use of Products by the Consumer," Journal of Marketing 39 (July), 48-55. Tse, David K. and Peter C. Wilton (1988), "Models of Consumer Satisfaction Formation: An Extension," Journal of Marketing Research 25 (May), 204-212. Valle, Valerie and Melanie Wallendorf (1977), "Consumers' Attributions of the Cause of Their Product Satisfaction and Dissatisfaction, in Consumer Satisfaction, Dissatisfaction and Complaining Behavior ed. Ralph L. Day, Bloomington, IN: Indiana University, 26-30. Vredenburg, Harrie and Chow-Hou Wee (1986), "The Role of Customer Service in Determining Customer Satisfaction," Journal of the Academy of Marketing Science 14 (Summer), 17-26. Vroom, Victor H. (1964), Work and Motivation New York, NY: John Wiley & Sons Wallendorf, Melanie and Eric J. Arnould (1988), "My Favorite Things: A Cross-Cultural Inquiry into Object Attachment, Possessiveness and Social Linkage," Journal of Consumer Research 14 (March), 531-547. Russell Belk, and Deborah Heisley (1988), "Deep Meaning in Possessions: The Paper," in Advances in Consumer Research Volume 15, ed. Michael J. Houston, Provo, UT: Association for Consumer Research, 528-530. Walster, Elaine (1964), "The Temporal Sequence of Post-Decision Processes," reprinted in Readings in Social Psychology eds. Jonathan L. Freedman, J. Merrill Carlsmith and David 0. Sears, Englewood Cliffs, NJ: Prentice-Hall, 1971, 517-528. Walton, John R. and Eric N. Berkowitz (1985), "Consumer Decision Making and Perceived Decision Freedom," in Advances in Consumer Research Volume 12, eds. Elizabeth C. Hirschman and Morris B. Holbrook, Provo, UT: Association for Consumer Research, 461-464.5.
448 Wanous, John P. and Edward E. Lawler III (1972), "Measurement and Meaning of Job Satisfaction," Journal of Applied Psychology 56 (Number 2), 95-105. Watts, William A. (1968), "Predictability and Pleasure: Reactions to the Disconfirmation of Expectancies," in Theories of Cognitive Consistency: A Sourcebook (eds.) Robert P. Abelson et al., Chicago, IL: Rand McNally and Company, 456-463. Webb, Eugene J., Donald T. Campbell, Richard D. Schwartz, and Lee Sechrest (1966), Unobtrusive Measures: Nonreactive Research in the Social Sciences Chicago, IL: Rand McNally College Publishing Co. Weick, Karl G. (1967), "Promise and Limitations of Laboratory Experiments in the Development of Attitude Change Theory," in Carolyn W. Sherif and Muzafer Sherif, eds., Attitude, Ego-Involvement and Change New York, NY: John Wiley & Sons, 51-75. Weiner, Bernard (1985), Human Motivation New York, NY: SpringerVerlag. (1980), "A Cognitive (Attribution)-Emotion-Action Model of Motivated Behavior: An Analysis of Judgments of Help-Giving," Journal of Personality and Social Psychology 39 (Number 2), 186-200. Dan Russell, and David Lerman (1979), "The Cognition-Emotion Process in Achievement-Related Contexts, Journal of Personality and Social Psychology 37 (Number 7), 1211-1220. Dan Russell and David Lerman (1978), "Affective Consequences of Causal Ascriptions," in New Directions in Attribution Research, Volume 2 eds. John H. Harvey, William Ickes and Robert F. Kidd, Hillsdale, NJ: Lawrence Erlbaum Associates, 59-90. Westbrook, Robert A. (1977), "A Study of Consumer Dissatisfaction Before Purchase," in Advances in Consumei Research: Volume 4 ed. William D. Perreault, Jr., Atlanta: Association for Consumer Research, 142-148. (1980a), "Intrapersonal Affective Influences on Consumer Satisfaction with Products," Journal of Consumer Research 7 (June), 49-54. (1980b), "A Rating Scale for Measuring Product/ Service Satisfaction," Journal of Marketing 44 (Fall), 68-72.
449 (1980c), "Consumer Satisfaction as a Function of Personal Competence/Efficacy," Journal of the Academy of Marketing Science 8 (Fall), 427-437. (1981), "Sources of Consumer Satisfaction With Retail Outlets," Journal of Retailing 57 (Fall), 68-85, (1983), "Consumer Satisfaction and the Phenomenology of Emotions During Automobile Ownership Experiences, in International Fare in Consumer Satisfaction and Complaining Behavior eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana University, 2-9. (1984), "On Satisfaction Responses and the Effects of Product Labels," in Advances in Consumer Research: Volume 11 ed. Thomas C. Kinnear, Provo, UT: Association for Consumer Research, 31 1-313. (1987), "Product/Consumption-Based Affective Responses and Postpurchase Processes," Journal of Marketing Research 24 (August), 258-270. and Joseph A. Cote, Jr. (1980), "An Exploratory Study oi Non-Product Related Influences Upon Consumer Satisfaction, in Advances in Consumer Research: Volume 7 ed. Jerry C. Olson, Ann Arbor, MI: Association for Consumer Research, 577-581. and Richard L. Oliver (1981), "Developing Better Measures of Consumer Satisfaction: Some Preliminary Results," in Advances in Consumer Research: Volume 8 ed. Kent B. Monroe, Ann Arbor, MI: Association for Consumer Research, 94-99. and Richard L. Oliver (1984), "Emotion and Consumer Satisfaction in Product Ownership/Usage," unpublished manuscript. and Michael D. Reilly (1983), "Value-Percept Disparity: An Alternative to the Disconf irmation of Expectations Theory of Consumer Satisfaction," in Advances in Consumer Research: Volume J_0, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Arbor, MI: Association for Consumer Research, 256-261 Wilkes, Robert E. (1978), "Fraudulent Behavior by Consumers," Journal of Marketing 42 (October), 67-75. Wiikie, William L. and Edgar A. Pessemier (1973), "Issues in Marketing's Use of Multi-Attribute Attitude Models," Journal of Marketing Research 10 (November), 428-441. Wilkinson, J. B. and J. Barry Mason (1979), "Dissatisfaction and Complaining Behavior as a Result of Unavailable Advertised Food Specials," Journal of Consumer Policy 3 (Spring), 43-55.
450 Wilson, David T. H. Lee Mathews, and James W. Harvey ( 1 975) "An Empirical Test of the Fishbein Behavioral Intention Model," Journal of Consumer Research 1 (March), 39-48. Wilton, Peter C. and David K. Tse (1983), "A Model of Consumer Response to Communication and Product Experiences," in Advertising and Consumer Psychology eds. Larry Percy and Arch G. Woodside, Lexington, MA: Lexington Books, 315-332. Winkler, Robert L. (1972), An Introduction to Bayesian Inference and Decision New York, NY: Holt, Rinehart and Winston, Inc. Winter, Frederick W. (1974), "The Effect of Purchase Characteristics on Postdecision Product Reevaluation, Journal of Marketing Research 11 (May), 164-171. Withey, Stephen B. (1977), "Integrating Some Models About Consumer Satisfaction," in Conceptualization and Measurement of Consumer Satisfaction and Dissatisfaction ed. H. Keith Hunt, Cambridge, MA: Marketing Science Institute, 120-149. (1974), "Values and Social Change," in Subjective Elements of Weil-Being ed Burkhard Strumpel, Paris: Organisation for Economic Cooperation and Development. Woodruff, Robert B., Ernest R. Cadotte, and Roger L. Jenkins (1983a), "Modeling Consumer Satisfaction Processes Using Experience-Based Norms," Journal of Marketing Research 20 (August), 296-304. Ernest R. Cadotte, and Roger L. Jenkins (1983b), "Charting a Path for CS/D Research," in International Fare in Consumer Satisfaction and Complaining Behavior eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana University, 118-123. Young, Paul T. (1961), Motivation and Emotion: A Survey of the Determinants of Human and Animal Activity, New York, NY: John Wiley & Sons, Inc. Zaichkowsky, Judith Lynne (1985), "Measuring the Involvement Construct," Journal of Consumer Research 12 (December), 341-352. (1987), "The Emotional Aspect of Product Involvement," Advances in Consumer Research: Volume 14 ed. Melanie Wallendorf and Paul Anderson, Provo, UT: Association for Consumer Research, 32-35. Zajonc, Robert B. (1984), "On the Primacy of Affect," American Psychologist 39 (Number 2), 117-123. (1980), "Feeling and Thinking: Preferences Need No Inferences," American Psychologist 35 (February), 151-175.
451 (1968), "Attitudinal Effects of Mere Exposure," reprinted in Readings in Social Psychology eds Jonathan L. Freedman, J. Merrill Carlsmith and David 0. Sears, Englewood Cliffs, NJ: Prentice-Hall, Inc., 1971, 93-131. Zaltman, Gerald, Rajendra K. Srivastava, and Rohit Deshpande (1978), "Perceptions of Unfair Marketing Practices: Consumerism Implications," Advances in Consumer Research: Volume 5 ed. H. Keith Hunt, Ann Arbor, MI: Association for Consumer Research, 247-253. Zeithaml, Valarie A. (1982), "Consumer Response to In-Store Price Information Environments," Journal of Consumer Research 8 (March), 357-369. (1981), "How Consumer Evaluation Processes Differ Between Goods and Services," in Marketing of Services eds. James Donnelly and William George, Chicago, IL: American Marketing Association, 186-190. Zillman, Dolf (1978), "Attribution and Misattribution of Excitatory Reactions," in New Directions in Attribution Research, Volume 2 eds. John H. Harvey, William Ickes and Robert F. Kidd, Hillsdale, NJ: Lawrence Erlbaum Associates, 335-368. Zimbardo, Philip G. (1960), "Involvement and Communication Discrepancy as Determinants of Opinion Conformity," Journal of Abnormal and Social Psychology 60 (Number 1), 86-94. Zimmer, Mary and Rohit Deshpande (1984), "The Development of Choice Tactics in Low Involvement Situations," in Proceedings of 1984 Winter Educators' Conference: Scientific Method in Marketing eds. Paul F. Anderson and Michael J. Ryan, Chicago, IL: American Marketing Association, 220-223.
BIOGRAPHICAL SKETCH Douglas R. Hausknecht was born August 10, 1957, in Baltimore, Maryland. He is the son of Norman C. and N. Elaine Hausknecht and the brother of Brian, Cheryl, and Pamela. His education began in Baltimore parochial schools, St. Augustine, and was continued in Jacksonville, Florida, where the family moved in 1970. He was graduated from Samuel W. Wolf son Senior High School in June, 1975. His college education followed immediately at the University of Florida and culminated in a Bachelor of Science in Agriculture degree (major: food science) in March, 1979. Subsequently, in June of 1981, he was graduated Master of Business Administration (major: marketing). He began doctoral studies in January, 1982. He has published articles in the Journal of Consumer Research and Ad vances in Consumer Research He began teaching at the University of Akron in January, 1986. 452
I certify that I have read this study and that in my opinion, it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Richard J. Lutz, Chairman Professor of Marketing I certify that I have read this study and that in my opinion, it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Us* /LJ/fse^ph W. Alba -Associate Professor of Marketing I certify that I have read this study and that in my opinion, it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. -/^a/C: Esam Ahmed Professor of Food Science and Human Nutrition This dissertation was submitted to the Graduate Faculty of the Department of Marketing in the College of Business Administration and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. December 1988 Dean, Graduate School
UNIVERSITY OF FLORIDA 3 1262 08285 323 4