EFFECTS OF ISSUE INVOLVEMENT AND
EGO-INVOLVEMENT UPON ACCEPTANCE OF A
STEVEN HUGH WILLIAMS
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 1987
I would like to take this opportunity to acknowledge and thank those individuals who have guided and supported me throughout the completion of my dissertation. My first thanks is to the Lord Jesus Christ for the life He has given me during the past six years.
I am grateful to Greg Neimeyer for guiding and encouraging me throughout my graduate studies and especially during the completion of my dissertation. Most of all, I am thankful for his time and his personal interest in me and my professional development. I would also like to thank the members of my committee, Franz Epting, Shae Kosch, Constance Shehan, and Mark Alicke, for their effort and insightful suggestions. It has been my pleasure to work with these outstanding individuals.
I would also like to thank all the people who were an integral part of this dissertation. Dr. Constance Shehan allowed me to utilize her classroom to solicit subjects. A special thanks goes to the undergraduate research assistants who were involved in the technical aspects of this study. Jean Callahan, Lisa Oglander, and John Guy were invaluable in their assistance. April Metzler also gave much time to this project.
I would also like to give a very special
acknowledgment to my wife, Barbara, for her constant love and support and the many sacrifices she has made in her life so that I may pursue my career. Barbara and my son, Adam, have been great sources of support and encouragement throughout this project.
TABLE OF CONTENTS
ACKNOWLEDGMENTS ............... ............ ......... ii
ABSTRACT o. vi
I INTRODUCTION ...... ......................
II REVIEW OF THE LITERATURE ... o-.......... 9
The Cognitive Response Approach to Persuasion. 9
The Elaboration Likelihood Model ..... .. 19
Cognitive Processing .............. 25
Objective Processing ....................... 28
Biased Processing .... ... ............... 32
Involvement and Persuasion ... 38
Involvement and Cognitive Processing ........... 44
Involvement Level in Counseling ......... 49
Hypotheses .... ............ .......... 51
III METHODS .................... ..... ... .......54
Research Design ......... .......... 55
Cognitive Responses to Arguments ................ 57
Description of the Sample ..................... 59
Instrumentation ...... ............. ..... ..61
Relationships Grid ......................... 61
Personal Problems Inventory.................. 66
Assertion Survey .......... ..... .... 69
Thought Listing Techniques .. ..... 70
Cognitive Assessment Questionnaire ......... 73
Procedures ................ .............. 73
Analyses of Data .............................. 75
Summary ......................................... 75
IV RESULTS ...................................... 78
Manipulation Checks ..................... 78
Argument Quality .......................... 78
Issue Involvement ..................... 79
Dependent Measures ............................ 80
Cognitive Favorability (CFCR) .............. 81
Attitude Rating ............................ 81
Behavioral Intention Rating ................ 86
Spontaneous Use of Cognitive Restructuring.. 86
Additional Analyses ......... .............. 87
V DISCUSSION ........ .......... .. ........... .... 89
Need for Cognition ............................ 96
Future Considerations ......................... 96
A RELATIONSHIPS GRID ............................ 101
B PERSONAL PROBLEM INVENTORY .................... 104
C ASSERTION SURVEY .............. 107
D THOUGHT LISTING TECHNIQUE .................... 109
E COVERT REHEARSAL TECHNIQUE ................ 112
F COGNITIVE ASSESSMENT QUESTIONNAIRE- ............ 114
G TRANSCRIPT OF TAPED INSTRUCTIONS ................ 116
REFERENCES. ......................................... .. 124
BIOGRAPHICAL SKETCH ................................... 135
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
EFFECTS OF ISSUE INVOLVEMENT AND
EGO-INVOLVEMENT UPON ACCEPTANCE OF A COUNSELING MESSAGE
STEVEN HUGH WILLIAMS
Chairman: Dr. Greg J. Neimeyer Major Department: Psychology
Within the general framework of the Elaboration Likelihood Model of persuasion, the present study investigated the influence of issue involvement, egoinvolvement, and argument quality upon the cognitive processing of a counseling message advocating the use of cognitive restructuring in the treatment of assertiveness problems. Issue involvement was measured by the total number of relevant thoughts governed by subjects of the Thought Listing Technique. It was predicted that there would be greater distinction of argument quality under conditions of high issue involvement than under conditions of low issue involvement.
Ego-involvement was measured by the proportion of total variance accounted for by a single construct vi
concerning assertiveness on a modified version of the Role Repertory Grid. It was predicted that there would be greater distinction of argument quality under high egoinvolvement than under low ego-involvement. In addition, it was predicted that under conditions of high issue involvement, high ego-involvement would exert a greater influence on the weak than strong arguments. Under low issue involvement, high ego-involvement would exert an equal influence on both strong and weak arguments.
Results of the 2 x 2 x 2 Analyses of Variance on
outcome expectation attitude ratings, behavioral intention scores', a general cognitive favorability index and covert rehearsal scores did not support the predictions. The analyses revealed significantly higher attitude ratings for the high issue involved group than the low issue involved group. They also revealed significantly greater cognitive favorability for strong than weak argument quality, and a similar trend for attitude rating and argument quality. No significant results were found for behavioral intention and covert rehearsal measures.
These results suggest that message variables strongly determine acceptance of counseling messages. The results of this study were also discussed in relation to the contributions of the Elaboration Likelihood Model to counseling research, and future directions were suggested.
Persuasion and resistance to change have been issues of theoretical speculation and empirical investigation throughout the past century (Freud, 1894/1959; Frank, 1961; Ellis, 1985). The extent to which a client yields and/or resists a therapist's persuasive messages can be influenced by a wide range of variables. Since Strong's (1968) initial theoretical conceptualization of counseling as an interpersonal influence process, considerable research has been generated concerning interpersonal influence variables in counseling and counselor power in particular (see Corrigan, Dell, Lewis, & Schmidt, 1980, and Heppner & Dixon, 1981, for reviews). Strong proposed a two-stage model of counseling. In the first stage, counselors attempt to increase their perceived power along the dimensions of expertness, attractiveness, and trustworthiness. In the second stage counselors use their influence to bring about client change. Virtually all of the research conducted so far on counseling as an interpersonal influence process has investigated the three
above counselor characteristics. Little attention has been paid to the influence of counseling message variables or client characteristics, particularly to those client characteristics which may be associated with resistance to change.
Research within the area of social psychology on attitude change has followed a similar route of development. Early work on attitude change explored communicator variables (such as prestige, expertise, and credibility) message content, and some focus on audience features. However, with the rapid development of a cognitive social psychology in the 1970s, attitude change research expanded to include the consideration of consistency, dissonance, and attributional processes in the communication recipient (Jones, 1985) With this expansion message variables and recipient characteristics were established as major variables in attitude change research (Heppner and Heesacker, 1982) Future research on the interpersonal influence process in counseling needs to more fully integrate the findings from this research in cognitive social psychology.
One social psychological theory that has relevance for counseling research is the Elaboration Likelihood Model (ELM: Petty and Cacioppo, 1981; Petty and Cacioppo, 1986). Petty and Cacioppo refer to elaboration as the extent to which a person thinks about the issue-relevant arguments
contained in a persuasion message. Elaboration likelihood would be high if a person is motivated and able to evaluate a message, and it would be low if a person is not motivated and/or not able to evaluate a message. People, of course, vary in their willingness and ability to evaluate messages due to individual and situational factors. The ELM began with Petty and Cacioppols attempts to account for the differential persistence of communication-induced attitude change. After reviewing the literature on attitude persistence, they hypothesized that two distinct routes to persuasion existed (Petty and Cacioppo, 1981). The first route, which they called the central route, involves an individual's thoughtful consideration and personal evaluation of the information presented in the communication. The second route, or peripheral route, involves an individual's selection from the persuasion context of some cue (e.g., counselor trustworthiness) that produces change without the individual having to consider or evaluate the information presented. The central and peripheral routes to persuasion correspond generally to high and low elaboration likelihood, respectively. Petty and Cacioppo found the central route to persuasion to be the more enduring of the two. They further postulate that "variables can affect the amount and direction of attitude change by : (A) serving as persuasive arguments, (B) serving as peripheral cues, and/or (C) affecting the extent
or direction of issue and argument elaboration" (Petty and Cacioppo, 1981, p. 132).
From the ELM viewpoint much of the interpersonal
influence research in counseling has focused on peripheral cues such as counselor characteristics. One aim of this study is to expand the focus of interpersonal influence research in counseling to include the study of message and recipient/client variables. In order to provide a context for the study which would approximate to a counseling analogue and provide a framework for the testing of these variables, a life problem that would likely be presented in counseling needed to be identified. The problem chosen concerns social assertiveness. In a needs assessment survey conducted on the Duke University Campus, Talley, Barrow, Fulkerson, and Moore (1983) found that of the 52 needs they asked college students to rate the current importance of the need to assertively stand up for myself was ranked number seven in overall importance. Thus, problems in personal assertion is very likely to be an issue that students will present to college counselors.
In counseling the therapist often presents the client with some task to perform in order to bring about therapeutic change. In the treatment of assertiveness problems, cognitive restructuring has been found to be an effective treatment technique (Kaplan, 1982). If the client acquires a favorable attitude toward the therapeutic
task (in this case cognitive restructuring), he/she would more likely actually perform the therapeutic task outside the counseling room. Two variables which could affect the client's attitude toward cognitive restructuring are the quality of the arguments for its use presented by the counselor and the degree of involvement the client possesses toward changing or retaining his/her present level of assertiveness.
In the process of getting a client to try a particular technique, such as cognitive restructuring, the counselor will present any number of arguments for its use. The quality of these arguments will influence the favorability of the client's attitude toward the therapeutic task. In order to provide a means to empirically test the ELM Petty and Cacioppo (1979b) devised a method for developing "strong" and "weak" messages. This method will be used in this study to provide the means to investigate the influence of two different types of subjects involvement upon their cognitive responses toward the therapeutic task of cognitive restructuring.
Because people would have a high level of involvement in a particular life problem before seeking counseling, it would seem that their elaboration likelihood would generally be high. According to the ELM they would thus be more likely to engage in central route processing rather than peripheral route processing. Petty, Cacioppo, and
Goldman (1981) provided evidence that under conditions of high involvement a thoughtful evaluation of message content is the most important determinant of attitude change. But is the "involvement" of the laboratory comparable to that brought to counseling or is it more complex? This study will attempt to explore the complexity of personal involvement and its influence upon the cognitive processing of a counseling technique.
In the research on susceptibility to influence,
personal involvement or relevance has been defined as the extent to which the issue under consideration is of personal importance (Petty and Cacioppo, 1979). Variables similar to this definition of personal involvement have been alternately referred to as "issue involvement" (Kiesler, Collins, and Miller, 1969), "personal involvement" (Sherif, Kelly, Rodgers, Sarup, and Tittler, 1973), and "ego-involvement" (Rhine and Severance, 1970; Greenwald, 1980). As mentioned earlier, any person seeking counseling is likely to be highly involved with the problem
or issue presented.
Another factor to consider is that people who are
personally involved with an issue have done some amount of prior thinking on the topic before seeking counseling. It is possible that this information has been organized into a schema or guiding principle for the perception of new information (Fiske and Taylor, 1984). Schema-driven
processing tends to be biased toward the maintenance of the guiding schema and may override the objective processing of externally provided communications (Ross, Leper, and Hubbard, 1975; Markus and Sentis, 1982). George A. Kelly's (1955) psychology of personal constructs proposes, as the fundamental postulate of the theory, that "a person's processes are psychologically channelized by the ways in which he anticipates events" (Kelly, 1955, p.46). A person uses his/her personal construct system to interpret and "anticipate" the world. Kelly's organization Collorary suggests that some constructs are more important and have more implications than others. More important or superordinate constructs may serve as schemas for the anticipation of incoming information. They also are more resistant to change (Hinkle, 1966/1965). Thus, if a person has developed a schema or superordinate construct for the dimension of assertiveness, that person's processing of information concerning counseling techniques for treating assertiveness problems may be biased.
Therefore, in this study two types of involvement will be distinguished. The first is issue involvement. This would correspond to the involvement seen in individuals seeking counseling. It concerns the presence or absence of perceived problems in assertiveness. High issue involvement should lead to greater argument elaboration. The second, which will be called ego-involvement, will
concern the presence or absence of a well-developed construct for assertiveness. The literature suggests that such a well-developed construct may bias individuals with assertiveness problems against change. Such a process may be one source of resistance in counseling.
REVIEW OF THE LITERATURE
There are several areas of literature and empirical research relevant to the current study. First, evidence for a cognitive response approach to attitude change and persuasion will be presented. Second, the Elaboration Likelihood Model, a general cognitive model of persuasion, will be presented. Next, the role of cognitive processing will be further explored. This will include both objective and biased modes of processing. At this point the variable of involvement will be introduced, and its contribution to the attitude change literature will be reviewed. The relationships between involvement and cognitive processing will then be explored, and its applications in a counseling context will be considered. Finally, specific hypotheses will be derived from this discussion.
The Cognitive Response Approach To Persuasion
Whenever people receive a persuasive message or any other communication, they will attempt to place the new information within the context of their existing knowledge of the topic. This is the basic postulate of the cognitive response approach presented by Greenwald (1968). In thinking about the message, the individual may consider
material not in the message itself. From their prior knowledge and evaluation of the communication the person may generate cognitions that agree, disagree, or are irrelevant to the persuasive message under consideration. To the extent that the message elicits supportive cognitive responses, attitude change in the advocated direction should be facilitated. To the extent that the message elicits negative cognitive responses, attitude change should be inhibited.
The concept that an individual's cognitive responses are an important mediator of attitudes is not a recent development in psychology. Freud's (1900/1939) method of free association was an early attempt at the measurement of cognitive responses and their role as mediators of attitudes in a clinical context. In the early study of attitude change Hovland (1951) suggested that the accurate and complete recording of an audience's thoughts as they listened to a communication would constitute the best method for investigating the internal process of change.
While the cognitive response approach to persuasion was not formally proposed until 1968, researchers have been concerned with cognitive responses in persuasion since the early research of attitude change. In a classic study investigating active versus passive participation in the persuasion process Lewin (1947) compared individual instruction with group discussion. The goal of both
treatments was to increase the consumption of unusual meats (hearts, kidneys, etc.) during World War II. Groups of housewives either listened to a persuasive lecture or participated in a group discussion on the material presented in the lecture. Survey results showed that only
3 percent of the women in the lecture group served one of the meats, while 32 percent of the women in the discussion group served them. While the results of this study are open to different interpretations, it may be that the selfgenerated arguments of the women in the discussion group were more persuasive than the arguments in the lecture.
In later research on the role of active and passive
participation in persuasion, Janis and King (1954) reported that subjects changed their attitudes more in the advocated direction when instructed to give an informal talk on a counterattitudinal topic than did other subjects who passively listened to the talks. In their experiment Janis and King used three different topics. Their results revealed significantly greater attitude change for two of these topics in comparison to the third. From their data and subject interviews they observed that for the two topics with greater change subjects improvised more in their talks. For the third topic subjects stayed close to the prepared outline made available for each topic. In this case improvisation, or self-generated cognition, may have operated as a critical factor in the greater attitude
change. In another experiment King and Janis (1956) had subjects either read a persuasive communication to themselves, read it into a tape recorder, or read and then give their own improvised version of the message. Those subjects who improvised changed their attitudes significantly more than subjects in either the oral or silent reading conditions. These results are supportive of the notion that one's own cognitive responses on an issue are the most compelling. In a further test of this hypothesis Greenwald and Albert (1968) had subjects improvise five arguments in response to instructions to advocate either career preparatory or general liberal arts undergraduate education. Subjects also read a set of provided arguments supporting the opposite side. Results indicated that subjects' attitudes tended to be in the direction of the content of their own cognition. moreover, subjects recalled significantly more of their own improvised arguments than provided arguments. They also evaluated their own arguments as more original than those
provided to them.
Another area of research suggesting that a person's own responses are important in mediating persuasion is inoculation theory. McGuire (1964) suggests that resistance to persuasion can be created by providing information and arguments supportive of an individual's original attitude or by "inoculating" the individual by
providing weak counterarguments to his or her attitude accompanied by refutations. McGuire states that the inoculation will pose a cognitive threat that will motivate the individual to generate supportive arguments and counterargument refutations for his or her original attitude. This practice should produce greater resistance to subsequent persuasion attempts. The point of interest for the cognitive response approach is, does inoculation lead to self-generated cognitive responses and thus greater resistance to persuasion. McGuire and Papagioris (1961) asked subjects to list thoughts in favor of their initial position one week after being exposed to either supportive defenses or refutational defenses (the inoculation treatment) Subjects who had been exposed to the refutational defenses listed more supportive thoughts than subjects who had been exposed to supportive defenses. In an experimental investigation of the effectiveness of active and passive participation in the defense of one's beliefs McGuire (1964) provided additional evidence for the importance of one's own cognitive responses in mediating attitudinal changes. In the active participation condition he assigned subjects the task of writing either a supportive or refutational defense. In the passive participation condition subjects simply read a defense provided by the investigator. The beliefs used in this study were cultural truisms (e.g., mental illness is not
contagious) which people generally have little practice in defending. According to inoculation theory individuals in the active condition would perform more poorly than those in the passive condition when exposed to immediate counterattitudinal attack. However, they should demonstrate increased resistance to attacks which are delayed. His results conformed to this predicted interaction. Active defenses increased in resistance when the attack occurred one week later, while passive defenses declined in resistance except for the passive refutationaldifferent defense. This passive defense, like the active ones, presumably posed a threat that motivated the subjects to generate additional cognitive responses in support of their beliefs.
Further support for the principle that a person's own cognitive responses are important in producing persuasion is offered by studies investigating the relation of attitudes to underlying beliefs and values. In the 1950s and 1960s widespread racial prejudice caused many psychologists to be concerned with changing prevailing attitudes. Two studies designed to assess the impact upon attitudes of changing underlying beliefs that illustrate the importance of cognition were conducted by Carlson (1956) and Stotland, Katz, and Patchen (1959). Carlson (1956) found that in moderately prejudiced college students more favorable attitudes toward racial integration could be
produced by making the students aware of how desegregation would help in the attainment of some important goals (e.g., greater American prestige in other countries) Stotland et al. (1959) had college students read a message designed to give insight into the psychodynamics of racial prejudice. They also assigned subjects to various manipulations intended to facilitate the internal restructuring of beliefs (e.g., ordering statements into cause-and-effect sequences). Attitude measures showed no immediate reduction in prejudice, but a significant reduction was found in follow-up measures three to four weeks later. These findings suggest that the restructuring of internal beliefs with subsequent attitude change takes some time to occur.
Thus far early research in support of a cognitive
response approach to persuasion has been presented from three areas: 1) active versus passive participation in persuasion, 2) inoculation theory, and 3) attitudeunderlying beliefs linkages. Cognitive response researchers have also provided empirical evidence for the importance of cognition in producing persuasion. They present four types of evidence for the cognitive response approach based on forewarning manipulations, issue-relevant thinking manipulations, argument quality manipulations, and psychophysiological measures (Cacioppo, Petty, & Stoltenberg 1985).
In an experiment utilizing the first approach, Petty and Cacioppo (1977) forewarned some subjects and did not forewarn others that they would be hearing a tape recording prepared by the "Faculty Committee on Academic Affairs" which recommended that seniors be required to pass a comprehensive exam in their major as a prerequisite for graduation. Half of the subjects completed a "thoughtlisting" technique (Cacioppo & Petty, 1981) in the latter
2.5 minutes of the 5 minute interval between the forewarning and the message. Attitude survey results indicated that those subjects who were forewarned, whether or not they listed their thoughts, showed greater resistance to persuasion than those who had not been forewarned. Analyses of the thought-listings showed that the forewarning elicited anticipatory counterargumentation. Forewarned subjects listed twice as many thoughts that were unfavorable than were favorable toward the recommendation. Subjects who had not been forewarned did not generate any issue-relevant thoughts, as they were unaware of the upcoming message's topic.
In the second experiment reported in Petty and Cacioppo (1977), they manipulated issue-relevant thinking. In this study students in an introductory psychology class were told that a guest lecturer, a psychologist from the University Counseling Center, would speak to their class that day, and in return for the visit they would complete a
questionnaire he had brought. Half of the students were forewarned that the psychologist would be advocating that all freshmen and sophomores be required to live in campus dorms, a topic which pretesting showed most students to be against. The remaining students were not forewarned of the topic. Half of all of the students were then asked to list their thoughts just prior to the speech. So far this study replicates the conditions of the forewarning study described above. The other half of the students, however, were asked to list all of their thoughts on the topic of requiring students to live in dorms. Thus, half of the subjects were instructed to engage in issue-relevant thinking whether or not they had been forewarned. Attitude measures revealed that unearned subjects who engaged in issue-relevant thinking demonstrated resistance to persuasion equal to that exhibited by the forewarned groups. It appears that just being instructed to think about a topic can produce cognitive responses which support one's beliefs and, subsequently, lead to greater resistance to persuasion.
In the third line of evidence argument quality has been operationally defined such that "strong" arguments elicit more favorable than unfavorable thoughts about a message, whereas "weak" arguments elicit more unfavorable than favorable thoughts concerning the message. In an illustrative study, Cacioppo and Petty (1985) exposed
subjects either one or three times to a set of strong or weak arguments in support of the recommendation that senior comprehensive exams be instituted. According to their theory, argument quality should differentiate people's attitudes more after three than after one message presentation. This should occur if it is the nature of people's issue-relevant thoughts rather than the number of different arguments they learn. Results indicate that argument-~quality did differentiate people's attitudes more after three exposures than after one. Subjects also recalled more arguments after three presentations than after one, but the amount of the message learned was unrelated to their attitudes.
The fourth source of evidence comes from the use of psychophysiological measures. Cacioppo and Petty (1979), used integrated electromyographic (IEMG) activity as a measure of subjects' silent language processing. In one experiment, IEMG activity was recorded as subjects sat quietly and as they listened to a tape-recorded message. One minute prior to the message, subjects heard one of three announcements about the message. One group only heard that a message would be presented in one minute; a second group was forewarned that they would hear a message recommending more lenient visitation hours be instituted in dormitories (a proattitudinal message according to pretesting) ; and a third group was forewarned that they
would hear a message recommending that stricter visitation hours be instituted (a counterattitudinal message) They found that IEMG activity increased in all conditions during the presentation of the message, which would be expected if IEMG activity indicated the presence of silent language processing. The finding of interest was that IEMG activity had increased following the forewarning of the upcoming counterattitudinal message.
In summary, early research prior to the formal
development of the cognitive response approach to pesuasion supported the importance of people's construing of the information in persuasive messages, rather than their hearing of the material per se, as the most important factor in determining their susceptibility or resistance to persuasion. The methodologies described above have all been used in research on a general framework for understanding persuasion called the Elaboration Likelihood Model (Petty & Cacioppo, 1986), which will be described next.
The Elaboration Likelihood Model
The Elaboration Likelihood Model (ELM) of attitude change is a general response theory, which arose out of Petty and Cacioppo's (1981) attempts to account for the differential persistance of communication-induced attitude change. They hypothesized that two distinct routes to persuasion existed, which they called the central and
peripheral routes. The central route involves a person's thoughtful consideration and evaluation of the information presented in the communication; whereas, the peripheral route involves a person's selection from the persuasion context of some cue (e.g., source trustworthiness) that produces attitude change without the person having to thoughtfully consider or evaluate the information presented. Petty and Cacioppo found the central route to be the more enduring of the two routes.
While people are neither universally thoughtful nor universally mindless in evaluating persuasive messages, they do vary in their motivation and ability to consider carefully the information and arguments that comprise a persuasive attempt. Petty and Cacioppo (1981) refer to elaboration likelihood as the degree to which a person thinks about the issue-relevant arguments contained in a persuasion message. If conditions produce or enhance an individual's motivation and ability to engage in issuerelevant thinking, then the elaboration likelihood is considered to be high. Under conditions of high elaboration likelihood a person allocates significant cognitive resources to considering and critically evaluating the presented message. If an individual's motivation or ability to process a message is reduced, then the elaboration likelihood is also decreased. When it is low, a person is likely to forego a careful consideration
of issue-relevant information and rely upon positive or negative cues in the persuasion context. A simple but reasonable decision rule could also be used in the persuasion context when elaboration likelihood is low.
In order to provide empirical support for distinct
central and peripheral routes to persuasion and the notion of elaboration likelihood, researchers have used the argument quality manipulation under high and low relevance conditions. In high relevance conditions subjects were led to believe that the issue had direct personal consequences for them, while in low relevance condition subjects were led to believe that the issue had few if any personal consequences. The issues and messages used were easy to understand so all subjects had the ability to think about the information presented. Such a design suggests that subjects in the high relevance conditions should follow the central route to persuasion and subjects in the low relevance conditions should follow the peripheral route.
Petty, Cacioppo, and Goldman (1981) conducted such a study in which college students were exposed to a counterattitudinal appeal (favoring senior comprehensive exams). In addition to personal relevance and argument quality, source expertise was also manipulated in this study. Under conditions of high relevance, students were led to believe that the exam policy would begin next year and thus affect them. Under conditions of low relevance,
they were led to believe that the exam policy would begin in 10 years and thus not affect them. For half of the subjects the message was attributed to a Princeton University Professor (high credible source), while for the other half it was attributed to a high school student (low credible source). Source credibility served as a persuasion context cue in this study. Attitude rating results revealed two significant interactions. First, a relevance X message quality interaction showed that argument quality was a more important determinant of persuasion for high than low relevance subjects. Second, a relevance X source credibility interaction suggested that the source cue was a more important determinant for persuasion for low than high relevance subjects. Thus, under conditions of high relevance (high elaboration likelihood), subjects exerted the effort to evaluate the issue-relevant arguments presented. Under condition of low elaboration likelihood, they were persuaded by a context cue and appeared to be unaffected by argument quality.
While in the above study message factors were prepotent under condition of central route processing and source factors were prepotent under conditions of peripheral route processing, the central/peripheral distinction is not one between message and source factors. It refers to whether issue-relevant thinking versus context cues or decision rules leads to attitude change. Different message factors
could have differing effects depending upon the level of elaboration likelihood. For example, the quality of the message arguments should have a greater influence when elaboration likelihood is high, while the actual number of arguments could have a greater influence when elaboration likelihood is low. In the latter situation, the person may employ a decision rule such as, "the more arguments the better."
To test these hypotheses Petty and Cacioppo (1984)
conducted two studies. In the first experiment, college students received a message on the issue of instituting senior comprehensive exams. Personal relevance was manipulated as outlined in the Petty, Cacioppo, and Goldman (1981) study described previously. Subjects received one of four messages in support of the exam proposal: 1) three strong arguments, 2) three weak arguments, 3) nine strong arguments, and 4) nine weak arguments. Attitude rating results revealed that number of arguments was a more important determinant of persuasion under low than high relevance but that quality of the arguments was more important under high than low relevance condition. In the second experiment, the message concerned a proposal to increase tuition, but relevance was manipulated by either stating that the proposal was for the student's own university (high relevance) or for a distant university (low relevance). In this study the message contained
either three strong argument, three weak argument, or six arguments (three strong and three weak). Results indicate that under high relevance conditions three strong arguments elicited more attitude agreement with the proposal than three weak arguments. The agreement for six arguments was greater than that for the three weak arguments but less than that for the three strong arguments. Under low relevance conditions, six arguments produced the most agreement followed by three strong and finally three weak. Thus, in these two studies argument quantity served as a cue under low relevance conditions, but quality of the arguments was more important under high relevance conditions.
This research suggests that two distinct routes to persuasion do exist and differ in their level of elaboration likelihood. The ELM provides a organizing framework for understanding these major cognitive processes underlying persuasion and how variables relate to these processes. Petty and Cacioppo (1986) present the ELM in postulate form as follows:
1. People are motivated to hold correct attitudes
2. Although people want to hold correct attitudes, the amount and nature of issue-relevant elaboration
in which people are willing or able to engage to
evaluate a message vary with individual and
situational factors (p. 128).
3. Variables can affect the amount and direction of attitude change by: (A) serving as persuasive arguments, (B) serving as peripheral cues, and/or
(C) affecting the extent or direction of issue and
argument elaboration (p. 132).
4. Affecting motivation and/or ability to process
a message in a relatively objective manner can do
so by either enhancing or reducing argument
scrutiny (p. 138).
5. As motivation and/or ability to process
arguments is decreased, peripheral cues become
relatively more important determinants of
persuasion. Conversely, as argument scrutiny is
increased, peripheral cues become relatively less
important determinants of persuasion (p. 152).
6. Variables affecting message processing in a
relatively biased manner can produce either a positive (favorable) or negative (unfavorable) motivational and/or ability bias to the issuerelevant thoughts attempted (p. 163).
7. Attitude changes that result mostly from
processing issue-relevant arguments (central route)
will show greater temporal persistence, greater
prediction of behavior, and greater resistance to
counterpersuasion than attitude changes that result
mostly from peripheral cues (p. 175).
While reviewing the accumulated empirical evidence for
the ELM is beyond the scope of this chapter, relevant
research will be presented in the discussions of cognitive
processing and the variable of involvement.
In the ELM Petty and Cacioppo (1981) propose that two
major cognitive routes, the central and the peripheral, are
used by people to process persuasive messages. The
understanding of these cognitive processes involves the
analysis of the structures and mechanisms that comprise
mental activity as well as the impact of various source,
message, and recipient variables upon mental activity.
Petty and Cacioppo (1986) further propose that variables
can affect the cognitive processing of message arguments so
that argument processing proceeds in a relatively objective
or a relatively biased manner. In objective processing, some variable "either motivates or enables subjects to see the strengths of cogent arguments and the flaws in specious ones, or inhibits them from doing do" (p. 136). In biased processing some variable "either motivates or enables subjects to generate a particular kind of thought in response to a message, or inhibits a particular kind of thought" (p. 136). objective processing appears to have much in common with "bottom-up" or "data-driven" processing, since the elaboration is primarily impartial and data focused. On the other hand, biased processing appears to share similarities with "top-down" or "theorydriven" processing, as the elaboration may be guided by prior knowledge, such as a relevant attitude schema (Landman & Manis, 1983; Fiske & Taylor, 1984).
Argument processing lies at the heart of the ELM
(postulate 3) along with two other critical constructs: argument quality and peripheral cues. While peripheral cues are not investigated in this study, argument quality is used as a manipulation to assess the degree to which variables of involvement affect cognitive processing in either a relatively objective or biased manner. In accordance with the ELM, a persuasive message with strong arguments should elicit more agreement when it is considered carefully (high elaboration) than when consideration is low (low elaboration), but a persuasive
message consisting of weak arguments should produce less agreement when consideration is high rather than low. By manipulating argument quality along with another variable, it is possible to assess whether that variable enhances or reduces argument processing in a relatively objective or biased manner. In relatively objective processing if the variable increases processing, subjects' attitudes and thoughts should be more clearly distinguished when the variable is present rather than absent. In the case of biased processing a variable will produce varying levels of effects depending upon the direction of the bias and argument strength. A variable that biases thinking in a positive direction should generally have a greater influence on a strong than a weak message, because it will be more difficult for an individual to generate favorable thoughts to weak than strong arguments. On the other hand, a variable that produces a negative bias should have a greater effect on a weak than a strong message. This should occur because it will generally be more difficult for an individual to generate counterarguments to strong than weak arguments. Research utilizing the argument quality manipulation has provided empirical evidence for the effect of numerous variables upon the extent of cognitive processing in both relatively objective and biased manners. A sampling of this research will be presented next.
In order to provide support for objective cognitive processing ELM researchers have studied the effects of a number of situational and motivational variables upon the extent of argument processing. These have included 1) distraction, 2) repetition, 3) personal responsibility, 4) need for cognition, and 5) personal involvement. Illustrative studies of the first four variables will be presented next, but research on personal involvement will be presented in a later section.
From the ELM viewpoint distraction is seen as affecting cognitive processing in a relatively objective way. Specifically, the ELM predicts a "thought disruption" hypothesis concerning the effects of distraction upon persuasion. The reasoning underlying this hypothesis consists of the following arguments. Distraction from information processing should enhance persuasion for a message consisting of weak arguments. This should occur because without distraction, weak arguments should generally elicit predominantly unfavorable thoughts. However, distraction should disrupt these unfavorable thoughts and thus enhance agreement. For a message consisting of strong arguments favorable thoughts would generally be predominant, and distraction would disrupt these thoughts leading to reduced agreement.
To test this hypothesis Petty, Wells, and Brock (1976) presented college students with either strong or weak arguments for the proposal that their tuition be raised by 20 percent accompanied by a distraction task. Their distraction task consisted of having subjects record the quadrant in which Xs flashed on a screen in front of them. Distraction varied according to the rate of presentation of the Xs with four levels: 1)no distraction, 2) low distraction (15 second intervals), 3) medium distraction (5 second intervals), and 4) high distraction (3 second intervals). After hearing the messages subjects completed attitude measures and listed their thoughts during the message. Attitude rating results revealed a significant argument quality X distraction interaction which is consistent with the thought disruption hypothesis. Under conditions of weak argument quality increasing distraction was associated with more favorable attitudes; however, under conditions of strong argument quality increasing distraction was associated with less favorable attitudes. Analyses of the thought listings showed that high distraction decreased counterargument generation for the weak message but not for the strong one. Moreover, high distraction tended to decrease the amount of favorable thoughts elicited by the strong message but not the weak one. For similar results see Lammers and Becker (1980).
Repetition is another variable which, in moderate
amounts, can increase cognitive processing of information in a relatively objective way. once again, Cacioppo and Petty (1985) used strong and weak arguments for the institution of a senior comprehensive exam; this time, however, half of their college student subjects heard the message once, and half beard the message three times in succession. Students showed greater attitudinal differentiation of strong from weak messages when the message was repeated three times rather than just presented once. Similar results on the impact of repetition on persuasion are reported by Petty and Cacioppo (1984).
Personal responsibility has also been found to enhance objective processing of issue relevant arguments. Social psychological research on the "social loafing" effect (Latane & Darley, 1970; Latane, Williams, & Harkins, 1979) suggests the classical finding of reduced personal responsibility in group performance in comparison to individual performance may result from a loss of motivation in the group setting. For the persuasion context this implies that if personal responsibility is increased, then motivation to engage in significant cognitive processing should also increase. In a test of this hypothesis Petty, Harkins, and Williams (1980) asked college students to provide peer feedback on editorial messages ostensibly written by journalism students. Subjects were led to
believe that they were either one out of a group of ten people who were evaluating the editorials or the only person responsible for the evaluation. Editorials contained either strong, weak, or a mixture of arguments for the institution of comprehensive senior exams. Attitude results revealed that individual evaluators were significantly more favorable toward the strong message and were significantly less favorable toward the weak message than group evaluators. Responsibility did not affect evaluations of the mixed message. Thus, as personal responsibility increased, argument quality became a more important determinant of the evaluation.
Finally, need for cognition, an individual difference variable, refers to the extent to which people need to structure information in reasonable, integrated ways (Cohen, Stotland, & Wolfe, 1955). Cacioppo, Petty, and Morris (1983) exposed high and low need for cognition college students to either strong or weak arguments for a proposal which would raise their tuition. After hearing the message, subjects were asked to evaluate the message and to give their personal opinion about the issue. Results indicate that high need for cognition students produced more polarized evaluations and attitudes for the strong and weak messages. Furthermore, a significantly higher correlation between argument evaluation and personal opinion was found for the high (r = .70) than the low
(r .22) need for cognition group. This would be expected since high need for cognition subjects should be more likely to deduce their attitudes from a careful consideration of the central arguments in the proposal.
it is now apparent that numerous variables can affect people's motivation and/or ability to process persuasive arguments in a relatively objective manner. As noted earlier, however, variables can also affect cognitive processing in a relatively biased way. Biased Processing
Studies of the impact of forewarning and prior
knowledge have provided evidence within the ELM framework for the relatively biased cognitive processing of persuasive messages. Forewarning has been seen to affect an individual's motivation to process arguments in a biased manner, whereas prior knowledge generally affects an individual's ability to engage in biased processing. Forewarning can be further distinguished as either warning of message content or warning of persuasive intent.
In a study of the forewarning of message content, Petty and Cacioppo (1977) manipulated argument quality, forewarning, and also issue-relevant thinking. Details and results of this study were presented earlier in the section on ELM research in support of the cognitive response approach. Briefly, if subjects engage in issue-relevant thinking, they resisted the message whether they were
forewarned or not. These results support the notion that it is not so much the warning but rather the accessing of attitude-supportive cognitions and the subsequent processing of the persuasive message in light of these cognitions which facilitates resistance. It appears that the forewarning served to motivate subjects to begin thinking about their beliefs and the information and thoughts that they already had to support those beliefs. In line with this reasoning, a content forewarning is more effective when there is some time delay between the warning and message to allow thinking (Hass & Grady, 1975; Petty &
Cacioppo, 1977) .
Hass and Grady (1975) also report that a forewarning of persuasive intent is just as effective when it immediately precedes a message as when it comes several minutes before a message. This suggests that an intent warning functions differently from a content warning. They propose that an announcement of an intent to persuade may arouse a psychological state of "reactance" that motivates a person to defend their freedom to hold a particular attitude (cf. Brehm, 1972). Such a state could produce biased processing. In a study designed to explore whether warnings of persuasive intent produced biased processing Petty and Cacioppo (1979b) told students that they would be evaluating radio editorials. Some students were given a warning that the editorials were intended to persuade
college students, while others were just told that the editorials were a journalism class project. Personal relevance was also manipulated in this study by telling the students that certain college regulations would change next year (high relevance) or ten years from now (low relevance). The editorial message used with all subjects consisted of five strong arguments for the institution of senior comprehensive exams as a requirement for graduation. Attitude measure results revealed a main effect for warning and a warning X relevance interaction. The interaction showed that warning significantly decreased attitude agreement only under the high relevance condition. Since the warning reduced agreement even though the arguments used were strong, this suggests that it induced biased rather than objective processing. As this effect was stronger for high than low relevance conditions the warning did not appear to function as a simple rejection cue but rather as a motivator for counterargument and resistance to change.
The common thread that runs through the operation of both types of forewarning effects is the presence of some organized structure of prior knowledge within the recipient of the persuasive attempt. In the social psychological literature a person's organized structure of knowledge is referred to as a schema (Landman & Manis, 1983; Markus & Zajonc, 1985). Markus and Zajonc (1985) state that the
processing of persuasive messages "may be seen as consisting of schema formation or activation, of the integration of input with these schemes, and of the updating or revision of these schema to accommodate new input" (p. 150). While it is possible that schemes would enable a person to process messages more objectively, the literature suggests that when processing is guided by a schema, it tends to be biased toward the perseverance of the existing schema (Ross, Lepper, & Hubbard, 1975; Crocker, Fiske, & Taylor, 1984; Fiske & Taylor, 1984). Thus, if a person has a great deal of prior knowledge, or a well-developed schema, he or she would tend to be able to more effectively counterargue messages opposing their initial attitudes and to cognitively argue for messages supporting their initial attitudes (Lord, Ross, & Lepper, 1979).
It should be noted that a schema as currently defined within the literature is conceptually similar to George A. Kelly's (1955) theorizing on personal constructs. Like the schema theorists he sees people as information processors interested in the organization of knowledge and prediction of future events. His Fundamental Postulate states that "a person's processes are psychologically channelized by the ways in which he anticipate events" (p. 46). In addition, the Organization Corollary states that, "each person characteristically evolves, for his convenience in
anticipating events, a construction system embracing ordinal relationships between constructs" (p. 56). Constructs are an individual's basic interpretations of the world and are bipolar in that they describe the way in which certain things are perceived as alike and different from others (Bannister and Mair, 1968) Constructs which are superordinate in a person's system subsume other constructs, have relatively more implications, and are generally based on more information. Thus, they function similarly to schemas.
In an investigation of the effects of prior knowledge
and schemas upon cognitive processing, Cacioppo, Petty, and Sidera (1982) provide evidence that prior knowledge does influence the processing of persuasive messages. Subjects who characterized themselves using trait adjectives as either "religious" or "legalistic" people were exposed to strong or weak proattitudinal messages, which were either schema-congruent or schema-incongruent. After presentation of the message subjects rated its persuasiveness and listed their thoughts. Results suggested that with proattitudinal messages subjects who received schema-relevant information were more positive about the quality of the communication's arguments and in their listed thoughts. Thus, it appears that self-schemas influence cognitive responses in a biased or top-down fashion.
With counterattitudinal message it would be expected that prior knowledge would increase a person's ability to counterargue the message. To test this hypothesis Wood (1982) divided subjects into high and low prior knowledge groups for the topic of environmental preservation. One to two weeks later subjects read a counterattitudinal message consisting of four arguments against environmental preservation. Attitude measure and thought listing revealed the subjects who had high prior knowledge changed less in the advocated direction than did subjects with low prior knowledge. In addition, high prior knowledge subjects generated more counterarguments and fewer favorable thoughts toward the message.
Wood, Kallgren, and Priesler (1985) extended this finding by adding argument quality and message length manipulations. They divided subjects into high, medium, and low prior knowledge about environmental preservation groups and exposed them to one of four persuasive messages. Two messages consisted of three strong arguments against preservation, and two messages consisted of three weak arguments against preservation. In the message length manipulation two versions of each type of argument were developed. one version contained short concise statements, while the other contained more wordy versions of the same arguments. Both versions were equal in argument strength and ease of comprehension. Attitude measure results reveal
that high prior knowledge subjects were overall more resistant to the different messages than low prior knowledge subjects and that this effect was stronger for weak than strong arguments. This result is consistent with the notion that prior knowledge would enhance a person's ability to counterargue an incongruent message. With weak arguments it would be easier to counterargue than with strong arguments. In addition, low prior knowledge subjects' attitudes were affected by argument length, but high prior knowledge subjects' attitudes were not.
In sum, research on forewarning and prior knowledge has provided support for the view that when a person has a well-developed level of organized knowledge on a topic, message processing will be biased. This will occur, because prior knowledge enables the counterarguing of incongruent messages and the strengthening of congruent ones. The presence of schemes or superordinate constructs are indicators of such well-developed levels of prior knowledge.
Involvement and Persuasion
In the preceding sections numerous variables (e.g.,
need for cognition, forewarning) which may affect cognitive processing in either a relatively objective or relatively biased manner have been discussed. one of the most important variables which can affect the extent and type of cognitive processing employed is personal involvement. In
past psychological research similar variables have been alternately referred to as "issue involvement" (Kiesler, Collins, and Miller, 1969), "personal involvement" (Sherif, Kelly, Rodgers, Sarup, and Tittler, 1973), and "egoinvolvement" (Greenwald, 1980). This has occurred because involvement can be judged in a variety of ways, such as the degree to which an issue has "intrinsic importance" (Sherif and Hovland, 1961) the centrality of the issue to a person's self, and the number of personal consequences of the issue.
Involvement became a central variable in research on persuasion with the advent of social judgment theory (Sherif and Hovland, 1961; Sherif, Sherif, and Nebergall, 1965). Social judgement theory views persuasion as a twostage process. First, one makes a judgment about the position of a persuasive message in relation to one's own position. Second, attitude change or no change occurs depending upon the assimilation or contrast effects evoked by the judged discrepancy between the message and one's own position. Central to the theory are two assumptions: (1) an individual's own stand on an issue serves as an internal anchor for judging messages, and (2) the more "involved" an individual is in the issue, the stronger the anchoring effects of the initial opinion. Greater involvement thus leads to more resistance to persuasion, because people were postulated to hold broader "latitudes of rejection" as
involvement increased. For Sherif and Hovland, involvement seemed to refer to both the intensity with which an attitude is held and the importance of that attitude for the self-identity.
Sheriff and Hovland (1961) described two field studies in support of social judgment theory. For illustrative purposes the "prohibition study" (Hovland, Harvey, and Sheriff, 1957) will be presented here. The prohibition study was conducted in Oklahoma shortly after a close referendum which favored prohibition in the final outcome. To insure that subjects were deeply involved with the issue, dry-stand subjects were recruited from Women's Christian Temperance Union groups, the Salvation Army, and strict denominational colleges. Wet-side subjects were selected from acquaintances of the experimenters. A group of moderate-stand subjects were also included in the study. Subjects were exposed to a 15 minute tape recorded message, which presented either a wet, moderately wet, or dry stand on prohibition. Dependent measures were obtained on the following: (1) the communication's estimated position, (2) subject's reactions to the communication in terms of fairness and impartiality, and (3) subject's preferred position and latitudes of acceptance and rejection.
In support of social judgment theory, Hovland, Harvey, and Sherif (1957) found that for the majority of the subjects with initially extreme stands, their latitudes of
rejection were broader than their latitudes of acceptance. Furthermore, for the majority of the moderate subjects, the width of their latitudes of acceptance exceeded the width of their latitudes of rejection. Thus, it appears that the two groups of subjects use very different reference scales as internal anchors.
Results for the dependent measures reflect the functioning of these different internal anchors. In judging the speaker's position, extremely dry subjects saw the message as advocating a much wetter position than it did, while extremely wet subjects judged it as advocating a drier position than it did. Moderate subjects were more accurate in judging it as advocating a slightly wet position. Moreover, the closer the communication was to the respondent's own position, the more likely that subject was to favorably evaluate the message in terms of fairness and impartiality. Attitude change results indicated that in comparison to moderate subjects, approximately twice as many extreme subjects remained unchanged by the message. Greater involvement thus leads to more resistance to persuasion.
Other early research has also found that increasing involvement was associated with resistance to persuasion. Miller (1965) found that high issue involvement consistently decreased the persuasive effect of a discrepant communication on attitude measures, but
involvement level did not affect subjects' latitudes of acceptance. According to social judgment theory, high involvement should reduce latitudes of acceptance. Eagly and Manis (1966) also report that highly involved subjects react more negatively toward persuasive messages that
contradict their beliefs.
other researchers, however, have not found results consistent with the notion that higher levels of involvement increase resistance to persuasion. Zimbardo (1960) defined involvement in terms of an individual's concern with the social consequences of his or her response in a given situation rather than the intrinsic importance of an issue for the individual. He found greater attitude change with highly involved subjects than with less involved subjects. Freedman (1964) also operationalized involvement in terms of concern about a response. In addition, he exposed subjects to messages which were slightly, moderately, or extremely discrepant from their initial positions. His results indicated that under low involvement, there was more change with greater discrepancy; but under high involvement, their relationship was nonmonotonic with maximum change occurring at moderate discrepancy. Moreover, under moderate discrepancy, both high and low involved subjects changed by a similar amount. Similar results were reported by Rhine and Sev erance (1970).
Contradictory results have also been reported by
researchers who have manipulated forewarning along with involvement. Apsler and Sears (1968) either forewarned or did not forewarn subjects that they would read a proposal calling for the replacement of professors by supervised teaching assistants. Involvement was manipulated by informing half of the subjects that the proposal would go into effect ten years from now. Results indicated a significant interaction with forewarning inhibiting attitude change under high personal involvement but facilitating change under low personal involvement. Dean, Austin, and Watts (1971) found, however, that forewarning inhibited attitude change for both high and low involvement issues and its effect was even more pronounced for the low involvement issue. On the other hand, Petty and Cacioppo (1979b) found the inhibiting effect of forewarning to be greater under high than low involvement conditions. In these latter two studies the forewarning given was one of persuasive intent not content, but all three studies manipulated involvement in similar ways.
one interesting finding in the Petty and Cacioppo
(1979b) study was that when no forewarning was given, high involved subjects tended to show more attitude change than low involved subjects. Similar findings were also reported by Apsler and Sears (1968) and Eagly (1967). Eagly (1967) gave subjects favorable or unfavorable discrepant
information about either themselves (high involvement) or another person (low involvement). She found that when favorable information was provided, high involvement subjects showed more attitude change than low involvement subjects. When unfavorable information was provided, the reverse was true.
In sum, research has produced contradictory findings on the effects of level of involvement. To some degree this has resulted from differing definitions of involvement, but the contrary results have persisted even when involvement was similarly defined. In hopes of reconciling these opposing results, involvement will now be examined in relation to cognitive processing.
Involvement and Cognitive Processing
From the ELM perspective Petty and Cacioppo (1979a)
propose that increasing involvement with an issue increases one's motivation to cognitively process issue-relevant information and can lead to either increased or decreased persuasion. The research presented in the previous section clearly demonstrates that high involvement can lead to either enhanced or inhibited persuasion. Petty and Cacioppo further postulate that whether persuasion is enhanced or inhibited depends upon the individuals initial opinion. If the persuasive message is contradictory to the individual's initial attitudes, it is likely that the individual is motivated and able to generate
counterarguments to the message. Therefore, as issuerelevant thinking increases, counterargumentation and resistance to persuasion also increases. On the other hand, if the persuasive message is congruent with the individual's initial attitude, it is likely that the individual is motivated and able to generate favorable cognitions. As involvement and issue-relevant thinking increases in this situation, more favorable thoughts might be generated, and increased persuasion would result.
In order to test these hypotheses Petty and Cacioppo
(1979a) conducted two experiments. In the first experiment subjects were exposed to either a proattitudinal or counterattitudinal message concerning coed visitation hours. Issue involvement was manipulated by stating that the changes in visitation hours would go into effect at their own university (high involvement) or at a distant university (low involvement). Dependent measures consisted of attitude scales and thought listings. Results on the attitude index revealed a significant interaction between involvement and type of message. Analysis of the interaction revealed that increased involvement increased agreement with the proattitudinal message but decreased agreement with the counterattitudinal message. The thought listing technique revealed similar results. Under high involvement conditions subjects produced more positive cognitions and fewer counterarguments to the proattitudinal
communication than to the counterattitudinal one. Message direction showed no significant effects on cognitive responses under low involvement conditions.
In the second experiment Petty and Cacioppo (1979a) again varied involvement, but they presented only a counterattitudinal message with either strong or weak arguments. The message in this experiment concerned the institution of senior comprehensive exams. The ELM predicts that increased involvement will lead to more issue-relevant thinking about the arguments presented in the message. For the message with strong arguments, which are difficult to counterargue, increased involvement should be associated with more persuasion. For the message with weak and easy-to-counterargue arguments, increased involvement will be associated with decreased persuasion. While high involvement may initially motivate an individual to reject a counterattitudinal message, the enhanced cognitive processing of the message should enable the virtues and flaws of the arguments to be objectively recognized. Results of the cognitive response measures supported these hypotheses. Under high involvement conditions subjects produced more favorable cognitions and fewer counterarguments to the strong than to the weak arguments. Argument quality produced no significant effects on cognitive responses under low involvement conditions. Furthermore, high involvement increased the
amount of counterarguments generated to weak arguments and increased the amount of favorable thoughts generated to strong arguments. These results show that increased involvement increases the distinction of strong and weak arguments.
This interaction of personal involvement and argument quality has been replicated several times (Petty, Cacioppo, and Goldman, 1981; Petty, Cacioppo, and Heesacker, 1981; Petty and Cacioppo, 1984) For example, in the context of examining the effects of source expertise under conditions of high and low involvement, Petty, Cacioppo, and Goldman (1981) found that under high conditions objective cognitive processing increased distinction of argument quality and that argument quality was the primary determinant of persuasion. under low involvement conditions, the peripheral cue of source expertise exerted a greater effect than argument quality.
Further support for the notion that increasing
involvement heightens message scrutiny was provided by studies in which involvement was manipulated along with source and message cues. Results of Chaiken's (1980) work indicate that high issue involvement subjects tend to use a systematic information processing strategy, whereas low involvement subjects tend to use a heuristic processing strategy. Subjects who use a systematic strategy focus on
the content of persuasive messages and carefully evaluate the content in a relatively objective manner.
Although the research cited above suggests that people become more likely to evaluate carefully and objectively the issue-relevant arguments in a persuasive message as personal involvement increases, circumstances may occur in which cognitive processing becomes biased as involvement increases. As suggested in the section on biased processing, the presence of a well-organized structure of prior knowledge may lead to biased processing. An individual who is involved with an issue has done some amount of prior thinking about the pool of issue-relevant arguments. Some individuals may have done a great deal of prior thinking and have a well-developed construct system or schema for the issue in question. Such individuals may have a greater ability to counterargue persuasive messages, a greater store of supportive arguments for their own beliefs, and/or little motivation to consider yet another persuasive appeal.
Ostrom and Brock (1968) suggest that when an issue is intimately connected with an individual's central values, its personal involvement may be intense enough to generate biased processing. Greenwald (1980) proposes that the ego is characterized by a cognitive bias of conservatism or resistance to change that serves to protect the self's organization of knowledge. If the ego is threatened, it
will respond with cognitions aimed at preserving the status quo. Thus, extremely high levels of involvement, or personal importance for a person's central aspects of identity, may also lead to biased processing. Biased processing may then consist of either a negative bias for counterattitudinal messages or a positive bias for proattitudinal messages.
Involvement Level in Counseling
Clear cases of high or low involvement probably do not exist in the therapy room. People generally seek help with issues in which they are involved to some significant degree. Prior to pursuing therapy, they will have given thought to how to deal with the issue and perhaps will have tried many self-generated change attempts. They may even have sought the advice of friends, family, and others.
Strong (1971) suggested that different types of
counseling may elicit different levels of involvement from clients, but it seems clear that clients may also bring different levels of involvement with them to counseling. Stoltenberg and McNeill (1984) showed that college students who were undecided about a career regarded the issue of career exploration as more personally involving than students who had already chosen a career. They also found undecided students agreed more with a message advocating a career exploration course than did decided students.
This finding suggests that more involved clients may see counseling as more beneficial and may attend more to the counseling process. Thus, clients with a high level of involvement should adhere more to central route cognitive processing of therapeutic messages, whereas clients with a low level of involvement may depend upon the peripheral route. Moreover, as central route processing appears to lead to more enduring attitude change (Petty and Cacioppo, 1981), it behooves the therapist to motivate clients to evaluate carefully the pros and cons of their present behaviors and to weigh the costs and benefits of change and various means of change.
An example of a highly involved client perhaps would
be the articulate, well-educated, and young business person seeking counseling concerning improving his or her relationships with co-workers and superiors. One would expect this person to consider carefully the issues and suggestions in counseling. The low involved client may be represented by the person forced to attend an alcohol therapy group due to a DWI conviction. He or she may rather be at home or even just spend the weekend in jail than attend therapy weekly for 3 months. Such an individual may elaborate very little on the issues raised in counseling but may be influenced by persuasion cues (e.g., counselor social attractiveness).
On the other hand, client factors may also serve to influence cognitive processing in a biased manner. An individual may be so extremely involved in an issue that he or she avoids thinking about it altogether (e.g., the denial of a person with alcohol dependence). Processing may also become biased in the service of one's own ego or for self-protection (e.g., Greenwald, 1980). If a person organizes his or her identity around being a certain type of individual (e.g., an easy to get along with, always ready to please, nonassertive person), he or she would probably be less objective in considering therapeutic messages designed to change that behavior. This would especially be the case if this identity had provided the individual with rewarding secondary gains in the past. Such a client may be well prepared to generate cognition in support of his or her present attitudes and behaviors and actively counterargue any therapeutic communications aimed at changing that basic identity. Thus, this biased processing would lead to greater resistance to change.
As outlined in the preceding sections of this review, cognitive processing of persuasive messages may follow either a central information-processing route (a careful consideration of the message content) or a peripheral information-processing route (a cue or decision rule basis for persuasion). Within the more central route, cognitive
processing may be relatively objective or relatively biased. As a result of this, individuals with different levels of involvement will engage in different types of cognitive processing.
The present study will screen for four groups of
subjects: (1) high ego and high issue involved, (2) high ego and low issue involved, (3) low ego and high issue involved, and (4) low ego and low issue involved. Subjects will be asked to consider a message advocating the use of the cognitive restructuring counseling technique for the treatment of assertiveness problems. The topic of assertiveness was chosen for this investigation because of its relevance for a college population. Using measures of CFCR, outcome expectation, behavioral intention, and spontaneous use of cognitive restructuring to assess the subjects' cognitive responses to strong and weak arguments advocating the use of the technique, the following predictions are made:
1. Replicating the findings of Petty and Cacioppo (1979b), there will be a significant interaction
between issue involvement and argument quality.
This effect will be such that there will be greater distinction between strong and weak arguments under
conditions of high issue involvement than under
conditions of low issue involvement.
2. A significant interaction is also predicted
between ego-involvement and argument quality with a
greater distinction between strong and weak messages occurring under high than low egoinvolvement.
3. A biasing effect is also predicted to be
reflected in a significant interaction between
issue involvement and ego-involvement. Under conditions of high issue involvement, high ego-involvement will exert a greater influence on the weak than the strong messages. Under low involvement, high ego-involvement will exert an equal influence on both strong and weak messages.
This study was designed to investigate the effects of ego-involvement and issue involvement upon the cognitive processing of strong and weak arguments for the use of cognitive restructuring as a treatment technique for assertiveness. The sample for the study consisted of 120 students drawn from the University of Florida introductory psychology subject pool and an undergraduate sociology class, Marriage and Family. Cognitive responses and acceptance of arguments was assessed across four groups: 1) high ego and high issue involvement, 2) high ego and low issue involvement, 3) low ego and high issue involvement, and 4) low ego and low issue involvement. Half of each group was exposed to strong arguments and half to weak arguments in favor of using cognitive restructuring. Levels of ego-involvement were assessed by using the Relationships Grid (see Appendix A), a form of the dyad grid (Ryle and Lunghi, 1970). Levels of issue involvement were measured through the use of the Simple Rathus Assertiveness Schedule (McCormick, 1984). The preceding instruments were administered to one General Psychology class and one Marriage and Family class.
Cognitive responses to strong and weak arguments for cognitive restructuring were assessed through the use of the Thought-Listing Technique (Cacioppo and Petty, 1981). Several seven-point Likert-type scales on which subjects rated their attitudes towa rd cognitive restructuring and their intentions to use it were also used to measure their acceptance or rejection of the arguments. An additional Thought-Listing Technique was used to assess the spontaneous use of cognitive restructuring by subjects during a covert rehearsal technique.
This chapter is divided into several sections. In the first section, the research design and variables of interest are discussed. In the second section, the population of interest and the study's sample and sample selection and recruitment procedures are discussed. In the third section, the instruments used in this study are presented. In the fourth section, the experimental procedure and data collection is described. In the fifth section, the method of data analyses are presented. Methodological limitations of the study and a summary of the methodology are presented in the final two sections.
This study involved a 2 x 2 x 2 (Ego-involvement: high or low X Issue involvement: high or low X Intervention Quality: strong or weak arguments) factorial design. Subjects were crossed on the first two factors according to
a median split for issue involvement and upper and lower divisions of a trichotomization of ego-involvement and then randomly assigned to either strong or weak intervention quality. Ego-involvement was operationalized in terms of a summed relationship score for the construct, "Assertive-Nonassertive" on the Relationships Grid. The summed relationship score (Bannister, 1965) is a measure of the total variance within a construct system accounted for by one construct. The greater the total variance for a construct the greater the implications and importance it holds for a person. The variable, ego-involvement, had two levels: high and low. The high ego-involvement level refered to students who scored in the uppermost division of a trichotomization of the sample's summed relationship scores for the construct, Assertive--Nonassertive. The low ego-involvement level refered to students who scored in the lowermost division for that score.
The second factor, issue involvement, was
operationalized by the use of an objective inventory assessing personal assertiveness, the Simple Rathus Assertiveness Schedule (McCormick, 1984). Subjects who scored above the median for the pool of subjects on the objective inventory (high trait assertiveness) were considered to have low issue involvement. As scores below the median on the objective inventory would indicate above
average concern with assertion, subjects who scored in the below the median range comprised the high issue involvement condition.
The third factor, argument quality, consisted of two levels, strong arguments and weak arguments. Strong and weak arguments for the use and effectiveness of the cognitive restructuring technique were developed by Greg Neimeyer and his research team (Unpublished data, 1986). Following the procedure outlined by Petty and Cacioppo (1981) for developing arguments for a topic, they generated strong and weak arguments for the use of cognitive restructuring in the treatment of eating disorders. In the present study these arguments have been adopted to the use of cognitive restructuring in the treatment of assertiveness.
Cognitive Responses to Arguments
The variables of interest in the present study
consisted of the subjects' cognitive responses to arguments for cognitive restructuring. Cognitive responses were operationalized in terms of cognitive favorability toward cognitive restructuring, outcome expectation and behavioral intention ratings, and judges' ratings of spontaneous use of cognitive restructuring by subjects during a covert rehearsal condition.
Cognitive favorability toward cognitive restructuring (CFCR) was derived from the Thought-Listing Technique
(TLT). The TLT is used to identify a person's subjective thoughts or reactions to a topic. In the TLT subjects were instructed to list the thoughts they had while listening to a taped message on cognitive restructuring (CR). Subsequently, they also rated whether their thoughts were in favor of using CR, opposed to using CR, or neither in favor of nor opposed to it. A general index of CFCR was calculated for each subject by subtracting the number of unfavorable thoughts from the number of favorable thoughts listed and then dividing by the total number of relevant thoughts yielding a ratio score (see Cacioppo and Petty, 1981). Higher CFCR scores indicated more positive or favorable thoughts toward CR, while lower or negative CFCR scores indicated less favorable thoughts or opposition to the use of the CR technique. It was expected that under conditions of high issue involvement individuals with high ego-involvement would have lower CFCR than individuals with low ego-involvement. It was also expected that under conditions of low issue involvement individuals with high ego-involvement would have greater CFCR than individuals with low ego-involvement.
Cognitive responses toward the CR technique were also operationalized in terms of outcome expectations and behavioral intention ratings. The outcome expectation asked subjects to indicate, "To what extent do you think the cognitive restructuring technique would be beneficial
to you, personally?" As a message of behavioral intention, subjects were asked to indicate "How likely are you to use the cognitive restructuring technique in the future in situations requiring assertion"?7 Subjects responded to those items as well as a number of manipulation check and ancillary items on a seven point Likert-type scale. It was expected that analysis of outcome expectation ratings would parallel the predictions for CFCR. In addition, under conditions of high issue involvement it was expected that subjects with low ego-involvement would have greater intentions to use the CR technique than subjects with high ego-involvement.
Finally, the spontaneous use of cognitive
restructuring during a covert rehearsal condition (see Appendix E) was also used to operationalize subjects' cognitive responses to the CR arguments. It was reasoned that subjects with greater favorability toward cognitive restructuring and greater behavioral intention to use it will generate more CR thoughts during a covert rehearsal of a situation requiring assertion.
Description of the Sample
The population of interest in this study included all undergraduate students attending the University of Florida, Gainesville, Florida.
The sample in this study was drawn from the University of Florida introductory psychology subject pool and from an
undergraduate sociology course on marriage and family at the University of Florida. During the first week of spring semester 1987, students in one section of Psychology 2013 (General Psychology) and Sociology 2430 (Marriage and Family) were asked to complete a pretest consisting of identifying information, the Relationships Grid, and the Personal Problems Inventory (see Appendices A & B).
on the basis of the scores from these two pretests, subjects were divided into four separate groups: 1) high ego and high issue involvement, 2) high ego and low issue involvement, 3) low ego and high issue involvement, and 4) low ego and low issue involvement. In order to assign students to these four groups the following procedure was followed. First, all students will be rank ordered according to their summed relationship scores for the construct, "Assertive--Nonassertive" on the Relationships Grid, and a trichotomization was performed on the basis of these scores. The uppermost division constituted the high ego-involvement group; the lowermost division constituted the low ego-involvement group. Second, all students were rank ordered according to their scores on the Simple Rathus Assertiveness Schedule, which is entitled the Personal Problem Inventory in this study. On the basis of these scores a median split was performed, and the upper half of the split consisted of the low involvement group, while the lower half consisted of a high issue involvement group.
Third, students were assigned to the appropriate group according to their high and/or low classifications on ego and issue involvement. This procedure resulted in the inclusion of 104 subjects (N=40, F=64). The percentage of males within each group is presented in Table 1. Mean egoinvolvement and issue involvement scores and standard deviations for all groups in this study are also listed in Table 1.
Six instruments were used in this study: 1) The
Relationships Grid, 2) Personal Problems Inventory (Simple Rathus Assertiveness Schedule), 3) Assertion Survey, 4) Thought Listing for Message, 5) Covert Rehearsal Technique, and 6) Cognitive Assessment Questionnaire. Relationships Grid
The Relationships Grid (RG; see Appendix A) is a form of the dyad grid (Ryle & Lunghi, 1970). The RG is an adaptation of the original repertory grid technique developed by G.A. Kelly (1955) to elicit and measure personal construct systems. "A grid may be defined as any form of sorting task which allows for the assessment of relationships between constructs and which yields these primary data in matrix form" (Bannister and Mair, 1968, p. 136). The basic components of a grid are elements and constructs. In the RG, like*other dyad grids, the elements used consist of relationships (e.g. my relationship with my
u IV CN r-I LO
m co Ln
4-4 u 4 r-i C4 r-i
ON 0 r-4
W v 0 co m
> r-4 C14 r--4 CN
w r--l r-4
44 > Lf)
ul 4-) u
W 04 En
w U) 00 CN ON C)
u :3 C) r- C14 co
4.) C C;
co r- r-4 1-4
4 4J r--4 r-4
E 1w 00 00 N
0) O)p C14 m Ln m
c > 0 CN lw co
0 c cn CN CN
.H 1-4 En
4-J I a)
(a 0 r-4 4J
.,q m Q C
> W m (1)
> 3: al 3:
> r-I 0 0
-4 41 PQ
a) (n >
spouse, my relationship with my father) instead of individuals as in the original repertory grid. Dyad grids use relationships as elements in order to gain information about a person's interpersonal behavior across a range of important relationships in his/her life. Dyad grids have been used in the past in clinical work with married couples (Ryle and Breen, 1972a and 1972b; Ryle and Lipshitz, 1975 and 1976) and in individual therapy to provide hypotheses for the conduct of therapy (Ryle and Lunghi, 1971; Ryle, 1979a; Ryle 1981) and to assess change (Ryle, 1979b; Ryle, 1980) Elements and constructs for this study's RG were adapted from Ryle's Relationships Grid (Ryle, 1985) and modified for the purposes of this study. For a complete listing of elements and constructs see Appendix A. The elements of interest consisted of the subjects' relationships with several important figures in their lives and how they saw themselves in those relationships. Constructs described interpersonal behavior and were presented in the classical bipolar form.
The RG, a ten by ten grid, involves having subjects rate how they see themselves in their relationships with ten interpersonal figures (e.g., mother, sibling, roommate) along ten provided constructs using a seven-point Likerttype scale. For example, a subject will be asked to rate his/her relationship with mother along the construct, Assertive +3 +2 +1 0 -1 -2 -3 Nonassertive. The subject
may believe him/herself to be moderately nonassertive in that relationship and place a -2 in the appropriate square in the grid. The left pole or side of each construct will be assigned a positive valence, while the right pole will be assigned a negative valence. The Relationships Grid and complete instructions are provided in Appendix A. Estimated time for completion of the RG is fifteen minutes.
Numerous measures can be derived from a grid, but the measure of interest for this study is the summed relationship score for the construct, "AssertiveNonassertive." The summed relationship score (Bannister, 1965) is a measure of the total variance within a construct system accounted for by a single construct. The greater the total variance accounted for by one construct; the more closely it is related to all other constructs. A construct with greater variance and more implications for the rest of an individual's construct system would be more superordinate and hold more importance for that individual than a construct accounting for less variance in the system. In this study subjects with higher summed relationship scores for whom the construct, AssertiveNonassertive, accounted for a greater amount of variance in their construct systems were designated as highly involved. Subjects for whom the assertiveness dimension accounted for less variance in the system were designated as low ego involved.
The calculation of variance scores is described by
Bannister (1965). Pearson product-moment correlations are first calculated between all constructs. Second, each correlation is squared, multiplied by 100, and the original sign retained. These scores are called relationship scores. The relationship scores for any one construct may be summed to give the total variance for that construct. In this study the relationship scores for the construct, "Assertive-Nonassertive," will be summed to give the total variance accounted for by that construct. If this score is then divided by the total variance in the system across all constructs, then it yields an index of the proportion of variance accounted for by the single construct AssertiveNon Assertive. A trichotomization was performed along the distribution of these scores. The uppermost division was designated as high ego-involvement and the lowermost as low ego-involvement.
Investigations into the psychometric properties of the Rep Grid are complicated by the varieties of grids and scorings in use. While the reliability and validity of summed relationship scores is not reported in the literature, psychometric properties of total variance scores (Bannister, 1960) are reported. Bannister (1962) reports an immediate retest reliability of 0.35 in a sample of 30 normal subjects. Honess (1977) reports the same reliability correlation in a rank order grid but a
correlation of 0.62 for a bi-polar implications grid. While reliability for total variance scores is generally low, this may be due to its sensitivity to construct system change as the individual and/or situation changes. Total variance scores have been shown to effectively discriminate thought-disordered schizophrenics from normals and other psychiatric groups (Bannister, 1962; Bannister and
Rather than using the total variance score, this study uses the proportion of variance accounted for by the single relevant construct. The psychometric properties of this score are not yet established. It seems reasonable to say that despite the instability of the overall variance in the system, the proportion of variance accounted for by any one construct might remain relatively constant. To test this assertion 30 students from an undergraduate psychology course were administered the RG twice with a 7 day interval between administrations. Proportion scores for the relevant construct were used to calculate a Pearson product-moment correlation. This test-retest correlation was .66, (p <.0001).
Personal Problems Inventory
The Personal Problems Inventory (PPI; see Appendix B) was used to assess issue involvement concerning assertion. It consisted of the Simple Rathus Assertiveness Schedule (SRAS) (McCormick, 1984). The SRAS is an easier-to-read,
revised version of the Rathus Assertiveness Schedule (Rathus, 1973). The SRAS is a 30-item self-report assertiveness inventory, which provides a global rating of perceived trait assertiveness. Respondents mark each item in terms of how characteristic the behavior is of the individual from 5 (very much like me) to 1 (very unlike me) A total score is obtained by summing item scores after correcting for reversed scoring weights.
In an investigation of the test-retest reliability of the SRAS, thirty undergraduate psychology students were administered the SRAS twice with a 7 day interval between adminstrations. A Pearson product-moment correlation yielded a coefficient of .91 (p <.0001).
McCormick (1984) provides evidence for a satisfactory degree of equivalence between the Simple Rathus Assertiveness Schedule (SRAS) and the Rathus Assertiveness Schedule (RAS) He reports a mean interim correlation of
0.79 between the two versions of the test with all correlations reaching statistical significance. The correlation between the total scores of the two tests was 0.94. He also reports a correlation of 0.90 between total odd and even item scores for both versions. Finally, in his sample of 116 undergraduate students he found an almost identical distribution of scores on both tests. Mean scores and standard deviations for the two versions were M=96.12, SD=23.85 (SRAS) and M=96.29, SD=24.94 (RAS).
Psychometric research on the RAS suggests that it has moderately high stability (test-retest reliability) and moderate to high homogeneity. Rathus (1973) presents a five-week test-retest reliability of 0.78 and a split-half Pearson product moment correlation of 0.77 between total odd and total even item scores. Vaal (1975) presents similar correlations of 0.76 for test-retest reliability over an eight-week period and 0.77 for split-half internal consistency. With college students seeking assertion training Heimberg and Harrison (1980) obtained over an eleven to fifteen day period a test-retest reliability correlation of 0.80 and a split-half internal consistency correlation also of 0.80. Other reports of internal consistency correlations have ranged from 0.69 to 0.86 (Mann and Flowers, 1978; Quillan, Besing, and Dinning, 1977; Futch, Scheirer, and Lisman, 1982).
A base of validity information has also been
established for the RAS. When comparing RAS scores with peer ratings on a seventeen-item semantic differential Rathus (1973) found that total RAS scores correlated positively with each of the items comprising the peerrating scale's assertiveness factor. Several studies have tested Wolpe's (1973) argument of an inverse relationship between trait anxiety and assertiveness; conformation of Wolpe's argument would provide indirect support for the construct validity of the RAS. Orenstein, Orenstein, and
Carr (1975) found RAS scores to be significantly correlated (range -0.60 to -0.75) with interpersonal fears as measured by the social factors from the Wolpe-Lang (1964) Fear Survey. Other studies by Morgan (1974) and Hollandsworth (1976) have found lower correlations between RAS scores and a social fear survey. In Morgan's study of college students the correlations ranged from 0.172 to -0.239 and in Hollandsworth's study a correlation of 0.436 was found. The combined data from these studies does appear to provide some support for the construct validity of the RAS.
Concurrent validity of the RAS has also been
established in several studies. Rathus (1973) reports a correlation of 0.705 between RAS scores and observer ratings of verbal behavior in response to five questions asking for assertive behavior. Frankel (1977) reports significant correlations between the RAS and the Conflict Resolution Inventory and between the RAS and the Assertion Inventory. MacDonald (1975/1974) also supports the concurrent validity of the RAS in her findings of a moderate relationship between RAS scores and behavioral measures of assertiveness. In addition, the RAS has been used in treatment outcome studies and has been shown to be a sensitive index of pretreatment to post treatment change (Rathus, 1973; Blanchard, Turner, Eschette, and Coury, 1977).
The Assertion Survey (AS; see Appendix C) is a
questionnaire developed by the researcher to assess outcome expectations. It consisted of eight 7-point Likert-type items. Outcome expectation was assessed by the item, "To what extent do you think the cognitive restructuring technique would be beneficial to you, personally?" Behavioral intention was assessed by the item, "How likely are you to use the cognitive restructuring technique in the future in situations requiring assertion?" As a check on the argument quality manipulation, subjects answered a seven-point Likert item asking them to rate the quality of the arguments in the intervention: very poor arguments (1) to very good arguments (7). In addition, as a general check on the issue involvement manipulation, subjects answered another Likert item asking them the extent to which the message has implications for them personally: not at all relevant to me (1) to very relevant to me (7). Ancillary items assessed other aspects of the message including voice quality, speaker qualification, and ratio of delivery.
Thought Listing Techniques
The Thought Listing Techniques used in this study were variations of the "thought-listing procedure" developed by Brock (1967) and Greenwald (1968). The procedure has been refined to become a suitable self-report technique for
obtaining a written listing of an individual's thoughts on a topic and a general index of CFCR toward the topic (Cacioppo and Petty, 1981). In the present study Thought Listing was employed under two conditions: 1) after listening to the taped arguments for cognitive restructuring and completing the Assertion Inventory and 2) as part of the covert rehearsal of a personal assertion situation. Under the first condition the Thought Listing Technique (see Appendix D) was used to compute a CFCR index, and under the second condition (see Appendix E) it was used to tally the spontaneous use of cognitive restructuring. Under the first condition subjects listed their thoughts concerning the arguments for CR according to directions similar to those of Petty and Cacioppo (1977). In addition, they rated their thoughts along the dimension of favorableness toward cognitive restructuring. The cognitive favorability index was calculated for each subject by subtracting the number of unfavorable thoughts from the number of favorable thoughts listed and then dividing by the total number of relevant thoughts (Cacioppo and Petty, 1981). Petty et al. (1976) report correlations between subjects' own rating and independent judges ratings of .82 for favorable thoughts and .79 for counterarguments. Under the second condition, two undergraduate psychology major students trained by the researcher and unaware of the hypotheses of the study and the involvement group
classifications of the subjects judged the thoughts listed for frequency of spontaneous use of cognitive restructuring. This consisted of the post-coding of each subject's listed thoughts into one of five categories: 1) self-instructional statement, 2) self-assuring statement, 3) self-doubting statement, 4) presentation content statement, and 5) other statement. Inter-judge reliability Pearson product-moment correlations were .81 for selfinstructional statements, .85 for self-assuring statements, .92 for self-doubting statements, .72 for presentation content, and .34 for other statements.
Support for the Thought Listing Technique (TLT) as a reliable measure can be found in a study by Cullen (1969/1968). She compared the split-half reliability and test-retest reliability of the TEIT with Likert attitude and Thurstone attitude scales on two topics, birth control and segregation. ,She found average split-half reliabilities of
0.78 for Thought-Listing, 0.83 for Likert scales, and 0.55 for Thurstone scales. The average test-retest reliabilities were 0.64 for the TEIT, 0.83 for Likert scales, and 0.53 for Thurstone scales.
In indirect support for the construct validity of the TLT, Petty, Harkins, and Williams (1980) found that the implied or real presence of others working on the same cognitive task decreased the thoughts listed by an individual in that implied or real presence of others when
compared to the individual working alone. This finding corresponds to the well-known "social loafing" argument (Latane and Dat ley, 1970).
Cognitive Assessment Questionnaire
The Cognitive Assessment Questionnaire (CAQ; see
Appendix F) was used to assess subjects' tendency to engage in and enjoy effortful cognitive endeavors. It consisted of the short form of the Need for Cognition Scale (NCS) (Cacioppo, Petty, and Kao, 1984) The short NCS is an 18item self-report inventory which provides a global assessment of need for cognition. Cacioppo, Petty, and Kao (1984) report a significant correlation (r = .95, p <,001) between subjects' scores on original and 18-item versions of the NCS. The NCS shows strong internal consistency (r.76, p<.001) and some evidence exists for its content and predictive validity (Cacioppo & Petty, 1982).
Procedures were set up to insure standardization of
administration. Instruments were presented to each subject by cassette tape (see Appendix G for a complete transcript of the tapes for each experimental condition).
Subjects were run in group administrations in a language laboratory classroom. They were seated and provided with a package of written materials, the measures of interest. They were then instructed by the researcher to place their headphones on and follow the instructions
that they heard on the tape carefully. All further instructions to the subject were on the taped message.
The taped message consisted of a general introduction to the study and the cognitive restructuring technique for all subjects. Half of the subjects in each involvement level group were randomly assigned to one of two experimental conditions, either strong arguments in favor of the effectiveness of cognitive restructuring or weak arguments favoring cognitive restructuring. (See Petty & Cacioppo, 1984, for a discussion of the development of these messages). The next section of the tape that the subjects heard presented either strong or weak arguments for cognitive restructuring depending upon the experimental condition being conducted. This was followed by instructions for the completion of the Assertion Survey followed by three minutes of silence to enable the completion of the questions. Next, instructions for the thought listing for the message were presented and followed by three minutes of silence during which subjects were instructed to "List all of the thoughts that came to mind as you listened to the taped message." Instructions for the Cogntive Assessment Questionnaire were presented next and followed by five minutes of silence during which the subjects completed the questionnaire. The tape resumed with instructions for covert rehearsal of a personal assertion situation and its associated thought listing,
again allowing for three minutes to complete the task. Finally, subjects were instructed to remove their headphones, turn in their materials to the researcher, and receive their experimental credit from the researcher.
All subjects participating in the study will be given a subject number code which will identify their responses while maintaining their anonymity. All subjects also received either experimental credit or extra class credit for their participation in the study.
Analyses of Data
The analyses compared the impact of high and low
levels of ego-involvement, issue involvement, and message quality on attitudes toward cognitive restructuring. Dependent variables included an index of cognitive favorability in response to the taped messages, outcome expectation attitude ratings, behavioral intention ratings, and post-coding of covert rehearsal thoughts. A series of 2 x 2 x 2 analyses of variance were performed to determine the effects of ego-involvement, issue involvement, argument quality, and interactions on these variables. Duncan's Multiple Comparison Procedure was used to evaluate the effects predicted in the research hypotheses.
This study was designed to test and compare the
varying levels of ego-involvement, issue involvement, and argument quality upon a subject's cognitive responses to
the cognitive restructuring technique presented in a message advocating its use in treating assertiveness problems.
The design was a 2 x 2 x 2 between subjects factorial design. The factors were ego-involvement (high and low), issue involvement (high and low), and argument quality (strong or weak). The dependent variables consisted of the following: 1) cognitive favorability index, 2) outcome expectations for the use of the CR technique, 3) behavioral intention toward cognitive restructuring, and 4) spontaneous use of CR during covert rehearsal.
The sample was drawn from the University of Florida introductory psychology subject pool and an undergraduate sociology class. It consisted of students pretested to form four distinct groups: 1) high ego involved, high issue involved; 2) high ego, low issue involved; 3) low ego, high issue involved; and 4) low ego, low issue involved.
Instruments used in this study included the
Relationships Grid, Personal Problems Inventory (which consisted of the Simple Rathus Assertiveness Schedule), Assertion Survey, Cognitive Assessment Questionnaire (Need for Cognition Scale), and Thought Listing Techniques for messages about cognitive restructuring and covert
rehearsal. Procedures for collection and evaluation of data were standardized in order to prevent bias between groups.
Analysis of data compared the four involvement groups and argument quality. A 2 x 2 x 2 ANOVA for each cognitive response variable determined the effects of egoinvolvement, issue involvement, argument quality, and interactions.
Before investigating whether the manipulations produced the desired effects on the dependent measures, the effectiveness of the manipulations should be ascertained. Manipulation checks were planned for argument quality and issue involvement and analyzed using a series of 2 x 2 x 2 analyses of variance.
To see whether subjects differentiated the strength of the arguments used in the message quality manipulation, subjects answered a seven-point Likert item asking them to rate the quality of the arguments in the intervention: very poor arguments (1) to very good arguments (7). Results of the Analyses of Variance indicate that the manipulation successfully influenced how strongly the subjects rated the quality of the two groups of arguments, F(1,96)=30.43, p<.0001: strong arguments M=5.0, weak arguments M=3.5. The involvement variables (issue and ego) did not significantly affect this manipulation check, either alone or in interaction with the argument quality manipulation.
In addition, as predicted by the ELM subjects who heard the strong argument quality message generated a significantly greater proportion of positive thoughts (M=.76) than did subjects who heard the weak message (M=.55), F(1,90) = 12.21, P<.0007. Furthermore, subjects who heard the message with weak arguments generated a significantly greater proportion of negative thoughts (M=.45) than did subjects who heard the strong message (M=.23), F(1,90) = 12.21, p<.0007. Issue Involvement
The other category of manipulati on checks investigated
the degree of issue involvement that subjects believed that they had with the use of the cognitive restructuring technique for treating personal problems in assertion. As an initial check on issue involvement a seven-point Likert item asking subjects to indicate the extent to which the message had implications for them personally was' included. An analysis of variance revealed a trend toward a main effect of issue involvement, F(1,96)=2.8, p=.10; high issue M=4.25, low issue M=3.72. All other variables (ego and argument quality) also did not significantly affect this manipulation check. As these results cast doubt on the effectiveness of the issue involvement manipulation, an alternative operationalization of involvement was pursued. Petty and Cacioppo .(1986) suggest that as personal involvement increases, people become more motivated to
process the issue-relevant arguments presented. One way in which argument processing could be measured would be by the number of relevant cognitions that a person generates. In order to assign subjects to high and low levels of argument processing (high and low issue involvement)-a median split was performed on scores of total number of relevant thoughts generated by subjects on the thought listing technique. The median score was deleted and subjects in the upper half of the distribution constituted the high issue involvement group, while subjects in the lower half constituted the low issue involvement group. Analysis of variance of the involvement manipulation check revealed that high issue involvement subjects believed that the message had significantly more implications for them than did the low issue involvement group, F(1,96)=6.17, p<.Ol: high issue involvement M=4.4, low issue involvement M=3.6. The other variables (ego-involvement and argument quality) did not significantly affect this manipulation check.
Data for each of the dependent measures was analyzed by a 2(High and Low Issue Involvement) x2(High and Low EgoInvolvement) x2(Strong and Weak Argument Quality) Analyses of Variance. Analyses for each dependent measure are presented separately.
Cognitive Favorability (CFCR)
The analysis of variance revealed a significant main effect of argument quality for the CFCR score, F(1,90)=12.21, p<.001, such that subjects who heard the strong arguments had a greater proportion score for overall cognitive favorability toward cognitive restructuring (M=0.54) than did subjects who heard the weak arguments (M=0.10) (see Table 2) Means and standard deviations for CFCR for all levels of the design are presented in Table 3. Attitude Rating
The analysis of variance revealed a significant main effect of issue involvement for the outcome expectation attitude rating, F(1,96)=9.23, p<.003, such that subjects in the high issue involvement condition believed that the cognitive restructuring technique would be significantly more beneficial to them (M=4.58) than did subjects in the low issue involvement condition (M=3.76) (see Table 4). Means and standard deviations for attitude rating are presented in Table 5.
The results also revealed a trend toward a main effect of argument quality for the attitude rating, F(1,96)=3.54, p=.06. These results suggest that the group that heard the strong arguments had a tendency to believe that the cognitive restructuring technique would be more beneficial to them (M=4.4) than did the group that heard the weak arguments (M=3.9).
N ON r-q r- Ln Ln CN
"W r-i 0 0 lw %D v
w 0 0 t.0 co V en
C14 r-I r-q r- v qw r-4
C14 0 C14 CN 0 ull a,%
DW 0 0 a 0 0 0
C) 0 C14 0 0 0 0
(n IV i a% o r- lw 0
00 0 r-4 0 rq N r- r-4
ON C> 0 r-4 0 CN m v
En 0 0 0 0
(D Ln 0 0 0 0 CD
fa w X
44 r-4 -4 r--i r--4 r-4 r-I 0
0 00 0 4 C) r-I C14 r- C14
u W aN 0 0 r-4 0 C14 m M
44 0 Ln 0 0 C) 0 w
44 w Cl >4
x x .,q
4J 4J 4J
as (D w a) >1 x
c > > > 41
r-4 r-4 r--i -4 0
0 0 0 0)
> > >
u (D (1) x Q) w
w :3 :3 :3 Z 0
:3 En 0 (a U) ri) 0 ul ).4
0 U) 0) 0) rA W
E-4 M C9 w w
0i m* 04 m aO~O
Ur~0 w 0 V$ LA %D co IV co
U t' 0 LA LA CN .4 U 0 0 0 0D 0 0;
44 > ci -4 r4 r
0i) v 4a
m cw C: Nd C
04 04-4 0 0 i 0 Mi 0
r_> -W J 4 .J 4 -' 3 4 -'
r-4 ::$ /)C) )U
E-i 4' 4J 0 '
04 r4a 0
> 0.JJ 04'9
ci) 0 .
0 0 0i
04 m Ln m 10 LO r-A 1w
0 v %D Ln C14 m
w m r- lw 00 CN
N r-4 Ln Ln w C%4 r--4
0) Clq 0 C>
r- 00 00 00 tn r-4 Ln
LO 0 LO r- m r- Nr 00
Ln m M-4 m Ln C14 C>
ON -W r- r-4 r-I a CN
ro r-I r-4 w
qw r- 00 00 co LO
ra ul C> LO r- m r- "W co
1-4 C14 Ln m m Ln N r-4
> 0 (a
4J -W -W
C r_ C:
w a) CR
Q) >1 x
> > > 4J
r--q r--4 -4 -4 0
0 0 0 -q (n
> > >
>1 0, x
W :3 0
:3 U) 0 (D Ul 0 U)
0 En w 0) (1)
La Ol -q W -q
r- Ln ko Ln C14 Ln "0
Ln C) ko "W r- 110 0 0)
ru CN r-4
%0 Ln .o
w W CN 10 cn "W as a% m
C4 0 41
ra 0 ra
:3 -H r-W 4J Q)
-4 a 04
>1 4J 0) 00 co co Ln a% co f-41 m r-I r- v Ln "W
ra r-4 m C14 0
U) >11 Ol n k.0 00 0
W E-4 4 Ln r-I N
m 4 a r-4 r-A r-A r-i
c ro 4-1
w 4-1 -14 Sd C: -14
01 0 0 m 0 co 0 ra 0 (13
C: 44 4-1 w 0) w a) w (1) w a)
0 al 4J :3: 4J 4J 3: 41 .11:
w U) U)
W 0) 4 Ol
fu w E
> u > 4-1
a) to c
ro 0 r. E
4 -4 >
ro a 0 V
0 0) 3: 3: ol 3:
> 0 0 0 0
4-) W a) r_ $-a --a -l
4J ra 0
3: 0 Q)
-1 > 4J
> r-4 c
Behavioral Intention Rating
The analysis of variance revealed no significant
findings for the behavioral intention score. means and standard deviations for the behavioral intention score are presented in Table 5.
Spontaneous Use of Cognitive Restructuring
Only the post-coding categories of self-instructional
(SI), self-assuring (SA), and self-doubting (SD) statements were deemed to have adequate inter-judge reliability (range .81 to .92). These three categories were analyzed by separate 2 x 2 x 2 Analyses by Variance.
Self-instructional statements. The analysis of variance revealed no significant findings for the SI score.
Self-assuring statements. The analysis of variance revealed no significant findings for the SA score. However, the results did reveal a trend toward an interaction effect of ego-involvement x argument quality for the SA score, F(1,89)=2.8, p=.10. Examination of the means suggests that under the high ego-involvement condition argument quality did not influence the number of self-assuring statements generated, but under the low egoinvolvement condition subjects who heard the message with strong arguments generated significantly less self assuring statements than subjects who heard the message with weak arguments.
Self-doubting statements. The analysis of variance revealed no significant findings for the SD score.
Correlations computed between eight of the design's
variables using Pearson product-moment correlations yielded several significant relationships. The correlations most germane to the present study included those between the need for cognition variable and the other variables in the study. These analyses yielded only one significant correlation between the need for cognition score and the score on the Simple Rathus Assertiveness Scale (SRAS), r .372, p<.0001. This result suggests that as subjects increase in need for cognition, they see themselves as more assertive. Perhaps even more interesting is the nonsignificant correlation between the need for cognition variable and the total number of relevant thoughts generated by the subjects. This nonsignificant correlation does not support Cacioppo and Petty's (1984) notion that individuals high in need for cognition would tend to think more extensively about messages presented to them. Correlations are presented in Table 6.
4J r-I r-I a 0) 00 CN cn m aN r--4 Ln 0 v 00
C 4J qw 0 00 r-I r-I Ln IV r-I a r- Ln a% -4 co
04 000 0 lw 000 CN 0 r-4 0 Ln 000
> 0) CD a 0 0 0 0 0 a 0
:3 v v v v v v
0 v 04 04 04 04 04 04
(n En v CD m r--i 0) C-4 Ln M CN 00 CN N 0
m (n CN m r-q lw co co v 110 Ln r-i r- C)
0 r- r-I q 0 CD 0 m 0 o 0110 cn C>
0 0 0 0 0 0 0 0
u I v v v I v I 1 1 1; 0 %,;
ul 04 04 04 04 04
w 0 co
0-4 w lw r- v r-4 CN C14 N -4 w qw r4-4 4J 0 m 14 m C%4 CN r-A 0 m CO m m
-4 0 r-I C14 -4 N 0 m l r-4 0
ro 0 0 0 0 0 0 0 0 *
w 0) 04 v v v
w 0 04 04 04 04 04
Ln v) rl- k.0 m C> O)o IV C>
%D N r-4 kO m C> Ln c CN 0
4 u a tn 0 OD (Y) C) CY) CD qw 0
lu V 0 0
> v 1 1; 1; 1;
4J 04 04 04 04
ra (a a 1-1
r. w 0 r-4 Ln w r- 00 C) 0
(D 0 M 00 Ln r- m C14 m 0
q 0 Lo m r- 0
> 0 0
M w v I v
,a 41 04 04
ra a) .
r. m 0-4
> ro a)
.,1 :3 C qw tn N a) C14
4J 4j -4 r- 0 r-4 Lf) 00 w
04 -ri 4.) C-4 0 r-4 CN 4 0
.r.4 4J rO 0 0 0 .
w 4J v v I v
44 (3) 41 r-4 OD r-4 00
0 E 0 00 C) 00
0 4 I v v
.H < Of 04 04
0 W r-4 r4
a) 0 > 04
4-J 0) r-q
> 4J 4J u
C c 1-1
Q) (1) M rE E 41 (13 Q) 4 0 w 0
Q) a) r_ >1 ra 0 -,q 0 -r-I
> > 0) -W :3 CP -4 4J 44 4.) 4
E 41 > C:
0 1 0 :1 Ix ra >
> 0 > m m 4-) -W a 41 0) al a)
U) r_ U) r_ w ::3 4.) CO Q) a) 0 %
J- W -4 F:4 Ol 4 % M -4 u z u (Tj
The primary purpose of the present investigation was to examine the effects of issue involvement and egoinvolvement upon the cognitive responses that individuals generate to arguments advocating the use of cognitive restructuring to treat problems in assertiveness. Hypotheses for these effects were developed from a review of the social psychological literature on the Elaboration Likelihood Model of Attitude Change. The present study attempted to extend this model to include messages that may be used in a counseling context.
In discussing variables that can affect a person's ability to process message arguments Petty and Cacioppo (1986) cite personal relevance/involvement as "perhaps the most important variable in this regard" (p. 144). Their ELM suggests that as issue involvement increases, people become more motivated to engage in the cognitive work necessary to evaluate the issue-relevant arguments presented in a message. Petty and Cacioppo (1979b) provide evidence consistent with this view. The first hypothesis of the present study was also based upon this view. more specifically, it predicted a significant interaction 89
between issue involvement and argument quality such that there would be greater distinction between strong and weak arguments under conditions of high issue involvement than under conditions of low issue involvement. The predicted issue involvement X argument quality interaction was not found to be significant for any of the dependent measures.
The second hypothesis made a similar prediction of a significant interaction between ego-involvement and argument quality such that there would be a greater distinction between strong and weak arguments under condition of high ego-involvement than under conditions of low ego-involvement. Once again, the predicted interaction was not found for any of the dependent measures.
The results do indicate that the strength of the
argument quality significantly differentiated scores on the CFCR dependent measure and demonstrated a strong tendency to differentiate scores on the attitude rating dependent measure. These results, coupled with the significant argument quality manipulation check, suggest that overall the content (arguments) of a counseling message made a major difference in its effectiveness in advocating cognitive restructuring for treating assertiveness problems. Also consistent with the ELM Model was the finding that strong messages produced more positive cognitive responses, whereas weak messages generated more negative thoughts. For the counseling situation this
implies that it may be advantageous for the therapist to screen out distractions, to repeat important messages over a number of sessions, and even to provide the client with a written summary of central points made during therapy, all manipulations that may also have the effect of enhancing client involvement.
This effect of the message variable also has other important implications for counseling. According to the ELM, clients who carefully evaluate message content are more likely to have enduring, behavior-related, centralroute attitude change than are those who attend only to peripheral cues, such as counselor personal characteristics. Message variables (such as argument quality) may play crucial roles in eliciting behavior change and in the maintenance of therapy gains. Future research should investigate message factors to determine those that are most helpful to counselors in bringing about client change.
Results also show that issue involvement produced a significant main effect for attitude rating such that subjects who generated more thoughts about the message believed that cognitive restructuring had more personal benefits than did subjects who generated fewer thoughts. Theoretically, an explanation of this finding could be found in Zajoncls (1980) theory of mere exposure, which describes changes in evaluation of objects as a result of
relatively primitive affective and associational processes. It may be that those subjects who thought more about the message associated it more with counseling and the common notion that counseling is beneficial. Because a main effect for issue involvement was not significant for the CFCR measure, it does not appear that high issue involvement subjects saw the cognitive restructuring technique itself as being better than did the low issue involvement subjects. In other words, greater favorability toward the technique was related to more thinking in general, but not to more favorable thinking per se.
This theoretical explanation further suggests some
methodological factors that may have affected the results of the study. It may be that the high and low levels of issue involvement in this study do not correspond to the high and low levels of issue involvement of previous ELM research reported in the social psychological literature. In previous research involvement manipulations were designed to induce relatively pure forms of the central and peripheral routes to persuasion. For example, in the 44
Petty, Cacioppo, and Goldman (1981) study high involvement subjects were told that the message they were to evaluate had relatively important consequences for them (i.e., if they did not pass the senior comprehensive exam that was to be instituted the next year, they would not graduate). In contrast to this, low involvement subjects were led to
believe that the proposal would have no personal consequences for them. These two groups of subjects represent relatively extreme levels of involvement. In the present study, and in the counseling situation in general, the issues of concern are applicable to almost everyone. That is to say, most topics discussed in counseling (for example, assertion, problems in personal relationships) are by definition highly involving since such issues are only presented if they are personally problematic concerns. Therefore, the levels of issue involvement in the present study might best be described as moderately high and moderately low. Further support for a moderate level of involvement came from the nonsignificant difference on Rathus scores between high and low issue involvement conditions. This could explain the current failure to duplicate the interaction between issue involvement and argument quality found in previous ELM research. This same argument could also be extended to the ego-involvement variable.
Another methodology question concerns the
operationalization of issue involvement in counselingrelated research. Past investigations have attempted to manipulate issue involvement in a number of ways. These attempts have included manipulation of clients' "perceived need" or requests for help (Dixon & Claiborn, 1981; Heppner & Dixon, 1978) and motivations for counseling (Heppner &