Citation
Gender schema and the tough message for political candidates

Material Information

Title:
Gender schema and the tough message for political candidates
Creator:
Bernstein, Arla G
Publication Date:
Language:
English
Physical Description:
xii, 137 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
Education ( jstor )
Educational evaluation ( jstor )
Gender politics ( jstor )
Gender roles ( jstor )
Political candidates ( jstor )
Political speeches ( jstor )
Stereotypes ( jstor )
Voting ( jstor )
Women ( jstor )
Womens studies ( jstor )
Dissertations, Academic -- Mass Communication -- UF ( lcsh )
Mass Communication thesis, Ph. D ( lcsh )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1998.
Bibliography:
Includes bibliographical references (leaves 127-136).
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Arla G. Bernstein.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
029530924 ( ALEPH )
41286068 ( OCLC )

Downloads

This item has the following downloads:


Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EIFF91A76_JEEOOD INGEST_TIME 2014-08-27T21:42:45Z PACKAGE AA00024974_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES



PAGE 1

GENDER SCHEMA AND THE TOUGH MESSAGE FOR POLITICAL CANDIDATES By ARLA G. BERNSTEIN 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 1998

PAGE 2

This dissertation is dedicated to my husband, Wayne, my partner in life, to my daughters, Monica and Deanna, who give me the best reason to be a role model, to my granddaughter, who inspires me to fight dragons, and to my parents, Adele and Paul Sonnenblick, who taught me the value of education.

PAGE 3

ACKNOWLEDGMENTS As an individual pursues his or her life's work, special working relationships are formed that have a significant impact on the character of that individual. As a doctoral student, the relationship with my committee members in the development of my dissertation has been one of the most significant in my professional life. Much of my growth as a scholar, I owe to my advisor, Dr. Mary Ann Ferguson, whose patience and enthusiasm have been paramount in my achievements. It has been a privilege to draw upon her exceptional scholarly wisdom. My co-advisor, Dr. Deborah Treise, has been instrumental in helping me to understand the multi-faceted nature of research and in giving me the moral support to get through some difficult times. Special thanks go to Dr. Michael Weigold, who has guided me through a deepening understanding of experimental design, and to Dr. Linda Hon, who insightfully guided me towards an empirical investigation of gender schema theory. To my mentor in the field of political science, Dr. Michael Martinez, I offer a deep appreciation for helping me to bridge the connection between mass communication and political behavior. Numerous others have helped me to complete this dissertation. I want to extend special recognition to Dr. Ken Wald for his expert guidance in research methods and to Jim Geason for generously sharing his statistical and computer expertise. To Alyse Gotthoffer, my friend and officemate, 1 want to give a special thanks for her support and encouragement. Thanks also to fellow doctoral students Cyndi Frisby, David Halpern, iii

PAGE 4

Brad Yates, and Kerry Crooks for allowing me to use their classes for my experiment. Most especially, I want to thank my family — my husband, Wayne, daughters Monica and Deanna, sisters Janice and Iris, and brother Harvey — for their belief in me. IV

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDGMENTS iii LIST OF TABLES viii LIST OF FIGURES x ABSTRACT xi INTRODUCTION 1 Nature of the Problem 2 Purpose of Study 4 Scope and Limitations of the Study 4 LITERATURE REVIEW 6 Gender Schema in Political Cognition 6 Political Cognition and Schema Theory 7 The Social Origins of Gender Schema 12 The Structure of Gender Stereotypes as Schema 17 Implications for Female Candidate Evaluation: Cognitive Modeling 17 Gender Schema and Political Stereotyping: A Cognitive Model for Candidate Evaluati on 18 Gender Schemata and Political Knowledge 19 Gender Schemata and Political Stereotyping in Candidate Evaluation 21 Implications for Female Candidate Evaluation: Constructing the Model 24 Gender and Political Credibility: A Function of Gender Schema and Stereotyping 25 Gendered Credibility: A Meta-Analytic Review 26 Results of M etaAnalysis 32 Implications for Female Candidate Evaluation: Gender and Competence Schema 35 Message Schematics and Strategies for Female Candidates 35 Leeper's Research 37 The Gender-Based Message Debate 40 Implications for Female Candidate Evaluation: Developing a Message Strategy 41

PAGE 6

Summary of Theoretical Foundations 43 Conceptual Definitions of Key Variables 43 Independent variables 43 Dependent variable 44 Gaps and Conflict in Prior Research 44 Laws/ Assumptions from Prior Research 44 Premises as Bases for Hypotheses 46 Hypotheses 47 Comparing female candidates to other female candidates or to themselves 47 Comparing female candidates to male candidates: 47 METHODS 51 Introduction 51 Operational Definitions of Key Variables 53 Manipulated Variables 53 Dependent Variables 57 Research Design and Validity 58 Internal Validity 58 Design Validity 58 Construct Validity 59 External Validity 60 The Experiment 61 The Participants 61 Treatment and Instrumentation 61 Message Variation 62 Data Collection 64 Data Analysis 64 FINDINGS 66 Study Participants 66 Indices 67 The Message Clarity Index 67 The Political Knowledge Index 68 The Gender Role Schema Index 71 Pilot Study 74 Manipulation Check 75 Research Questions 76 Hypothesis Testing 77 Female Candidates: Main Effects and Interactions for Message Theme and Clarity 78 Comparing Female and Male Candidates: Main Effects of Candidate Gender 82 Effects of Subject Characteristics 84 VI

PAGE 7

Post Hoc Analysis of the General Model for Policy Competence 88 Post Hoc Analysis of Message Clarity Effect for the Male Candidate 89 Post Hoc Analysis of Candidate Gender for Messages on Education 90 Post Hoc Analyses of Subject Characteristics 90 Summary of Findings 92 DISCUSSION 95 Limitations of the Study 95 A Comparison of Predicted and Observed Results 98 Message Theme and Clarity for Female Candidates 98 The Effects of Candidate Gender on Candidate Evaluation 99 Effects of Political Knowledge, Media Use, and Gender Schema on Candidate Evaluation 101 Implications for Theory Development 103 The Gender-Based Message Debate Revisited 103 Gender Schema Theory 104 Candidate Evaluation Modeling 105 Implications for Political Campaign Strategy 107 Suggestions for Future Research 108 CAMPAIGN MESSAGES HI QUESTIONNAIRES 119 REFERENCES 127 BIOGRAPHICAL SKETCH 137 Vll

PAGE 8

LIST OF TABLES Table page Table 2.1: Studies Included in the Meta-Analysis 49 Table 2.2: Comparing Policy Areas Across Studies 50 Table 3.1: Comparison of Leeper's and Bernstein's Studies 52 Table 3.2: Message Variation by Cells 63 Table 4. 1 Factor Analysis of Message Clarity Index with Principal Components Analysis 68 Table 4.2: Percent of Correct Political Knowledge Questions by Female Respondents... 70 Table 4.3: Percent of Correct Political Knowledge Questions by Male Respondents 70 Table 4.4: Factor Analysis of Gender Role Schema Index with Principal Components Analysis 73 Table 4.5: Analysis of Variance for Preexperiment Manipulation Check of Message Clarity 75 Table 4.6: Analysis of Variance for Manipulation Check of Perceived Message Clarity... 76 Table 4.7: Means for Perceived Message Clarity by Message Theme for Clarity Index Items 76 Table 4.8: Mean Competence Scores for Female Candidate with Message on Crime 80 Table 4.9: Mean Competence Scores for Clear Message on Crime by Candidate Gender 83 Table 4.10: Mean Competence Scores for Unclear Message on Crime by Candidate Gender 83 Vlll

PAGE 9

Table 4.11: Correlations Between Political Knowledge Index and Policy Area Competence S6 Table 4.12: Correlations Between Print Media Exposure and Policy Area Competence... 86 Table 4. 13: Analysis of Variance: Test for Interaction between Gender Role Schema Category and Candidate Gender 88 Table 5: Cell Sizes for Test for Interaction Between Gender Role Schema and Candidate Gender 102 IX

PAGE 10

LIST OF FIGURES Figure page Figure 4.1: Political Knowledge Index by % of Subjects 71 Figure 4.2: Exposure to News Stories in Newspapers 87 Figure 4.3: Political Knowledge by Respondent Sex 91 Figure 4.4: Gender Role Schema Index by Respondent Sex 93

PAGE 11

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 GENDER SCHEMA AND THE TOUGH MESSAGE FOR POLITICAL CANDIDATES By Aria G. Bernstein August 1998 Chair: Dr. Mary Ann Ferguson Cochair: Dr. Deborah Treise Major Department: Mass Communication Gender has been, and continues to be, a political barrier in that the sex of a communication source makes a difference in the persuasiveness of political communication. Female candidates are frequently perceived as less credible than their male counterparts, especially in races for higher political offices, and research has indicated that female candidates face tougher standards in establishing credibility. This dissertation explores a key dimension of credibility — perceived competence — and its relationship to the gender of political candidates. An experiment with 209 college students at the University of Florida was conducted in 1997 in order to investigate the relationship between candidate gender, message characteristics, and perceived competence of two hypothetical candidates, one female and one male. It was hypothesized that a "tough" message consisting of a clear position on a stereotypically male policy area — crime — would be advantageous to female XI

PAGE 12

candidates. It was also hypothesized that political knowledge, media exposure, and gender role schema would relate to how subjects evaluated the two candidates on competence in nine policy areas. The hypotheses were partially supported in that the tough message was advantageous for some policy areas but not for all of them. No significant relationships were found between political knowledge and candidate evaluation or between media exposure and candidate evaluation. Also, no interactions were found between the gender role schemas of subjects and the gender of the candidate. However, post hoc analyses indicated relationships among the subject characteristics of political knowledge, media exposure, and gender schema. It was found that the sex of subjects was associated with both their level of political knowledge and gender role ideology. It was also found that high users of television news were more likely to be traditional in their gender role schema than were subjects who watched television news at lower rates. xn

PAGE 13

CHAPTER ONE INTRODUCTION Gender has been considered a political communication barrier in that the sex of a communication source can make a difference in the persuasiveness of political communication (Hodgson and Pryor 1984; Raid 1981; Renton 1989; Miller 1981; Williams 1 984). This study has investigated the effects of candidate gender in relation to its persuasive elements. While some gender differences have been found, they have presented themselves in some unexpected ways. The process of persuasion is the core of political communication (Nimmo and Sanders 1981) and the driving force of political election campaigns (Trent and Freidenberg 1995). Persuasive message effects are related to source credibility, media types, and individual differences of the receiver, as well as, characteristics of the message itself (Raid 1981; McGuire 1985; Miller 1981; O'Reefe 1990). This study is concerned primarily with a key dimension of source credibility-perceived competence--and its relationship to the gender of the political candidate as the communication source. Perceived competence is conceptually defined as expertise, knowledge and experience as perceived by the evaluator (Bernstein 1995; Lumsden and Lumsden 1996; Miller, et al. 1986; O'Reefe 1990; Severin and Tankard 1992; Williams 1984). The relationship between gender and perceived competence was built into a model of candidate evaluation which integrates

PAGE 14

candidate gender, perceived competence, message themes and clarity, subject gender, gender schema, levels of political knowledge, and media use. Nature of the Problem Female candidates have been generally perceived as less credible than their male counterparts, especially in races for higher level offices (Huddy and Terkildsen 1993; Johnston and White 1994; Rosenwasser and Dean 1989). Research has indicated that female candidates face tougher standards in establishing credibility than male candidates (Allen, 1990; Johnston and White 1994) and that gender stereotyping of political candidates is frequently associated with issue expertise or competence (Alexander & Andersen, 1993; Huddy & Terkildsen, 1993; Kahn and Gordon 1997; Leeper, 1991; Kahn, 1992, 1994; Rosenwasser, et al., 1987; and Sapiro, 1982). For these and other reasons, women have made limited progress as political candidates. Even in 1992, known as the "Year of the Woman, " only 29 female candidates were elected to Congress (Wilcox 1994). As calculated by Wilcox, women would need more than ten "Years of the Woman" to attain parity in the Senate, and more than seven "Years of the Woman" to reach equality in the House. Additionally, only 20 percent of the seats in state legislatures are occupied by women. To what degree do gender stereotypes influence the outcome of political races? Despite recent gains, women still win political office much less frequently than men (Kahn 1992, 1994; Darcy, Welch, & Clark; 1987; Wilcox, 1994; Witt, Paget, & Matthews, 1994). The 1992 elections produced a House of Representatives which was only 1 1 percent

PAGE 15

female and a Senate which was only seven percent female (Alexander & Andersen 1993). Nearly all female challengers to House and Senate incumbents lost, often by large margins. A critical element of the struggle for female political candidates is credibility, and more specifically, perceived competence (Allen, 1990; Cantor, Bernay, and Stoess, 1992; Kahn, 1992, 1994; Pearson, West, & Turner, 1995; Witt, Paget, & Matthews, 1994). However, a number of studies have indicated gender differences in perceived competence (Alexander & Andersen 1993; Brown et al. 1993; Hedland, Freeman, Hamm, & Stein, 1979; Hodgson & Pryor, 1984; Huddy & Terkildsen, 1993; Kahn, 1992, 1994; Kenton, 1989; Leeper, 1991; Matland, 1994; Nicotera & Rancer, 1994; Norris 1997; Rosenwasser, Rogers, Fling, Silvers-Pickens, & Butemeyer, 1987; Sapiro, 1982; Sigelman & Sigelman, 1982; Whittaker & Meade, 1967; Williams, 1984; and Witt, Paget, & Matthews, 1994). When the evaluation of candidate competence is based not on the merits of the individual but on stereotypes of the social group of which the individual is a member, there may well be a threat to democratic choice. One may argue that using gender as a cue is not entirely irrational. Indeed, many people have suggested that stereotypes often contain a "kernel of truth. " The problem is that while stereotypes may provide some truth, they provide only a kernel of truth at best. "The effect of stereotype-based cue taking is to turn what we define as a process in which success is supposed to be based on merit into one in which success is based on ascribed characteristics" (Sapiro 1982, 75). Evaluations based on stereotypes serve as barriers toward advancement of equal opportunity in the political realm.

PAGE 16

Purpose of Study The purpose of the proposed study is to investigate the effects of gender schema on candidate evaluation. While previous research has addressed this relationship, there is a dearth of research incorporating the particular set of variables to be integrated here in a gender-based schematic model of candidate evaluation. In spite of the obstacles to women's electoral success, very few studies have attempted to identify strategies that might be effective for female candidates (Johnston and White 1994). Exploration of the influence of mediators such as gender schema, political knowledge, message themes, and media use in an empirical study would be useful to political theorists and strategists. This study addresses the question: "To what extent and under what conditions does gender of candidates affect their credibility?" While credibility is most often associated with both competence and trustworthiness, this study focuses on the competence dimension of credibility. The gender differences in perceived competence documented in the studies cited point to a serious and significant problem for female candidates and the ways in which they are evaluated. Scope and Limitations of the Study In constructing a gender schematic model of candidate evaluation, various components of persuasion have been considered, including source credibility, message characteristics, media types, and individual differences of the receiver of the message. Each of these components branches off in a variety of dimensions and sub-components. For

PAGE 17

example, source credibility contains the two key dimensions of competence and trustworthiness (Bernstein 1995; McGuire 1985; O'Keefe 1990); message characteristics contains a variety of dimensions pertaining to number of arguments, one-sided versus twosided arguments; types of appeal, ambiguity of position, and theme (McGuire 1985; O'Keefe 1990); mass media types include a variety of print and electronic media (O'Keefe 1990); and individual differences of the receiver include gender, race, age, personality traits, socioeconomic status, intelligence, knowledge, and involvement (O'Keefe 1990). For the purposes of this study, the independent variables selected are candidate gender, message themes and ambiguity of position, subject gender, gender schema, media types, and political knowledge of the receiver. The dependent variable of interest is perceived competence of the candidate. These variables are (1) central to the candidate genderperceived competence relationship and (2) of immediate interest to the researcher. It is expected that other variables will be investigated in future phases of investigation by this researcher.

PAGE 18

CHAPTER TWO LITERATURE REVIEW This literature review consists of two major components: (1) a conceptual discussion of gender cognition and schema theory and (2) a meta-analytic review of previous studies regarding candidate gender and perceived competence by evaluators or voters. The metaanalysis is a statistical review which is useful in terms of summarizing the results of a specific collection of studies. Four specific areas to be discussed in this literature review are: (1) the role of gender schema in political cognition; (2) the interaction among gender schema, political knowledge, and media use and the effects of this interaction on candidate evaluation; (3) the effects of gender schema on the perception of candidate competence; and (4) the interaction of gender schema and message characteristics and its effects on candidate evaluation. Gender Schema in Political Cognition While much previous research on the effects of candidate gender has been atheoretical, more recently, researchers have begun to investigate voters' stereotypical reactions to candidates in terms of gender schematic processing, and have applied gender schema theory in empirical investigations (Hitchon, et al. 1997). Gender schemas are conceptually defined as knowledge structures which are based on sex-linked associations, such as those related to traits or roles associated with males or females (Bern 1981).

PAGE 19

Hitchon, et al. (1997) describe the application of theories specific to gender, such as gender schema theory, as an important milestone in the study of voter perceptions of female politicians and in the study of the gender construct as significant in theories of mass communication. Most research regarding gender differences in candidate evaluation tap into the schema that voters use upon learning a candidate is female-referred to as gender schema or gender-based schema (Bernstein 1996; King and Matland 1997). Gender-based schema can be used in three ways to evaluate candidates: (1) evaluation is gender-neutral, gender is not a cue and gender schemata are not used; (2) gender schemata are used only to fill in gaps of missing information; and (3) schemata are used as a filter, filtering in schema-consistent information, and filtering out schema-inconsistent information. The third instance is the case in which stereotypes are strong and respondents are frequendy oblivious or resistant to new information. Gender schema figure in most prominently in evaluating candidates when other information about the candidate is lacking (King and Matland 1997; Leeper 1991; Sapiro 1982). Political Cognition and Schema Theory In political science, schema theory is used primarily in three research domainspolitical communication, voting behavior, and belief systems (Wilcox 1990). This study focuses on political communication, with specific emphasis on campaign communication of political candidates. The field of political communication has, as have other domains of political science and mass communication, recognized the need for information processing theory development in order to explain and predict the effects of communication on political

PAGE 20

8 campaigns (Miller 1981). An important focus of political information processing or cognition is the perception of political candidates by the voting public. Cognitive theories, which have been imported from cognitive and social psychology, have been applied to studies of political perception and political behavior (Lau and Sears 1986a). Political cognition is the information-processing approach to explaining political behavior (Lau and Sears 1986b). The central concept and focus in both social and political cognition is cognitive schema"a hierarchical organization of knowledge in a particular domain, which includes a category label, generic descriptions of the stimulus domain, particular instances of it, and interconnections of these" (Lau and Sears 1986b, 349). The concept of schema is that "one's prior knowledge about some particular domain is organized in memory in a coherent structure that influences what one sees and remembers and how one interprets reality" (Hamill and Lodge 1986). The hierarchically organized network consists of concepts and attributes-thought of as nodes-that are interconnected through associations or relationships-thought of as links (Lau and Erber 1985). These links allow for varying degrees of inference, depending on a person's schema complexity. When a person has a welldeveloped schema, the inference is top-down or "theory-driven" (Lau and Erber 1985). The Purpose of Schema . Schema theory helps researchers to address the question: How do people think about political candidates and issues? Schemata serve two basic functions: (1) guiding the processing and storing of incoming information and (2) guiding the recall and interpretation of information in memory (Lau and Sears 1986b). People actively process and interpret incoming information according to their preexisting knowledge structures-their schemata (Lau and Erber 1985).

PAGE 21

Political schemata allow us to explain more specifically candidate evaluations and voting decisions (Lau 1986). How might schematic processing influence political perception? Generally, schematic models of candidate evaluation are used to explain candidate perception and judgment (Lau and Sears 1986b). Schemata aid information processing by determining what information is attended to, how it is encoded and stored, and how it is utilized" (Garramone, Steel, and Pinkleton 1994, 312). Schemata determine what types of information are relevant, and what information is noticed, stored and available for recall and judgment. The same information may be stored by one individual, but ignored by another (Lau and Sears 1986b). "Due to limitations in cognitive capacity and motivation, individuals often are unable or unmotivated to deliberately process all the information that is relevant to a judgment or decision. Instead, heuristics are employed at a number of stages of information processing" (Iyengar and Ottati 1994). A number of assumptions about the schema concept provide a basis for theoretical exploration: (1) People are forced to be cognitive misers because of limited ability to process information; (2) Old information affects the manner in which we interpret new information; (3) Schemata influence the abstraction, interpretation, and integration of new information (Hamill and Lodge 1986; Conover and Feldman 1986; Lau and Erber 1985; Medvic 1996; Wicks 1992; Reeves, Chaffee and Tims 1982). Concepts represented by a schema vary in the degree of abstraction. On a more general level, one can hold a generic politician schema, perhaps in which the target is believed to be power-hungry. On a more specific level, on can have a schema of a congressperson, and on the next, even more specific level, one can have a schema of a particular member of Congress (Lau and Erber 1985).

PAGE 22

10 Schema Development and the Mass Media. There is general agreement among political behaviorists that explaining vote choice requires attention to the processes that voters use in evaluating political candidates (Medvic 1996; Rahn 1993). Schema theory helps us to explain how we process the information used in candidate assessment. In candidate assessment, a voter categorizes the candidate as an instance of a particular schema. As the individual is processing incoming information, the person is building or developing a specific schema of the candidate. Candidate schemas, used to make inferences about candidates, are derived from four general types of schemata: (1) role schema; (2) social group schema; (3) party schema; or (4) ideological schema (Conover and Feldman 1986). Gender schema is a type of social group schema, but is closely tied to role schema in terms of sex role stereotypes (Ottati and Wyer 1993). Gender schemas are acquired through both socialization processes and individual experience (Medvic 1996). The mass media play a crucial role in the development of political schemata, including candidate schema. Direct political experience for most individuals is minimal and we derive our political information primarily through the media (Brians and Wattenberg 1996; Fiske 1986; Fiske, Lau and Smith 1990; Garramone, Steele, Pinkleton 1994; Goldenberg and Traugott 1987; Iyengar and Ottati 1994; Lau and Sears 1986b; McLeod, Kosicki, and McLeod 1994; Miller and Asp 1985; Miller, Wattenberg, and Malanchuk 1986; Reeves, Chaffee, and Tims 1982; Trent and Friedenberg 1995; Wicks 1992). The mass media are instruments for increasing levels of political knowledge and sophistication (Drew and Weaver 1990; Emig 1991; Pan et al. 1994). As such, they contribute to the development of schemata which are utilized in the candidate assessment

PAGE 23

11 process. In fact, political schema sophistication is frequently operationalized in research using measures of political knowledge, interest, behavior, and media exposure (Fiske, Lau and Smith 1990; McGraw and Lodge 1996; McGraw and Pinney 1990; Zaller 1990). Research indicates variability in political knowledge associated with "differential media usage." Media exposure is conceptualized as types and frequency or amount of media viewed, heard, or read (Brians and Wattenberg 1996; Chung 1984; Emig 1991; Fiske, Lau, and Smith 1990; Goldenberg and Traugott 1987; Lau and Erber 1985; Robinson and Levy 1996; and Zhao and Bleske 1995). Not only are political experts expected to use media sources more frequently than novices, but studies have also indicated that political experts are more likely to rely on newspapers for political information, while novices are more likely to rely on television (Lau and Erber 1985; Miller and Asp 1985). Differential media usage has implications for political learning (Ansolabehere, Behr, and Iyengar 1991; Drew and Weaver 1990; Pan et al. 1994), including the depth of information processing associated with candidate evaluation. Research indicates an association between particular types of mass media usage and the amount of issue-based reasons underlying particular choices (Brians and Wattenberg 1996). A 1978 study by Clarke and Fredin found that television news viewers offered fewer issue-based reasons for candidate choices than did newspaper readers (Brians and Wattenberg 1996). In more recent research, it was found that the strongest predictors of high political information levels were associated with regular newspaper and newsmagazine reading and with regular watching of educational and public television (Drew and Weaver 1990; Pan et al. 1994; Robinson and Levy 1996). It is reasonable to conclude that

PAGE 24

12 differential media usage is linked to variations in cognitive involvement levels among the media (Brians and Wattenberg 1996). How do we know which political schema will be cued (or utilized) at any given time? According to Lau and Sears (1986b), people have multiple political schemata that can be applied to any political stimuli-for example, a speech, advertisement, or debate. There are also contextual or external factors that influence cueing of different schemata. Most political perceptions are thought to be mediated through the mass media (Fiske, Lau, and Smith 1990; Kenski 1996; Lau and Sears 1986b, Miller and Asp 1985), and political experts gain most of their information from the electronic and print media. In fact, media exposure would be considered both a cause and a consequence of expertise or knowledge (Conover and Feldman 1986; Fiske, Lau, and Smith 1990; Lau and Erber 1985; Miller, Wattenberg, and Malanchuk 1986). The Social Origins of Gender Schema Social cognitions of gender have become "routinized" and "rationalized" in our society, much in the social and political traditions described by sociologist Karl Mannheim in his famous work on "The Prospects of Scientific Politics" (1985). Gender cognitions are part of an institutionalized structure within the realm of political practice, including the arena of candidate races for political office. To what degree do gender cognitions (viz. , stereotypes) influence the outcome of political races? A survey taken in 1990 indicated that 65 percent of women and 61 percent of men believe that the primary reason for paucity of women in higher office (such as the United States Senate) is that many Americans are not ready to elect a woman to higher office (McGlen and Sarkees 1993). Another survey taken

PAGE 25

13 in 1994 found that two-thirds of voters believe that women have a tougher time than men do getting elected to office (Seltzer, Newman, and Leighton 1997). Several explanations have been offered for women's lack of success: (1) inadequate access to political resources; (2) gender stereotyping by voters; and (3) inequitable treatment by the press (Iyengar, et al. 1997; Kalm 1992; Kahn and Gordon 1977). This study focuses primarily on gender stereotyping of political candidates by voters. However, it should be noted that gender stereotypes-a type of schemata-tend to develop based to a significant extent on what is selected for dissemination by media gatekeepers (Beasley 1997; Carroll and Schreiber 1997; Kahn and Gordon 1997; Wicks, 1992). This literature review will explore the conceptualization and use of gender stereotypes as a social cognitive process in candidate evaluation. Humans are constantly evaluating the world around them. This reality is part of the nature of human conduct and survival. As human beings, we are constantly faced with the task of making judgments and decisions resulting from these judgments. Our sociopolitical universe requires us to do so. Within our universe, psychologists and sociologists have attempted to explain female-male relations. They often begin with the basic questions: "What is a man? What is a woman?" As a result of scientific interest, a large body of work on gender stereotypes has emerged. "Stereotypes are abstract knowledge structures linking a social group to a set of traits or behavioral characteristics" (Hamilton and Sherman 1994, 3). The categorization associated with stereotyping is a mechanism for cognitive efficiency (Hamilton and Sherman 1994). Groups that deviate from what is perceived as a cultural norm are salient. In American politics, the white male is the norm, and female candidates are deviations from that norm. Social scientists are

PAGE 26

14 interested in how women and men are socially constructed and have focused on two major areas: (1) the content of beliefs regarding the sexes and (2) the relative desirability of attributes ascribed to one or the other gender. Furthermore, they have focused on "how the sexes and issues pertinent to female-male relations are evaluated" (Ashmore 1986, 69). Research in social cognition, including gender cognition, has stimulated considerable interest on the part of political scientists. In the field of political science, numerous studies have cited the use of cognitive schemata for candidate assessment (Conover and Feldman 1986; Miller et al. 1986; and Rahn et al. 1990). Voters are perceived as cognitive misers who are faced with coping with ambiguity in campaigns. Cognitive structures or schemata play a key role in aiding information processing in an uncertain political environment (Conover and Feldman 1986; Rahn et al. 1990). The candidate is assessed using schema to fill in missing data. For example, various role or social group schemata are likely to have a part in guiding the inferences that voters make about candidates. Schemata associated with various social categories-party, ideology, gender, and race-may help structure the perception of candidates (Conover and Feldman 1986; Rahn, et al. 1990). Theory regarding candidate assessment has become less reliant on party identification as a factor and increasingly focused on person and issue schema; that is, voters may focus on personal characteristics of a candidate to acquire important information relevant to assessing how an individual will perform in office (Miller et al. 1986; Garramone et al. 1994). Political cognition by voters has increasingly been attributed to schematic assessment of candidates. Schema theories suggest that voters use prototypes or stereotypes to organize their cognitions about candidates, and that candidate schemas are generally based on

PAGE 27

15 categorization of characteristics of candidates (Miller et al. 1986). These theories suggest that voters do not evaluate each political candidate de novo, but rather in terms of their inferences based on schematic perceptions (Miller et al. 1986). In that different schemata vary in their content, schemata will vary in the type and amount of inferences they produce. There are also individual differences in the use of schemata to make inferences. An individual's level of political knowledge, political involvement, and attention to the media will affect the complexity of the person's schemata, thereby affecting the inferences made about the candidate (Conover and Feldman 1986; Rahn, et al. 1990). Particularly, higher levels of political knowledge are associated with more complex schema development, meaning that the person not only has more information, but also that the individual's schema are more tightly organized, more hierarchically structured, and more integrated (Conover and Feldman 1986; Lau and Sears 1986b; Milburn 1991). More recently, research on gender stereotypes has shifted its focus from global gender stereotypes (e.g. women) to analyzing specific categories or subtypes (e.g. career women, black women, female politicians). Gender stereotypes are complex cognitive structures based on structured sets of beliefs about personal attributes of women and men, and these sets of beliefs are categorized by descriptive and prescriptive elements within different social contexts. The notion that gender stereotypes are multilevel cognitive stereotypes is based on the assumptions that (1) there are nonhomogeneous gender cognitions which relate to subtypes; (2) subtypes relate to traits, role behaviors, physical characteristics, and occupations; and (3) each subtype has a distinct component profile (Eckes 1994; Hamilton and Sherman 1994).

PAGE 28

16 Researchers have studied the cognitive processes of gender stereotyping in order to elicit culturally relevant subcategories of females and males, determine associated beliefs related to these subcategories, and identify the principles which underlie the cognitive categorization of females and males (Eckes 1994). Beliefs about gender primarily relate to traits and role behaviors associated with the sexes. For example, a commonly accepted stereotype of women is that they are emotional and intuitive (Belenky et al. 1997). Stereotypes of this nature have significant implications for female political candidates, which will be explored in this paper. A 1972 review article by Broverman, et al., which summarized numerous studies on gender stereotypes, presented three major conclusions: (1) there is wide consensus-a shared perspective-regarding the content of gender stereotypes across groups that differ in sex, age, religion, marital status, and educational level; (2) attributes associated with males are valued more positively than those associated with females; and (3) positively valued masculine traits include competence, rationality, and assertion, while positively valued feminine traits indicated are warmth and expressiveness (Ashmore et al. 1986, 79). In Women's Ways of Knowing, Belenky, et al. (1997) depict this masculine bias, particularly as associated with the third conclusion, in referring to the devaluation of women's minds and contributions in a patriarchal culture. An issue relevant to the study of candidate evaluation is how these traits and trait attributions affect such evaluation. This issue will be revisited later in this paper. Ashmore, et al. (1986) have posited three major goals for gender stereotype research: (1) determining the degree to which characteristics associated with females and males are differentially evaluated; (2) delineating the content of gender stereotypes; and (3)

PAGE 29

17 documenting the pervasiveness of beliefs about the sexes (Ashmore, et al. 1986, 83). An overview of the structure of gender stereotypes provides a context for these research objectives. The Structure of Gender Stereotypes as Schema The structure of gender stereotypes is intricately connected to the cognitive processes in which they are utilized for evaluation. The structural elements of the process are interrelated and parallel, not sequential. They consist of the processes of attention, unitizing, categorization, inference, attribution, and response selection (Ashmore, et al. 1986). Attention requires some degree of focus on, and is frequently associated with arousal by, a target of perception-such as a political candidate. Unitizing (also referred to as chunking) is a partitioning of stimuli into meaningful units, while categorization is the identification of an individual as a woman or man and as a particular subtype, such as a career woman or female candidate. Inference is the drawing of conclusions about the target, based on categorization and attribution. Attribution is the process of using a target's gender to assign traits to a target and explain or account for her/his behavior. Finally, response selection consists of behavioral intention and overt action-such as intention to vote for and actually casting a vote for a candidate (Ashmore et al. 1986). Implications for Female Candidate Evaluation: Cognitive Modeling The investigation of gender schemata helps us to understand how voters attend to, encode, store, retrieve, and interpret information related to a candidate's gender. The same information may be used in different ways by different individuals, and may even be ignored by some. The use of gender schema in candidate evaluation has important

PAGE 30

18 implications for female candidates, because campaigns in which a woman faces a male opponent would likely invoke gender-based inferences. It is also likely that the use of gender schema, especially stereotypes, is related to individual-difference factors, such as demographics and levels of political knowledge. The investigation of general demographics is beyond the scope of this study, but individual differences in voters' levels of political knowledge have been examined empirically. In order to "get inside" the cognitive process, it is important to examine the relationship between gender schema and political knowledge of individual voters. Furthermore, since most political information is mediated through the mass media, it is important to examine the relationship of media use to the processing of gender schema and to levels of political knowledge in voters. Gender Schema and Political Stereotyping: A Cognitive Model for Candidate Evaluation Given the gender barriers to political office, it would be advantageous to female political candidates to have available a cognitive model of candidate evaluation which is sensitive to gender schema. A key purpose of such a model would be to help understand how people use gender schema in assessing candidates. To this purpose, an understanding of how gender schema interact with other political variables, such as political knowledge, in the development of candidate perception would be quite useful for the planning of campaign communication.

PAGE 31

19 Gender Schemata and Political Knowledge All schemata are knowledge structures; indeed, the hallmark of a schema is knowledge (Hamill and Lodge 1986). Thus, it follows that both gender schema and other political schema are affected by the characteristics of knowledge held by an active processor of schema, such as an evaluator of candidates or voter. The relevancy and consistency of information in relation to existing schema of an individual are a function of the level of expertise or development of the schema of that individual (Lau 1986). Theoretically, if schemata guide the processing of political information, they should also determine the type of information on which candidate evaluations are based (Lau 1986). For example, if a person holds a strong party schema, that schema will most likely form the basis for evaluation of candidates (Iyengar and Ottati 1994; Lau and Sears 1986b); if a person holds a strong issue schema, that will affect the perception of policy stands for a candidate (Lau and Sears 1986b). Similarly, if an individual holds a strong gender schema, the individual's candidate perception will be affected by that schema. The schemata which are most important to an individual are crucial to the information processing activity of that individual (Lau 1986). Individual differences in knowledge about politics-often referred to as political sophistication or expertise-are the most important sources of heterogeneity in political information processing (Emig 1991; Fiske, Lau, and Smith 1990; Krosnick 1990b; Lau and Erber 1985; McGraw and Lodge 1996). Individuals differ in how well their schemata are developed and in how easily their schemata are activated. Experts in political matters should have more elaborate, more structured, more integrated, and more easily activated

PAGE 32

20 schemata than novices in politics (Fiske, Lau, and Smith 1990; Lau and Erber 1985; Lau and Sears 1986b). Researchers have found that people who are political experts process information about a cued schema in a less simplistic, mechanical way than do people who are political novices. Experts have been found to be quicker at processing new information and are better able to recall the information (Krosnick 1990b; Lau and Erber 1985). Having fewer and larger chunks of information enables experts to retrieve information from their schemas more easily and efficiently (Lau and Erber 1985). Because experts have more interconnected schema, more information should be available for candidate evaluations; that is, the information held in the schema of an expert is more well-developed in terms of its utility for candidate evaluation (Lau and Erber 1985). Furthermore, experts remember schema-consistent and schema-inconsistent information equally well, whereas novices remember mainly schema-consistent information (Lau and Erber 1985; Lau and Sears 1986b). For example, experts should be better able to perceive specific information about candidates that is inconsistent with party identification (Lau and Erber 1985). In addition to the speed, organization, and consistency of information, another aspect of candidate evaluation affected by political knowledge or expertise is the significant difference in the number of issues used in evaluating the candidates. Generally, experts use three to five issues, while novices use only one or two (Lau and Erber 1985). Due to the superior "chunking" of information by experts, they are able to use more information in evaluating candidates. Researchers have found that experts are less susceptible to selective misperception and are more deeply embedded in past political socialization. In fact, the development of

PAGE 33

21 political knowledge or expertise can be viewed as a sequential process of schema development or schema elaboration (Lau and Sears 1986b). Political knowledge is conceptualized as information pertaining to political ideology, partisanship, officeholders and candidates, laws of government, and current events (Fiske, Lau, and Smith 1990; Krosnick 1990a, 1990b; Milburn 1991; Zaller 1990). Higher levels of political knowledge are associated with (1) a higher degree of ability to think about politics with multiple, integrative perspectives (Milburn 1991); and (2) a higher ability to organize dissimilar information within the same schematic structure (Fiske 1986). Quite importantly, as people become more expert, they become more aware of inconsistencies between data and schema and more cautious in applying prototypical or stereotypical attributes of a schema to specific cases (Lau and Sears 1986b). Information that is inconsistent with a person' s schema-such as a gender-based schema-would especially be noticed and weigh heavily in the candidate evaluation process. In fact, empirical models of candidate evaluation generally predict that candidate evaluation is primarily determined by schematic processing. Schematic models are valuable in that they make specific predictions and are inherently testable (Lau and Sears 1986b). Gender Schemata and Political Stereotyping in Candidate Evaluation Gender stereotyping has been indicated in previous studies as an important component of candidate evaluation (Bernstein 1996). "Of the various explanations explored for women's disparate numbers as elected officials, the gender lens (Bern 1993) that filters our conception and interpretation of our world, or beliefs about the v natures' of men and

PAGE 34

22 women, merits further consideration in a broadened understanding of the effects of gender on candidate evaluation" (Alexander 1995, 26). Gender stereotyping of political candidates is frequently a matter of perceived issue expertise (Alexander & Andersen 1993; Huddy & Terkildsen 1993; Leeper 1991; Kahn 1992, 1994; and Sapiro 1982) and perceived ability to argue for policies or political views (Matland 1994). Voters have been shown to have differing expectations about the types of issues handled well by female and male politicians. The connection between gender role schemata and issue schemata is important for political information processing (Wicks 1992). For example, the use of specific issue themes in a campaign message is a type of purposive schema activation, priming the voter to think about a candidate in terms of specific issue schemata (Garramone 1992). Studies have indicated that issue priming, as by the media, activates schemata which affect candidate evaluation (Iyengar, Shanto, Peters, and Kinder 1982). Female politicians are persistently portrayed as better able to handle "compassion" issues-poverty, education, and health policy issues~but worse at dealing with perceived "male" issues such as the economy, military, or defense issues (Alexander & Andersen 1993; Leeper 1991; Kahn 1992, 1994; Norris 1997; and Sapiro 1982). Huddy & Terkildsen (1993) state that "differing expectations among voters about the types of issues handled well by male and female politicians have proved the most consistent form of political gender stereotyping." In addition to problems of perceived issue competence, female candidates are often perceived as less competent in their ability to argue for policies or political views (Matland 1994)-a finding consistent with studies addressing the ability to assert oneself and argue effectively for positions in general (Infante 1985; and Nicotera & Rancer, 1994). Huddy

PAGE 35

23 and Terkildsen (1993) further assert that while perceived gender differences in political expertise have been substantially documented, explanation of the phenomenon has received inadequate attention. Political cognition, including candidate evaluation, involves the processing of vast amounts of external stimuli. As with other types of information-processing, individuals tend to use cues, such as gender and race, to assist in the evaluation of candidates (Conover and Feldman 1986: Fiske 1986; Miller, Wattenberg, and Malanchuk 1986). These cues include gender stereotypes, meaning that a candidate's gender elicits memory of certain traits or characteristics that the evaluator associates with a particular gender. In this sense, stereotypes are one type of schema-a cognitive structure that contains a category's attributes and the links among those attributes. As Fiske explains, to respond schematically (such as to evaluate a candidate), a person draws upon organized prior knowledge to aid the understanding of new information. A new person, event, or issue is treated as an exemplar of a pre-established, familiar category or schema (a schema contains attributes most typical of category members, such as female or males). An example of the use of a gender schema in candidate evaluation is the way people thought about Geraldine Ferraro in her run for the vice-presidency. She could have been viewed as an exemplar of a "northeastern urban Democrat, a typical glad-handing politician, a harebrained female, or a pragmatic feminist" (Fiske 1986, 42). Whichever schema a person applied, it enabled that individual to focus on schema-relevant information in forming an impression of the candidate. The use of gender schemata to evaluate a political candidate has important implications for the impression one forms about a candidate's issue positions and competence in dealing with policy decisions. In this

PAGE 36

24 candidate gender-policy integration, an individual is connecting a gender schema with political competence. While most of the previous research has focused on aggregate effects of gender schema on candidate evaluation, there has been at least one study that has also addressed individual differences in gender role schema held by voters. In Alexander's (1995) study on "The Influence of Gender Beliefs in Constructing Candidate Images," it was found that not only were gender role beliefs consequential in the formation of candidate images, but also that individual differences in the type of gender beliefs held were significant. In a twowave panel survey of 98 voters conducted in 1990 in Syracuse, New York, it was found that "egalitarians" favored female candidates and "traditionalists" favored male voters. In other words, individuals with egalitarian gender role schema differed from individuals with traditional gender role schema in their evaluation of candidates. Implications for Female Candidate Evaluation: Constructing the Model Given the relationship between levels of political knowledge and an individual's ability to use a variety of information (including gender) about a candidate, this study proposes the following premises as bases for the hypotheses to be tested: PI: Individuals with higher levels of political knowledge will rate female candidates higher for competence than individuals with lower levels of political knowledge. Given that higher levels of political knowledge are associated with (1) more frequent use of mass media in general; (2) higher use of print media than electronic media; and (3) better ability to process schematic inconsistencies; and (4) less reliance on gender stereotypes, this study proposes the following premise as a basis for hypotheses to be tested:

PAGE 37

25 P2: Individuals who use print media at a higher frequency will rate female candidates higher for competence than individuals who use print media at a lower level of frequency. Finally, given that Alexander's study found individual differences in how a gender role schema affects candidate evaluation, the following premise is provided as a basis for hypothesis to be tested: P3: Egalitarian voters appear to favor female candidates, while traditional voters appear to favor male candidates. Gender and Political Credibility: A Function of Gender Schema and Stereotyping This section explores research concerning the relationship between gender schemas and candidate evaluation within a meta-analytic context; that is, previous research (Bernstein 1996) has been reviewed through a meta-analytic process in order to systematically synthesize previous empirical results (Table 1). Research has indicated that women and men are often rated unequally for credibility despite equal performance (Hodgson & Pryor 1984; Kenton 1989; and Williams 1984). Likewise, research also indicates that people differentiate between identically qualified male and female candidates (Eckstrand & Eckert 1981; Kahn 1992, 1994; Leeper 1991; Sapiro 1982; and Sigelman & Sigelman 1984). This inequality has serious repercussions for women-especially those in disadvantaged groups who hope to gain status and improve their group's position in society. If there is indeed a gender gap in credibility, it will be difficult for women to achieve equal opportunity in business, in education, and in politics. It will be

PAGE 38

26 difficult for them to aspire to levels of authority and power proportionate to their existence in our society. This concern for gender equality is particularly problematic in the political domain, where underrepresentation still widely exists. Gendered Credibility: A Meta-Analytic Review Studies dealing with gender differences in candidate evaluation either manipulate gender to test the effects on voting or willingness to vote for a candidate, or they focus on whether male or female candidates are perceived to be more competent in dealing with specific policy areas (Bernstein 1996; King and Matland 1997). These approaches to studying gender differences in candidate perception are consistent with the general use of gender stereotypes in various social settings subscribed to by social psychologists (King and Matland 1997). Study Selection Criteria. The literature search strategies for the meta-analytic review (Bernstein 1996) consisted of (1) a computer search of abstract databases, including PsycINFO, ERIC, Dissertation Abstracts International, and the INFOTRAC General Academic Index; (2) a computer search of the Social Sciences Citation Index; (3) references or footnotes in books and articles written by researchers in the field [Cooper & Hedges (1994) refers to this technique as "footnote and reference chasing"]; and (4) browsing through library shelves. Given time constraints, a decision was made to include published works only; therefore, references included in ERIC or Dissertation Abstracts International were used only for the purpose of finding published works. Computerized databases were searched using the following key terms: gender, women, politics, credibility, competence, expertise, communication, public opinion, and persuasion.

PAGE 39

27 In order to review articles relevant to the observed relationship between candidate gender and perceived competence, articles were retrieved which included gender of candidate as the independent variable and one or more measures of perceived competence as the dependent variable(s). A total of 13 articles were found which reported the candidate gender-competence relationship in a statistical format adaptable to a meta-analytic review. Given the modest number of studies, two additional studies were included pertaining to the candidate gender-willingness to vote for male/female relationship, based on the assumption that an individual's "willingness to vote for" could be used as a surrogate measure for perceived competence of a candidate. Of the 17 studies selected addressing candidate gender in relation to perceived competence, four studies included general competence only, nine studies included specific policy area competence only, two studies included willingness to vote for male/female candidates, and two studies included a combination of general competence and specific policy area competence measures. Overall, 13 research reports were found, a collection which yielded 17 empirical studies. Of these studies, 13 were based on experiments and four were based on survey data. Criteria for including studies in the meta-analysis were: (1) the independent variable was gender of candidate (or in two cases, communicator); (2) the dependent measure(s) were perceived policy competence, perceived general competence, or willingness-to-vote for a male/female candidate; and (3) the candidate/communicator evaluation was based on speeches (actual or hypothetical) or descriptions of a candidate. Studies selected could include one dependent measure or multiple dependent measures. For each study, the relationship between candidate gender and competence was analyzed using a statistic which

PAGE 40

28 could be transformed to correlation coefficient r for the purpose of meta-analytic statistical procedures. In order to make comparisons among studies, the following variables were coded for each study selected: author(s), source of publication, year of publication, type of sampling, sample size, type of subjects or participants, method as experimental or survey, dependent variable(s) measured, level of political race, mode of communication, types of effect sizes reported, magnitude of effect sizes, and direction of effect. Sources of publication were journal articles or books; types of sampling were random or convenience; types of subjects were college students, high school students, voters, or general public; levels of political race were national, state, or local; modes of communication were speech or description; types of effect sizes reported were population percentages or proportions, F statistic, t test, or correlation coefficient; direction of effect included either "males perceived more competent" or "females perceived more competent. " Study Attributes . A brief description of key attributes of each study used in the meta-analysis provides a sense of the nature of the research which has addressed the candidate gender-perceived competence relationship. Of particular interest are (1) the types of dependent variables used; and (2) the types of questions asked about the "candidates." In the Alexander and Andersen (1993) study, the researchers examined voters' perceptions regarding the issue or policy area competence of candidates for national and state office. Candidates were rated on competence in the following policy areas: agriculture, defense, foreign trade, taxes, environment, economy, education, and health. The types of questions asked were: "Tell me whether you think the man or the woman [candidate] would, most of the time, do a better job dealing with the issue when in office. "

PAGE 41

29 In the Brown et al. (1993) study, the researchers examined voter attitudes toward male and female candidates for entry-level city council positions. Candidates were rated on the following policy areas: youth, health, race, recreation, parks, economic development, business development, garbage, police, traffic, streets, and fire service. Questionnaire items included: "City government is really made up of a number of different service areas, such as fire, police, health, etc. I am going to read you a list of these services. Do you think men or women on City Council are more capable of dealing with the following service areas..." (Brown et al 1993, 10). Responses were requested in terms of selecting either women or men as more capable. In the Eckstrand and Eckert (1981) study, the researchers examined voters' willingness to vote for one of two candidates-one male and one female-hypothetically running for governor. Candidate sex was implied by name of candidate. Subjects were given candidate profiles and issue statements to read. The statements pertained to three issues: environment, unemployment, and taxation. Subjects were asked to "vote" for one candidate or the other. In the Hedlund et al. (1979) study, the researchers examined voters' likelihood of voting for a hypothetical candidate for either school board or judicial office. Question wording was "If die candidate for (office) is a female, would you be more likely to vote for this person or less likely to vote for this person?" (Hedlund et al. 1979). In die Huddy and Terkildsen (1993) study, the researchers investigated the origins of voters expectations for candidate competency on a variety of policy issues. Policy areas included military/police, economy, compassion, and women's issues. Question wording

PAGE 42

30 was "How well would Ms. /Mr. McGuire handle each of the following issues: Very well, somewhat well, not very well, not well at all?" (Huddy and Terkildsen 1993). In the Kahn (1992) study, the researcher investigated the evaluations of hypothetical candidates for the U.S. Senate in relation to their gender. Policy issues for which candidates 's competence were evaluated included education, health, women, economy, farm, and military. Question wording was "What is your best guess about (candidate's name, male or female) competence in dealing with military issues?" (Kahn 1992). Competence in other policy areas was worded the same way, with a 7-point, respondent scale from "very competent" to "very incompetent." Kahn's 1994 study was similar to her 1992 study, with the addition of ratings for a hypothetical gubernatorial candidate. In the Leeper (1991) study, the researcher investigated evaluations of candidate competence for a gubernatorial candidate. Subjects were asked to read a speech by the hypothetical candidate who they were told was either "Jim Preston" or Elizabeth Preston. " Policy areas rated included education, assisting the poor, health, consumer protection, economy, crime, big business, unions, and farm issues. Question wording included: "Based on your impressions drawn from the speech, please provide your best guess as to the candidate's competence regarding the following issues:" (as listed). Subjects were also asked "How well do you think Mr /Mrs. Preston understands the issues discussed in the speech?" "How clear are Preston's stands on the issues?" and similar types of questions. In the Matland (1994a) study, the researcher tested for "gender schemata" in subjects' evaluation of a hypothetical candidate regarding competence in the following policy areas: defense, agriculture, foreign policy, economy, tax policy, unemployment, health care, transportation, education, elderly care, environmental policy, child care policy,

PAGE 43

31 and women's rights. Subjects were asked to rate either a male candidate (by name) or a female candidate (by name) on a 7-point scale from "very competent" to "incompetent" (Matland 1994). In the Nicotera and Rancer (1994) study, the researchers investigated sex differences in self-reported and generalized levels of argumentativeness and verbal aggressiveness as competence traits. Question wording was not provided. In the Rosenwasser et al. (1987) study, the researchers investigated the perceived competence of hypothetical male and female candidates for the office of president. Policy areas included terrorism, military, education, racial problems, care of the disabled and handicapped, and elderly care. Based on a brief description of the candidate, participants were asked to rate on a 5-point scale how effective they believed the candidate would be if he/she became President of the United States (Rosenwasser et al. 1987). In the Sapiro (1982) study, the researcher investigated the perceived competence of a hypothetical candidate (male or female) in the following policy areas: military, economy, assistance to the poor, big business, education, environment, consumer interests, health, crime, labor unions, and farm issues. Respondents were asked to rate the candidate's competence, using a 7-point scale ranging from "very competent" to "very incompetent" Sapiro 1982). In the Sigelman and Sigelman (1984) study, the researchers examined the effect of gender, race, and age on voting behavior. Participants were given randomly-assigned "ballots" describing two candidates for Mayor and were instructed to "vote" for one. Five experimental candidates were matched in random pairs and were described as middle-aged white female, middle-aged black female, a thirtyish white male, a middle aged black male,

PAGE 44

32 or an elderly white male. Candidates were assigned careers as either lawyers or business executives. Question wording was not provided. Finally, the Whittaker and Meade (1967) study investigated cross-cultural attitudes regarding gender and communication competence. Subjects listened to recorded speeches regarding the United Nations with either a male speaker or a female speaker. Question wording included: "Did the speaker do a good or a poor job in giving the facts on the UN?"; "Were the conclusions justified by the facts?"; and "Was the speaker convincing?" (Whittaker and Meade 1967). It is interesting to note that in some studies gender was implied by the candidate/speaker's name, while in other studies, the candidate/speaker's gender was overtly stated. Perhaps, the level of gender-cueing has some effect on perceived competence rating. In the studies for which gender is overtly stated, there may be increased effect calling for a socially desirable answer, thereby biasing the responses regarding competence. Future research could address this issue. Effect Sizes and Competence Categories. The effect size estimate for each study was an estimate of the correlation between candidate/communicator gender and perceived competence. Effect sizes were calculated as correlation coefficients. Effect sizes were transformed from population percentages, f-statistics, or F-statistics, and were adjusted for bias associated with small sample sizes. Results of MetaAnalysis For the overall analysis, the mean effect size across all studies, weighted for sample sizes, was r=(-) 0.03. The negative sign indicates that, overall, males were perceived as

PAGE 45

33 more competent, but the magnitude of the overall effect size is small. Several variables appear to be moderating and counterbalancing the overall effect. These variables include subject population, category of competence area, and level of political race. Of particular interest, effect sizes for specific policy competence areas were significantly larger. Students comprise 71 % of die subjects across all 17 studies. The remaining 29% were members of the general public. A data analysis of the two subject populations indicated diat the mean effect size for students, when weighted by sample size, was r= (-) 0.08, and the mean effect size for general public, when weighted for sample size, was r= 0. 12. A comparison of the two groups of the subject population, using Rosenthal's (1991) formula for comparing effect size estimation between two groups, indicates a significant difference between the effect sizes (p_ <_ .001). In other words, the type of subjects is interacting with the relationship between gender and perceived competence of candidates. There were seven studies which indicated that, on average, females were perceived more competent, while males were perceived as more competent for ten studies. Of the seven studies favoring females, three were based on national races, two were based on state races, and two on local races. Of the ten studies favoring males, four were national races, two were for local races, and four were non-specific. Of the four non-specific treatments, three pertained to speeches regarding the importance of the United Nations, and a fourth was generalized to speeches focusing on verbal assertiveness as a competence measure. Gender effects for perceived competence were found for several competence categories. Competence in agricultural policies was addressed in four studies, competence in defense/military policies was addressed in eight studies, economy in eight studies, education in eight studies, health in nine studies, and general competence in nine studies.

PAGE 46

34 Mean effect sizes (r) were agriculture = (-) 0.09, defense = (-) 0.11, economy =0.05, education =0.19, health =0.18, and general competence =(-) 0.17. For direction of effect, positive is for "females perceived more competent" and negative is for "males perceived more competent. " As indicated in Table 2, when comparing the defense policy area with the education policy area, the defense area is perceived as a male domain (r=(-) 0. 1 1), while the education policy area is perceived as a female domain (r=0. 19). When the two policy areas are compared across all relevant studies, a significant difference for gender is found at g j< .01. Furthermore, when comparing two sets of policy areas-that is, comparing the combination of agriculture and defense as the "male" policy areas with the combination of education and health as the "female" policy areas-across all relevant studies, a significant difference for gender is found at the p_ j< .01 level. As predicted in previous research and based on the meta-analytic review (Bernstein 1996), certain policy areas are considered as typically "male", while others as typically "female. " Agriculture and defense are considered as policy areas in which males are perceived as more competent, while education and health are considered as policy areas in which females are perceived as more competent. Furthermore, when comparing the two sets of policy areas-one in which males are favored and the other in which females are favored-the gender difference is found to be statistically significant. That is, not only is the agriculture-defense policy set male-favored and the education-health policy set femalefavored, but also the difference between the two policy sets in perceived competencies by gender is statistically significant atp_ _< .01.

PAGE 47

35 Implications for Female Candidate Evaluation: Gender and Competence Schema What do the previous research results imply for male and female candidates? The research suggests that male candidates may need to make an extraordinary effort to be perceived as competent in health and education policy areas, while female candidates may need to make an extraordinary effort to be perceived competent in agriculture and defense policy areas. Given the significant underrepresentation of females in Congress, female candidates who want to overcome gender stereotypes and bias with regard to agriculture and defense issues may need the support of organizations such as the National Women's Political Caucus to become proficient in mastering discussion of these policy areas. Given the results of the meta-analytic review of previous empirical research regarding the candidate gender-perceived competence relationship, the following premises are proposed as bases for the hypotheses to be tested: P4: Female candidates are generally rated as more competent than male candidates in dealing with education-a stereotypically "female" policy area. P5: Male candidates are generally rated as more competent than female candidates in dealing with crime-a stereotypically "male" policy area. Message Schematics and Strategies for Female Candidates People develop normative expectations about the appropriateness of communication behavior that differ for males and for females, and when these expectations are violated, the result may be a decrease in persuasiveness (Burgoon 1983). Although a female candidate's speech is generally evaluated as less credible than that of a male candidate, women may be

PAGE 48

36 penalized when they adopt a more masculine style (Johnston and White 1994; Trent and Freidenberg 1995). This is what Ruth Mandel, former director of the Eagleton Institute Center for Women and American Politics at Rutgers University, has long referred to as the "double bind" (Jamieson 1995; Johnston and White 1994; Sullivan and Turner 1996) women running for office must face. The double bind for female candidates is that if they're perceived as too feminine, it damages their perceived competence, and if they are perceived as competent, they don't conform to the expected norms of femininity (Wadsworth et al. 1987). In spite of this obstacle to women's electoral success, very few studies have attempted to identify strategies that might be effective for female candidates (Johnston and White 1994). At least one study suggests that message tones that either reinforce traditionally feminine traits, such as warmth and compassion, or that emphasize traditionally masculine traits, such as aggression, are best avoided by female candidates (Hitchon et al. 1997). Instead, the researchers find that using a neutral tone— that is, emphasizing issues or voting record— is more beneficial to the female candidate. The neutral-toned message is less likely to cue stereotypical associations and can facilitate communication of the candidate's worth (Hitchon et al. 1997). Another approach, recommended by Sullivan and Turner (1996), is to use a careful blend of masculinity and femininity in political messages, thereby showing they are both competent and nurturing. How do specific issue or policy areas attributed as important message themes for political office interact with and influence the relationship between candidate gender and perceived candidate competence? Previous research indicates that certain policy areas are considered as typically "male", while others are considered typically "female." Agriculture

PAGE 49

37 and defense, for example, are considered as policy areas in which males are perceived as more competent, while education and health are considered as policy areas in which females are perceived as more competent. With regard to message characteristics, some studies have indicated that message theme, such as political issues emphasized, can influence the use of gender stereotypes or cues in candidate evaluation (Huddy and Terkildsen 1993; Leeper 1991). Leeper found that the "tough" message-which he defined as one which is unambiguous in the candidate's issue positions and that emphasizes "masculine" policy areas such as the economy and crime-can minimize the use of gender cues in evaluation of candidates. He suggests that if a female candidate is ambiguous in her campaign themes and issue stands, the voters may infer that she has feminine traits with regard to her competence, and that she could be adversely affected in her chances for electoral success. Leeper's Research Leeper 's research on gender prejudice regarding female candidates is focused on (1) gender differences in candidate evaluation based on candidate attributes, such as masculine and feminine traits; and (2) the usefulness of what he calls the "tough" message as employed by female candidates in political campaigns. In contrast with researchers who assert that female candidates should capitalize on stereotypically "female" issues in their campaign messages (Iyengar et al. 1997; Kahn 1991), Leeper proposes the use of stereotypically "male" issues as an electoral advantage. He conducts an experiment with college students in an introductory political science course, asking them to evaluate hypothetical candidates on the basis of a written speech. One-half of the students are given a speech attributed to "Jim Preston," a hypothetical candidate, and one-half are asked to

PAGE 50

38 evaluate a speech attributed to "Elizabeth Preston, " also a hypothetical candidate. The students were told that they were participating in a study which would be used to measure information gleaned from campaign speeches, and they were asked to assess the candidate's competence based on their perceptions as drawn from the speech. The campaign speech contained what Leeper refers to as "tough," masculine messages related to crime and the economy. While the perception of the economy as a policy issue has been somewhat gender-neutral (Bernstein 1996; Huddy and Terkildsen 1993), crime is frequently viewed as a masculine policy area (Huddy and Terkildsen 1993; Leeper 1991; Trent and Freidenberg 1995). In some campaigns, female candidates have been challenged as to whether they are strong enough to deal with the issue of crime (Jamieson 1995); indeed, women are presumed to be soft on crime (Witt, Paget, and Matthews 1994). In recent years, crime has been an important issue for the American public, and during the 1994 Congressional elections, a significant obstacle to the campaigns of most women was the issue of rising crime (Trent and Freidenberg 1995). Leeper refers to other researchers-for example, Rosenwasser and Seale-who have said that women may be more successful in running for office when they are perceived as having masculine attributes, but that there is a possibility that "masculine" women could be penalized for violating traditional norms. Voter expectations based on gender create problems for female candidates in that stereotypes are troublesome in their potential for a backlash effect in the campaign efforts of those women whose communication challenges normative standards (Trent and Freidenberg 1995). Backlash may occur when a female candidate exhibits so-called masculine communication behavior and thereby violates the traditional view of women as deferential, soft, and feminine (Trent and Freidenberg 1995).

PAGE 51

39 Leeper cites as the purpose of his study to investigate whether voters actually look to masculine women with disfavor and punish them at the polls. He continues the convention of calling the so-called "feminine" traits of being passive, emotional, and caring, as those which are less desirable than "masculine" traits of being active, aggressive, and competent. He also investigates whether gender-specific inferences are made when the candidate clearly uses masculine themes in her message. In terms of his operational ization of the "tough" message, he uses a speech which he describes as being a display of "toughness" and somewhat "masculine in demeanor" in that he emphasizes "masculine" policy areas. Message toughness is conceptualized as message characteristics of (1) themes of traditionally male policy areas; and (2) clarity of issue positions (Leeper 1991). By juxtaposing toughness and masculinity, he conjures up a gender-stereotyped description as a means for evaluating the candidate. He describes the hypothetical, female candidate as "not shrouding herself in perceived, stereotypical, feminine themes." He asks of his subjects, "How well do you think Mrs. Preston understands the issues discussed in die speech?" Leeper reports in his findings that he has established evidence that voters may infer that "tough, aggressive women" are generally as competent as male candidates, and that female candidates do not have to emphasize traditional "female" areas of strength. "It appears," he says, "that female candidates can employ tough messages without being rebuked by startled voters." (Leeper 1991, 254). He also finds that voters can even infer "latent (stereotypical) warmth" in tough, aggressive women. For traditionally female policy areas not mentioned in the candidate's speech, such as healtii and education, respondents rated the female candidate more competent.

PAGE 52

40 For the issue of crime, traditionally perceived as a masculine policy area, the female candidate was able to "pull even" with the male candidate by using the "tough" message in her speech (Leeper 1991). Leeper concludes that voters are just as willing to accept a woman's tough themes as they are a man's. Leeper leaves us with the suggestion that future research should focus on "what sort of cognitive baggage voters persist in bringing to the polls, and how voters respond to various types of female candidates with different kinds of messages" (Leeper 1991, 257). The Gender-Based Message Debate Campaign managers for female candidates consistently focus on messages containing stereotypical "female" issues more frequently than do the campaign managers for male candidates (Alexander and Andersen 1993; Kahn and Gordon 1997). According to the authors, women can capitalize on common stereotypes by emphasizing issues where voters believe they are more competent— environment, health, education (Iyengar, Valentino, Ansolabehere, and Simon 1997; Kahn and Gordon 1997). Based on their studies, Iyengar, et al. (1997) and Kahn and Gordon (1997) suggest that candidates are best advised to emphasize issues that resonate gender stereotypes. Their premise is that campaign communication is most persuasive when it conforms to voters' previous dispositions (Iyengar et al. 1997). For example, they suggest that people need to infer that women are less supportive of stronger criminal sentencing, because "if voters link message content with a candidate's gender to inform their vote choices, then we expect support for candidates to be highest when they emphasize issues which resonate with gender stereotypes" (Iyengar et al. 1997, 80). Specifically, in Iyengar et al.'s study, they found

PAGE 53

41 that when U.S. Senatorial candidates Barbara Boxer and Dianne Feinstein campaigned on gender-consistent issues the results were positive, but when they emphasized the issue of crime, they experienced little or no advantage. Their studies found that when their advertisements were concerned with so-called women's issues, their vote share increased more than double the increase from an advertisement concerned with crime. For the studies included in the meta-analysis, most relied on descriptions of candidates to relay information relevant to evaluating the candidate. Only those studies conducted by Leeper, Matland, Sapiro, and Whittaker and Meade contained experiments that used speeches as the mode of communication. Of these, Leeper and Sapiro used the message as a key focus of the studies, with Leeper using the "tough" message and Sapiro using a vague message about the economy with no policy stands expressed. Only Matland used more than one form of speech, with variability based on partisanship; his study indicated a slightly more favorable evaluation for female candidates by those who read the speech of the liberal party over those who read the speech of the conservative party. However, he had liberals reading the liberal speech and conservatives reading the conservative speech, so the speeches conformed to predispositions of the readers. In any case, for both forms of the speech, Matland found gender-based stereotyping in the evaluations, with females found more competent in traditionally female policy areas, and males found more competent in traditional male policy areas. Implications for Female Candidate Evaluation: Developing a Message Strategy There is an obvious difference of findings and conclusions among researchers addressing the message effects related to gender stereotypes for political candidates. Some

PAGE 54

42 researchers support the strategy of female candidates using messages that expose voters to the candidate's stereotypical strengths; others believe that female candidates are caught in the double bind of being perceived more competent when emphasizing male policy areas, but that they will be damaged by a backlash effect for violating sex role expectations. Leeper, on the other hand, found that female candidates can benefit from using counterstereotypical messages. While Leeper ignores individual differences of the receiver as mediating factors in gender stereotyping, Iyengar, et al, have pointed to possible confounding effects related to a variety of contextual (e.g. political environment) and individual-difference factors (e.g. partisanship and respondent gender). None of these researchers, however, appear to have acknowledged variations in political knowledge as an important difference that can affect individuals' use of gender schemata. Further research is needed to determine the most effective ways for female candidates to present themselves as equally or more competent than their male counterparts. One approach is to study message characteristics and the interaction with gender effects. If we are to achieve parity in gender representation at the national level, we must break through die barriers of gender stereotypes by voters and the media, while also increasing women's access to political resources. Strategies for neutralizing gender barriers are important. The female candidate who is perceived as highly competent in a variety of policy areas, even the stereotypically male policy areas, undermines negative stereotypes. The challenge is for female candidates to assert themselves as competent in traditionally male policy domains, such as military defense and crime, without sacrificing their positive qualities in addressing traditionally female policy areas, such as education and health.

PAGE 55

43 Based on Leeper's findings regarding the "tough" message, the following premise is proposed as a basis for hypotheses to be tested: P6: When female candidates use a "tough" message, they are perceived as more competent than when they use a "soft" message. Summary of Theoretical Foundations The theoretical foundations derived from the review of previous research are summarized here according to the theoretical benchmarks of conceptual definitions, gaps and conflict in prior research, and laws and assumptions from prior research. These conceptual foundations provide a basis for further empirical investigation regarding the gender-perceived competence relationship. From these benchmarks, a number of hypotheses are proposed for testing as described in Chapter Three. Conceptual Definitions of Key Variables Based on the literature reviewed, the gender-perceived competence relationship has been investigated in terms of the following key variables: The independent variables are candidate gender, message theme, message clarity, gender role schema, political knowledge, and media exposure. The dependent variable is competence. Independent variables Candidate gender is conceptualized as sex of the candidate; that is, male or female. Message theme is public policy issue that is stereotypically male or female. Message clarity is specificity of issue positions. Gender role schema is a knowledge structure based on sexlinked associations with gender roles in society. Political knowledge is information

PAGE 56

44 pertaining to political matters, such as political ideology, partisanship, officeholders and candidates, laws of government, and current events. Media exposure is types and frequency or amount of media viewed, heard, or read. Dependent variable. Perceived competence is conceptualized as expertise, knowledge, experience pertaining to politics, as perceived by the evaluator. Gaps and Conflict in Prior Research 1. Previous research on gender and candidate evaluation has not integrated the variables of candidate gender, message themes, political knowledge of voters/respondents, and media exposure with regard to the effects on candidate evaluation. This study proposes a gender-based schematic theory that integrates these components /determinants of candidate evaluation. 2. Few studies have identified message strategies for female candidates based on empirical research, and the few studies that have addressed message strategies pertaining to "male" and "female" issues have conflicted. Laws/ Assumptions from Prior Research 1. People are cognitive misers, who use schemata to assist in encoding, storing, retrieving, and interpreting incoming information. 2. Old schema (e.g. gender schema) influence new schema (e.g. candidate schema). 3. Political knowledge, including political schema, is derived primarily from the mass media.

PAGE 57

45 4. Higher levels of political knowledge are associated with the use of print media, while lower levels of political knowledge are associated with electronic media usage. 5. Increased political knowledge is associated with greater complexity of schemata used in candidate evaluation. 6. Political experts (individuals with higher levels of political knowledge) are better able to process schematic inconsistencies and to organize dissimilar information within a schematic structure; therefore, they are less reliant on gender stereotypes when assessing a candidate. 7. When using gender schema in candidate evaluation, individuals associate gender with issue expertise and perceived competence. 8. Male candidates are generally perceived more competent in dealing with traditional "male" policy areas and female candidates are generally perceived more competent in dealing with traditional "female" policy areas. 9. Message themes, such as the political issues emphasized, can influence the use of gender stereotypes in perceived candidate competence. 10. The policy area of crime is considered a traditional "male" policy area. 1 1 . Female candidates can use "tough" messages (that is, consisting of stereotypically "male" policy areas and clear positions on the issues) and be rated equally with male candidates who use a "tough" message. 12. Voters with egalitarian gender role beliefs (schema) are more likely to favor female candidates over male candidates, while voters with traditional gender role beliefs appear to favor male candidates.

PAGE 58

46 Premises as Bases for Hypotheses Based on the assumptions derived from previous research, the following premises are proposed as bases for the hypotheses to be tested: PI. Individuals with higher levels of political knowledge will rate female candidates higher for competence than individuals with lower levels of political knowledge. P2. Individuals who use print media at a higher frequency will rate female candidates higher for competence than individuals who use print media at a lower level of frequency. P3: Egalitarian voters appear to favor female candidates, while traditional voters appear to favor male candidates. P4. Female candidates are generally rated as more competent than male candidates in dealing with education-a stereotypically "female" policy area. P5: Male candidates are generally rated as more competent than female candidates in dealing with crime-a stereotypically "male" policy area. P6: When female candidates use a "tough" message (that is, consisting of stereotypically "male" policy areas and clear positions on the issues), they are perceived as more competent than when they use a "soft" message (that is, consisting of stereotypically "female" policy areas and unclear positions on the issues). Premises one and two provide a basis for testing associations of political knowledge or media use with the gender-message relationship. Premise three provides a basis for testing individual differences in gender role schema and their effect on candidate evaluation. Premises four and five provide a basis for the "main effects" between candidate gender and

PAGE 59

47 perceived competence. The sixth premise provides a basis for testing any interactions among gender, message theme, and message clarity. Hypotheses Comparing female candidates to other female candidates or to themselves HI. When female candidates use a message on crime (stereotypically male issue) that has a clear (unambiguious) position on the issue, they are perceived as more competent than when they use a message on a crime that has an unclear position (premise 6). H2. When females use a clear message on education (stereotypically female issue), they are perceived as less competent than when they use a clear message on crime (premise 6). H3. When female candidates use a n unclear message on crime , they are perceived as less competent than when they use a clear message on education (premise 6). H4. When female candidates use a clear message on crime , they are perceived as more competent than when they use an unclear message on education (premise 6). Comparing female candidates to male candidates: H5. When female candidates use a message on crime that has a clear p osition on the issue, they are perceived as less competent than male candidates who use a message on crime that has a clear position (premise 5). H6. When female candidates use a message on crime that has an unclear position on the issue, they are perceived as less competent than male candidates who use a n unclear message o n crime (premise 5). H7. When female candidates use a clear message o n education , they are perceived as more competent than male candidates who use a clear message o n education (premise 4).

PAGE 60

48 H8. When female candidates use an unclear message on education , they are perceived as more competent than males who use an unclear message on education (premise 4). H9. When female candidates use a clear message on crime , they will be rated as more competent by people with higher levels of political knowledge than by people with lower levels of knowledge (premise 1). H10. When female candidates use a clear message on crime , they will be rated as more competent by people who use print media more frequently than by those who use it less frequently (premise 2). HI 1 . Individuals with egalitarian gender role schema favor female candidates, while individuals with a traditional role schema favor male candidates (premise 3).

PAGE 61

49 Table 2.1: Studies Included in the MetaAnalysis Study Alexander & Andersen (1993) Brown, Heighberger (1993) Hedland, Freeman, Hamm, & Stein (1979) Huddy & Terkildsen(1993) Kahn(1992) Kahn(1994) Leeper(1991) Matland(1994) Nicotera & Rancer (1994) Rosenwasser, Rogers, Fling, Silvers-Pickens, & Butemeyer(1987) Sapiro(1982) Sigelman & Sigelman (1982) Whittaker & Meade (1967) Subject Method population voters voters voters students survey survey survey students experiment general public experiment general public experiment students experiment students experiment students survey students experiment students experiment students experiment experiment Direction of effect f>m f>m f>m m>f f>m f>m f>m m>f m>f f>m m>f a)m>f b)m>f a)m>f b)m>f c)m>f m>fmales perceived more competent f>m = females perceived more competent

PAGE 62

50 Table 2.2: Comparing Policy Areas Across Studies defense r n (.06) 217 (.07) 98 (.14) 297 (.16) 209 (.16) 143 (.17) 286 (.21) 204 .08 117 education r n (.01) 204 .06 143 .08 209 .15 286 .19 217 .23 142 .39 117 .42 98 total n 1571 1416 n weighted mean r = (-)0. 1 3 = 0.16 comparison of two policy areas (z=8.03, g <.0001) agriculture health r n r n (.03) 98 .00 209 (.12) 217 .15 286 (.12) 204 .18 142 (.17) 143 .19 401 .09 209 .20 143 .22 217 .31 117 .42 98 (.04) 204 total n 871 1817 n weighted mean r = (-)0. 10 = 0.19 comparison of two policy areas (z=7.44, p <.0001) ( ) indicates males perceived more competent

PAGE 63

CHAPTER THREE METHODS Introduction The primary research in this dissertation comprises an experimental study of 209 undergraduate students at the University of Florida. The central focus of the experiment was the testing of eight campaign messages for two hypothetical candidates based on Mark Leeper's (1991) concept of the "tough" message. The expansion of Leeper's concept in this research represents both a theoretical and empirical expansion of his study, as well as the studies described in the meta-analytic review discussed in Chapter Two. Incorporating the concepts and methodology used in previous research regarding the relationship between candidate gender and perceived competence contributes to the body of research in this area by increasing the understanding of theoretical and operational constructs and relationships, thereby building on the scientific work that has been conducted previously (King 1995). It enables a researcher to determine whether generalizations supported under one set of conditions are also supported under others (Herrnson 1995). The theoretical construct that Leeper proposes as central to his research-message toughness-is an unique and interesting offering to the field of candidate evaluation. However, his methodology is limited both in the conditions tested and the data analysis techniques used. His experiment has been replicated in terms of key constructs and basic 51

PAGE 64

52 experimental approach; however, several components have been revised, modified, or expanded (as depicted in Table 3.1); most importantly, (1) the number of independent variables have been expanded; (2) the dependent variable of interest is limited to "perceived competence" of the hypothetical candidate (vote expectations and intentions will be used in future studies); (3) the treatment component has been expanded; and (4) the data analysis has been expanded. Table 3.1: Comparison of Leeper's and Bernstein's Studies Leeper Bernstein Subjects Dependent Variables Treatment Variables Subject Variables college students (n=142) 1) competence x issue 2) likelihood of vote 3) believed likelihood of win 1) candidate gender none Data Analysis r-test college students (n=209) 1) competence x issue 1) candidate gender 2) campaign message theme 3) campaign message clarity 1) subject gender 2) gender role schema 3) political knowledge 4) media exposure ANOVA, r-test, correlation

PAGE 65

53 In order to collect data regarding the influence of policy area focus and other message characteristics on the gender-credibility relationship, an experiment has been conducted encompassing a survey of attitudes regarding hypothetical candidates for Congress. The dependent variable of interest is "perceived competence of candidate" as applied to nine different policy areas. The independent variables include candidate gender, message characteristics, and subject characteristics. Operational Definitions of Key Variables Based on the literature reviewed, the gender-perceived competence relationship will be investigated through hypothesis testing utilizing the following variables. The independent, manipulated variables are candidate gender, message theme, and message clarity; and the subject variables of gender role schema, political knowledge, and media exposure. The dependent variables include perceived competence of the candidate in each of the nine policy areas-the arts, business, crime, economy, education, farm, health, poverty, and employee unions. Manipulated Variables Candidate gender is the sex of the hypothetical candidate; that is, male or female. Each of the campaign speeches used as a treatment condition was identified as a speech by either Elizabeth Preston or Jim Preston. Message theme is the issue that characterizes the political theme of the message and that is stereotypically male/masculine or female/feminine. Each of the campaign speeches used as a treatment condition contained either a message on crime or a message on education. Message clarity is the unambiguity

PAGE 66

54 or specificity of issue positions. Each of the campaign speeches contained a message that was either clear or unclear with respect to the issue of either crime or education. Subject Variables Gender role schema is a sex-linked association related to gender roles measured by a gender role ideology scale that was used to determine an individual's gender role schema in terms of egalitarian, moderate, and traditional beliefs. Examples of items asked of participants in the study are: Do you agree or disagree that "It is more important for a wife to help her husband's career than to have a career herself; "Most men are better suited emotionally for politics than are most women"; "Women should be able to compete with men for jobs that have traditionally belonged to men, such as firefighters"; and "It is a good idea for a husband to stay at home and care for the children while his wife is employed outside the home." Questions used for the gender role ideology scale were taken from Alexander's (1995) study and Alexander and Andersen's (1993) study, in which the reliability coefficient (Cronbach's alpha) was found to be 0.79, and a principal component factor analysis of the variables extracted only one factor, supporting the unidimensionality of the scale. The six-item scale used by Alexander and Andersen was derived from the 36-item, sex-role orientation scale developed by Brogan and Kutner in the 1970s (Alexander and Andersen 1993; Brogan and Kutner 1976). A second gender role beliefs scale 1 is the Attitudes Toward Women Scale developed by Spence and Helmreich in the early 1970s (McHugh and Frieze 1997; Spence and Hahn 1997). This scale has been widely used in recent decades to ascertain

PAGE 67

55 individuals' beliefs and attitudes toward women's rights. The original version consisted of 55 items, but a more recent version consists of 15 items (Spence and Hahn 1997). Sample items from this scale include: "Under modern economic conditions, with women active outside the home, men should share in household tasks such as washing dishes and doing laundry," "Women should worry less about their rights and more about becoming good wives and mothers," "Women should assume their rightful place in business and all the professions along with men," "Economic and social freedom is worth more to women than acceptance of the ideal of femininity, which has been set up by men," and "Women should be given equal opportunity with men for apprenticeship in the various trades" (Spence and Hahn 1997, 24). Another widely used gender role beliefs and attitudes scale is one constructed by King and King in the 1980s (King and King 1990 and 1997; McHugh and Frieze 1997), which has been referred to as the "Sex-Role Egalitarianism Scale." This scale has been tested extensively for construct validity and reliability on both college students and heterogeneous population (King and King 1990; King et al. 1994). The nature of the question items included in this scale are similar to that of the Alexander and Andersen scale, but the test is administered in either 25-item or 96-item forms. Sample questions include: "The husband should be the head of the family," "It is more appropriate for a mother, rather than a father, to change the baby's diapers," "Women can handle job pressures as well as men," and "It is wrong for a man to enter a traditionally female career." 1 In the literature regarding gender roles, the terms beliefs, attitudes, and schemas are frequently used interchangeably. Refer to Ashmore, et al. (1986), Bern (1981), and McHugh and Frieze (1997) for examples of overlapping definitions.

PAGE 68

56 An important issue with regard to the various scales is the population segment or segments that have been sampled. The Alexander and Andersen research was conducted with a sample from the general voting population. The King and King scale was tested for consistency between sample types across two studies, one with college students and a second with the general population, so as to provide a basis for comparison of results between the two samples. The researchers found consistency in construct validity between the two samples (King and King 1997). The Spence and Helmreich scale has been tested on several decades of college students. While the King and King scale and the Spence and Helmreich scale have been tested extensively (King and King 1997; McHugh and Frieze 1997), the Alexander and Andersen scale was selected for this study due to its time efficiency (shorter length) and previous application to political candidate evaluation. Given that the Alexander and Andersen scale was found to be both valid^ and reliable by Alexander and Andersen (Alexander and Andersen 1993; Alexander 1995), a replication of its use in this study of candidate evaluation was appropriate. Political knowledge is the ability to respond accurately to a knowledge set-including questions regarding the following: (a) the name of the political party with the majority seats in the U.S. House of Representatives; (b) the political offices held by such political figures as Clinton, Gingrich, Rehnquist, and Greenspan; (c) the names of U.S. House of Representatives in the given district; (d) identification of Clinton as either a liberal or conservative; (e) and the ideology of the Democratic and Republican parties on the issues of 2 Validity was supported by testing relationships of scale with political ideology, party affiliation, and gender trait schemata.

PAGE 69

57 defense spending, aid to Central America, and government services. These items were taken from the 1986 American National Election Studies as recommended in John Zaller's "political knowledge" scale (Zaller 1992), in which a reliability coefficient of .80 was found. These items provide a good mix of easy-to-difficult measures of political knowledge, as recommended by Zaller. As an index of the political knowledge construct, they have been found to provide a good predictor of attitude characteristics (Zaller 1992). Media exposure is operationalized as the amount of time spent (a) watching television news; (b) reading news stories in the newspaper; and (c) listening to radio news. Dependent Variables Perceived competence is defined operationally as competence in an individual policy area, expressed as a rating of candidate's competence regarding each of the following issues: (a) strengthening the economy (neither stereotypically male nor female); (b) assisting the poor (stereotypically female); (c) dealing with big business (stereotypically male); (d) improving our educational system (stereotypically female); (e) dealing with crime (stereotypically male); (f) dealing with health care (stereotypically female); (g) dealing with employee unions (stereotypically male); (h) encouraging the arts (stereotypically female); and (i) making decisions on farm issues (stereotypically male); The nine individual policy area measures were derived from the meta-analytic review described in Chapter Two, and in particular, from Leeper's research.

PAGE 70

58 Research Design and Validity In planning and conducting the experimental procedures for this study, the following issues were addressed in order to improve the validity of results generated: Internal Validity Intrinsic to the concept of internal validity is the question: "Did my experiment make a difference?" The researcher must be concerned with the existence of causality between the independent and dependent variables and must try to eliminate, or at least mitigate, variables that may confound the effects of the treatment variables. Threats to internal validity have been addressed through random assignment of treatments; that is, each subject had an equal chance of receiving each of the treatments. Within each class setting, the eight versions of the campaign speech were distributed in a random manner. Given the random seating of students, each student had an equal chance of receiving any of the eight versions of the speech. Design Validity In this study, two specific issues of concern for design validity were (1) the relationship between manipulated and perceived variation in message clarity; and (2) the priming of subjects about gender as an issue. Concern for message clarity variation focused on the need for subjects to perceive varying degrees of clarity in the eight speech versions administered as part of the posttest phase of the experiment. In order to address this concern, a preexperiment manipulation check was conducted by recruiting subjects in a

PAGE 71

59 telecommunication class of approximately 150 students. This design test is further discussed in Chapter Four in the sections titled "Pilot Test" and "Manipulation Check." Given the preexperiment-posttest design of this study, the second particular concern pertaining to design validity was the possibility of inadvertently priming subjects about gender. Because the preexperiment test included questions that examined gender role attitudes, there was concern for alerting the subjects to the researcher's testing for gender stereotypes and the possible effects on evaluation of female and male candidates in the posttest phase of the experiment. In order to minimize the association between the preexperiment and posttest phases of the experiments, a one-to-two week interval was used between the two phases, the questionnaire design was changed for the posttest, and the posttest phase was administered (in most cases) by a different person than the preexperiment phase. These steps were taken in order to give the appearance that the two phases were two different studies. While attrition occurred between the preexperiment test and posttest, only matched cases were used in the data analysis. Matching depended on the attendance of students in both sessions for each of the three classes from which subjects were recruited. The matching of cases required a sampling of approximately 370 students in order to obtain 207 matched cases for the analysis. Construct Validity In order to ensure that the operationally defined variables represent the theoretical concepts used in this study, a thorough review of the literature was conducted to ascertain: (a) the common denominators among different studies in relation to conceptual and

PAGE 72

60 operational definitions; and (b) applicability of definitions to the key relationships in this study. Specific information regarding the use of the key variables in previous studies can be found in Chapter Two. In addition, for the indices used as variables in this study, factor analyses were conducted as described in Chapter Four. An additional approach taken in this study to ensure construct validity is the use of multiple measures of constructs (on message clarity, political knowledge, gender role schema, and candidate competence), which are described in Chapter Four. Furthermore, in order to make it difficult for subjects to guess at the hypotheses regarding gender differences, the gender of candidates was not identified specifically, but rather was only reflected in the candidates' names. External Validity External validity is perhaps the weakest element of validity in this study. As is commonly practiced, college students in classroom settings were used as subjects; therefore, no attempts have been made to generalize the findings to other target populations or settings. It is believed, however, that the testing of college students regarding issues of political candidate evaluation is appropriate, because college students are generally a segment of the voting-age population. Future research based on the theoretical concepts in this study will attempt to include other segments of the voting-eligible population. With regard to the timing of data collection for this study, the data was collected during an off-year for national elections in the geographic area. It is possible that during an election campaign, exposure to actual candidates could alter the salience of certain

PAGE 73

61 characteristics or issues that might influence responses pertaining to the hypothetical candidates used in the treatment conditions of this research. The Experiment The Participants Participants for the experiment were drawn from a population of approximately 370 students in three classes (political science, telecommunication, and agricultural communication) at the University of Florida. Matching of preexperiment-posttest phases of the study generated 207 subjects for data analysis purposes. Subjects were told that they were participating in a study regarding "Politics News and Views." A first phase, preexperiment instrument was administered in which subjects were "tested" for level of political knowledge and gender role ideology. Later, a posttest was administered in which students were asked to read a campaign speech and fill out a questionnaire regarding their perceptions about the hypothetical candidate and the speech. A 2 x 2 x 2 (candidate gender x message theme x message clarity) between-subjects, factorial design was used to test the hypotheses listed in Chapter Two. There were eight messages, assigned randomly across subjects. The number of subjects in each of the eight treatment conditions was 21 , 25, 26, 27,27,27, 28, and 28, with a mean cell size of 26 (Table 3.2). Treatment and Instrumentation In Leeper's study, each subject was given a campaign speech to read, varied only by the assignment of either a candidate with a female name or a candidate with a male name. In all other respects, speech content was the same for all subjects. In the current study,

PAGE 74

62 each subject was given one version of a campaign speech to read, with eight versions of the campaign speech randomized across the subject pool. Each speech condition varied as to candidate gender (female or male), message theme (crime or education), and message clarity (clear or unclear). After reading the speech, each subject was asked to fill out a questionnaire-type instrument. In terms of the posttest instrument used (Appendix B), subjects were asked six questions regarding clarity of the speeches and nine questions regarding their perception of candidate competence on the nine policy competence areas, including economy, poverty, business, education, crime, health, employee unions, the arts, and farm issues. Questions pertaining to speech clarity were 7-point, Likert type questions regarding (1) the candidate's understanding of the issues presented in the speech; (2) the clarity of the candidate's position on the issue (of crime or education); (3) the vagueness of the speech, (4) the persuasiveness of the speech; (5) the specificity of the candidate position on the issue (of crime or education); and (6) the understandability of the candidate's position on the issue (of crime or education). Message Variation In order to test the effects of candidate gender, one-half of the subjects were given a speech to read for which a hypothetical male candidate's name (Jim Preston) had been assigned and one-half of the subjects were given a speech to read for which a hypothetical female candidate's name (Elizabeth) had been assigned. In order to test the effects of message theme, one-half of the subjects read a message in which the issue was crime (a stereotypically male issue) and one-half of the subjects read a message in which the issue

PAGE 75

63 was education (a stereotypically female issue). Finally, in order to test the effects of message clarity, one-half of the subjects read a message that had a clear position on the issue of either crime or education, and one-half of the subjects read a message that had an unclear or ambiguous position on the issue of either crime or education. Variation in candidate gender, message theme, and message clarity was randomized across subjects by assigning subjects to one of the following eight cells: Table 3.2: Message Variation by Cells (1) female candidate (2) female candidate (3) female candidate (4) female candidate clear message on unclear message on clear message on unclear message crime crime education on education (n=28) fa-27) (n-27) (n=26) (5) male candidate (6) male candidate (7) male candidate (8) male candidate clear message on unclear message clear message on unclear message crime on crime education on education Cn-28) (n-27) (n=21) (n=25) The same introduction and closing of the speech, with vision statements and references to jobs, were used as paragraphs one and three in all versions of the speech (Appendix A). However, specific references to either crime or education (paragraph two) varied among versions of the speech. The clear message on crime refers to a specific proposal for funding the building of prisons through a 20 cents per pack cigarette tax. The unclear message on crime mentions being tough on crime, but with no specific proposals for the treatment of criminals or funding for prisons. The clear message on education advocates funding for education through a 20 cents per pack of cigarettes tax. The unclear message on education refers to having a better future through education, but with no specific proposals.

PAGE 76

64 Data Collection In order to obtain participants for the experiment, students from the University of Florida in three classes (political science, telecommunication, and agricultural communication) were asked to participate voluntarily. Classes were selected for data collection based on diversity of student majors, class size, and availability of access for the experiment. Each student received two extra credit points for participating. The process contained the following steps: (1) Participants were asked to complete a questionnaire on "Political News and Views" that contained questions regarding political knowledge and gender role ideology; (2) Between one and two weeks after the "Political News and Views" phase, participants were asked to read a campaign speech (described previously in section "Message Variation"), randomly assigned from the eight conditions previously described; and (3) Participants were asked to complete a questionnaire-type instrument (described previously in section "Treatment and Instrumentation") after reading the speeches. Data Analysis Data analysis was designed to address the following research questions: (1) Does message theme have a main effect on competence ratings for female candidate evaluation? (2) Does message clarity have a main effect on competence ratings for female candidate evaluation? (3) Do message theme and clarity combined interact with each other in their effect on competence ratings for female candidates? (4) Does candidate gender have a

PAGE 77

65 significant effect on candidate evaluation? (5) Is the level of political knowledge of respondents associated with the competence ratings for female candidates? (6) Is the level of newspaper exposure associated with competence ratings for female candidates? And (7) Is an individual's gender role schema associated with and/or interact with the effect of candidate gender on competence rating? In order to test the relationships depicted in the research hypotheses stated in Chapter Two, the following statistical tests were conducted: (1) t-tests for effects of message clarity (clear vs. unclear) on competence ratings for female candidates (Hypothesis 1); (2) t-tests for effects of candidate gender (female vs. male) on competence ratings (Hypotheses 5 and 6); (3) correlations for associations between (a) political knowledge of subjects and competence ratings of female candidates (Hypothesis 9); and (b) newspaper exposure of subjects and competence ratings of female candidates (Hypothesis 10); (4) analysis of variance for interaction between gender role schema and candidate gender as related to candidate competence ratings (Hypothesis 1 1). In addition to the testing of hypotheses, several post hoc analyses were conducted: (1) a t-test for the effect of message theme (crime vs. education) on competence rating for female candidates; (2) an analysis of variance for candidate competence rating as a function of interaction between candidate gender and message clarity; (3) a t-test for difference between female and male candidate using messages on education; and (4) correlational analyses for relationships between political knowledge and gender role schema, between subject gender and political knowledge, between subject gender and gender role schema, and between gender role schema and exposure to television news.

PAGE 78

CHAPTER FOUR FINDINGS This chapter includes a demographic description of the study participants, a description of the indices used for data analysis, and results of die analysis. Analysis of variance (ANOVA) and t-test procedures were used to test hypotheses about the effects of candidate gender, message theme, and message clarity on ratings of candidate competence. Subject variables, including political knowledge, media exposure, and gender schema were tested for association with the manipulated variable of candidate gender. Properties of the indices, results of the manipulation checks, and results of the hypothesis tests will be discussed. Study Participants This study drew approximately 370 college students from the following undergraduate classes: American Federal Government, Introduction to Telecommunication, and Writing for Agriculture and Natural Resources. Pretest-posttest matching resulted in 209 matched cases for data analysis. The students, including 109 females and 100 males, had a mean age of 20 with the most frequently occurring age being 18. In terms of political party affiliation, 39 percent identified themselves as Democrats, 25.6 percent as Republicans, and 17.6 percent as independents. Sixteen percent responded that they had no 66

PAGE 79

67 party affiliation. When asked to describe their political ideology, 44.3 percent identified themselves as liberal, 32.4 percent as conservative, and 23.3 percent as middle-of-the-road. Indices This analysis generated tiiree indices for the purposes of manipulation checks and hypotheses testing. As will be indicated in the sections describing the political knowledge index and the gender role schema index, the indices for political knowledge (Zaller 1990, 1992) and gender role schema (Alexander and Andersen 1993; Alexander 1995) have been shown to be reliable in previous research. The index for message clarity was developed for this study, as described in the following section. The Message Clarity Index The message clarity index included six items measured by 7-point Likert-type questions. Respondents were asked questions regarding the clarity of the campaign speech and the candidate's position, including "How well do you think (Elizabeth/Jim) understands the issues discussed in the speech?" "How clear are (Elizabeth/Jim) 's stands on the issue of (crime/education)?" "How specific is Preston on the issue of (crime/education)?" "Do you think that Preston's speech is vague?" "How persuasive is the speech?" and "Is Preston's position on (crime/education) easy to understand?" A principal components analysis^ on the six items resulted in a one-factor solution, and an internal consistency analysis of the six items resulted in a Cronbach's alpha of .83. 3 A second analysis using principal axis factoring as the extraction method yielded a similar factor matrix for a one-factor solution, but explained only 43.3% of the variance.

PAGE 80

68 The average inter-item correlation for the six variables was .45. The factor loadings for the six variables were all .67 or above (Table 4. 1). The one factor solution explained 54.3 percent of the variance. Factor scores were used as variable data for the manipulation check on message clarity, as discussed in the section "Manipulation Check." Table 4. 1 Factor Analysis of Message Clarity Index with Principal Components Analysis Item How clear candidate's stand is on policy How well candidate understands the issue How specific candidate is on issue How vague is speech How easy to understand candidate's position How persuasive is speech Factor Loadings Factor 1 Mean* SD .81 4.5 1.8 .77 3.9 1.4 .76 3.9 1.6 .71 3.0 1.5 .70 4.8 1.6 .67 3.7 1.6 % of variance explained Eigenvalue 54.3% 3.26 * on a 1-7 scale The Political Knowledge Index A 9-item 4 political knowledge index used to measure political knowledge regarding prominent national and state political figures; general party ideology, party ideology on defense, foreign affairs, and government spending issues; and majority party status in the 4 The questionnaire contained 18 questions pertaining to political knowledge; however, 4 questions were deleted because they represented halves of paired questions for which each pair received one score; one question was excluded from the index because over 90% of the subjects had correct responses; and 4 questions were deleted because less than 15% of the subjects had correct responses. One question that received 5% correct answers was arbitrarily retained for the index so as to include an especially difficult question.

PAGE 81

69 U.S. House of Representatives was derived from the 1986 American National Election Studies, as recommended in John Zaller's The Nature and Origins of Mass Opinion (1992). The index score for each subject is a summative score based on six single items pertaining to recognition of political officeholders and three sets of paired questions on policy issuesdefense spending, aid to Central America, and federal spending for government services. The paired questions pertain to ideological positions on the liberal-conservative continuum regarding the three policy issues. Each pair asks for the ideological position of the Democratic and Republican parties on a given issue (Appendix B). The political knowledge index was intended to be diverse as to types of political information and was not well represented by a factor analytic procedure. Items in the political knowledge index were scored binomially as either " 1 " for correct or "0" for incorrect, making it problematic to develop a factor solution for the index. An attempt to obtain a factor solution for the nine items in the index resulted in a five-factor solution, each with eigenvalues greater than 1.0. Therefore, a summative score was used as an index score for each subject, based on the number of correct answers (ranging from 0-9) for each subject. The mean score for all subjects was 4.9 with a standard deviation of 1.5, and the most frequent number of correct answers was 5 (Figure 4. 1). A comparison of female and male respondents indicates that males scored higher on 7 out of 9 items (Tables 4.2 and 4.3).

PAGE 82

70 Table 4.2: Percent of Correct Political Knowledge Questions by Female Respondents N Mean Std. Deviation Democratic Party position on US in Central America Democratic Party position on defense spending Gingrich's political office Democratic Party position on spending for government services Greenspan's job title Political ideology of Democratic Party Majority party in House of Representatives Rehnquist's job title Name of member of U.S. House of Representatives from Alachua County district 109 109 109 109 109 109 109 109 109 .24 .70 .82 .87 .05 .91 .72 .35 .04 .43 .46 .39 .34 .21 .29 .45 .48 .19 Table 4.3: Percent of Correct Political Knowledge Questions by Male Respondents N Mean Std. Deviation Democratic Party position on US in Central America Democratic Party position on defense spending Gingrich's political office Democratic Party position on spending for government services Greenspan's job title Political ideology of Democratic Party Majority party in House of Representatives Rehnquist's job title Name of member of U.S. House of Representatives from Alachua County district 97 98 98 98 98 98 97 98 98 .35 .82 .84 .80 .26 .83 .82 .47 .07 .48 .52 .37 .41 .44 .38 .38 .50 .26

PAGE 83

71 political knowledge index Figure 4. 1 : Political Knowledge Index by Percent of Subjects The Gender Role Schema Index A 6-item gender role schema index was designed to measure attitudes regarding gender roles in society. The purpose of the index was to create a measure by which the gender schema of individuals could be tested for interaction with candidate gender when subjects were asked to evaluate the candidates on policy area competence. The underlying concept for the gender role schema index, derived from Alexander and Andersen (1993) and Alexander (1995), is gender role egalitarianism, with low scores representing "traditional" views and high scores representing "egalitarian" views. Alexander (1995) posited that gender role beliefs reflect critical voter predispositions that affect the attribution of stereotypic traits to candidates. The

PAGE 84

72 traditional-egalitarian index is used to represent the concept that gender role beliefs based on traditional stereotypes have a limiting effect on the role of women in politics, while gender role egalitarianism reflects the view that women are equal to men in their political competencies. In order to confirm the reliability of the gender role schema index used in this study, a factor analysis was conducted. A principal components analysis 5 on the six variables resulted in a one-factor solution and internal consistency analysis of the items resulted in a Cronbach's alpha of .83. The average inter-item correlation for the six items was .45. All the items loaded on the factor at .69 or higher (Table 4.4). The one-factor solution explained 54.7% of the variance. In this study, the index score for each subject is a summed average score based on six variables 6 representing six gender-role related topics — child care, husband's careerversus-wife's career, nontraditional careers for women, emotional suitability of men and women for politics, competitive salaries in two career couples, and male-versus-female caregiving in the home. Respondents were asked to agree or disagree on a 6-point scale with the following statements: "It is more important for a wife to help her husband's career than to have a career herself," "A preschool child is likely to suffer if his or her mother works," "It is much better for everyone involved if the man is the achiever outside the home and the woman takes care of the home and family," "Most men are better suited emotionally for politics than most women," "Women should be able to compete with men 5 A second analysis using principal axis factoring as the extraction method yielded a similar factor matrix for a one-factor solution, but explained only 49.6% of the variance.

PAGE 85

73 for jobs that have traditionally belonged to men, such as firefighters," "A husband should not feel uncomfortable if his wife earns a larger salary than he does." The first four statements are traditional in gender role ideology, while the last two are egalitarian in gender role ideology. The last two questions were reverse coded prior to factor analysis so that the highest egalitarian ideology was represented by a response of "very strongly agree" as a number "6." For the four traditional statements, the highest egalitarian position was represented by a "very strongly disagree." The no opinion/don't know responses were coded as missing values so that they would not corrupt the summative index scores. Table 4.4: Factor Analysis of Gender Role Schema Index with Principal Components Analysis Item Factor Loadings Factor 1 Better if man is achiever outside home and woman is caregiver in .81 home Husband should not be uncomfortable if wife earns higher salary * .79 More important for wife to help husband's career than have own .73 Women should be able to compete for jobs traditionally for men * .73 Preschool child likely to suffer if mother works .69 Men better suited emotionally than women for politics .69 % of variance explained 54 . 7 % Eigenvalue 3.28 These items were reverse coded prior to factor analysis in order to maintain consistency of egalitarian concept for the six statements. In keeping with the underlying concept of egalitarianism as a categorical variable, the gender role schema index scores were converted into a categorization of traditional, 6 A seventh variable based on the question "It is a good idea for a husband to stay at home and care for the children while his wife is employed outside the home" was dropped from the index because it loaded as a separate factor in the factor analysis.

PAGE 86

74 moderate, and egalitarian schema. Then index scores were recoded so that "1" was traditional, "2" was moderate, and "3" was egalitarian based on the location of scores relative to the midpoint of the 1-6 scale, which was 3.5. Mean scores <3 were categorized as traditional, mean scores >3 but <5 were categorized as moderate, and scores >5 were categorized as egalitarian. The categorical scores were later used for hypothesis testing related to gender role schema. Pilot Study In order to develop campaign messages of varying clarity for the hypothetical candidates in the study, a pilot test was conducted. The purpose of the test was to develop messages that were different from each other in clarity of candidate's position on either crime or education. Each subject was asked to read one of eight versions of a campaign speech and then to fill out a questionnaire that included six questions related to the clarity of the speech. Questions referred to clarity, specificity, understandability, vagueness, and persuasiveness of the eight different speech messages regarding crime and education. An analysis of variance (Table 4.5) for the pilot test on message clarity indicated that respondents did not perceive significant variation in clarity ( F (1 , 155) = 2.25, p <_. 14) among the eight versions of the message; therefore, the speeches were revised to increase variation. The mean factor score for unclear messages was .1 and for clear messages, the mean factor score was -.1. As described in Chapter Three, messages were revised so that intentionally unclear messages made general references to the problem of either crime or

PAGE 87

75 education without specific policy proposals, while messages that were intended to be clear were revised so that they included specific policy proposals. Table 4.5: Analysis of Variance for Preexperiment Manipulation Check of Message Clarity Source df Perceived clarity Version clarity 1 2.25 Error 155 (.99) Note. Value enclosed in parentheses represents mean square errors. Manipulation Check A manipulation check was based on a set of clarity-related questions included in the posttest phase of the actual experiment. These questions were the same clarity-related set used in the pre-experiment, pilot study. The posttest phase consisted of the respondents reading campaign speeches and filling out a questionnaire pertaining to the speech and hypothetical candidate to whom each speech was attributed. The six items in the questionnaire pertaining to speech clarity were reduced to a one-factor solution that represented respondents' perception of message clarity (described in the previous section on the message clarity index). An analysis of variance (ANOVA) procedure was used to test the relationship between (1) the manipulated clarity of the four speech versions on crime and their perceived clarity (Table 4.6); and (2) the manipulated clarity of the four speech versions on education and their perceived clarity (Table 4.6). Additional description of campaign speech versions is contained in Chapter Three.

PAGE 88

76 Table 4.6: Analysis of Variance for Manipulation Check of Perceived Message Clarity JF Perceived Clarity Source df Crime message df Education message Version clarity 1 7.66 * 1 .05 Error 105 (.99) 96 (.97) Note. Values enclosed in parentheses represent mean square errors, indicates p < .01 In order to identify variations in perceived clarity (Table 4.7) of the campaign messages, a manipulation check was performed for each of the message themes, crime and education. Results indicated a significant relationship between perceived clarity and manipulated clarity for the messages on crime (M =-.3 for unclear and M = 2 for clear, F (1, 105)=7.66,j) <,007), but no significant relationship for the messages on education (M =.04 for unclear and M =-.004 for clear, F(l,96) =.05, p_ <.82). Given the lack of subjects' perceived variation in clarity for the messages on education, it was expected that, for the purpose of hypothesis testing, there would be no effect for clarity when the message was on education. Research Questions This study addresses two major questions: (1) To what extent and under what conditions does gender of candidates affect their perceived competence? and (2) How do

PAGE 89

77 Table 4.7: Means for Perceived Message Clarity by Message Theme for Clarity Index Items Crime Messages Education Messages Item for Perceived Clarity of Message clear Mean SD unclear Mean SD cleai Mean SD unclear Mean SD How clear candidate's stand on policy 5.0 1.6 3.8 1.8 4.6 1.8 4.5 1.7 (7= very clear) How well candidate understands issue 3.9 1.4 4.1 1.5 3.8 1.4 4.1 1.4 (7= very well) How specific candidate is on issue 4.6 1.4 3.2 1.5 3.9 1.6 3.7 1.7 (7= very specific) How vague is speech 3.2 1.6 2.6 1.5 3.3 1.6 2.8 1.4 (7= not vague at all) How easy to understand candidate 5.1 1.5 4.5 1.7 4.6 1.8 4.9 1.5 position (7= very easy) How persuasive is speech 3.7 1.4 3.8 1.8 3.4 1.7 4.0 1.6 (7= very persuasive) message theme and message clarity affect the perceived competence of female candidates? The conditions that are manipulated in the experiment are candidate gender, message theme, and message clarity^, while the subject characteristics that are analyzed for association with the candidate-competence relationship are gender role schema, level of political knowledge, and level of media exposure. The research hypotheses are centered around these manipulated and subject variables with a key interest in the perceived competence of the hypothetical candidates for United States Congress— Elizabeth and Jim Preston. Hypothesis Testing Eleven general hypotheses were originally proposed for significance testing at thep <_ .05 level. However, as previously stated, the manipulation check for message clarity

PAGE 90

78 indicated that subjects perceived significant differences in clarity for messages on crime, but that there was a problem with the perceived clarity of the education messages. As a result, Hypotheses 2,3,4,7, and 8 could not be tested due to the problem of construct validity for clarity of education messages. Therefore, hypotheses for which clarity of education messages was a component were dropped from the analysis. Of the original hypotheses, the first four hypotheses address the effects and interactions of message clarity and message theme for female candidates only. The next four compare female and male candidates in association with message theme and clarity, and the last three are related to the association between subject characteristics and their effects on candidate evaluation. Within each general hypothesis, nine subhypotheses were tested for the nine policy areas. The nine dependent measures for individual policy competence areas are: competence in the arts, competence in business, competence in crime, competence in economy, competence in education, competence in farm, competence in health, competence in poverty, and competence in (employee) unions. Female Candidates: Main Effects and Interactions for Messag e Theme and Clarity One general hypothesis was proposed to test the question: Does message clarity have a main effect on subjects' perception of female candidate competence? Hl(a-i). When female candidates use a message on crime that has a clear position on the issue, they are perceived as more competent than when they use a message on a crime that has an unclear position. For Hypothesis 1, it was expected that subjects in the high message clarity condition would evaluate female candidates more favorably than subjects in the low message clarity 7 Message clarity was successfully manipulated for crime messages, but not for education messages.

PAGE 91

79 condition, and that this would result in higher competence ratings. Dependent measures for candidate competence include the two key measures for individual policy competence areas that relate directly to the messages tested-competence in crime and competence in educationas well as competence in the arts, business, the economy, farming, health care, poverty, and unions. In order to test Hypothesis 1, subjects who read a female candidate's clear message on crime (n=28) were compared to subjects who read the female candidate's unclear message on crime (n=27) on the nine dependent measures: (a) When Hypothesis 1 was tested for the dependent measure of "competence in crime," a significant difference was found for message clarity (Table 4.8). The t-test for competence in crime indicated that the treatment group for subjects who read a speech with a clear, crime message (M= 5.0) rated the female candidate significantly higher than the control group— subjects who read an unclear, crime message (M= 3.9, t_(53)= 2.72, p <. .01). Therefore, die hypothesis was supported for competence in crime. (b) For competence in education (Table 4.8), however, there was no significant difference between the clear on crime message group (M =4.96) and unclear on crime message group (M =3.89) in their ratings of the candidate (t_(53) = .13, p <_.90). Therefore, the hypothesis was not supported for competence in education. (c) thru (i) For competence in health, poverty, the arts, farming, the economy, employee unions, and business issues, no significant main effects for message

PAGE 92

80 clarity were found (Table 4.8). Therefore, the hypothesis was not supported for these policy areas. Table 4.8: Mean Competence Scores for Female Candidate with Message on Crime Condition: Condition: Clear message on crime Unclear message on crime Policy Competence Area Mean SD n Mean SD n t arts 3.4 1.7 28 2.8 1.4 27 1.45 business 3.0 1.2 28 3.0 1.2 27 .00 crime 5.0 1.4 28 3.9 1.5 27 2.72* economy 3.5 1.3 28 3.4 1.1 27 .06 education 3.8 1.4 28 3.8 1.2 27 .13 farm 3.0 1.4 28 2.6 1.2 27 1.12 health 3.0 1.4 28 3.0 1.3 26 .00 poverty 3.4 1.3 28 3.6 1.2 27 .49 unions 2.9 1.3 28 2.7 1.3 27 .54 Note. * indicates significance at 2 <. -05 One general hypothesis was proposed to test the question: "Does message theme (crime vs. education) have a main effect on perceived competence of the female candidate? " H2(a-i): When females use a clear message on crime , they are perceived as more competent than when they use a clear message^ on education . It was expected that subjects in the clear on crime message condition (n = 28) would evaluate female candidates more favorably than subjects in the clear on education message condition ( n = 27), and that this would result in higher competence ratings for the candidate with the crime message. However, due to the lack of perceived clarity variation for the education messages, this hypothesis was not testable.

PAGE 93

81 Two general hypotheses were proposed to address the question: "Do message theme and message clarity have an interaction in association with their effects on perceived candidate competence when the candidate is female?" H3(a-i): When female candidates use a clear " message on education , they are perceived as more competent than when they use an unclear message on crime . For Hypothesis 3, it was expected that subjects in the "clear" message on education group would evaluate female candidates more favorably than subjects in the unclear message on crime group. The assumption was that a clear message would have a more favorable effect on candidate's competency rating than an unclear message, even if the message was education instead of crime. However, due to the lack of perceived clarity variation for the education messages, this hypothesis was not testable. H4(a-i). When female candidates use a clear message on crime , they are perceived as more competent than when they use an unclear message on education . For Hypothesis 4, it was expected that subjects in the group with the clear message on crime would evaluate female candidates more favorably than subjects who read the unclear message on education. The assumption was that a clear message on crime would be favored for both clarity and theme. However, due to the lack of perceived clarity variation for the education messages, this hypothesis was not testable. ° The "clear" message is the message intended to be clear by the researcher, although subjects did not seem to perceive variation in clarity for education messages. " Refer to footnote 6.

PAGE 94

82 Comparing Female and Male Candidates: Main Effects of Candidate Gender Four general hypotheses were proposed to test the question: "Do message theme and clarity affect female and male candidates differently? H5(a-i). When female candidates use a message on crime that has a clear position on the issue, they are perceived as less competent than male candidates who use a message on crime that has a clear position. It was expected that male candidates with a clear message on crime would be rated higher in competence than female candidates with the same message. In order to test this hypothesis, t-tests were conducted for the nine dependent measures on competence (Table 4.10). (a) For the t-test on competence in crime (Table 4.9), no significant difference was found for candidate gender (t_(55)=.09, p_ _< .93); therefore, the hypothesis was not supported for competence in crime. (b) For competence in education, the t-test indicated a near-significant gender difference for candidates (t (55) = 1.96, p _< .06). However, the female candidate had a mean of 3.8, while the male candidate had a mean of 3. 1 (Table 4.9). Therefore, the hypothesis was not supported. (c) Given a clear message on crime, there were no significant effects of candidate gender for competence in business, economy, health, or unions. There were, however, significant mean differences for candidate gender for the policy competence areas of the arts, farm issues, and poverty (Table 4.9), with the female candidate rated higher. When significant gender differences were found,

PAGE 95

83 they were in the opposite direction from the hypothesis. Therefore, the general hypothesis was not supported. Table 4.9: Mean Competence Scores for Clear Message on Crime by Candidate Gender arts business crime economy education farm health poverty unions Condition: Condition: Female candidate Male candidate Mean SD n Mean SD n t 3.4 1.7 28 2.2 1.2 29 3.06* 3.0 1.2 28 3.1 1.6 29 .19 5.0 1.4 28 4.9 1.6 29 .09 3.5 1.3 28 3.2 1.5 29 .79 3.8 1.4 28 3.1 1.5 29 1.96*** 3.0 1.4 28 2.1 1.0 29 2.74* 3.0 1.4 28 2.7 1.1 29 1.04 3.4 1.3 28 2.7 1.2 29 2.27** 2.9 1.3 28 2.8 1.4 29 .28 Notes. * indicates significance at p _< .01 **indicates p _< .05 ***indicates n <. .06 The scale for the means is 1-7, 7 being most competent. H6. When female candidates use a message on crime that has an unclear position on the issue, they are perceived as less competent than male candidates who use an unclear message o n crime . When t-tests were conducted for all of the dependent measures, no significant differences were found between the male and female candidates for an unclear message on crime (Table 4.10). Therefore, the hypothesis was not supported. H7. When female candidates use a " clear " 10 message on education , they are perceived as more competent than male candidates who use a " clear " message on education . I" The manipulation check indicated no perceived variation in clarity for education messages.

PAGE 96

84 Table 4.10: Mean Competence Scores for Unclear Message on Crime by Candidate Gender Condition: Condition: Female candidate Male candidate Policy Competence Area Mean SD n Mean SD n t arts 2.8 1.4 27 2.6 1.6 26 .49 business 3.0 1.2 27 3.5 1.6 26 1.20 crime 3.9 1.5 27 4.4 1.8 27 1.16 economy 3.4 1.1 27 4.1 1.5 27 1.72 education 3.8 1.2 27 3.9 1.8 27 .18 farm 2.6 1.2 27 2.7 1.7 27 .18 health 3.0 1.3 26 3.3 1.5 26 .69 poverty 3.6 1.2 27 3.4 1.5 27 .61 unions 2.7 1.3 27 3.2 1.6 26 1.14 Due to the lack of perceived clarity variation for the education messages, this hypothesis was not testable. H8. When female candidates use an unclear message on education , they are perceived as more competent than males who use an unclear message on education . Due to the lack of perceived clarity variation for the education messages, this hypothesis was not testable. Effects of Subject Characteristics The next three hypotheses address die proposed associations between subject characteristics and the effects of message theme and clarity on perceived competence of female candidates. H9. When female candidates use a clear message on crime , there will be a positive association between political knowledge and perceived competence of the candidate.

PAGE 97

85 For Hypothesis 9, it was expected that individuals with higher levels of political knowledge would rate the female candidate higher than individuals with lower levels of knowledge. A correlational analysis between the political knowledge index and competence measures for female candidates with a clear message on crime was conducted. This was tested for subjects who read the message that was clear on crime by a female candidate (n = 28). Results indicated no significant correlations (atp_ _< .05) for any of the dependent measures (Table 4.11). Therefore, the hypothesis was not supported. H10. When female candidates use a clear message on crime , there will be a positive association between print media use and perceived competence of the candidate. For Hypothesis 10, it was expected that individuals who had higher amounts of print media exposure would rate the female candidate higher than individuals with lower levels of print media exposure. A correlational analysis was conducted between the number of hours subjects read newspapers per week and competence measures for female candidates Table 4. 12). This was tested for subjects who read the message that was clear on crime by a female candidate (n = 28). Despite the variation in news exposure (Figure 4.2), results indicated no significant correlations (atp <_ .05) for the numbers of hours per week an individual read a newspaper and the competence ratings the individual assigned to the female candidate with the clear message on crime.

PAGE 98

86 CD o a a, £ o U CO < _o O P-, -o a 3 a> -a a 4) to -a o o o d 4) CD i 4) CQ 3 .9 jS b o U , r = w to cu c CD O Q-'E o r E => o C o — c > CM CD "C T— Zi CD £ > gO E a 9 = o •— |£ 00 o CD CD Q. CD y) p" "53 £ Q. CD E,e o CJ CD CB •a cu CD a u O a..* Jh CD O a cu a. S o O CD < "3 Oh "(3 a H> In O c_ a ecj o c CD 35 CD CQ e* G «S u O U CN _2 E*— • .22 c * o "55 o Q-'E E = c c o — c >< «7 CD -C T— "St CD 95 > 2" o E a 9 c o — "£ j= a> cd £ "S 5 Q. 0) E^= o c u c * JS E CN a. .2 E e c o •o o CO CD c c CO l_ CD O T— o O lei c £ 5 o o 35 co o cd co CD 0. -t-J xr S E ^ 03 ._ o Q..5 c 1 E ° 8 8 c „, OJ 53 S OJ CD .S I o — o C OT <* CD 10 CN « C C S i fi ^2 u c CD _g U) ^ "53 "E Q. co E.S o o CD i_ 3 ca in print medi expo

PAGE 99

87 %of subject 8 9 10 12 40 hours/week read newspaper Figure 4.2: Exposure to News Stories in Newspapers HI 1. Individuals with egalitarian gender role schema favor female candidates, while individuals with a traditional role schema favor male candidates. For this hypothesis, it was expected that female candidates would be rated higher than male candidates by individuals with egalitarian gender schema, and that male candidates would be rated higher than female candidates by individuals with traditional gender schema. In order to operationalize gender role schema, a gender schema index score (a summative score) was converted to a categorical index and analyzed in relation to the candidate gender-competence relationship. The categorical index for gender role schema consisted of (1) traditional (n = 11) (2) moderate (n =127); and (3) egalitarian (n =69).

PAGE 100

88 To test the hypothesis that gender role schema would interact with candidate gender in its effect on competence ratings, gender role schema categorical scores were entered into the analysis of variance model, and a series of analysis of variance procedures were conducted for the dependent measures. Results indicated no statistically significant interaction (at the p _< .05 level) between gender role schema and candidate gender when measuring effects for candidate competence ratings. Table 4.13 is illustrative of these findings of statistical non-significance. The test for interaction between gender role schema and candidate gender for competence in crime yielded an F of .19, p _< .83; for competence in education, an F of .86, p _< .43. The other policy competence measures yielded similar non-significant results. Post Hoc Analysis of the General Model for Policy Competence A post hoc analysis of the overall model for perceived policy competence was based on an analysis of variance procedure, testing the following model for interactions: competence = intercept + candidate gender + message theme + message clarity + (candidate gender * message theme) + (candidate gender * message clarity) + (message theme * message clarity) + (candidate gender * message theme * message clarity) Results indicated that the only significant interaction was the interaction between message theme and message clarity for candidates' perceived competence in crime (F (1, 208) = 6.97, p < .01. Candidates with a clear message on crime (M = 4.9) were rated higher than candidates with an unclear message on crime (M = 4.2), but for candidates using an education message, clarity did not make a significant difference. Underlying this

PAGE 101

89 Table 4.13: Analysis of Variance: Test for Interaction between Gender Role Schema Category and Candidate Gender Source df Competent in Crime Competent in Education Candidate gender (CG) 1 1.42 1.51 Gender schema category 2 .65 79 (GS) CGxGS 2 .19 .86 Error 201 (2.89) (2.75) Note. Values enclosed in parentheses represent mean square errors. interaction is the problem described earlier regarding the lack of perceived variation in clarity for the education messages. No significant interactions were found for the other competence areas. Post Hoc Analysis of Message Clarity Effect for the Male Candidate Given the finding that message clarity was important for the female candidate using a message on crime, it was decided that a comparison with the male candidate would be useful for determining if candidate gender made a difference in approaches to communication effectiveness. First, an analysis of variance was conducted to determine if there was an interaction between candidate gender and message clarity when evaluating the candidates' competence in crime. Results indicated no statistically significant interaction (F = .97 (1,106), p < .33) between candidate gender and message clarity for evaluation of the candidates' competence in crime. Next, a t-test was conducted to compare the male candidate using the clear message on crime with the male candidate using the unclear message on crime. Results indicated no effect on evaluation of the male

PAGE 102

90 candidate's competence in crime for message clarity of the crime message (t (53) = 1.08, p_ < -29). Post Hoc Analysis of Candidate Gender for Messages on Education A post hoc analysis for both female and male candidates using messages on education 1 * indicated no significant differences between female and male candidates when education messages were used. A t-test for competence in education (t (97) = 1.10, p < .28) indicated that the female and male candidates using messages on education (M = 4.9 for the female candidate and 4.5 for the male candidate) were perceived equally competent. Post Hoc Analyses of Subject Characteristics While no statistically significant relationships between the subject characteristics (political knowledge, media exposure, and gender role schema) and candidate competence were found, a post hoc analysis of relationships among the subject characteristics indicated some significant findings. Firstly, political knowledge was found to be negatively associated with the gender role schema index (r=-.22,j> j< .002), indicating that individuals with higher political knowledge tended to be traditional in their gender schema. These statistical findings indicate that lower levels of political knowledge were associated with egalitarian gender role schema. As discussed previously, male respondents tended to be higher in political knowledge and traditional in gender role schema, while female respondents tended to be lower than males in political knowledge and egalitarian in gender role schema (Figures 4.2 and 4.3). A t-test was conducted to test the relationship between

PAGE 103

9! level of political knowledge and respondent sex. Results indicated a significant difference (t = 2.58, p_<.01) between females (M =4.7) and males (M =5.2). %of subjects 12 3 4 5 6 political knowledge index Figure 4.3: Political Knowledge by Respondent Sex Secondly, gender role schema was negatively associated with exposure to television news (r =-.14, p _< .05), indicating that individuals with higher exposure to television news tended to be traditional in their gender schema. To test the relationship between the gender role schema index and respondent sex, a t-test was conducted between gender role schema 1 1 Group means for subjects reading either of the messages on education were collapsed into one category for each of the candidates.

PAGE 104

92 index scores and respondent sex. Results indicated a significant difference in level of gender role schema for female and male respondents. For an index from 1 to 6, female respondents were found to be significantly more egalitarian (M =4.9) than male respondents, who were found to be more traditional (M =4.1, t (205) =7.24, p_ <.001) in gender role ideology. Based on a categorization of the gender role schema index into three groups for traditional, moderate, and egalitarian respondents, approximately two percent of females were traditional, 49 percent were moderate, and 49 percent were egalitarian. Of the males, 8 percent were traditional, 76 percent were moderate, and only 16 percent were egalitarian. The low percent of subjects in the traditional group created a problem of low statistical power in testing Hypothesis 1 1 . Given that there were small cell sizes for subjects with traditional gender role schema (n =7 evaluating the female candidate and n = 4 evaluating the male candidate), it is not unexpected that no significant interaction was found between the schema and candidate gender, in terms of the subjects' evaluation of candidate competence. Summary of Findings One hypothesis was supported for individual policy competence ratings. It was found that when comparing clear and unclear messages on crime (Hypothesis 1), female candidates were rated more competent for competence in crime when the message was clear.

PAGE 105

93 traditional moderate egalitarian gender role schema category Figure 4.4: Gender Role Schema Category by Respondent Sex Statistical tests for one hypothesis indicated significant findings in the opposite direction of the hypothesis. When comparing female and male candidates using a clear message on crime (Hypothesis 5), the female candidate received a higher rating than the male candidate for competence in education, but a virtually equal rating to the male candidate for competence in crime. In fact, for all messages, females were perceived equal to males for competence in crime— a stereotypically male issue. This result was similar to that of Leeper's research, in which the female candidate using the message on crime— the "tough" message— was perceived equal to the male candidate for competence in crime, while still being perceived more competent than the male candidate in dealing with education issues.

PAGE 106

94 For the subject variables of political knowledge and newspaper exposure, no significant associations with competence ratings were found. For the subject variable of gender role schema, no significant interaction with candidate gender effects was found.

PAGE 107

CHAPTER FIVE DISCUSSION This dissertation investigated the effects of candidate gender, message characteristics, and subject characteristics on the evaluation of political candidates. The research was designed to test variable conditions of candidate gender, message theme, and message clarity in their effects on the perception of candidate competence with regard to nine public policy areas. The previous chapter addressed the results of statistical tests conducted to analyze the data collected. This chapter discusses limitations of the study, the comparison of predicted results with observed results, the implications of the findings, and suggestions for future research. Limitations of the Study As in all research, some limitations in experimental design have occurred. Results in this study may have been affected by such limitations as selection of the subject pool, attrition of participants, priming of subjects, length of the gender role schema instrument, themes and clarity of messages, and type of medium used. As is commonly practiced, college students in classroom settings were used as subjects; therefore, no attempts have been made to generalize the findings to other target populations or settings. For example, with regard to the gender role schema index, college students tend to be more egalitarian than the general population (Huddy and Terkildsen 95

PAGE 108

96 1993; McHugh and Frieze 1997), thereby skewing the distribution of responses on the egalitarianism scale. One mitigating factor in the selection of subjects, however, was that they were drawn from several different classes within two colleges, thereby providing some heterogeneity of academic and career interests. Relative to the distribution of subjects on the egalitarianism scale is the instrument that was used to test the gender role schema of the subjects. The scale used was based on a six-item instrument such as the one used in the Alexander and Andersen (1993) study. However, the other scales discussed, the Spence and Helmreich scale and the King and King scale, were based on 15 to 95-item instruments. The shorter length of the scale used in this study may have limited the variability of results in measuring gender role beliefs. Related to the issue of length of the scale is that of the types of questions included in the instrument. The longer instruments may have included more questions particularly relevant to college students than could feasibly be included in a six-item scale. As discussed in Chapter Four, there was attrition (from 370 to 209) in the subject pool due to the absence of some students in the second session, or posttest, of each class. The drop-off of some students may have influenced the characteristics of the students who participated in both phases of the experiment; however, random assignment of the students participating in the posttest phase should minimize any possible bias of subjects due to selection. As discussed in Chapter Three, a particular concern pertaining to external design validity was the possibility of inadvertently priming subjects about gender in the preexperiment test that included questions pertaining to gender role attitudes and there was concern for alerting the subjects to the researcher's testing for gender stereotypes and the

PAGE 109

97 possible effects on evaluation of female and male candidates in the posttest phase of the experiment. Despite procedures used to minimize the association between the preexperiment and posttest phases of the experiments, including the scheduling of a one-totwo week interval between the two phases, changing the questionnaire design for the posttest, and administering the posttest by a different person than the preexperiment phase, some priming of subjects about gender may have occurred. However, if some priming occurred, it was not apparent in the relationship between gender role schema and evaluation of candidates by gender. While the policy themes of the messages tested were crime and education, message effects were tested for seven additional policy areas. The purpose of the use of messages on crime and education was to test the effects of a stereotypical male policy area and a stereotypically female policy area. However, limiting the messages to one of each stereotype may be inadequate to test the effects of theme. Related to message theme was the problem with manipulating clarity for the messages on education. Because subjects perceived variation in clarity for messages about crime but did not perceive variation in clarity for messages about education, some of the hypotheses were not testable. Therefore, findings regarding message clarity were limited to crime messages. The messages used in this study were limited to print messages; specifically, written campaign speeches. The use of video or audio messages might be more effective in activating gender schema. One final issue that should be considered in retrospect is the name of the hypothetical female candidate. Given the prominence of Elizabeth Dole in the 1996

PAGE 110

98 presidential election, usage of the name Elizabeth in this study could possibly have triggered some association with Elizabeth Dole. Future studies should be cautious of possible name associations. A Comparison of Predicted and Observed Results In general terms, the research hypotheses predicted that (1) message theme and message clarity would affect the evaluation of female candidates; (2) candidate gender would affect the evaluation of female and male candidates; and (3) subject characteristics— political knowledge, print media use, and gender schema— would be associated with candidate evaluation. Message Theme and Clarity for Female Candidates It was expected that a clear message on crime would have positive results for female candidate evaluation, and it was found that the evaluation of policy competence for female candidates was indeed affected by the theme and clarity of a message. The clear message about crime affected the perception of the candidate's competence in crime (a stereotypically male policy area) in a positive manner. It appears that a message on crime has an advantage for female candidates in terms of breaking the gender barrier on so-called male policy areas, while still allowing them to be perceived as competent in policy areas typically associated with females. It was expected that a clear message would have a more positive effect than an unclear message on female candidate evaluation. This was found to be true only when the message was about crime, and only for the consideration of competence in crime. A post

PAGE 111

99 hoc analysis for the male candidate, however, indicated no significant effect on perceived competence in crime for message clarity of the crime message. Why might message clarity for the crime message be important for the female candidate but not for the male candidate? Perhaps clarity is important for the female candidate using the counter-stereotypical message, but it is not as important for the male candidate using a stereotypical message. This would indicate that female candidates need to be particularly attentive to clarity of their position in developing campaign messages that employ stereotypically male policy areas, such as crime and defense. For the female candidate using messages about education, findings may have been affected by the subjects' lack of perceived variation in clarity for the messages on education. Given that the education messages were written in language analogous in clarity to the crime messages, one might wonder if clarity was only important or salient to the subjects for stereotypically male policy areas. If so, is it because female candidates are expected to be competent in education, but not in crime? Since messages only about crime and education were tested, it is impossible to address the effects here of other types of messages. Furthermore, the type of data generated in this study does not reveal why clarity was important or salient for the male policy area, but not for the female policy area. Future studies will address this issue. The Effects of Candidate Gender on Candidate Evaluation It was predicted that the male candidate would be perceived more competent in the policy area of crime, while the female candidate would be perceived more competent in the area of education. However, results indicated that the female candidate was perceived to be

PAGE 112

100 on par with the male candidate for competence in crime. Findings indicated that the male and female candidate were generally perceived equal in competence, except when using a clear message on crime. For the unclear message on crime, no differences for candidate gender were found. The fact that the female candidate with the clear message on crime was perceived to be on par with the male candidate for competence in crime is a significant finding. Furthermore, for the clear message on crime, female candidates were favored in the policy areas of education, the arts, farming, and poverty. It is significant that the clear message on crime improved the perceived competence of the female candidate in farming issues— a stereotypically male policy area. As predicted in the section regarding Leeper's "tough message," the female candidate was advantaged by the clear message on crime in two ways: (1) she was rated higher than die male candidate in a stereotypically male policy area— farming; and (2) she was rated higher than the male candidate in three stereotypically female policy areas— the arts, education, and poverty. For the unclear message on crime, there was no apparent advantage for die female candidate, even in stereotypically female policy areas. A post hoc analysis of the effect of candidate gender on candidates using messages on education indicated no significant difference between the competence evaluation of female and male candidates. One possible explanation for this lack of gender effect is that the gender of the candidate may not have been well noted by the respondents. This could be due to the fact that candidate gender was only presented by the name identification of the hypothetical candidates, or it could be due to a lack of interest on the part of the subjects with regard to the candidate's gender. Perhaps, the subjects reading the messages on crime were more conscious of candidate gender, but why? Is it because subjects were more

PAGE 113

101 attentive to candidate gender when the message was counter-stereotypical for female candidates? In any case, the data collected in this study is not sufficient to answer these questions. Effects of Political Knowledge, Media Use, and Gender Schema on Candidate Evaluation It was predicted that higher levels of political knowledge would be associated with higher levels of print media use, and that both would be associated with higher ratings for female candidates. However, no statistical associations were found to support these hypotheses. Nor was a correlation found between political knowledge and print media use. However, in post hoc analysis, an association was found between political knowledge and respondent sex. Male respondents had a higher level of political knowledge than female respondents. Males were also found to be more traditional than females in their gender role schema, while females were found to be more egalitarian than males in their gender role schema. Perhaps the higher level of political knowledge by male respondents is counterbalanced by their more traditional gender role schema, thereby negating the predicted relationship between higher levels of political knowledge and higher ratings of female candidates. Post hoc analyses indicated a negative association between gender role schema and television news exposure, in that egalitarians were found to have lower levels of television news exposure and traditionalists were found to have higher levels of television news exposure. However, no statistically significant association was found between print media use and gender role schema. This is consistent with earlier findings regarding differential effects of media usage by media type.

PAGE 114

102 It was expected that subjects' gender role schemata would interact with candidate gender in its effect on candidate evaluation; however, this relationship was not supported by the data. A possible explanation for this finding pertains to the small cell sizes for the two traditionalist groups, one reading messages from the female candidate and the other reading messages from the male candidate. The small cell sizes would account for low statistical power of the analysis. Observed statistical power, in this case, represents the probability of detecting differences between the female and male candidates for individuals with egalitarian versus traditional gender role schema. Since the cell sizes (number of subjects in each cell) were quite small (Table 5), it is believed that cell sizes were inadequate for the analysis. It is also likely that the within-group variability was large in relation to the between-group variability, given that each of the cell groups contained both male and female subjects and that male and female subjects were found to have differences in their gender schema. Table 5: Cell Sizes for Test for Interaction Between Gender Role Schema and Candidate Gender Male Candidate Female Candidate Gender Schema Category Traditional 7 Moderate 68 Egalitarian 5 33 59 36

PAGE 115

103 Im plications for Theory Development The Gender-Based Message Debate Revisited Findings of this study indicate that when females use a clear message on crime, they are perceived equal or superior to males for competence in crime and farming, stereotypically male policy issues, while still considered superior to male candidates in several stereotypically female policy areas— education, the arts, and poverty. This result was similar to that of Leeper's research, in which the female candidate using the clear message on crime— the "tough" message— was perceived equal to the male candidate for competence in crime, while still being perceived more competent than the male candidate in dealing with education issues. Some researchers support the strategy of female candidates using messages that expose voters to the candidate's stereotypical strengths; others believe that female candidates are caught in the double bind of being perceived more competent when emphasizing male policy areas, but that they will be damaged by a backlash effect for violating sex role expectations. According to several authors, women can capitalize on common stereotypes by emphasizing issues where voters believe they are more competent— environment, health, education (Iyengar, Valentino, Ansolabehere, and Simon 1997; Kahn and Gordon 1997) and suggest that candidates are best advised to emphasize issues that resonate gender stereotypes. Their premise is that campaign communication is most persuasive when it conforms to voters' previous dispositions (Iyengar, et al. 1997). Leeper (1991), on the other hand, found that female candidates can benefit from using counter-stereotypical

PAGE 116

104 messages. This study supports the findings of the Leeper study in favor of the tough message for female candidates. Gender Schema Theory Gender schema theory suggests that gender-based stereotypes influence the processing of incoming information— such as campaign messages— concerning political candidates. Schemata are used as a filter, filtering in schema-consistent information, and filtering out schema-inconsistent information. When stereotypes are strong, respondents are frequently resistant to new information. According to gender schema theory, gender stereotypes would influence the evaluation of candidates based on the gender schema held by subjects. Results of this study do not indicate that gender role schema interacted with candidate gender to have an effect on candidate evaluation. This may simply be due to the small number of subjects found in this study to be traditional in gender role schema. While college students tend to be more egalitarian than the general population, other studies have been able to detect significant variability along the gender role egalitarianism scale. One possible explanation for this is the length of the instrument used to test the gender role egalitarianism of subjects. As discussed earlier and has been the case in other studies, a gender role egalitarian scale that includes more items may detect greater variability within the subject pool. Another explanation for the failure to demonstrate the hypothesized interaction may be due to the manner in which candidate gender is presented in this study. Or it may be a function of how gender role schema are activated in the candidate evaluation process. In other words, the messages read by the subjects may not activate gender schemata, even

PAGE 117

105 when they are existent within the subjects. The addition of video or audio messages could possibly increase the salience of candidate gender. Candidate Evaluation Modeling A major purpose of this study was the integration of specific variables related to persuasive communication into a model of candidate evaluation. In this sense, this study was partially successful. The proposed model for predicting evaluation of a candidate's competence has been, at least partially, supported for the following components: candidate gender, message theme, and message clarity (at least in relation to the crime messages). Predictions stated in the research hypotheses regarding the subject characteristics were not supported. That is, political knowledge was not positively associated with female candidate evaluation; print media use was not significantly associated with female candidate evaluation; and gender schema did not significantly interact with candidate gender for its effect on candidate evaluation. However, particular relationships (and non-relationships) among the subject characteristics have been identified: (1) political knowledge was negatively associated with gender role egalitarianism; (2) political knowledge was not associated with print media use or television news watching; and (3) gender role egalitarianism was negatively associated with watching television news. Previous research has indicated that as people become more expert, they become more aware of inconsistencies between data and schema and more cautious in applying prototypical or stereotypical attributes of a schema to specific cases (Lau and Sears 1986b; Milburn 1991). Therefore, it was expected in this study that higher levels of political knowledge would be associated with more egalitarian gender schema. It was found,

PAGE 118

106 however, that the opposite was true of the subjects from which the data was collected. The higher levels of political knowledge were associated with more traditional gender schema. This may be due to the counterbalancing effect of respondent sex, in which male respondents (who tended to be more traditional) also tended to be more knowledgeable about politics, while female respondents tended to be less knowledgeable politically, but more egalitarian in their thinking about gender roles in society. With respect to the findings regarding the political knowledge-media use relationship, while other researchers have indicated that the mass media are instruments for increasing levels of political knowledge and sophistication (Drew and Weaver 1990; Emig 1991; Pan et al. 1994), the data in this study do not support a relationship between political knowledge and media use. Previous research indicates variability in political knowledge associated with "differential media usage," in that political experts are more likely to rely on newspapers for political information, while novices are more likely to rely on television (Lau and Erber 1985; Miller and Asp 1985). This study, however, indicates no differential media use with respect to level of political knowledge. The negative association between gender role egalitarianism and television news exposure indicated in this study is consistent with previous research. In recent research, it was found that the strongest predictors of more complex political schemata and candidate evaluation were associated with regular newspaper and newsmagazine reading, but not with watching television news (Drew and Weaver 1990; Pan et al. 1994; Robinson and Levy 1996). Because individuals with less complex candidate schemata are more reliant on gender stereotypes, it follows that gender role egaltarianism would be negatively associated

PAGE 119

107 with watching television news; that is, that high users of television would be more traditional in their gender schema ideology. While the complete candidate evaluation model proposed in the research hypotheses of this study was not supported, several elements were supported by the data. Candidate gender, message theme, and message clarity were shown to have effects on candidate evaluation. Political knowledge, media exposure, and gender role schema were not shown to have effects on candidate evaluation. Implications for Political Campaign Strategy Female candidates have the potential to be both compassionate and strong, nurturing and tough. The question for political strategists is "Which types of leadership traits should be reflected in their campaign images?" Some strategists believe that women running for office do best to emphasize their unique "womanly" qualities. Others believe that women campaigning for office who tout their gender differences from men create their own political glass ceiling. This research indicates that female candidates can use a "tough" message to emphasize strength in stereotypically male policy areas 1* such as crime, while still being perceived as possessing strength in traditionally female policy areas, such as education and assisting the poor. While the tough message was found to be more potent than other messages for the hypothetical female candidate, female candidates and their political strategists should keep in mind the importance of political context — an issue that goes beyond the scope of this study. When faced with a choice regarding messages for female candidates, they should

PAGE 120

108 consider the relative value of the tough message within the context of a specific campaign. For example, is crime or some other stereotypically male policy area an important and salient public issue within the campaign? Or perhaps, have stereotypically female policy areas, such as education and poverty, emerged as prominent issues in the campaign? Furthermore, have gender role stereotypes been transformed by societal trends? As society changes, our conceptualization of gender roles and political leadership changes, and women running for office would do well to consider these trends when developing their political communication strategies. In the mean time, the tough message may serve the female candidate well as a vehicle of credibility. Suggestions for Future Research The following suggestions for future research are intended to provide avenues for strengthening theory about the relationship among candidate gender, message characteristics, and perceived candidate competence. In this dissertation, candidate gender was represented only by identification of candidate's name on the written campaign speeches. Future studies that include video or audio messages should be incorporated into experimental designs in order to test the possibility of higher sensitivity to candidate gender. This approach would integrate additional source characteristics not included in this study. Therefore, the experimental design would need to increase in complexity in order to 12 Because male and female stereotypes for policy areas used in the current study were based on previous studies, future studies should readdress the nature of these stereotypes, particularly as they relate to both college and general populations.

PAGE 121

109 distinguish between sex of the candidate and other characteristics, such as style of message delivery and appearance of the candidate. A second factor that could be added to future study in gender-based candidate evaluation is more diversity in message theme. Messages involving additional policy areas— such as the economy, farming, and poverty— could be tested for effects on candidate evaluation. In relation to message clarity, additional research is needed to determine the differential effects of clarity for different message themes. Message theme and clarity could be tested for interactions in relation to additional policy areas. If stereotypically male and stereotypically female policy areas (in addition to crime and education) are found to have the same effects as crime and education, more generalized conclusions could be supported. Third, with respect to the interaction between message theme and clarity, not only is it important to determine what the effects are on candidate evaluation, but also to seek answers to the question of why clarity might be more important or salient to the female candidate for stereotypically male policy areas than for female policy areas. Relatedly, messages should be extensively tested, perhaps using focus groups, to further examine perceptions of clarity. Fourth, future research in candidate evaluation should test longer, more inclusive instruments for measuring gender role egalitarianism. While previous studies have compared 15-item and 25-item scales with longer scales, there is a need to compare the validity of even shorter scales, such as the six-item scale used in the current study and in the Alexander and Andersen study, with the longer scales used in other studies. Finally, with respect to the subject characteristics of political knowledge and print media exposure, it would be useful if future research further addressed the relationships

PAGE 122

110 between (1) political knowledge and gender schema; and (2) media exposure and gender schema. A generalized assessment of political knowledge of voters may not be as relevant to female candidate evaluation as specific domains of political knowledge and interest, and these may relate to contextual political environments within specific election campaigns. As for media exposure, generalized levels of exposure may not be as significant as selectivity of attention to specific issues and events. Therefore, future research should investigate the relationship between selectivity of media attention and gender schemata as it pertains to candidate evaluation. Future research should also draw subjects from populations that would have a higher representation of gender role traditionalists than college students are likely to have. In this way, more significant comparisons between traditionalists and egalitarians could be made. This study has provided a significant foundation for the construction of a theoretical model of gender-based candidate evaluation. Significant findings pertaining to message effects should be beneficial to female candidates and their advisors. In addition, findings regarding differences in gender role schemata between female and male respondents should be considered in the development of communication campaigns for candidates. Given the limitations of scope in this study, the possibilities that gender schema effects on candidate evaluation may exist should be addressed in future research.

PAGE 123

APPENDIX A CAMPAIGN MESSAGES Elizabeth Preston for Congress Campaign Speech by Elizabeth Preston-a Hypothetical Candidate for Congress I have a vision of a better nation. I want to tell you about my dream, and ask you to help me make it a reality. My vision is of a nation with strong, healthy people living in a strong, healthy environment. My vision is of a nation with good jobs for all who will work for a better life. The fight against drugs and other ills of our society must be waged on many fronts: by our law enforcement agencies, by help from the media, by our schools, by our mothers and fathers, and by our religious institutions. Our youth must be taught that crime doesn't pay. Specifically, we should enact policies and laws that (1) treat the serious juvenile offender as a criminal in need of incarceration; (2) provide for building more prisons to keep criminals off the street; and (3) levy a special "20 cents a pack" cigarette tax to raise dollars for that purpose. On the economy, the number one task for a brighter future is to create more and better jobs. We can do as much for citizenry as we dare to dream. This is not a pipedream...It is a vision for the immediate future. To achieve this dream, IElizabeth Preston-need your help, and that of your friends and neighbors. I need the help of all the people of this District, working with diligence and with unity to help make it an even greater nation. [condition 1] 111

PAGE 124

112 Elizabeth Preston for Congress Campaign Speech by Elizabeth Preston ~a Hypothetical Candidate for Congress I have a vision of a better nation. I want to tell you about my dream. . .and ask you to help me make it a reality. My vision is of a nation with strong, healthy people living in a strong, healthy environment. My vision is of a nation with good jobs for all who will work for a better life. The fight against drugs and other ills of our society must be waged on many fronts: by our law enforcement agencies, by help from the media, by our schools, by our mothers and fathers, and by our religious institutions. We should get tougher on crime. The crime rate is too high, the prisons are crowded, and our youth is prey to drugs. We have a moral obligation to do what needs to be done to rid our society of the evils of drugs and violence. Strong family values and education offer the best prospects for a society of law-abiding citizens. On the economy, the number one task for a brighter future is to create more and better jobs. We can do as much for citizenry as we dare to dream. This is not a pipedream...It is a vision for the immediate future. To achieve this dream, I— Elizabeth Preston-need your help, and that of your friends and neighbors. I need the help of all the people of this District, working with diligence and with unity to help make it an even greater nation. [condition 2]

PAGE 125

113 Elizabeth Preston for Congress Campaign Speech by Elizabeth Preston--a Hypothetical Candidate for Congress I have a vision of a better nation. I want to tell you about my dream... and ask you to help me make it a reality. My vision is of a nation with strong, healthy people living in a strong, healthy environment. My vision is of a nation with good jobs for all who will work for a better life. The fight for better jobs through better education must be waged by all sectors of our society-by our government, by our business community, by our schools, by our mothers and fathers, and by our religious institutions. Our youth must be taught that education is important for their future. Specifically, we should enact policies and laws that (1) treat education as a priority; (2) provide for building more schools to alleviate overcrowding; and (3) levy a special "20 cents a pack" cigarette tax to raise dollars for these purposes. On the economy, the number one task for a brighter future is to create more and better jobs. We can do as much for citizenry as we dare to dream. This is not a pipe dream... It is a vision for the immediate future. To achieve this dream, I-Elizabeth Preston-need your help, and that of your friends and neighbors. I need the help of all the people of this District, working with diligence and with unity to help make it an even greater nation. [condition 3]

PAGE 126

114 Elizabeth Preston for Congress Campaign Speech by Elizabeth Preston-a Hypothetical Candidate for Congress I have a vision of a better nation. I want to tell you about my dream. . .and ask you to help me make it a reality. My vision is of a nation with strong, healthy people living in a strong, healthy environment. My vision is of a nation with good jobs for all who will work for a better life. The fight for better jobs through better education must be waged by all sectors of our society-by our government, by our business community, by our schools, by our mothers and fathers, and by our religious institutions. Education is the path to a better future. Our youth should be taught that education is important for their future. Our schools are in disrepair and are overcrowded. We have a moral obligation to do what needs to be done to make our schools safe and functional. Strong family values and education offer the best prospects for a society of productive citizens. On the economy, the number one task for a brighter future is to create more and better jobs. We can do as much for citizenry as we dare to dream. This is not a pipedream...It is a vision for the immediate future. To achieve this dream, IElizabeth Preston-need your help, and that of your friends and neighbors. I need the help of all the people of this District, working with diligence and with unity to help make it an even greater nation. [condition 4]

PAGE 127

115 Jim Preston for Congress Campaign Speech by Jim Preston-a Hypothetical Candidate for Congress I have a vision of a better nation. I want to tell you about my dream. . .and ask you to help me make it a reality. My vision is of a nation with strong, healthy people living in a strong, healthy environment. My vision is of a nation with good jobs for all who will work for a better life. The fight against drugs and other ills of our society must be waged on many fronts: by our law enforcement agencies, by help from the media, by our schools, by our mothers and fathers, and by our religious institutions. Our youth must be taught that crime doesn't pay. Specifically, we should enact policies and laws that (1) treat the serious juvenile offender as a criminal in need of incarceration; (2) provide for building more prisons to keep criminals off the street; and (3) levy a special "20 cents a pack" cigarette tax to raise dollars for that purpose. On the economy, the number one task for a brighter future is to create more and better jobs. We can do as much for citizenry as we dare to dream. This is not a pipedream...It is a vision for the immediate future. To achieve this dream, IJim Preston-need your help, and that of your friends and neighbors. I need the help of all the people of this District, working with diligence and with unity to help make it an even greater nation. [condition 5]

PAGE 128

116 Jim Preston for Congress Campaign Speech by Jim Preston -a Hypothetical Candidate for Congress I have a vision of a better nation. I want to tell you about my dream. . .and ask you to help me make it a reality. My vision is of a nation with strong, healthy people living in a strong, healthy environment. My vision is of a nation with good jobs for all who will work for a better life. The fight against drugs and other ills of our society must be waged on many fronts: by our law enforcement agencies, by help from the media, by our schools, by our mothers and fathers, and by our religious institutions. We should get tougher on crime. The crime rate is too high, the prisons are crowded, and our youth is prey to drugs. We have a moral obligation to do what needs to be done to rid our society of the evils of drugs and violence. Strong family values and education offer the best prospects for a society of law-abiding citizens. On the economy, the number one task for a brighter future is to create more and better jobs. We can do as much for citizenry as we dare to dream. This is not a pipedream...It is a vision for the immediate future. To achieve this dream, IJim Preston-need your help, and that of your friends and neighbors. I need the help of all the people of this District, working with diligence and with unity to help make it an even greater nation. [condition 6]

PAGE 129

117 Jim Preston for Congress Campaign Speech by Jim Preston-a Hypothetical Candidate for Congress I have a vision of a better nation. I want to tell you about my dream... and ask you to help me make it a reality. My vision is of a nation with strong, healthy people living in a strong, healthy environment. My vision is of a nation with good jobs for all who will work for a better life. The fight for better jobs through better education must be waged by all sectors of our society--by our government, by our business community, by our schools, by our mothers and fathers, and by our religious institutions. Our youth must be taught that education is important for their future. Specifically, we should enact policies and laws that (1) treat education as a priority; (2) provide for building more schools to alleviate overcrowding; and (3) levy a special "20 cents a pack" cigarette tax to raise dollars for these purposes. On the economy, the number one task for a brighter future is to create more and better jobs. We can do as much for citizenry as we dare to dream. This is not a pipedream...It is a vision for the immediate future. To achieve this dream, IJim Preston-need your help, and that of your friends and neighbors. I need the help of all the people of this District, working with diligence and with unity to help make it an even greater nation. [condition 7]

PAGE 130

118 Jim Preston for Congress Campaign Speech by Jim Preston-a Hypothetical Candidate for Congress I have a vision of a better nation. I want to tell you about my dream. . .and ask you to help me make it a reality. My vision is of a nation with strong, healthy people living in a strong, healthy environment. My vision is of a nation with good jobs for all who will work for a better life. The fight for better jobs through better education must be waged by all sectors of our society-by our government, by our business community, by our schools, by our mothers and fathers, and by our religious institutions. Education is the path to a better future. Our youth should be taught that education is important for their future. Our schools are in disrepair and are overcrowded. We have a moral obligation to do what needs to be done to make our schools safe and functional. Strong family values and education offer the best prospects for a society of productive citizens. On the economy, the number one task for a brighter future is to create more and better jobs. We can do as much for citizenry as we dare to dream. This is not a pipedream...It is a vision for the immediate future. To achieve this dream, IJim Preston-need your help, and that of your friends and neighbors. I need the help of all the people of this District, working with diligence and with unity to help make it an even greater nation. [condition 8]

PAGE 131

APPENDIX B QUESTIONNAIRES Political News and Views 1 . What are the last 5 digits of your social security #? _ _ (This is for form ID purposes only. Your answers are completely confidential). We want to see how much information about public figures gets out to the public from television, newspapers, etc. What job or political office does each one hold? The names are 2. Clinton 3. Gingrich_ 4. Cohen 5. Kerry 6. Rehnquist_ 7. Greenspan_ Do you happen to remember the names of the members of the U.S. House of Representatives in Washington from this district (Alachua County)? yes no If yes, who are they? 8. 9. 10. 119

PAGE 132

120 Below is a seven-point scale on which the political views that people might hold are arranged from extremely liberal to extremely conservative. 1 1 . Where would you place the Democratic Party on this scale? Circle one. 1. extremely liberal 5. slightly conservative 2. liberal 6. conservative 3. slightly liberal 7. extremely conservative 4. moderate; middle of the road8. don't know 12. Where would you place the Republican Party on this scale? Circle one. 1 . extremely liberal 2. liberal 3. slightly liberal 4. moderate; middle of the road 5. slightly conservative 6. conservative 7. extremely conservative 8. don't know 13. Where would you place Clinton on this scale? Circle one. 1 . extremely liberal 2. liberal 3. slightly liberal 4. moderate; middle of the road 5. slightly conservative 6. conservative 7. extremely conservative 8. don't know Some people believe that we should spend much less money for defense. Others feel that defense spending should be greatly increased. Here is a scale for defense spending. 14. On a scale of 1 to 7, where would you place the Democratic Party on this scale? Circle one. 1. 2. 3. 4. 5. 6. 7. greatly decrease greatly increase defense spending defense spending

PAGE 133

121 15. Where would you place the Republican Party on this scale? Circle one. 1. 2 3. 4. 5. 6. 7. greatly decrease great l y in Crease defense spending defense spending Some people think the U.S. should become more involved in the internal affairs of Central America. Others believe we should become less involved. 16. Where would you place the Democratic Party on this scale? Circle one. ! 2. 3. 4. 5. 6. 7. much less much more involved involved 17. Where would you place the Republican Party on this scale? Circle one. !• 2. 3. 4. 5. 6. 7. much less much more involved involved Some people think the government should provide fewer services. Other people feel it is important for the government to provide many more services even if it means an increase in spending. 18. Where would you place the Democratic Party on this scale? Circle one. !• 2. 3. 4. 5. 6. 7. provide many provide many fewer services more services 19. Where would you place the Republican Party on this scale? !• 2. 3. 4. 5. 6. 7. provide many provide many fewer services more services 20. Do you happen to know which party has the most members in the House of Representatives in Washington?

PAGE 134

122 Republicans Democrats don't know Now we would like to ask you if you strongly agree, agree, disagree, strongly disagree, or have no opinion about the following statements. 21 . It is more important for a wife to help her husband's career than to have a career herself. 1 2 3 4 5 6 7 ver y very no opinion/ don't strongly strongly know agree disagree 22. A preschool child is likely to suffer if his or her mother works. 1 2 3 4 5 6 7 ver y very no opinion/ don't strongly strongly know a g ree disagree 23. Women should be able to compete with men for jobs that have traditionally belonged to men, such as firefighters. 1 2 3 4 5 6 7 ver y very no opinion/don't strongly strongly know a S ree disagree 24. A husband should not feel uncomfortable if his wife earns a larger salary than he does. 1 2 3 4 5 6 7 ver y very no opinion/don't strongly strongly know agree disagree 25. It is much better for everyone involved if the man is the achiever outside the home and the woman takes care of the home and family. 1 2 3 4 5 6 7 ver y very no opinion/don't strongly strongly know a S ree disagree

PAGE 135

123 26. Most men are better suited emotionally for politics than most women. 12 3 4 5 6 7 . Ve . ry very no opinion/ Str0ngly strongly don't know agree disagree 27. It is a good idea for a husband to stay at home and care for the children while his wife is employed outside the home. 12 3 45 very strongly agree Finally, I'd like to ask you a few background questions for statistical purposes. 28. What is your age in years? 29. Are you (1) female (2) male 6 7 very strongly disagree noopinion/ don't know 30. What is your political party affiliation? Democratic Republican Independent don't know Thank you for your participation in this study.

PAGE 136

124 Campaign Speech 1 . What are the last 5 digits of your social security # ? (This is for form ID purposes only. Your answers are completely confidential). 2. When you were reading Elizabeth/ Jim's Preston's speech, what thoughts came to mind? 3. How well do you think Elizabeth/Jim Preston understands the issues discussed in the speech? 12 3 4 5 6 7 Very Not well well at all 4. How clear are Elizabeth/Jim Preston's stands on the issue of crime/education? 12 3 4 5 6 7 Very Not clear clear at all 5. If Elizabeth/Jim Preston's proposals are put into action, what kind of effect do you think they would have? 12 3 4 5 6 7 Very Very positive negative

PAGE 137

125 6. How much do you agree or disagree with Elizabeth/Jim Preston? 12 3 4 5 6 7 Strongly Strongly agree disagree 7. What is the likelihood that you would vote for Elizabeth/Jim Preston? 12 3 4 5 6 7 Very Very likely unlikely 8. What is the likelihood that Elizabeth/Jim Preston will win the election? 12 3 4 5 6 7 Very Very likely unlikely 9. List three specific qualities or characteristics that you believe a U.S. Senator should possess. a) b) O 10. Based on your impressions drawn from the speech, please provide your best guess as to Elizabeth Preston's competence regarding the following issues: 1 = very 7 = very competent competent a. Strengthening the economy 12 3 4 5 6 7 b. Assisting the poor 12 3 4 5 6 7 c. Dealing with big business 12 3 4 5 6 7 d. Improving our educational system 12 3 4 5 6 7 e. Dealing with crime 12 3 4 5 6 7 f. Dealing with health care 12 3 4 5 6 7

PAGE 138

126 11. g. Dealing with employee unions h. Encouraging the arts i. Making decisions on farm issues 12 3 4 5 6 7 12 3 4 5 6 7 12 3 4 5 6 7 Please indicate what you believe to be Elizabeth Preston's general ideological position: (1) very liberal (2) somewhat liberal (3) slightly liberal (4) middle of the road (5) slightly conservative (6) somewhat conservative (7) very conservative , 12. Please indicate what you believe to be your own ideological position: (1) very liberal (2) somewhat liberal (3) slightly liberal (4) middle of the road (5) slightly conservative (6) somewhat conservative (7) very conservative 13. How many hours in an average week do you watch T.V. news/ hours a week 14. How many hours in an average week do you read news stories in the newspaper? hours a week 15. How many hours in an average week do you listen to radio news? hours a week Thank you for participating in this study.

PAGE 139

REFERENCES Agresti, A. and Finlay, B. 1986. Statistical Methods for the Social Sciences. San FranciscoDellen. Alexander, D. 1995. The Influence of Gender Beliefs in Constructing Candidate Images. A Ph.D. Dissertation at Syracuse University. Alexander, D. and Andersen, K. 1993, September. Gender as a Factor in the Attribution of Leadership Traits. Political Research Quarterly 46(3) 527-545. Allen, C. 1990. Political Campaigning: A New Decade. Washington, D.C.: National Women's Political Caucus. Ansolabehere, S., Behr, R., and Iyengar, Shanto. 1991. Mass Media and Elections. American Political Quarterly 19(1) 109-139. Ashmore, R.D., Del Boca, F.K., and Wohlers, A. J. 1986. Gender Stereotypes. In Ashmore and Del Boca's (eds.) The Social Psychology of Female-Male Relations: A Critical Analysis of Central Concepts. Orlando: Academic Press, Inc. Babbie, E. 1995. The Practice of Social Research. Belmont: Wadsworth. Barker, H. R. and Barker, B.M. 1984. Statistical Power. In Multivariate Analysis of Variance: A Practical Guide to Its Use in Scientific Decision Making, pp. 31-33. University, AL: University of Alabama Press. Beasley, M. 1997. How Can Media Coverage of Women be Improved? In Pippa Norris's (ed.) Women, Media, and Politics. NY: Oxford University Press. Belenky, M. F., Clinchy, B. M., Goldberger, N. R., and Tarule, J. M. 1997). Women's Ways of Knowing: The Development of Self , Voice, and Mind. NY: BasicBooks. Bern, S.L. 1993. The Lenses of Gender: Transforming the Debate on Sexual Inequality. New Haven: Yale University Press. Bern, S.L. 1981. Gender Schema Theory: A Cognitive Account of Sex Typing. Psychological Review 88 354-64. 127

PAGE 140

128 Berch, N. 1996. The "Year of the Woman" in Context. American Politics Quarterly 24(2) 169-193. Bernstein, A. G. 1996. Gender and Political Credibility: A Meta-Analytic Review. A paper presented to the annual meeting of Northeastern Political Science Association, Boston, MA. Bernstein, A. G. 1995. Who Should We Get to Endorse the Proposition? The Persuasive Effect of Source Credibility in Ballot Proposition Campaigns. A paper presented to the annual meeting of the International Communication Association, Alberquerque, NM. Bourdieu, P. 1991. Language and Symbolic Power. Cambridge: Harvard University Press. Brians, C. L. and Wattenberg, M.P. 1996, February. Campaign Issue Knowledge and Salience: Comparing Reception from TV Commercials, TV News, and Newspapers. American Journal of Political Science 40(1) 172-193. Brogan, D. and Kutner, N.G. 1976, February. Measuring Sex-Role Orientation. Journal of Marriage and the Family 38 (1) 31-40. Brown, C, Heighberger, N. R., and Shocket, P. A. 1993. Gender-Based Differences in Perceptions of Male and Female City Council Candidates. Women and Politics 13(1)1-17. Burgoon, M., Dillard, J. P., and Doran, N. E. 1983, winter. Friendly or Unfriendly Persuasion: The Effects of Violations of Expectations by Males and Females. Human Communication Research 10(2) 283-294. Cantor, D.W., Bernay, T., and Stoess, J. 1992. Women in Power: The Secrets of Leadership. New York: Houghton Mifflin. Carroll, S. J. and Schreiber, R. 1997. Media Coverage of Women in the 103rd Congress. InPippa Norris's (ed.) Women, Media, and Politics. NY: Oxford University Press. Chung, M.S. 1984 The Relationship of Media Exposure, Need for Cognition, and Media Dependency. A thesis presented to the University of Florida. Conover, P. and Feldman, S. 1986. The Role of Inference in die Perception of Political Candidates. In Richard R. Lau and David O. Sears (eds.) Political Cognition. Proceedings from the 19th Annual Carnegie Symposium on Cognition. Hillsdale, NJ: Erlbaum.

PAGE 141

129 Cook, T. D. and Campbell, D. T. 1979. Quasi-Experimentation: Design & Analysis Issues for Field Settings. Boston: Houghton Mifflin. Cooper, H. and Hedges, L. V. (Eds.). 1994. The Handbook of Research Synthesis. New York: The Russell Sage Foundation. Darcy, R., Welch, S., and Clark, J. 1987. Women, Elections, and Representation. New York: Longman. Deaux, K. and Lewis, L. L. 1984. Structure of Gender Stereotypes: Interrelationships Among Components and Gender Label. Journal of Personality and Social Psychology 46 (5) 991-1004. Delli Carpini, M. X. and Keeter, S. 1996. What Americans Know about Politics and Why It Matters. New Haven: Yale University Press. Delli Carpini, M. X. and Keeter, S. 1993, November. Measuring Political Knowledge: Putting First Things First. American Journal of Political Science 37 (4) 1 179-1206. Drew, D. and Weaver, D. 1990, Winter. Media Attention, Media Exposure, and Media Effects. Journalism Quarterly 67 (4) 740-748. Eagly, A. H. and Steffen, V. J. 1984. Gender Stereotypes Stem From the Distribution of Women and Men into Social Roles. Journal of Personality and Social Psychology 46 (4) 735-754. Eckes, T. 1994. Explorations in Gender Cognition: Content and Structure of Female and Male Subtypes. Social Cognition 12 (1) 37-60. Eckstrand, L. E. and Eckert, W.A. 1981. The Impact of Candidate's Sex on Voter Choice. Western Political Quarterly 34 78-87. Emig, A. G. 1991. Media Use Patterns, Political Sophistication, and Political Discontent. Unpublished doctoral dissertation, University of Florida, Gainesville. Fiske, S. T., Lau, R. R., and Smith, R. A. 1990, Spring. On the Varieties and Utilities of Political Expertise. Social Cognition. (Special Issue). Thinking about Politics: Comparisons of Experts and Novices. 8(1) 31-48. Fiske, S.T. 1986. Schema-Based Versus Piecemeal Politics: A Patchwork Quilt, But Not a Blanket, of Evidence. In Richard R. Lau and David O. Sears' Political Cognition. Hillsdale, NJ: Erlbaum. Fowler, F. J., Jr. 1993. Survey Research Methods. Newbury Park: Sage.

PAGE 142

130 Garramone, G. M., Steele, M.E., and Pinkleton, B. 1994. The Role of Cognitive Schemata in Determining Candidate Characteristic Effects. In Frank Biocca (Ed.). Television and Political Advertising. Hillsdale, NJ: Erlbaum. Garramone, G. M. 1992. A Broader and Warmer Approach to Schema Theory. In Stanley A. Deetz (Ed.) Communication Yearbook 15. Goldenberg, E. N. and Traugott, M.W . 1987. Mass Media Effects on Recognizing and Rating Candidates in U.S. Senate Elections. In Jan Pons Vermeer (Ed.) Campaigns in the News: Mass Media and Congressional Elections. NY: Greenwood Press. Graziano, A. M. and M.L. Raulin. 1989. Research Methods: A Process of Inquiry. NY: Harper and Row. Hamill, R. and Lodge, M. 1986. Cognitive Consequences of Political Sophistication. In Richard R. Lau and David O. Sears' Political Cognition. Hillsdale, NJ: Erlbaum. Hamilton, D. L. and Sherman, J.W. 1994. Stereotypes. In Robert S. Wyer, Jr. and Thomas K. Srull (Eds.) Handbook of Social Cognition Vol. 2: Applications. Hillsdale, NJ: Lawrence Erlbaum Associates. Hedlund, R. D., Freeman, P. K., Hamm, K. E., and Stein, R. M. 1979. The Electability of Women Candidates: The Effects of Sex Role Stereotypes. The Journal of Politics 41: 513-524. Herrnson, P.S. 1995, September. Replication, Verification, Secondary Analysis, and Data Collection in Political Science. PS: Political Science and Politics 28 (3) 452-455. Hitchon, J.C., Chang, C.C., and Harris, R. 1997. Should Women Emote? Perceptual Bias and Opinion Change in Response to Political Ads for Candidates of Different Genders. Political Communication 14 49-69. Hodgson, S. and Pryor, B. 1984, October. Sex Discrimination in the Courtroom: Attorney's Gender and Credibility. Psychological Reports 55(2) 483-486. Howell, D.C. 1992. Statistical Methods for Psychology. Belmont, CA: Wadsworth. Huddy, L. and Terkildsen, N. 1993, February. Gender Stereotypes and the Perception of Male and Female Candidates. American Journal of Political Science 37(1) 1 19-147. Infante, D.A. (1985, Spring). Inducing Women to Be More Argumentative: Source Credibility Effects. Journal of Applied Communication Research. 13(1): 33-44.

PAGE 143

131 Iyengar S Peters M. and Kinder, D.R. 1982. Experimental Demonstrations of the NotSo-Mmimal Consequences of Television News Programs. American Political ScienceReview 81: 848-858. Iyengar S and Ottati, V. 1994. Cognitive Perspective in Political Psychology. In Robert S. Wyer, Jr. and Thomas K. Srull (Eds.) Handbook of Social Cognition Vol 2Applications. Hillsdale, NJ: Lawrence Erlbaum Associates. Iyengar, S Valentino, Nicholas A., Ansolabehere, S., and Simon, A.F. 1997 Running as a Woman: Gender Stereotyping in Political Campaigns. In Pippa Norris's (ed ) Women, Media, and Politics. NY: Oxford University Press. Jamieson K.H. 1995. Beyond the Double Bind: Women and Leadership. NYOxford University Press. Johnston, A. and White, A. B. 1994., summer. Communication Styles and Female Candidates: A Study of the Political Advertising During the 1986 Senate Elections Journalism Quarterly 7 1 (2) 32 1 -329. Kahn, K.F. 1994, February. Does Gender Make a Difference'? An Experimental Examination of Sex Stereotypes and Press Patterns in Statewide Campaigns. American Journal of Political Science 38(1) -.162-195. Kahn, K.F 1992, May. Does Being Male Help? An Investigation of the Effects of Candidate Gender and Campaign Coverage on Evaluations of U.S. Senate Candidates. The Journal of Politics 54(2) 497-517. Kahn, K.F. and Gordon, A. 1997. How Women Campaign for the U S Senate In Pippa Norris's (ed.) Women, Media, and Politics. NY: Oxford University Press. Kahn, K.F. 1991. Senate Elections in the News: Examining Campaign Coverage Legislative Studies Quarterly 16: 349-374. Kaid, L.h. 1981. Political Advertising. In D.D. Nimmo and K.R. Sanders' (Eds ) Handbook of Political Communication. Beverly Hills: Sage. Kate, P Silvern, L and Coulter, D.K. 1990. Gender Processing and Person Perception Social Cognition 8 (2) 186-202. Kenton SB. (1989 >, Spring). Speaker Credibility in Persuasive Business Communicationtn^l^f EXplalnS ° ender Differences Journal of Business Communication. 26(2): 143(15). Kerlinger, F.N. 1986. Foundations of Behavioral Research. Fort WorthHolt Rinehart and Winston.

PAGE 144

132 King, L.A. and King, D.W. 1997, March. Sex Role Egalitarianism Scale. Psychology of Women Quarterly 2 1 : 7 1 -87. King, L.A., King, D.W., Carter, D.B., Surface, C.R., and Stepanski, K. 1994. Validity of the Sex Role Egalitarianism Scale: Two Replication Studies. Sex Roles 31-339348. King, D.C. and Matland, R.E. 1997, March. Partisanship and the Impact of Candidate Gender in Congressional Elections: Results of an Experiment. Paper presented at the meeting of the Southwestern Social Science Association, New Orleans. King, G. 1995, September. Replication, Replication. PS: Political Science and Politics. Kirk, R.E. 1982. Experimental Design: Procedures for the Behavioral Sciences. Monterey C4:Brooks/Cole. Kraemer, H.C. and Thiemann, S. 1987. How Many Subjects? Statistical Power Analysis in Research. Newbury Park: Sage. Krosnick, J.A. 1990a, Spring. Lessons Learned: A Review and Integration of Our Findings. Social Cognition 8(1) 154-158. Krosnick, J.A. 1990b, Spring. Expertise and Political Psychology. Social Cognition 8(1) 18. Lau, R.R. 1986. Political Schemata, Candidate Evaluations, and Voting Behavior. In Richard R. Lau and David 0. Sears' (Eds.) Political Cognition. Hillsdale, NJ: Erlbaum. Lau, R.R. and Erber, R. 1985. Political Sophistication: An Information-Processing Perspective. In Sidney Kraus and Richard M. Perloff s (Eds.) Mass Media and Political Thought. Beverly Hills: Sage Publications. Lau, R.R. and Sears, D.O. 1986a. An Introduction to Political Cognition. In Richard R. Lau and David 0. Sears' (Eds.) Political Cognition. Hillsdale, NJ: Erlbaum. Lau, R.R. and Sears, D.O. 1986b. Social Cognition and Political Cognition: The Past, the Present, and the Future. In Richard R. Lau and David O. Sears' (Eds.) Political Cognition. Hillsdale, NJ: Erlbaum. Leeper, M.S. 1991. The Impact of Prejudice on Female Candidates: An Experimental Look at Voter Inference. American Politics Quarterly 19:248-261.

PAGE 145

133 Lumsden, G. and Lumsden, D. 1996. Communicating with Credibility and Confidence. NY: Wadsworth Publishing Company. Manheim, J.B. and Rich, R.C. 1995. Empirical Political Analysis: Research Methods in Political Science. White Plains: Longman. Mannheim, K. 1985. The Prospects of Scientific Politics." In Ideology and Utopia: An Introduction to the Sociology of Knowledge. San Diego: Harcourt, Brace & Co. Matland, R.E. 1994a. Putting Scandinavian Equality to the Test: An Experiment of Campaigns in the News: Mass Media and Congressional Elections. Unpublished manuscript. Matland, R. E. 1994b. Putting Scandinavian Equality to the Test: An Experimental Evaluation of Gender Stereotyping of Political Candidates in a Sample of Norwegian Voters. British Journal of Political Science 24 273-292. McGlen, N.E. and Sarkees, M.R. 1993. Societal Factors: Sex Roles and Stereotypes. In Women in Foreign Policy: The Insiders. NY: Routledge. McGraw, K.M. and Lodge, M. 1996. Political Information Processing: A Review Essay. Political Communication 13 131-142. McGraw, K.M. and Pinney. 1990. General and Domain-Specific Expertise. Social Cognition 8(1). McGuire, W.J. 1985. Attitudes and Attitude Change. In G. Lindzey and E. Aronson (Eds.) Handbook of Social Psychology: Vol. II. NY: Erlbaum. McHugh, M.C. and Frieze, I.H. 1997, March. The Measurement of Gender-Role Attitudes. Psychology of Women Quarterly 21:1-16. McLeod, J.M., Kosicki, G.M., and McLeod, D.M. 1994. The Expanding Boundaries of Political Communication Effects. In Jennings Bryant and Dolf Zillman (Eds.) Media Effects: Advances in Theory and Research. Hillsdale, NJ: Erlbaum. Medvic, S.K. 1996. 1996, November. A Theory of Deliberate Priming. Paper presented at the meeting of the Northeastern Political Science Association, Boston, MA. Milburn, M.A. 1991. Persuasion and Politics: The Social Psychology of Public Opinion. Pacific Grove, CA: Brooks/Cole Publishing. Miller, A.H. and Asp, K. 1985. Learning about Politics from the Media: A Comparative Study of Sweden and the United States. In Sidney Kraus and Richard M. Perloff (Eds.) Mass Media and Political Thought. Beverly Hills: Sage Publications.

PAGE 146

134 Miller, A.H., Wattenberg, M.P., and Malanchuk, 0. 1986, June. Schematic Assessments of Presidential Candidates. American Political Science Review. 80(2) 521-540. Miller, D.C. 1991. Handbook of Research Design and Social Measurement. Newbury Park: Sage. Miller, R.E. 1981. Experimental Studies. In D.D. Nimmo and K.R. Sanders' (Eds.) Handbook of Political Communication. Beverly Hills: Sage. Myers, J.L. and Well, A.D. 1995. Research Design and Statistical Analysis. Hillsdale, NJ: Erlbaum. Nicotera, A.M. and Rancer, A.S. 1994, Fall. The Influence of Sex on Self-Perceptions and Social Stereotyping of Aggressive Communication Predispositions. Western Journal of Communication 58:283-307. Nimmo, D.D. and Sanders, K.R. 1981. Introduction. In D.D. Nimmo and K.R. Sanders' (Eds.) Handbook of Political Communication. Beverly Hills: Sage. Norris, P. 1997. Introduction: Women, Media, and Politics. In Pippa Norris (Ed.) Women, Media, and Politics. NY: Oxford University Press. O'Keefe, D. J. 1990. Persuasion: Theory and Research. Newbury Park: Sage. Ottati, V.C. and Wyer, R.S., Jr. 1993. Affect and Political Judgment. In Shanto Iyengar and William J. McGuire (Eds.) Explorations in Political Psychology. Durham: Duke University Press. Pan, Z., Ostman, R.E., May, P., and Reynolds, P. 1994, Spring. News Media Exposure and Its Learning Effects During the Persian Gulf War. Journalism Quarterly 71 (1) 7-19. Pearson, J.C., West, R.L., and Turner, L.H. 1995. Gender and Communication. Dubusque, IA: Brown and Benchmark. Presser, S. and Schuman, H. 1980. The Measurement of Middle Positions. Public Opinion Quarterly 44(1): 108-123. Rahn, W.M., Aldrich, J.H., Borgida, E., and Sullivan, J.L. 1990. A Social-Cognitive Model of Candidate Appraisal. In John A. Ferejohn and James H. Kuklinski (eds.) Information and Democratic Processes. Chicago: University of Illinois Press. Rahn, W.M. 1993, May. The Role of Partisan Stereotypes in Information Processing about Political Candidates. American Journal of Political Science 37(2) 472-496.

PAGE 147

135 Rakow, L.F. 1987, Fall. Looking to the Future: Five Questions for Gender Research. Women's Studies in Communication. 10:79-86. Reeves, B., Chaffee, S., and Tims, A. 1982. Social Cognition and Mass Communication Research. In Michael E. Roloff and Charles R. Berger (Eds.) Social Cognition and Communication. Beverly Hills: Sage Publications. Robinson, J.P. and Levy, M.R. 1996, Spring. News Media Use and the Informed Public: A 1990s Update. Journal Of Communications 46(2) 129-135. Robinson, LP., Rusk, J.G., and Head, K.B. 1972. Measures of Political Attitudes. Chapter 10. Ann Arbor: Institute for Social Research, University of Michigan. Robinson, J.P., Shaver, P.R., and Wrightsman, L.S. 1991. Measures of Personality and Social Psychological Attitudes. Volume 1. San Diego: Academic Press. Rosenthal, R. 1991. Meta-Analytic Procedures for Social Research. Newbury Park: Sage Publications. Rosenwasser, S.M., Rogers, R., Fling, S., SilversPickens, K., and Butemeyer, J. 1987. Attitudes Toward Women and Men in Politics: Perceived Male and Female Candidate Competencies and Participant Personality Characteristics. Political Psychology 8(2): 191-200. Rosenwasser, S.M. and Dean, N.G. 1989. Gender Role and Political Office. Psychology of Women Quarterly 13 77-85. Sapiro, V. 1981-82, Summer. If U.S. Senator Baker Were a Woman: An Experimental Study of Candidate Images. Political Psychology 61-83. Seltzer, R.A., Newman, J., and Leighton, M.V. 1997. Sex as a Political Variable: Women as Candidates and Voters in U.S. Elections. Boulder: Lynne Rienner Publishers. Severin, W.J. and Tankard, J.W. 1992. Communication Theories: Origins, Methods, and Uses in the Mass Media. White Plains, NY: Longman Publishing. Sigelman, L. and Sigelman, C.K. 1984. Sexism, Racism, and Ageism in Voting Behavior: An Experimental Analysis. Social Psychology Quarterly 45:263-269. Spence, J.T. and Hahn, E.D. 1997, March. The Attitudes Toward Women Scale and Attitude Change in College Students. Psychology of Women Quarterly 2\\ 17-34.

PAGE 148

136 Sullivan, P.A. 1993, Fall. Women's Discourse and Political Communication: A Case Study of Congressperson Patricia Schroeder. Western Journal of Communication 57 530545. Sullivan, P.A. and Turner, L.H. 1996. Politics, Power, and Gender in From the Margins to the Center: Contemporary Women and Political Communication. Westport, CT: Praeger Publishers. Trent, J.S. and Friedenberg, R.V. 1995. Political Campaign Communication: Principles and Practices. Westport, CT: Praeger. Wadsworth, A.J., Patterson, P., Kaid, L.L., Cullers, G., Malcomb, L. 1987. "Masculine" vs. "Feminine" Strategies in Political Ads: Implications for Female Candidates. Journal of Applied Communication Research 15(1-2) 77-94. Whittaker, J.O. and Meade. 1967. Sex of Communicator as a Variable in Source Credibility. Journal of Social Psychology 72: 27-34. Wicks, R.H. 1992. Schema Theory and Measurement in Mass Communication Research: Theoretical and Methodological Issues in News Information Processing. In Stanley Deetz (Ed.) Communication Yearbook 15. Newbury Park, CA: Sage Publications. Wilcox, C. and Williams, L. 1990. Taking Stock of Schema Theory. The Social Science Journal 27(4) 373-393. Wilcox, C. 1994. Why Was 1992 the v Year of the Woman'? Explaining Women's Gains in 1992. In Elizabeth A. Cook, Sue Thomas, and Clyde Wilcox (Eds.). The Year of the Woman: Myths and Realities. Boulder: Westview. Williams, J.L. 1984, April. Gender and Credibility: A Study of Relationships. A Ph.D. Dissertation at Indiana University. Witt, L., Paget, K.M., and Matthews, G. 1994. Running As A Woman: Gender and Power in American Politics. NY: The Free Press. Zaller, J. 1990, Spring. Political Awareness, Elite Opinion Leadership, and the Mass Survey Response. Social Cognition. Thinking about Politics: Comparisons of Experts and Novices. 8(1) 125-153. Zaller, J. 1992. The Nature and Origins of Mass Opinion. Cambridge University Press. Zhao, X. and Bleske, G.L. 1995. Measurement Effects in Comparing Voter learning from Television News and Campaign Advertisements. Journalism and Mass Communication Quarterly 72 (1) 72-83.

PAGE 149

BIOGRAPHICAL SKETCH Aria Gail Bernstein was born in New Brunswick, New Jersey, the youngest of four children to Adele and Paul Sonnenblick. In 1977, she received her bachelor of arts degree from Florida Atlantic University and in 1982, she received her master of arts degree in urban and regional planning from the same university. Aria worked as a community planner and community development director within the state of Florida for sixteen years. While a doctoral student at the University of Florida, Aria has been an instructor for courses in mass communication and agricultural communication. She has conducted seminars and workshops in strategic planning and project management for the National Aeronautics and Space Administration, the Women's Leadership Network of Marion County, Florida, and Marion County government, and in fundraising for Central Florida Community College. She has accepted a position at Pennsylvania State University as Assistant Professor of Communication, beginning in Fall, 1998. 137

PAGE 150

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy^// /) fiL^JL lu Mary Ann Ferguson, Chair Professor qf Journalism and Communications I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. AjljamJLi , v' — "Deborah M. Treise, Cochair Associate Professor of Journalism and Communications I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Michael Weigold Associate Professor of Journalism and Communications I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Linda Hon Assistant Professor of Journalism and Communications I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. it Jia^Cb it^c^l Michael Martinez Associate Professor of Political Science k

PAGE 151

This dissertation was submitted to the Graduate Faculty of the College of Journalism and Communications and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. August 1998 Dean, College of Joun&lism and Communications Dean, Graduate School