5.” ...-am}. VFW. ”.12ng . .u ...? . a... . ... if ., . . V . V V . . “4%.. ”ram. ... (z.- . x, 5.0% . . ...; 3 .52... .23.... ) 6!... I)! )353: $2: ,. ‘v n... . ii! .3. . (...-Ha. ( . ...—gang... ‘its‘tas 2 2‘; .... 2* O 7 A 1. ,. .71.! .. $.35! . 1.65.! : -..! infliltiino. .2. .. 0.. than , 3.5... w LE.» 9.. .l . -. in! .- D.‘ . . (RM lug .. .33.. . ...... . .....fli...fl! r: .....fiayinntfl i. 3.12. i... ; Evan... ...: Pia ll I59. I .. 2...): $.13 .. z ...z. .... a.-.» ... . . __ e its... 233w”... 5)! tst slittatvilf n. (...: 3.5.3.1.. ‘5‘! 51. . It; «I (91(4- 8! ....a....u:i§ s . I :13 $1... .5. 1:33.34 sand. “Huh. fi...flu.t-....anh..u a..::r.......v;\.l...um. . . .. . . . . . .. . (hih .Y in...- f.» ill... "Liar ...? V I it}. meme ||||lllllllllllllllllllllllllllllllllllllll||||ll|||lll C1951 to) 1 31293 01421 5051 This is to certify that the dissertation entitled The Effects of Incidental and Integral Affect upon the Expression of Prejudice presented by Lori Ann Sheppard has been accepted towards fulfillment of the requirements for Ph.D. degree in PSYCthng WWW/1 Major professor MS U is an Affirmative Action/Equal Opportunity Institution 0-12771 _——7-—.——‘A.——-———.__ -._.__..—..—-—_ A—‘A *“HnW—‘m ,_ ...—h.” LIBRARY Michigan State University PLACE IN RETURN BOX to roman this checkout {mm your record. TO AVOID FINES mum on or him duo duo. DATE DUE DATE DUE DATE DUE MSU loAn Affirmative Adm/Emu] Opponmlty Inflation Wyn-9.1 THE EFFECTS OF INCIDENTAL AND INTEGRAL AFFECT UPON THE EXPRESSION OF PREJUDICE BY Lori Ann Sheppard A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1995 ABSTRACT THE EFFECTS OF INCIDENTAL AND INTEGRAL AFFECT UPON THE EXPRESSION OF PREJUDICE BY Lori Ann Sheppard Two experiments were conducted in order to examine how mood state and prejudice level interact to influence the expression of prejudice in both personal standards and indications of behavior in intergroup contexts. In the first experiment, participants scoring high and low in homophobia experienced a mood induction procedure aimed at inducing affect incidental to the target social group, gay men. They then rendered social judgments about this group and indicated their affective responses to the judgments. In the second experiment, high and low homophobic participants experienced a procedure attempting to induce affect integral to the social group in question (i.e., gay men). Confederates posing as either ostensibly gay or heterosexual men engaged in an interaction during the experiment aimed at inducing the desired emotion. Subsequent to the mood induction, participants completed the social judgment task used in the first experiment. Results were similar across both experiments and indicated that regardless of mood, in general, men and high- prejudice participants endorsed the most negative personal standards and behaviors. Low-prejudice participants tended to respond more negatively to nongay targets than to gay targets. High-prejudice respondents, however, responded to both targets with approximately the same amount of negativity. High- prejudice participants endorsed more behaviors that transgressed their personal standards than did low-prejudice participants. All participants demonstrated a larger discrepancy between standards and behaviors when responding to nongay targets. Interestingly, all participants, but especially those low in prejudice, reported a greater feeling that their endorsed behaviors corresponded to their standards when to responding to gay as opposed to nongay targets. In addition, low-prejudice participants perceive a greater match between their responses than do high-prejudice participants. Contrary to the ‘work_ of previous researchers, perceived discrepancies between standards and behaviors were not associated with feelings of compunction by the low-prejudice participants nor with feelings of discomfort on the part of high-prejudice participants. Results are interpreted in light of research concerning affect and cognition as well as modern theories of prejudice. This dissertation is dedicated to my family: my parents: Kenneth Ernest Sheppard Jr. and Patricia Myers Sheppard, my dearest Aunt Louise Myers Stanley, my sisters: Lisa Michelle Sheppard and Aimee Louise Sheppard and finally, my maternal grandparents: Jack T. Myers and Ethel W. Myers: without whose support, I would have never finished this degree, much less have even attempted it. iv .‘C IE' ' A An. 5.“. H A :3- 2.4 in“ AKNOWLEDGEMENTS I would like to express my sincere appreciation and respect to Dr. Galen Von Bodenhausen, my dissertation advisor, for his never ending patience, guidance, and insight for the past years. He has shown me the highest standard of what a true scholar, teacher, and mentor can be. He has also stood by me and supported me, when I didn't necessarily deserve it. I would also like to thank Dr. Thomas H. Carr, Dr. Norbert L. Kerr, and Dr. Linda A. Jackson for serving on my dissertation committee and for their recommendations regarding this project. Special thanks are due to my many 490 undergraduate research assistants for all of their hard work, enthusiasm, and support for the duration of this project. In particular, I thank leila Ashely, Deborah Howard, Brian vesnaugh, and Robert Eder for serving as experimenters and data scorers as well as Paul Kolarik and Brad Maliszewski for serving as confederates in the second experiment. As far as preparation of the manuscript, I would like to express my sincere appreciation and respect for the staff of the psychology department, in particular: Suzi Pavick, Vicki Alexander, and Seka Remsing. I also express thanks to Keitha Kasel and Nancy Ehret in the Psychology Undergraduate Office for all of their thee c c; r '.l . 1“ ‘ ‘5' I628 1“} rt. 6“ cheery support. I am also grateful for the encouragement and cooperation from the Chairman of the Psychology Department, Dr. Gordon Wood and Roger Hally, Head of the Psychology Business Office. Finally, I would like to express personal thanks for the receipt of friendship and strength from: Kenneth Edward Allhands, Wilbur Nesta Sheppard, Jeannene E. Allhands, Kristen J. Cavanaugh, Ron Coen, Sandra M. Cross, John Dooling, Dr. Lynda G. Farley, Donna A. Gerard, Valerie Kaye, Susan J. Kennedy and George Kennedy, Dr. Todd D. Nelson, Dr. Miguel Angel Quinones, Dr. Kurt. Morgan Ribisl, Todd and April Scherer, Devin Garrett Silloway, Marlo Fallon Steinhauser, Dr. Margaret Van Meter, and everyone at El Azteco Restaurant, East Lansing, Michigan. vi TABLE OF CONTENTS CHAPTER I: GENERAL INTRODUCTION . . . . . . . . . . 1 Affect and Cognition . . . . . . . . . . . . . . 4 Mood Congruency Processes . . . . . . . . . . . 5 Mood and Information Processing . . . 8 Explanations for the Effects of Mood upon Cognitive Processing . . . . . . . . . . . . . . . . . . . 15 Emotion and Stereotyping . . . . . . . . . . . 18 Modern Approaches to Stereotyping and Prejudice 28 Limitations of Previous Research . . . . . . . . 4O Addressing the Limitations of Previous Research 45 Statement of the Problem . . . . . . . . . . . 47 Independent Variables . . . . . . . . . . . . . 47 Assessment of Dependent Measures . . . . . . . 49 Amendments to Devine et a1. "Should-Would" Measure. . . . . . . . . . . . 49 Primary Dependent Variables . . . . . . . 53 Secondary Dependent Variables . . . . . . 55 CHAPTER II: EXPERIMENT I . . . . . . . . . . . . . 57 Introduction and Predictions for Experiment I . 57 Potential Effects of Sadness . . . . . . . 57 Potential Effects of Anger . . . . . . . . 60 Potential Effects of Happiness . . . . 64 The Effects of Mood upon Secondary Dependent Variables . . . . . . . . . . . . . . . . ..68 The Attribution Questionnaire . . . . 68 Method of Experiment I . . . . . . . . . . . . 73 Participants and Selection . . . . . . . . 73 Procedure and Materials . . . . . . . . . 74 Design . . . . . . . . . . . . . . . . . . 77 Statistical Analyses for Experiment I . . . . . 78 Preliminary Analyses. . . . . . . . . . . . . 78 Mood Manipulation Checks . . . . . . . . . . 78 School Analyses . . . . . . . . . . . . . 82 Verb Order . . . . . . . . . . . . . . . 82 Analysis of Primary Dependent Variables . 87 Total Should and Would Ratings . 87 Discrepancy Scores . . . . . . . . . 94 Respondents’ Perception of the Match Between "Should" and "Would Scores" . 97 vii F‘" I: b. ..‘1. kw‘ “.5. ...“ Affect Indices . . . . . . . . . . . 100 Correlates of Emotion Scores . . . . 106 Analysis of Secondary Dependent Measures . 110 Attribution Questionnaire . . . . . . 110 Summary of Findings for Experiment I . . . . . 118 Discussion of Experiment I . . . . . . . . . . 121 Incidental Affect . . . . . . . . . . . . 122 General Findings . . . . . . . . . . . 128 Replication of Devine et al. (1991) . . . 135 conCIuSion O O I O I I O O C O O O O O O O O O 140 CHAPTER III: EXPERIMENT II . . . . . . . . . . . . 142 Introduction . . . . . . . . . . . . . 142 Positive Integral Affect . . . . . . . . . 144 Negative Integral Affect . . . . . . . . . 148 Statement of Problem . . . . . . . . . . . 152 Method of Experiment II . . . . . . . . . . . . 154 Design . . . . . . . . . . . . . . . . . . 154 Participants and Selection . . . . . . . . 154 Procedure and Materials . . . . . . . . . 155 Statistical Analyses for Experiment II . . . . 160 Mood Manipulation Checks . . . . . . . . . 160 Sex Effects . . . . . . . . . . . . . . . 163 Verb Order . . . . . . . . . . . . 164 Analysis of Primary Dependent Variables . 165 Total Should and Would Ratings . 165 Discrepancy Scores . . . . . . . 171 Match Between Should and Would Scores 177 Affect Indices . . . . . . . . . . . 182 Correlates of Emotion Scores . . . . 193 Analysis of Secondary Dependent Measures . 201 Attribution Questionnaire . . . . . . 201 Discussion of Experiment II . . . . . . . 206 Effects of Integral Moods . . . . . . 206 General Findings . . . . . . . . . . 210 CHAPTER IV: GENERAL DISCUSSION . . . . . . . . . . 218 Manifestations of Prejudice . . . . . . . . . . 218 Mood Induction Procedures . . . . . . . . . . . 221 Order of Questionnaire . . . . . . . . . . . . 225 Conclusions and Implications . . . . . . . . . 228 APPENDIX A: Heterosexuals' Attitudes toward Homosexuals Scale Bouton et al., 1987) . . . . . . . . . . . . . . 229 APPENDIX B: Experimenter Protocol Experiment I . . 230 APPENDIX C: Consent Form for First Experiment of Experiment I O O O O O O O O O O O O I O O O O O 235 APPENDIX D: The Life Experiences Inventory Measure 237 viii 1:?" V “J. APPEJI ASPEN: Ififl‘vl A!" 1 A to I HIF‘H I or: b“. I!‘ v HIEI. YYRH bibi APPENDIX E: Consent Form for Experiment II . . . . 240 APPENDIX F: The Adapted Should-Would Measure . . . 242 APPENDIX G: Experimenter Protocol for Experiment II 255 APPENDIX H: Participant Information Sheet for Experiment I . . . . . . . . . . . . 260 APPENDIX I: Participation Information Sheet for Experiment II . . . . . . . . . . . . 262 APPENDIX J: Analyses Regarding Sex Effects from Experiment II 0 O O O I I O O O O O O O O O O I 264 APPENDIX K: Analyses Regarding Verb Order Effects from Experiment II . . . . . . . . . . . . 272 LIST OF REFERENCES . . . . . . . . . . . . . . . . . 288 ix Table Table Table Table Table Table Table Table Table Table 10 LIST OF TABLES Summary of Affect Predictions, in Response to Gay Targets inExperiment I67 MeanSelf-ratingsofMood........................80 Mean Total "Would" Scores as a Function of Verb Order of Questionnaire, Induced Mood, and Prejudice Level of Respondent. . . . . . . . . . . . . . . . . . . .84 Mean Scores of Respondents’ Perception of the Match between their "Should" Scores and "Would" Scores as a Function of Verb Order of Questionnaire, Target in the Scenarios, and Gender of Respondent.............................86 Mean Total "Should" Scores as a Function of Prejudice Level of Respondent and Target in the Scenarios........................................89 Mean Total "Would" Scores as a Function of Prejudice Level of Respondent and Target in the scenariOSIIOOOOOOOO......OOOOOOOOOOOOOOO00.0.000090 Mean Total "Would" Scores as a Function of Gender, Induced Mood, and Prejudice Level of RespondentOOOOOOOOOO......OOOOOOOOOIOOO0.0.00.0..92 Respondents' Mean Perceptions of the Match Between their "Should" and "Would" Scores as a Function of Target in the Scenarios, Induced Mood, and Prejudice LevelofRespondent..............................99 Mean Positivity Scores of as a Function of Induced Mood and Prejudice Level of Respondent. . . . . . . . . . . 105 Significant Correlations between Emotion Scores and Participants' Perception of the Match Between Their Total "Would" and Total "Should" Scores and Between Emotion Scores and the Actual Discrepancy Between Participants' Scores for eachTarget.....................................108 'l'2'I 4“- , . 1a: 9);. . but: 0.1:“ CCU . l T P o¢~ Tat: Table Table Table Table Table Table Table Table Table Table Table 11 12 13 14 15 16 17 18 19 20 21 Significant Correlations between Emotion Scores and Participants' Perception of the Match Between Their Total "Would" and Total ”Should" scores and Between Emotion Scores and the Actual Discrepancy Between Participants' Scores for eachPrejudiceLevel............................109 Mean Number of Attributions Made to the Target in the Scenario for "Would" Scores as a Function of Induced Mood and Prejudice Level of Respondent.......................................114 Mean Number of Attributions Made to Nontarget- Related Factors in the Scenarios as a Function of Target in the Scenario, Gender, and Prejudice Level ofRespondent....................................117 Mean Scores on Dependent Variables for Gay Targets for Experiment ICC...0............OOOOOOOOOOOO..0125 Mean Total "Should" Scores as a Function of Prejudice Level of Respondent and Target in the scenariOSOOOOOOOIOOO.......OOOOOOOOO0.0.0.0000000169 Mean Total "Would" Scores as a Function of Prejudice Level of Respondent and Target in the scenarios.......OOOOIOOOOOOO0.0.0.0...0.000.000.0170 Mean Amount of Discrepancy between Total "Should" and Total "Would" Scores as a Function of Prejudice Level of Respondent and Target in the Scenarios...... ........ . ....... ..................174 Mean Amount of Discrepancy between Total "Should" and Total "Would" Scores as a Function of Prejudice Level of Respondent, Induced Mood, andTargetintheScenarios......................176 Respondents' Mean Perception of the Discrepancy Between Total "Should" and "Would" Scores as a Function of Prejudice Level of Respondent, Induced Mood, and Target in the Scenarios. . . . . . . . . . . . . . . .181 Mean Amount of Positivity as a Function of Prejudice Level of Respondent and Target in the scenariOSOOOOOO0.0....O......-00.0.00000000000000186 Mean Amount of Negative Feelings Towards Others as a Function of Prejudice Level of Respondent and Target in the Scenarios..........................187 xi Table Table Table Table Table Table Table Table Table Table 22 23 24 25 J1 J2 J4 K2 Mean Amount of Positivity as a Function of Confederate, Induced Mood, and Target in the scenariOOOOO...OOOOOOOOOOOOOOO0.0.0.000...0......191 Significant Correlations between Emotion Scores and Participants' Perception of the Match Between Their Total ”Would" and Total "Should" Scores and Between Emotion Scores and the Actual Discrepancy Between Participants' Scores for eachflTarget......................................195 Significant Correlations between Emotion Scores and Participants' Perception of the Match Between Their Total "Would" and Total "Should" Scores and Between Emotion Scores and the Actual Discrepancy Between Participants' Scores for each Prejudice Level.............................196 Significant Correlations between Emotion Scores and Participants' Perception of the Match Between Their Total "Would" and Total "Should" Scores and Between Emotion Scores and the Actual Discrepancy Between Participants' Scores 'for each Confederate.................................197 Mean Total "Should" Scores as a Function of Gender of Respondent and Target in the Scenarios. . . . . . . . 265 Mean Total "Would" Scores as a Function of Target, Induced Mood, and Gender of Respondent. . . . . . . . . . . 266 Mean Discrepancy between Total "Would" and Total "Should" Scores as a Function of Gender and Prejudice Level of Respondent....................268 Mean Number of Attributions Made to Nontarget- Related Factors in the Scenarios as a Function Gender, Prejudice Level of Respondent, and Target in the:Scenarios..................... ..... ..........271 Participants' Mean Perception of the Match Between Their "Should" and "Would" Scores as a Function of Prejudice Level of Respondent, Target in the Scenarios, and Verb Order of Questionnaire. .' . . . . . 275 Mean Number of Attributions Made to Nontarget- Related Factors in the Scenarios as a Function of Verb Order of Questionnaire and Prejudice Level of Respondent.......................................278 xii Table K3 Table K4 Table K5 Table K6 Mean Amount of Negative Feelings Towards the Self as a Function. of ‘Target in the Scenario, Induced M006, and verb orderIOOIOOOO0.0.0.0.0000000000000281 Mean Amount of Discomfort as a Function of Target in the Scenario, Induced Mood, and Verb Order. . . . . . 282 Mean Amount of Threat as a Function of Target in the Scenario, Induced Mood, and Verb Order. . . . . . . . . . . 284 Mean Amount of Negative Feelings Towards Others as a Function of Target in the Scenarios, Confederate, and verb order.I...OOOOOOOOOOOOOOOOOOOOO000......285 xiii 5., at. 5““: INTRODUCTION In the 1960's in America, national attention was brought to the issues of prejudice, racism, and discrimination. Demands were made that all people in this country be treated equally regardless of race. Since that time, the legal and judicial systems of America have tried to remove and render illegal any systematic forms of racism (Gaertner & Dovidio, 1986; McConahay, 1986). It appears that on a national level, America has declared that racism is unacceptable. However, despite such sweeping reform, many researchers feel that evidence abounds in support of the contention that prejudice still exists. In the present, for example, minorities, particularly African-Americans, are still disadvantaged socially, economically, and educationally (Jhally & Lewis, 1992; Katz, Hass, & Wackenhut, 1986; McConahay, 1986). It is not just racial and ethnic minorities who suffer the consequences of prejudice. Other targets of negative attitudes and behaviors include the obese, lesbians, and gay men (Bouton, Gallaher, Garlinghouse, Leal, Rosenstein, Young, 1987; Crandall, 1994). In fact, these latter groups may suffer even more consequences of prejudice, given that there has not been as concerted a national effort to remove or can-fie: 1111:! (ms? rh to: 1 penal C3218: Illtic “We 51:he t h s fi‘tey k $353»- \ 2 condemn the expression of bias towards them equivalent to that launched against prejudice concerning racial minorities (Crocker, Voelkl, Testa, & Major, 1991). Therefore, despite rhetoric about appropriate intergroup attitudes, prejudice perhaps continues to be multifaceted and commonplace in contemporary society (Crosby, Bromely, & Saxe, 1980). This same discrepancy between what is said and what is actually done has also been documented on an individual level. Social psychological researchers have been very effective in documenting this individual-level ambivalence. For example, individuals consistently report that they consider themselves to be non-prejudiced, but their actions reveal subtle biases (Crosby, Bromely, & Saxe, 1980; Gaertner & Dovidio, 1986). In addition, individuals often indicate a discrepancy between what their personal standards dictate they should do in an intergroup situation and what they feel they actually would do (Devine, Monteith, Zuwerink, & Elliot, 1991). It is important to understand why this disparity between condoned actions and actual behavior exists as well as the extent to which it does. One reason for the need to study these issues lies in the diversity of modern society. Because we live in a multicultural society, it is necessary for different types of people to come into contact with one another in a variety of situations. Consequently, it is of interest to study the patterns of interaction among various types of people. Research, for’ example, might explore factors influencing IF in .fi? ‘5. 1”? *II' "F“ H'- 3 whether social interaction reflects the pull of injunctive norms to be egalitarian or nonracist (Gaertner & Dovidio, 1986). A related issue concerns the factors that predispose prejudiced interpersonal behavior. In other words, it might be possible to identify common precursors to behavior indicative of prejudice. Social psychological researchers have examined such antecedents as peoples' degree or level of prejudice towards the specific groups involved (Jackson & Sullivan, 1988). Also examined empirically has been the relation between participants' personal standards regarding appropriate behavior in intergroup contexts and indications of actual behavior in these situations (Devine, Monteith, Zuwerink, & Elliot, 1991). Researchers have also examined.how one’s transitory mood states affect social judgments.regarding others, specifically the degree to which these judgments are indicative of the use of stereotypes (Bodenhausen, 1993). Research has not addressed how these various lines of inquiry might converge. In other words, how might variables such as mood state and prejudice level interact to influence the expression of prejudice in both personal standards and indications of behavior in intergroup contexts? The present experiments aim to achieve the integration suggested above. These studies also hope to extend previous research involving affect and cognition. The two experiments outlined have several objectives. Firstly, these experiments represent an attempt to explore how various affective states pr: Ste: :Efe I.. AHLE 4 might impact upon responses indicative of personal standards regarding behavior and indications of likely or possible behavior in a social intergroup context. Secondly, these experiments endeavor to distinguish between two types of affect and their respective effects upon expressions of prejudice. Incidental affect refers to an emotion that arises for reasons that are irrelevant to an intergroup contact situation, but is brought into such a situation. Most of the previous research concerning affect and cognition focuses upon the ramifications of incidental affect for various cognitive processes or strategies such as, the propensity to use stereotypes in social judgment. Integral affect, however, refers to an emotion that arises due to factors related to an intergroup context and is brought to bear on a present situation (Bodenhausen, 1993). The effects of this type of affect upon cognitive strategies such as the propensity to utilize stereotypes or to express prejudice have largely not been empirically established. W211 The research of social psychology has grown increasingly cognitive since the 1970's (Hamilton, 1981). Recently, however, researchers have begun integrating the cognitive perspective with one that takes into account the importance of affective states. Current research examines the primary affective states of happiness, sadness, and anger and attempts to delineate their respective ramifications for the human inform with or questic the no: new di: he'd m: stereot INCA; Cl 5 information processor. This review of the literature begins with one of the oldest theoretical formulations regarding the question of how affective states influence cognition, namely the mood-congruency hypothesis, and concludes with a specific new direction in the affect and cognition literature, namely how moods influence one's propensity to utilize social stereotypes when forming judgments about other people. 00 o ue c ses One of the oldest. guiding hypotheses regarding the effects of mood upon mental processes postulated a unidirectional influence of 'mood. In this view, moods influence cognitive processing in a manner that is congruent with their valence. If a person is experiencing a positive mood, for example, then subsequent cognitive processes (e.g., memory retrieval, judgments) will be biased in a positive direction. A substantial amount of research has supported this view regarding the effects of mood state. The presence of positive feelings, for example, has indeed been documented to cue positive material in memory and promote creativity in problem solving and negotiation (Isen, 1993). Moods also influence the impact of mood-congruent parts of persuasive messages: happy readers are more influenced by the positive parts of a mixed message, unhappy readers by its negative parts (Bower, 1991). The researcher whose work is perhaps most closely associated‘with mood congruency processes is Gordon Boweru He has f'nd infl One attiv \g.‘ “OE‘U 6 has developed an emotion-network theory that can account for findings in several domains in which social judgments are influenced in a mood-congruent direction (Bower, 1981, 1991). One of the strengths of this model is that it allows for how an individual might remember global evaluations or the final judgment while forgetting the evidence that led to those evaluations. The basic presumption guiding this theory is that there are roughly six basic emotions hardwired into the brain. A number of innate and learned environmental situations turn on a specific emotion node. ‘When this node is turned on due to certain recognition rules, it spreads activation to a variety of indicators. These indicators include facial expressions and memories of events associated with that emotion. Bower has found that temporarily induced moods of happiness, sadness, and anger lead to mood-congruent biases in peoples’ free associations, imaginative fantasies (TAT stories), and their snap judgments of acquaintances“ His research has also shown that when given a choice, happy or sad participants prefer to be exposed to happy or sad stimuli respectively. Other research has documented a wide range of mood- congruency effects. People, for example prefer to affiliate with others in the same mood state or the same situation (Schachter, 1959). Depressed people seek out more information about sad people than about happy people (Gibbons, 1986). Also when forced to select people with whom to become better as; A] P “l ”V .131 ter; effi sang “WM 3‘1‘ 7 acquainted, depressed people prefer to meet unfortunate, unhappy people rather than happy people (Wenzlaff & Prohaska, 1989). Mood congruent effects have also been found for subsequent, immediate evaluations and social judgments. People who had been given a small gift reported, a few moments later, that their cars and televisions worked better than people who had not received a small gift (Isen, Shalker, Clark, & Karp, 1978). Kavanagh and Bower (1985) found that temporary mood states appeared to affect one’s sense of efficacy regarding task performance. Relative to control participants, happy participants reported a heightened sense of efficacy, whereas sad people experienced a lowered sense of efficacy. Although a large amount of research supports the mood- congruency view, some findings regarding negative affect have failed to support this hypothesis. Researchers have often found that negative affect does not lead to effects that parallel those seen with positive affect (Isen, 1993). It appears that negative affective states do not lead to unidirectional mood congruent responses with the same frequency or degree as positive affective states. Therefore, in recent years, the mood-congruency stance has been somewhat challenged and qualified. Bodenhausen’s (1993) work, for example, indicates that sadness and anger, both types of negative affect, can have very different effects upon cognitive processing. Sadness appears to lead to more 8 systematic processing and anger to more cursory processing (Bodenhausen, Sheppard, & Kramer, 1994) . It remains for future research to delineate the parameters that define when and if moods lead to congruent cognitive outcomes. Researchers have, however, formulated alternative explanations for mood-congruency effects rather than the associative network put forth by Bower (1991). Schwarz and Clore (1983, 1988), for example, have postulated that the effects of mood upon judgment, have little to do with spreading activation processes in memory, but instead arise due to the utilization by people of their mood states as a form of information in the judgment process. In other words, the mood serves as information about how the individual feels about the target of the judgment. Therefore, when asked to make an evaluative judgment, people ask themselves how they feel and their prevailing mood is used as an indication of feelings toward the target. What is happening, according to these researchers, is that people are misattributing their current mood as a reaction to the to-be-judged stimulus. This mood-as-information ‘theory' represents one attempt. by researchers to reconcile demonstrated mood-congruency effects with current contradictory research findings. ood d a ' n c 55' Given the recent dominance of a cognitive perspective in social psychology, investigators have also been interested in specifying how mood affects social information processing. in“ d). Ffl‘o‘ ”VI. 9 Examples of the topics studied from this cognitive-affective standpoint include problem-solving, social judgment tasks, and persuasion (Isen, 1984; Isen, Means, Patrickg 8 Nowicki, 1982; Worth 8 Mackie, 1987). For example, in regards to positive moods, it has been found that participants in whom a positive mood has been induced tend to utilize rapid and less taxing problem-solving strategies. People in a positive mood prefer simple, intuitive solutions to problems (Isen et al., 1982), rely upon judgmental heuristics (Isen et al., 1982), use broad and inclusive rather than specific categories in classification tasks (Isen 8 Daubman, 1984), and make decisions both more quickly and on the basis of less information (Isen 8 Means, 1983) than their neutral mood counterparts. It is perhaps in the domain of attitude change or persuasion that some of the most definitive work has been done regarding the general effects of affective states upon cognitive processing. Some of the earliest empirical work in this vein was conducted by Worth and Mackie (Worth 8 Mackie, 1987; Mackie and Worth, 1989). In one of their first studies, they examined the impact of positive mood upon the cognitive processes, mediating attitude change, in response to a persuasive appeal. Participants, in whom a positive mood, specifically happiness, was induced, were found to evidence attitude shifts and cognitive responses that indicated an absence of thoughtful, systematic processing. The attitude 10 change of participants in a good mood, for example, was less influenced by the argument quality of the persuasive appeal and tended to be based more upon the presence or absence of a simple, heuristic persuasion cue. Heuristic persuasion cues are those that recipients can use as a basis for their response that are not integral to the message quality or content per se of the persuasive appeal. Heuristic cues that recipients often use include characteristics of the source of the persuasive appeal, such as, physical attractiveness and putative expertise (Petty 8 Cacioppo, 1986). Therefore the cognitive responses of participants in a good mood, differentiated less between weak and strong arguments and more between whether or not the heuristic cue was present. In addition, happy participants recalled less of the message content, than their neutral mood counterparts. In other experiments, these researchers attempted to determine the specific effects the positive mood was having upon responses to a persuasive message (Mackie 8 Worth, 1989) . These experiments showed that people in a positive mood, when given unlimited time to process the persuasive appeal, took longer than.their neutral mood counterparts, but.processed the message as systematicallyu Therefore participants in a positive mood. were able to prevent the usual cognitive deficiencies associated with the mood by taking more time to complete the judgment task. The researchers’ concluded therefore that positive moods lead to a reduction in cognitive ca; he: 11 capacity available to process the persuasive message, and hence the cursory cognitive strategies associated with them. The various explanations for the effects of emotion will be addressed later in this document, but for now it is of interest to discuss additional findings of empirical investigations into the relationship between emotion and cognition. Schwarz and his colleagues (e.g., Schwarz, Bless, Bohner, 1991) expanded upon the previous research concerning mood and persuasion by examining both negative and positive moods. Moreover, their research proposed a different explanation of mood effects, namely that the presence of certain moods would interfere with the motivation to elaborate the content of the persuasive 'message. 131 a series of experiments, these researchers found results that corroborated the earlier findings of Mackie and Worth (1989) by showing that happy individuals tended not to engage in systematic processing of a persuasive message. In these experiments, happy individuals were equally persuaded by strong and weak arguments unless specifically instructed to pay attention to the content of the message. Sad participants, however, differentiated the messages based upon argument quality and their attitudes were influenced, only if the message presented strong arguments. In further support of the view that moods affect persuasion via their impact on message elaboration, sad participants did not evidence the argument quality differentiation while proc tap; the 12 processing the message simultaneously with a distracter task. On the other hand, the distracter task did not have any significant effects upon the processing of the happy participants, suggesting, as the researchers concluded, that happy participants did not engage in message elaboration from the start. These experiments suggest that people in a bad mood are more likely to engage in effortful processing than people in a happy mood. Schwarz et al. (1991) also proposed that the effects of mood were perhaps more motivational in nature rather than due to a reduced amount of cognitive capacity. In other words, happy participants perhaps were less motivated to process the message, whereas sad participants were more motivated to process the message relative to neutral control groups. It is possible as these researchers, propose that happy participants are less likely to process the message due the desire to maintain their positive mood via off-task rumination. On the other hand, the motivation of sad participants to process the message might stem from the desire to alleviate the sad.mood that they are feeling by focusing on other matters. As was stated eariler, these various explanations will be considered later in this introduction. Having investigated happiness and sadness, research interest turned to the question of whether all moods of the same valence would produce similar effects upon social information processing. Accordingly, Bohner and Hauschildt (1992 upan effec ratic 13 (1992) compared the effects of happiness, sadness, and anger upon the persuasion process. It was hypothesized that the effects of anger would parallel those of happiness. The rationale for this hypothesis was that the antecedents of anger would be more likely to gain attention and thus reduce the likelihood of systematic cognitive processing devoted to anything other than the anger-eliciting factors” Elicitors of anger might, for example, signal to the organism to prepare for action, detracting from the willingness to engage in effortful cognitive strategies. These researchers interpreted their results as indicating that perhaps mood had different effects upon the encoding and judgment stages of processing a persuasive appeal, respectivelyu Angry participants appeared to use less systematic strategies at the stage of forming a judgment, and were willing to give rather extreme evaluations of the message’s position, on the basis of an overall evaluation of argument strength. These possibilities aside, the original hypothesis was supported in that overall, angry participants appeared to employ processing strategies similar to happy participants, in general, tending to avoid cognitive elaboration. Bodenhausen, Sheppard, and Kramer (1994) conducted two experiments also examining the effects of anger in a persuasion context. These experiments also investigated participants’ utilization of source characteristics, established by previous research as influential in the pets; ms: Abra? 14 persuasion setting, namely source expertise and source trustworthiness (Kelman 8 Hovland, 1953; Walster, Aronson, 8 Abrahams, 1966), while under the influence of certain moods. These cues, mentioned earlier, are commonly known as peripheral or heuristic cues. A reliance upon these cues, in message-related judgments, is thought to indicate less thoughtful processing of a persuasive message (Petty 8 Cacioppo, 1986). In the first experiment, a message advocating raising the legal driving age was attributed to either transportation policy experts at Princeton University, a high expertise source, or to a group of college students at a community college, a low expertise source. Holding message quality constant, angry participants agreed with the higher expertise source to a greater extent than their sad or neutral counterparts. The sad and neutral participants did not significantly differ fromleach other; In the next experiment, the message topic concerned the banning of meat in the university dining halls at breakfast and lunch. The message was presented either by the "Student Government League," a trustworthy source, or the "Student Vegetarian League," a less trustworthy source. Angry individuals again evidenced a strong trend of agreeing more with the source possessing greater credibility, regardless of message quality. Therefore it appears that people in an angry mood, when asked to respond to a persuasive appeal, instead of processing it systematically, might prefer to formulate an opinion based 15 upon peripheral cues. These results suggest that people in an angry mood tend to use simplistic cognitive strategies, similar to people in a happy mood. so .1. 91 o 2- -- - . .o-d .. c.. T 'v- ' . -s . Various theoretical approaches have been propounded to account for the effects of mood upon cognitive processing. The persuasion process serves as the paradigm, in which to couch these explanations for a substantial amount of the definitive work regarding the effects of mood has utilized responses to a persuasive appeal as the dependent measures. Schwarz, Bless, and Bohner (1991) have reviewed five major explanations for the effects of mood states, specifically upon the persuasion process. The first set of explanations centers on the notion that mood may serve as a peripheral cue itself. Research examining the mood as cue premise was primarily conducted within learning theory approaches to attitude change. Results indicated that if the mood became associated with the attitude object or with the source of the message, it could serve as a basis for subsequent attitudes (Zanna, Kiesler, 8 Pilkonis, 1970). Schwarz and his colleagues have extended the mood as cue idea into a hypothesis that assumes that affective states can serve informative functions (Schwarz, 1990) . According to this viewpoint, mentioned earlier, people simplify complex judgmental tasks by using their affective response to the attitude object as a basis for or information upon which to 16 base a response. In other words, people utilize a heuristic response strategy based upon how they feel about the attitude object. This heuristic strategy confounds preexisting feelings with reactions to the current stimulus; whereby these feelings are mistaken as reactions to the stimulus. Research has shown, however, that people can avoid mistaking their preexisting feelings as a reaction to a current stimulus, if they are directed to question their feelings as a legitimate basis upon which to structure a response (Schwarz 8 Clore, 1983). In other' words, when research participants are reminded as to the true origin of their feelings, these feelings are subsequently not assumed to be their judgments. A second hypothesis concerns the notion of mood congruency effects. In terms of persuasion, this hypothesis postulates that moods influence recipients’ elaborations generated during exposure to the persuasive appeal due to the increased availability of mood-congruent material stored in memory (Bower, 1981, 1991). Consequently, elaborations in response to the appeal and reactions to peripheral persuasion cues will be valenced.in the direction of the prevailing mood. A third explanation regarding the effects of mood postulates that recipients’ mood may influence the criteria used to evaluate the quality of the current stimulus, or specifically of the persuasive appeal. For example, people in a bad mood may tend to use harsher criteria to evaluate a persuasive message than people in a good mood (Schwarz, Bless, 8 Bohner, 17 1991, p. 165). Therefore, individuals in a bad mood might tend to evaluate the message more negatively and should evidence less attitude change than individuals in a good mood. A fourth explanation proposes that various affective states might have ramifications upon recipients’ motivation to engage in more effortful cognitive strategies or specifically, elaborate upon the content of a persuasive message. People in a good mood might be less likely to engage in systematic or effortful processing of a message and prefer more heuristic processing, one reason being to maintain their positive mood. Participants in a bad mood, however, would be more likely to engage in effortful processing, perhaps to focus attention on something that might alleviate the negative mood. Another motivational explanation relies on the assumption that emotions serve to signal the state of the organism’s world, in particular what conditions merit response. In line with this view, negative moods would inform the individual that the current environment is problematic and deserves attention, thus leading to more cognitive effort being expended towards processing available stimuli. Positive moods, by the same token, would signal that all is right in the individual’s world and that simplification strategies are sufficient for addressing available stimuli. A growing body of literature supports this view (see Schwarz, 1990, for a review). Finally, as previously noted, a cognitive capacity viewpoint has been proposed. This position holds that mood- i1“ 04“ p .ECE Ste: 9‘: 18 related thoughts will decrease research participants’ amount of available cognitive capacity and thus interfere with the elaboration of the persuasive message or other cognitive tasks. In addition, affective states have various levels of arousal associated with them and this arousal itself, could influence the amount of cognitive capacity available for use. The validity of these various hypotheses will be addressed later in light of research that attempts to test them (Bodenhausen, 1993). Before addressing empirical support for the aforementioned hypotheses, and more relevant to the present proposal, is the research that represents a particular extension of the work concerning affect and cognition. This research is that which examines emotion and stereotyping. o ‘o n t Similar to the field of social psychology as a whole, in recent years, research concerning the topics of prejudice and stereotyping has focused upon cognitive issues such as the information-processing' mechanisms that facilitate and perpetuate these biases (Hamilton, 1981). However, researchers have begun to assert increasingly that emotion is a central component of intergroup relations and thus exerts effects upon the cognitive processes involved in these relations (Bodenhausen, 1993; Mackie 8 Hamilton, 1993). Consequently, an increasing number of empirical studies have been conducted examining the effects of various emotional states upon processes integral to prejudice and is: St: I9! SIE ’21“ Q . ‘m‘ 19 discrimination. Research, for example, has investigated how mood states affect the propensity to utilize social stereotypes when making social judgments. One program of research examining mood and stereotyping is that of Bodenhausen and his colleagues (e.g., Bodenhausen, Sheppard, 8 Kramer, 1994). This line of research assumes that stereotypes can serve as judgmental heuristics for the social perceiver. People can use stereotypes to simplify interactions with a complex stimulus environment. In other words, individuals can utilize stereotypes as a means to simplify the judgments they make. Although it is beyond the scope of the present paper to render a complete theoretical account of the stereotyping process, the research evidence regarding ways in which various mood states have been shown to affect the use of stereotypes in social judgment will be discussed. Bodenhausen (1993) has identified the components of the stereotyping process for which mood states may have the most ramifications. The first component concerns the cognitive strategy utilized by the social perceiver when involved in an intergroup situation. In these situations, although some evidence suggests that stereotypes are automatically activated (Devine, 1989), rarely is the stereotype a sufficient element Upon which to base a response. Therefore the social perceiver is left to form a judgment from some sort of integration based uDon the situational and stereotypic information (Bodenhausen, 1993, p. 1990; ii acontin 'top-dow stereoty situatio play a 1 processi exazple, consiste- Processil ‘v’i< earlier certain ll Processi; We curs pIOCESSi; a"iailable iiigzem . bee n 655. , ., 'B'denhau. 'bottOm‘ul b! or £00,) in the N: “ii-ed ”it 20 1993, p. 5). Bodenhausen, similar to other theorists (Fazio, 1990; Fiske 8 Neuberg, 1990; Petty 8 Cacioppo, 1986) , proposes a continuum of social information processing that ranges from ”top-down" decisions based solely and heuristically upon stereotypes to a ”bottom-up" strategy based upon the situational data at hand. A particular affective state could play a role in determining the type of social information processing utilized by the perceiver. A certain mood, for example, might predispose perceivers to form judgments consistently indicative of "top-down" or "bottom-up" cognitive processing. Evidence to support this role of affect has been reviewed earlier in this paper, for it has been established that certain moods do appear to lead to specific types of cognitive processing (e.g. Schwarz, 1990), either more systematic or more cursory. cursory or "top-down" processing information processing would be that which utilized primarily the available stereotype-relevant information in forming a social judgment. The affective states of happiness and anger have been associated with this type of cognitive processing (Bodenhausen, 1993; Bodenhausen, Sheppard, 8 Kramer, 1994). Systematic cognitive strategies would be indicative of more "bottom-up" processing or cognitive effort not driven solely by or focusing primarily on the stereotypical elements present h1 the judgment situation. The emotion of sadness has been linked with more systematic cognitive strategies (Bodenhausen, i153" eff ace. res. (Fis cert 21 1993). Zrt appears therefore that certain moods lead to a greater or lesser tendency to base decisions upon stereotypes. As has also been discussed earlier, affective states can have implications for the motivation needed to engage in effortful cognitive processing. Indeed, it is a widely accepted premise that humans tend to be "cognitive misers," resisting expending any more cognitive effort than necessary (Fiske 8 Taylor, 1991; Simon, 1967). The experiencing of a certain emotion might lead one to avoid the effort of systematic cognitive processing even more than usual or, on the other hand, might motivate one to engage in such processing. Evidence to support this notion has also been presented earlier, in the discussion of the empirical work of Schwarz and his colleagues (e.g., Schwarz, Bless, 8 Bohner, 1991), which demonstrates that happy research participants appear not to be motivated to engage in effortful message elaboration of a persuasive appeal whereas sad participants do appear to be motivated to process material effortfully. Finally, another component in the stereotyping process in which emotions could play a role concerns the final overt response the social perceiver makes. Given that the emotional state might have influenced the cognitive strategies utilized and by inference the final judgment formulated, overt responses might also differ as a function of emotional state. People in happy or angry moods, for example, might be more prche t they 11' thcught: consuzp axiom tespcnse less th experiet ‘v'cid be tereoty 1E5 per} f9$pcnse lfla‘l kem Prej‘Jdic. Giv. if- the ; indeed 01 Stetegt}. 5059:1th MStion first. '. View“ the labo st" ‘1 . “5!: 11 t X‘ h“ N. me! “item! 22 prone to render a prejudiced response due to the fact that they' might be less likely to engage in the presumably thoughtful modification of their responses for public consumption. In other words, due to the tendency for these emotional states to lead to cursory processing, overt responses, while under the influences of these moods, might be less thoughtful and censored. On the other hand, people experiencing a sad emotional state might be more prone to avoid basing a response upon personal prejudice or a simple stereotype due to their tendency to think more systematically and perhaps respond in a likewise manner. Therefore the responses of sad research participants might be more censored and kept in line with prevailing social norms to avoid prejudice (Gaertner 8 Dovidio, 1986). Given the various ways in which emotion could play a role in the stereotyping process, the question remains of how indeed certain emotions affect the tendency to utilize social stereotypes when forming social judgments. The work of Bodenhausen and his colleagues directly addresses this question. Research regarding negative moods will be discussed first. The typical paradigm in this line of inquiry employs a two—experiment ruse. Participants are told upon arriving at the laboratory that due to the brevity of the advertised study, it will be piggy-backed with another short, unrelated experiment. The purported first experiment is described as concerning mood and memories and constitutes the mood It:1 as: |“‘ i" Pitt] | Rare- 4‘1 23 manipulation procedure. In the second purported experiment, participants are given a disguised mood manipulation check and the primary dependent variables, which usually consist of some measure of stereotyping or a social perception task. In a series of experiments investigating the differential impact of negative moods (Bodenhausen, Sheppard, Kramer, 1994), research participants’ moods were manipulated by a procedure the researchers labeled the Life Experiences Inventory (LEI). This procedure asks respondents to recall and write about a life event that produced a particularly intense emotional experience at the time it occurred. Participants are told to reimagine the event as it occurred in as great detail as they can. This mood induction procedure has been widely used and empirically substantiated by many researchers (e.g., Strack, Schwarz, 8 Gschneidinger, 1985). The second purported experiment and the measure of stereotyping was labeled "The Students Court." In this measure, participants read about cases of alleged student misconduct and are asked to make judgments about the cases, including the primary dependent measure, the likelihood of the target person’s guilt. The target person is, or is not portrayed as, representative of a socially stigmatized (i.e., stereotyped) group. In regards to the primary dependent measure of perceived guilt, the results of this experiment demonstrated that participants, who wrote about angry life events, demonstrated 24 a greater perception of the guilt of the stereotyped target than the participants who wrote about sad or neutral events. In other words, angry participants appeared to utilize stereotypes in forming their judgments of guilt more so than sad or neutral mood participants. Moreover, the pattern of judgments was interesting in that it supported the notion that anger is associated with more heuristic processing, similar to happiness. Therefore different types of negative affect, specifically sadness and anger, can possibly have very different effects upon social information processing. Experiments such as this one are noteworthy in that they cast serious doubt on simple congruency hypotheses regarding mood effects upon, cognitive processing. Experiencing a negative mood, for example, such as sadness does not appear to predispose social perceivers to render negative judgments or by the same token to utilize negative social stereotypes in forming decisions. Moreover, experiencing a positive mood such as happiness does appear to increase the probability of utilizing stereotypes in social judgments, many of which are negative in content (Jackson, Hodge, Gerard, Ingram, Ervin, 8 Sheppard, 1994). It appears that specific negative moods might respectively lead to very different effects, and future research will need to delineate between emotion-specific and mood valence-congruent effects. Other less obvious differences between emotions might need to be delineated, such as their respective arousal components (Bodenhausen, 1993). the e stere Suess atte: effec ape: well studi Bedenj 25 Another series of experiments systematically investigated the effects of happiness upon the propensity to utilize social stereotypes in cognitive processing (Bodenhausen, Kramer, 8 Suesser, 1994). These experiments bear mentioning for they attempt to tease apart the various explanations for the effects of happiness upon social judgments“ The first experiment employed the two-experiment ruse with the LEI as well as the ”Students Court" materials, similar to other studies conducted by Bodenhausen and colleagues (e.g. Bodenhausen, 1993) . This experiment also provided evidence to suggest that happy research participants are prone to engage in simplified cognitive strategies. The results indicated that happy participants were more likely to render judgments of guilt when the target person was a member of a stereotyped group than he/she was not. The second and third experiments ruled out explanations for the mood effects that propose cognitive capacity deficits created by intrusive happy thoughts or the energetic arousal that might occur in conjunction with happiness. In order to rule out the cognitive capacity hypothesis, participants were exposed to a mood induction procedure that would not involve a cognitive load. Specifically, a facial feedback procedure ‘was utilized. :Previous research. had demonstrated that the contraction of facial muscles into poses that were associated with the expression of a certain emotion could elicit that emotion even when the facial posturing was 26 introduced via an elaborate cover story (Strack, Martin, 8 Stepper, 1988). It was assumed that if this procedure produced an equivalent amount of happiness in participants, but without the accompanying increased level of stereotyping, then support would be provided for the notion that it is distracting happy thoughts that are producing the greater reliance upon stereotypes in happy research participants rather than happiness per se. In other words, it could be concluded that the experiencing of the emotion of happiness results in a reduction in the cognitive capacity that can be devoted to judgment processes, therefore these processes tend to be simplified and one result of this simplification would be reliance upon stereotypes. Results did not support this hypothesis. Participants in whom happiness was elicited by facial feedback mechanisms were not likely to generate random distracting thoughts that might diminish cognitive capacity. At the same time, however, happy participants were more likely to give judgments that corresponded with stereotypes than were control participants. In the third experiment, moods that varied in terms of the arousal associated with them were induced in research participants. Specifically, a musical mood induction procedure was used. Participants were exposed to musical selections that had been established via pretesting to induce either an energetic happiness or more of a calm, serene happiness. It was hypothesized that if arousal per se was that vast the more reliance hpcthes. found. I participa of the s tereotyl ills in iha‘ pende he??? sta A l EX‘lahat. ElPlanat happy pe< the elite lilntena right a atentio "lemme i”cell-tit: Ifhappy engaheal f°r the. “ml-t9 Articip 27 what was driving the happiness and stereotyping effect, then the more arousing musical selection would result in a greater reliance upon stereotypes. Contrary to the arousal disruption hypothesis, only a main effect of stereotype activation was found. In other words, both the calm happy and excited happy participants demonstrated the same difference between ratings of the stereotyped and nonstereotyped targets and utilized stereotypes to the same degree. It was concluded that the bias in happy participants’ cognitive processing was independent of the autonomic arousal associated with their happy state. A final experiment then turned to the motivational explanations for mood effects upon cognitive processing. This explanation centers on the notion, as was stated earlier, that happy people are less motivated to think systematically about the external environment, perhaps due to desire for mood maintenance or due to their notion that everything is all right and the environment thus merits less systematic attention (Isen, 1993; Schwarz, 1990). Consequently, in this e>1tperiment, a manipulation was employed that would provide an ihcentive for participants to engage in systematic processing. I It happy participants under the influence of such an incentive thaged in systematic processing, then support would be found for the motivational explanations for mood effects. Results 3\apported these motivational explanations in that happy participants, when informed that they would be held er. C02 IE! the To Be 28 accountable for their judgments, did not evidence the processing deficits seen in the other experiments. Therefore it appears that happy individuals can avoid the influence of stereotypes in their judgments when a situational element motivates them to avoid such biased processing. WWW Although Bodenhausen and his colleagues have conducted experiments that illustrate an integration between research concerning affect and that involving stereotyping, most modern aresearch has not been as integrative. As previously noted, Iresearch concerning prejudice and stereotyping has documented the coexistence of contradictory reactions in individuals, specifically the professing of non-prejudiced beliefs and the enactment of prejudiced responses. Indeed although old- :fashioned forms of prejudice and racism are still presumed to (exist, they appear to have given way for the most part to more subtle forms of prejudice (Crosby, Bromely, 8 Saxe, 1980). Devine (1989) has.proposed one interpretation that attempts to resolve the disparity between people’s verbal reports and their nonconsciously monitored overt responses. She has provided a model that accounts for how individuals might renounce prejudice, yet still continue to experience prejudiced thoughts and feelings that might be expressed behaviorally. Devine postulates a continuum of cognitive processing involved in the expresssion of stereotypes and prejudice. She proposes that prejudiced thoughts or stereot‘ of hen} thereof the a: consci: {Devine the at nonprej express flhhitit ecre con allows f¢ Mil/ate: n°iprejug prejudice Person, Dev; hfihprejut th3"“‘9‘1'1ts to be a indicati. ehtrien‘ kelihgs Che: . UH‘ l 1 ct l Oldegre 29 stereotypes are activated automatically when in the presence of members of a stereotyped group or stimuli suggestive thereof. Nonprejudiced responses require the inhibition of the automatically activated negative material and the conscious intentional activation of nonprejudiced beliefs (Devine et al. , 1991, p. 817) . These beliefs would contradict the automatically activated material and facilitate a nonprejudiced response. Therefore stereotyping and the expression of prejudice as well as their inhibition involve cognitive processes that are both automatic and those that are more controlled. This automatic versus controlled distinction in 1 lows for the coexistence in the same person of automatically activated stereotypes and more cognitively controlled nonprejudiced beliefs. Therefore, nonprejudiced beliefs and prejudiced thoughts and feelings can coexist within the same person. Devine’s model asserts that although, one may adopt lnonprejudiced beliefs, this may not eliminate all prejudiced thoughts. Stereotype-based knowledge structures may continue to be activated in response to group members or stimuli indicative of outgroup members. As a result, individuals may QItcperience a conflict between their self-reported thoughts and feelings and endorsed beliefs. Devine proposes that this Q<>nflict is especially characteristic of people who possess a low degree of prejudice. According to the aforementioned researchers (Devine, Henteit who see (e.g. T internal stereoty personal can be 1 internal iniivi‘u Prejudic thEil’ lot hese pe these in Corforta 30 Monteith, Zuwerink, 8 Elliot, 1991; Monteith, 1993), people who score as low in prejudice on common paper-and—pencil tests (e.g. The Modern Racism Scale, McConahay, 1986) have well internalized standards and beliefs that decry prejudice and stereotypes. These people experience compunction when these personal standards are transgressed. In contrast, people who can be characterized as high in prejudice do not have well internalized standards that exclude prejudice. These individuals, when confronted with possible expressions of prejudice on their part, evidence a different reaction from their low-prejudice counterparts. Instead of feeling guilty, these people experience a diffuse discomfort. Clearly, for these individuals prejudiced behavior may not be especially comfortable, but is not unduly distressing (Devine et al., 1991). In order to document the findings discussed above, Devine and her colleagues have designed and implemented a paper-and- pemcil test that measures the reactions of people in interpersonal situations with members of socially stigmatized groups. This measure has been labeled the "Should-Would" h‘easure (Devine, Monteith, Zuwerink, 8 Elliot, 1991; Monteith, 1993). Participants are asked to imagine themselves in Various scenarios that bring them into contact with a target bear-son who is a member of such a group (e.g., African- AI'llericans, gay men). Research participants indicate first what they M do in these situations. In other words, of res; closely standar regardin This las residual usually discrepa; Expe research of Piju: The r8511: Name I: dlScrepaI the$e dis heme 0: affectiv, fur PEOp Bent'il'lflEc has ”to 31 individuals indicate what their personal standards dictate as appropriate behavior. Secondly, participants are asked to imagine the same situations again and this time indicate what they E91115! actually do if indeed placed in that situation. Participants are also asked how well they think these two sets match up (i.e., how and "would") with their personal of responses ("should" closely their behavior coincides standards). Finally, participants indicate their feelings regarding the degree of fit between the two sets of responses. This last measure is used specifically to assess individuals’ The affect indicated on this measure is residual affect. any attributed to perception of u sua 1 1y participants ’ discrepancy between their standards and behavior. Experiments have been conducted using this measure with research participants scoring either high or low on measures Modern Racism Scale, McConahay, 1986). Of prejudice (e.g., The results, from these experiments, indicate that although People both low and high in prejudice tend to indicate discrepancies between their "should" and "would" responses, these discrepancies are smaller for those possessing a lower Furthermore, the degree of prejudice (Devine'et al., 1991). affective consequences of these discrepancies are different As for people representative of the two prejudice levels. lmentioned earlier, peOple low in prejudice felt guilty about Moreover, this guilt transgressing their personal standards. People high in prejudice was often directed at the self. definin with 1 tram gr people Evidenc individ interna counter theate because Standar feeling 0t Hire“ Sterem t0 mair CORCept “mites stigma: PattiCL this De PTOpOSE 32 simply felt a diffuse discomfort over these discrepancies. Devine concludes, as was mentioned earlier, that low-prejudice people have highly internalized personal standards. These people feel that these standards are important and self- defining and they feel an obligation to respond consistently with them. When these internalized standards are transgressed, then the nonprejudiced self-concepts of these people are threatened, resulting in feelings of compunction. Evidence also supported the conclusion that high-prejudice individuals' standards regarding prejudice were not as well internalized or self-relevant as those of their low-prejudice counterparts. Therefore their self-concepts perhaps were not threatened by transgressions of their standards. Instead, because high-prejudice people appear to derive their personal standards from society's standards, discrepancies led to feelings of discomfort (Devine et al., 1991, p. 827). Other contemporary models of prejudice have also addressed the conflicting reactions people have to members of stereotyped.groups (McConahay, 1986) as well as the motivation to maintain nonprejudiced beliefs and a nonprejudiced self- concept. Although a majority of these models assume that Whites manifest an ambivalence in their reactions to socially stigmatized groups, the Theory of Aversive Racism appears particularly relevant not only to the research proposed in this paper, but also to Devine's model as well. The theory proposes a two-part distinction that can be compared to the distint and Ice a dis? Althou; propose cozpari hwothe Tr racLsm This th in a ccnjunc this a' \J “‘9.“ «fit 33 distinction that Devine draws between people possessing high and low degrees of prejudice. Aversive Racism Theory proposes a distinction between aversive racists and nonracists. Although the distinctions that these two lines of research propose do not overlap entirely, they are worth noting and comparing because they enrich the ability to generate hypotheses in the present research. The Theory of Aversive Racism outlines one subtle form of racism (Dovidio, Gaertner, Anastasio, & Sanitioso, 1992). This theory conceptualizes aversive racism as the coexistence, in a person, of negative affect towards minorities in conjunction with an egalitarian value system. The result of this ambivalence is 'that aversive racists express their prejudices only in subtle and covert ways. The parameters of the situation, determine ‘when negative, and or prejudiced responses are enacted. Due to the fact that aversive racists consciously recognize and endorse egalitarian values and ideals, they will not act in a discriminatory manner in situations in which norms prescribing appropriate behavior are clear, and these rmmms prescribe nondiscriminatory behavior. Discriminatory behavior, in these situations, would threaten aversive racists' egalitarian image both to self and others. When the normative structure of a situation is weak or conflicting, and discriminatory behavior can be attributed to some other non- minority related factor and thus be disguised, prejudice will tend t these of the to soc: pusher" theory The al.'s t present to main‘ Devine 2 that peg when t} Prejudic their eg are corn; that it diSCrimj in Devir the: Co- congemE ”Perm, “ting egalita] or at} ‘ we“ion 34 tend to be expressed. Camouflaged discriminatory behavior in these situations will not threaten the egalitarian self-image of the aversive racist. Across a variety of paradigms common to social psychology (e.g. bystander intervention, "wrong- number" technique), researchers have found support for this theory (Gaertner & Dovidio, 1986). The tenet of this theory most applicable to Devine et al.’s (1991) research, and one that will be tested in the present research, concerns the motivation for aversive racists to maintain an egalitarian image both to self and others. Devine and her colleagues demonstrated in their experiments that people who were low in prejudice felt a sense of guilt when they violated their personal standards regarding prejudice, or in terms of aversive racism theory, violated their egalitarian self-image. Low-prejudice people perhaps are comparable to aversive racists in the sense that they feel that it is important to act egalitarian and not engage in discriminatory behavior. Therefore the low-prejudice people in Devine et al.'s research might engage in processes akin to what could be expected of aversive racists. One question concerns the high-prejudice people in Devine et al.'s (1991) experiments. These individuals did not feel guilty about acting in a prejudiced manner. In these people, perhaps egalitarian ideals were not as established and the maintenance of an egalitarian self-image was not as important. The question that arises concerns how these individuals would be classj words, concer high prejud T above. are lc person ftrthe egalitl CCflpun. (“Wino are the intern; Th emere ‘k “‘6 We effect CR‘Pre 5:1, \nua, ‘tiw 35 classified according to Aversive Racism Theory. In other words, are only low-prejudice people aversive racists and thus concerned with maintaining an egalitarian self-image? People high in prejudice might not even care if they appear prejudiced to others or if they act in prejudiced ways. Two issues spring from the summarization and comparison above. Firstly, according to Devine et al. (1991) , people who are low in prejudice are concerned with adhering to their personal non-prej udiced standards behavioral 1y , and furthermore in intergroup contact situations activate their egalitarian belief structures. Finally, these people feel compunction over any transgressions of their belief structure (Devine et. a1, 1991). By the same token, aversive racists are motivated to maintain an egalitarian self-image, both internally to themselves and externally to others. Implied in both of these theoretical orientations is perhaps a motivation of people to censor their responses to members of socially stigmatized groups. These issues of compunction and self-censorship are explored in the present research in several ways. Firstly, the question is addressed as to how various mood states might effect the responses found in Devine’s research using the ”Should-Would Measure". The question of interest regarding low-prejudice participants is whether or not moods will attenuate or heighten their tendency to follow their personal standards regarding prejudice, in their responses, and to 36 experience compunction over discrepancies between their scores. The experience of sadness, with its induced thoughtfulness, for example might heighten the aforementioned responses of the low-prejudice participants. The cognitive effort associated with sadness might facilitate the cognitive access of one's personal beliefs regarding prejudice, and by the same token contribute to a resistance to base responses on a heuristic strategy of utilizing social stereotypes. In addition, the increased thoughtful state in conjunction with the valence of the sad mood, might also heighten the sense of compunction these participants feel, or perhaps render them more prone to experience compunction. It seems likely that compunction would be a thought-producing emotion as well as one more self-directed similar to the usual conceptualization of sadness (Bower, 1981). Therefore these two emotions might be easily experienced together. In contrast to sadness, happiness and anger might attenuate the usual responses rendered by low-prejudice participants. These emotions due their induced thoughtlessness and potential preponderance of mood-related thoughts firstly, might.undermine‘thetability'of low-prejudice participants to cognitively access their personal beliefs about prejudice thus leading them to express less egalitarian responses and perhaps a greater discrepancy between "should" and "would" scores (Bodenhausen, Sheppard, Kramer, 1994. Schwarz, 1990). Secondly, in addition, the cognitive effects 37 of these two moods might attenuate any compunction experienced as a result of any discrepancies between responses. Both of these aforementioned outcomes could occur as a result of a mood-induced lack of attentiveness to the "should-would" task whether due to general thoughtlessness or a focus upon mood congruent or related thoughts. Speculation of how moods might affect the responses of high-prejudice participants is tenuous at best. The degree of prejudice expressed by high-prejudice participants in their "should" and "would" responses might correlate with the valence of the prevailing mood. Perhaps similar to the low-prejudice participants, sadness will heighten feelings of discomfort, experienced as a result of discrepancies between scores, with anger and happiness attenuating any discomfort. Therefore, the present research aims to address the question of how an initial affective state might undermine the compunction and discomfort usually felt as a result of the "Should-Would" measure. A second issue concerns the ramification of various affective states for the proposition of Aversive Racism Theory regarding the tendency or the ability of individuals to monitor their responses so as to avoid appearing prejudiced (Gaertner & Dovidio, 1986); and the analogous tenet of Devine et al.’s model (1991) the tendency or the ability to avoid transgressing personal standards. One process that might be indicative of the censoring of responses or adhering to personal standards is the attributions that people offer 38 regarding their responses. Aversive Racism Theory predicts that aversive racists will often make use of nonracial or non- minority-related factors to justify a negative response. In other words, people will look for other factors to which to attribute their response, in an intergroup situation, other than the social group membership of the person with whom they are interacting. This attribution strategy is labeled the search for nonracial justification factors (Gaertner & Dovidio, 1986) . It is possible that this attribution strategy might distinguish between high and low-people as well as indicating processes of self-censorship. It is of interest to examine whether a particular affective state might have effects on the attributions that participants will make. In terms of the present research, participants were asked to make attributions about their "would" responses or their indications of actual behavior. They were given the opportunity to endorse attributions that range from nonessential factors in the scenarios to attributions centering on the social group membership of the targets in the scenarios. It is possible that certain emotions might interfere with the self-protecting attributions of aversive racists for different reasons. Anger, and its associated cursory cognitive strategies, could perhaps lead to less self- censorship and more overt expression of prejudice. One result of this uninhibited prejudice so to speak might be a greater preponderance of attributions directed at the targets. 39 Sadness, on the other hand might also undermine the censoring of attributions by increasing thoughtfulness. One result of this increased thoughtfulness might be more accurate self- assessments, and an honest expression of prejudice. Therefore participants will not hesitate to attribute their behavior to the social group membership of the targets. The present experiments aimed to investigate such issues by attempting to determine the types of attributions favored by high and low- prejudice individuals experiencing certain mood states. One additional comparison between Devine et al.'s (1991) work and the Theory of Aversive Racism bears mentioning. It is not clarified in the theory how this attribution strategy coincides with one’s racism level. Perhaps what distinguishes low-prejudice individuals from those high in prejudice, in addition to the compunction found by Devine and her colleagues, and in line with the predictions formulated by the Theory of Aversive Racism concerning self-image concerns, is this search for and utilization of nonracial or nonminority attribution outlets. Therefore, in predicting the responses of the individuals in the present research, in conjunction with the Theory of Aversive Racism, individuals differing in levels of prejudice might also differ in their use of nonminority related justifications for their "would" responses. Therefore predictions can be made that, in accordance with previous findings (Devine et al., 1991), individuals differing in levels of prejudice will differ in st re th di: he: so: the pre tar 40 their feelings regarding discrepancies between their indicated standards and indicated behavior. Moreover, respondents representative of the two levels of prejudice might differ in their use of nonminority-related justification factors when discrepancies do arise. High-prejudice people might not hesitate to attribute their responses to a target person's social group membership as such attributions might not violate their personal standards (Devine et al., 1991). High- prejudice research participants might therefore indicate more target-related attributions than low-prejudice participants. However, low-prejudice research participants might favor another attribution strategy. Given their personal stake in egalitarian personal standards, in an effort to project a nonprejudiced image, low-prejudice participants might make more attributions to nonminority-related justification factors. Low-prejudice people might therefore be expected to avoid target attributions and tend to attribute their hypothetical behaviors to factors not related to social group membership. Although the affect and cognition research reviewed herein has been systematic in demonstrating the effects of certain emotions upon specific types of cognitive processing, certain limitations and issues for future research are evident. Three basic limitations in the emotion and stereotyping research can be noted and have been expanded upon 41 by Bodenhausen (1993). The investigation of these limitations will serve to integrate the affect and cognition reviewed earlier with that of the approaches to prejudice just discussed. The first limitation concerns the type of affect that has been investigated in conjunction with cognitive processing. Bodenhausen (1993) has proposed a distinction between two types of affect. "Integral" affect refers to the emotions elicited by the social group itself (Bodenhausen, 1993, p. 3). It is associated with the usual conditions and contexts with which the group is associated and encountered. For many stereotyped groups, the affect that is integral to situations involving them is negative (e.g., Jackson & Sullivan, 1988; Stephan & Stephan, 1985). In addition, there is also what Bodenhausen labeled "incidental affect." This affect is that which is elicited by situations unrelated to an intergroup context. This affect, for example, might be elicited in a prior context and be brought into the intergroup context. One limitation of the affect and cognition research concerning stereotypes is that it has tended to focus principally upon the effects of incidental affect on the propensity to engage in stereotypical processing. Given the notion that intergroup contexts may tend to be biased in the direction of negative emotions, differing predictions might be made for the effects of integral versus incidental affect. For instance, in an intergroup contact situation, negative acce stez cone not {85; 42 affect such as anxiety and anger, which have typically been associated with heuristic processing (see Bodenhausen, 1993), might instead lead to more systematic processing because such processing might alleviate the negative state and help to improve the situation (Schwarz, 1990). However, as Bodenhausen speculates, if indeed such negative states are accompanied by sufficient arousal, a greater reliance upon stereotypes might still be observed. A second limitation of the previously reviewed research concerns the question of individual differences. Research has not addressed how moods might differentially affect the responses of individuals between whom reliable differences can be established, 'When asked to form judgments about members of certain social groups, individuals might give different responses as a function of certain identifiable differences between them. One class of individual differences that Bodenhausen (1993) points to as relevant concerns individual differences in stereotypic beliefs and their cognitive accessibility. Although many social stereotypes are commonly known, they may differ in their accessibility across individuals. Indeed as one line of research has shown, the extent and nature of the impact of an attitude upon perception, cognition, and behavior is a function of that attitude's accessibility in the individual's memory (Fazio, 1986, 1989). As Bodenhausen (1993) asserts, different predictions about the impact of emotions upon cognitive 43 processing could be made for individuals differing in their accessibility of stereotypes. Individuals for whom stereotypes would be highly accessible would obviously have the stereotypes ready to be used as an heuristic cue. On the other hand, individuals for whom stereotypes are not as readily accessible might be forced to look for other cues upon which to base a quick response. If it could be shown that happiness or anger led to an increase in stereotypic judgments for the stereotype-accessible individuals, but not for the stereotype-inaccessible individuals, then the support would be found for the view that the reason these emotions lead to an increase in stereotyping is because they encourage a reliance upon more heuristic thinking (Bodenhausen, 1993, p. 29). Another individual difference variable that would merit investigation and might be indirectly related to the issue of stereotype accessibility concerns individuals’ level of prejudice regarding the social group about whom a judgment is to be made. Different predictions about the impact of emotions upon cognitive processing could be made for individuals differing' in ‘their level of prejudice. If prejudice is related to stereotype accessibility, then individuals who have a high degree of prejudice might have stereotypes that are more readily able to be used as a heuristic cue. Therefore, these people would be more likely to enact a response based upon stereotypes. On the other hand, individuals who demonstrate a lower level of prejudice 44 might not have stereotypes as readily accessible and.therefore might look for other cues upon which to base a quick response. It could possibly be shown therefore that happiness or anger leads to an increase in stereotypic judgments for the high- prejudice individuals, but not for the low-prejudice individuals. Indeed, prejudice level might moderate the effects of mood. A final limitation that can be noted regarding previous research is the reliance upon cognitive dependent variables. The mood research reviewed herein does not systematically examine the effects of various moods upon indications of personal, overt, intergroup behavior. Current research into prejudice documents that individuals possessing a low degree of prejudice are more motivated to censor their responses and base them less upon stereotypes than individuals representative of a higher prejudice level (Gaertner & Dovidio, 1986; & Devine, Monteith, Zuwerink, & Elliot, 1991; Monteith, 1993). Indeed, it has been shown that individuals low in prejudice work to avoid the influence of stereotypes upon their behavior (Devine, 1989, 1990) and actively try to behave in a nonstereotyped manner. Researchers have also suggested that people will work to maintain egalitarian overt responses so as to avoid appearing prejudiced (Gaertner & Dovidio, 1986). Moods might have ramifications for these censorship and concealment activities. If moods impose constraints upon cognitive capacity or motivation, then 45 experiencing these moods might result in more stereotypic judgments and behavior simply because the moods prevent the corrective or self-censorship actions people might usually undertake. It is also possible that certain moods might therefore interact with.one’s prejudice level to influence the propensity to stereotype. It is clear that the effects of mood pose important questions for modern theories of prejudice. s v' se The proposed research aimed to address [the three limitations just presented. The strategy for addressing each of these limitations will be discussed briefly below and taken into further consideration in the methodology sections of the experiments. Firstly, this research aimed to compare effects due to incidental affect versus those due to integral affect. Incidental affect was induced, in the first experiment, by what has been labeled, the Life Experiences Inventory or LEI (Bodenhausen, Sheppard, & Kramer, 1994). This procedure has been reliably used to induce happiness, sadness, and anger (Bodenhausen, 1993). Integral affect was induced, in the second experiment, by having participants witness an interaction between the experimenter and two confederates, one of whom could have been identified as a member of the social group in question. The emotional context of this interaction was intended to provoke either a happy, angry, or neutral #— 46 reaction in participants. Secondly, the proposed experiments intended to examine the issue of individual differences in the effects of mood upon social information processing. It was investigated how individual differences in degree of prejudice towards the target social group interacted with current moods to influence expressions of prejudice. A common pretest measure was utilized as a measure of participants' prejudice level toward the target social group. This pretest allowed the researchers to distinguish participants as either high or low in prejudice. It was of interest to determine how various transitory emotional states might interact with one's personal prejudice to impact reactions to members of stigmatized social groups. Finally, this research aimed to examine, to a certain degree, behavioral dependent variables. Current research has designed and implemented a paper-and-pencil test (i.e., the Should-Would measure) of an individual’s prejudiced behaviors (Devine, Monteith, Zuwerink, & Elliot, 1991; Monteith, 1993). As was described earlier, it assesses individuals' reports of their personal standards and hypothetical behaviors regarding various inter-group contact situations. Although the "Should- Would" measure is not as desirable a measure as a direct behavior observation, it still holds merit and lent itself nicely to laboratory research. This measure was amended in several ways in order to expand its value in the proposed . r.- -';r J‘«.a;. 0t he E8 up: in: 47 research. Some of these amendments will be made according to Devine and her colleagues (1991) in order to render the measure usable with various socially stigmatized groups. Other amendments were made in order to endeavor to improve the measure. These amendments are discussed more fully in the methods section of this paper. fitatemeat_ef_the_£rehlem The: two experiments. conducted aimed. to explore the effects of incidental and integral affect and prejudice level upon the expression of personal standards appropriate to and indications of behavior appropriate to an intergroup context. The discrepancy between these two measures and affect due to this discrepancy were also measured. The aforementioned dependent variables constitute the "Should-Would" measure of Devine et al. (1991). The socially stigmatized group to which the participants were to respond was gay men. The measure of prejudice was the seven—item Attitudes Toward Homosexuals (ATHS) scale devised by Bouton, Gallaher, Garlinghouse, Leal, Rosenstein, & Young (1987). Mood was manipulated via the LEI and a situational paradigm described below. The independent and dependent variables as well as the methodology for these two experiments are described in detail in the sections below. W The experiments described herein explored the effects of the following independent variables. First, in line with Devine et al. (1991), individuals’ level of prejudice toward 48 the target social group was assessed. This predictor variable enabled the comparison of the responses of people high versus low in prejudice. Moreover the inclusion of this measure allowed for the replication of Devine and her colleagues' results. Conceptually, this measure tapped into a general overall attitude towards lesbians and gay men. Components of this attitude include the following issues, the morality of being a gay man or a lesbian, the civil rights of gay men and lesbians, the legality of homosexuality, and the degree to which homosexuality is considered disgusting. The second independent variable examined was the induced mood state. The mood manipulation procedure attempted to induce a particular affective state (specifically, anger, happiness, sadness, or a neutral mood). As was discussed earlier, certain moods appear to predispose social judgment makers to engage in cognitive strategies characterized either by a greater or lesser degree of systematicity, as well as a tendency to utilize or to avoid the use of social stereotypes in forming a judgment. IMoods, due to their influence upon the use of stereotypes, by extension, may exert an effect on whether judgments are inadvertently' rooted in jprejudice. Therefore it was predicted that an induced mood state would exert effects upon individuals' expression of prejudice. A final main independent variable was the target person(s) in the scenarios about which participants formed judgments. This target person(s) was either characterized as DE. re: Th: Va] util 49 a gay man or no information about his sexual orientation was provided. This variable allowed the comparison between responses to members of socially stigmatized groups versus responses to people who are apparently not members of such groups. Secondary independent variables examined included the gender of the respondent as well as the verb order of the questionnaire, a variable described in detail below. ss ss e o e e su es The dependent variables assessed in both experiments represented an amalgamation of the "Should-Would" measure of Devine and colleagues (1991). This measure will be henceforth referred to as the "amended should-would measure" or the ASW. This measure and its amendments as well as its dependent variables are discussed below. d t vi e t . 9 " u -W u " Measure. The first amendment to the measure was that, in contrast to Devine and her colleagues, it was introduced via a cover story. Instead of informing respondents up front of what.the.questionnaire‘was assessing; a cover story concerning urban congestion and cultural diversity was utilized. This extension is warranted given the subtlety of modern racism and the inhibitions people feel about expressing prejudice (Crosby, Bromely, & Saxe, 1980). Furthermore, more recent research demonstrates that people enact inhibitions to prevent utilizing stereotypes (Macrae, Bodenhausen, Milne, Jetten, 1994) . questie charact experin that i relatic assessn cover 5 partici rendere one ano A Wouldu an ele 50 1994). Also, lacking a cover story, the measure raises questions about demand characteristics. The issue of demand characteristics is of more relevance in the present experiments than perhaps in Devine et al.'s (1991) work in that it is important that participants not realize the relation between the mood manipulation procedure and the assessment of the dependent variables. Therefore utilizing a cover story was helpful in that it not only may have decreased participants’ inhibition of responses, it also will have rendered the purported "two" experiments very different from one another. A second adaptation to the previously utilized "Should- Would" measure was the inclusion in each of the scenarios of an element that could justify a negative response on ostensibly nonprejudiced grounds. This element was a factor in line with the cover story of population congestion and diversity. A typical situation utilized by Devine et al. (1991) is that of a minority couple moving in next door. An example of the nonminority element included in this scenario would be that the moving in of the couple next door would necessitate the destruction of a former wooded lot. This second adaptation to Devine et al.'s measure is closely related to the third adaptation, which was an attribution questionnaire following Devine et al.'s (1991) measure that assessed reasons for research participants' "would" responses. This questionnaire was included for several reasons. First, 51 they allowed for an expansion upon the data usually collected by previous researchers and allowed for the assessment of attributions participants made for their "would" responses. Secondly, there existed theoretical grounds to propose that the attributions favored by participants might vary as a function of the independent variables. In line with the previous research conducted by the Devine laboratory, it was of interest to see to what low-prejudice and high-prejudice individuals would attribute their behavior. In addition, in accordance with the theory of aversive racism, in an attempt to portray a nonprejudiced image, research participants might also tend to attribute responses to other factors in the scenario in order to justify their responses. It is a tenet of the Theory' of Aversive Racism. that people look for nonracial or nonminority justifications for their behavior (Gaertner & Dovidio, 1986). The scenarios all contained such a nonminority related element that could be used to justify a prejudiced response. It was of interest to see whether participants would make use of the non-minority element in explaining their "would" responses. A fourth and final alteration to Devine's measure was a counterbalancing of the order of the measures. A possible flaw in Devine et al.'s (1991) research might lie in having participants always indicate their "should" responses before their "would" responses. Consequently, in these experiments, half of the participants gave their "should" responses first 52 and the other half gave their ”would" responses first. The rationale for this counterbalancing concerned the possibility of order effects. It is possible that when individuals are asked first to indicate what they should do in a particular situation, they begin to activate concerns regarding the maintenance of an egalitarian self-image and the wrongfulness of prejudiced behavior. Consequently, their subsequent "would" responses might not be as extreme as they would be in the absence of the prior "should" responses. By the same token, consistently indicating "would" responses before "should" responses might also attenuate "should" responses. Indeed, after indicating what they would do, participants might indicate "should" responses that are not as nonprejudiced as when the "should" responses are given first. Finally, similar to Devine et al. (1991), the measure contained an index of participants' feelings about any discrepancies between their "should" and "would" responses. The situations utilized included two that Devine et al. (1991) have used with gay targets (feeling bothered that a gay person sits next to you on a bus and feeling upset about a gay couple moving in next door). Two new situations were added after pretesting established that they constituted normally occurring situations in everyday life that involve intergroup contact. The first of these two new situations involved being stuck in traffic backed up due to an unannounced rally for gay 53 (or nongay) people. The second situation involved reaching a cab at the same time as a gay (or nongay) person who offers to share the cab. These situations were pretested on individuals selected from the same population as those participating in the main experiments. These two situations were chosen because they did.not elicit.extreme responses from individuals and were seen as being fairly realistic. Similar to Devine et al. (1991), some of the situations in the measure were phrased in terms of how the person should (or would) feel and others in terms of how the person should not (or would not) feel. Primary Dependent Variables. The primary dependent variables contained in the ASW and that were assessed in the experiments were as follows. The first variable was labeled "Total Would" (TW). This score was an average of participant's "would" ratings across the four situations. This measure represented an index of the behaviors that participants indicated they would enact in the hypothetical intergroup contact situation. Ratings were reverse scored when necessary'sowthat lower ratings were associated with less negative hypothetical behaviors and higher ratings with more negative hypothetical behaviors (Devine et.al., 1991, p. 819). Conceptually this measure was indicative of the degree of prejudice that participants would allow in their behavior with a higher score reflecting responses that are more prejudiced towards the target person(s). 54 The second dependent variable was labeled "Total Should" (TS). This score was an average of participants' "should" ratings across the four situations. This measure represented an index of the behaviors that participants indicated were preferable or acceptable, according to their personal standards, in an intergroup contact situation. Ratings were reverse scored when necessary, as with "would" responses, so that lower ratingS‘were associated with less negative "should" standards and higher ratings with more negative standards (Devine et al., 1991, p. 819). Conceptually this measure was therefore indicative of the degree of prejudice inherent in participants' personal standards for behavior. A higher score was be representative of standards that allowed a greater degree of prejudice towards the targets person. The third primary dependent measure was the discrepancy between participants' total "would" and total "should" scores (D). This discrepancy index was calculated by subtracting participants' "should" ratings from their "would" ratings for each situation and summing the discrepancies across the four situations (Devine et al., 1991, p. 819). Conceptually this score represented the discrepancy individuals express between their standards of acceptable behavior and the behavior that they are most likely to enact. This discrepancy therefore represented the degree to which indicated behavior deviates from expressed personal standards. 55 Secondary Dependent Variables. One secondary dependent measure involved the series of feeling indexes that participants completed after the ASW. These indexes assessed individuals' current feelings at the time. Conceptually, these measures are usually assumed by Devine and her colleagues to be indicative of the affect that people feel as a function of completing the "Should-Would" measure or in other words, the affect that is felt due to any discrepancy expressed between standards and behavior. In terms of the present research, and in light of the "baseline" effects documented by Devine et al. (1991), it was of interest to see whether the affective state induced prior to the ASW, would lead to an attenuation of the compunction felt by low- prejudice participants and the discomfort felt by the participants possessing a higher level of prejudice. A final secondary dependent measure included was the attribution questionnaire participants completed about their "would" responses. This questionnaire yielded two composite measures. The first measure was the degree to which participants attributed their "would" responses to factors other than the target in the scenario. The second measure constituted the extent to which participants attributed their intended behaviors to the target person(s) in the scenario. Conceptually these measures represented the attributions favored by particular participants. It was hoped to find support for the predictions made by modern theories of 56 prejudice as was stated earlier e.g. Gaertner & Dovidio (1986), specifically that people representing various levels of prejudice might tend to be motivated to avoid viewing their actions as based in prejudice. Moreover, these people will look for a non-intergroup related factor(s) to which to attribute their response. EXPERIMENT I 'on e ' ' s o ' t In the first experiment participants who scored high and low in prejudice towards gay men and lesbians were induced to feel happy, sad, angry, or neutral via a memory elicitation procedure and then, as part of an ostensibly unrelated study, completed the ASW. Several predictions might be made regarding how a person's prejudice level and incidental mood state might interact to influence responses to the ASW. The following sections propose hypothetical effects of the various moods, for individuals of high and low prejudice respectively. e a e ss As previously described, the emotion of sadness has been associated with greater thoughtfulness and systematic thinking. People, in whom a sad mood has been induced, are more attuned to the quality of arguments contained in a persuasive appeal (Schwarz et al., 1991) and appear to rely less upon social stereotypes when forming social judgments (Bodenhausen, 1993). These results point to the conclusion that one main effect upon cognitive processing of sadness is that it appears to induce more systematic cognitive strategies. It is possible that these effects of sadness might 57 58 interact with 'prejudice level in the following ways to influence responses to gay targets on the ASW. Sad, high- prejudice individuals' dominant response on the ASW might be more negative than both low-prejudice participants' responses and other high-prejudice participants' responses to scenarios not containing gay targets. It is predicted that the effects of the sad mood (i.e., systematic cognitive strategies) might lead the high-prejudice participants to accurately identify their negative feelings towards gay men and they might not hesitate to indicate these feelings. The responses of these individuals might replicate those of the high-prejudice participants in Devine et al.'s (1991) experiments. Hence, these participants will indicate negative "should" and "would" scoresas compared to the their low-prejudice counterparts. Accordingly, there might be little discrepancy between these two sets of scores“ Moreover, similar to the results documented by Devine and her colleagues, sad, high-prejudice participants might indicate in their affect ratings little in the way of compunction in reference to the "should" and ”would" scores they have indicated. Despite any mood-induced rumination of their negativity towards gay men, it could be expected that high-prejudice participants will not hesitate to express this negativity, nor will they suffer affectively as a consequence of it. This lack of affective consequences as a result of expressed negativity perhaps stems from the fact that this negativity might not violate high-prejudice people's 59 personal standards (Devine et al., 1991). Low-prejudice participants, when sad, due to the effects of the mood might also indicate more negative responses than their low-prejudice counterparts in the happy and neutral mood conditions. Sadness, as well as leading to more thoughtful, systematic cognitive processing, is also considered to be a more self-directed emotion (Forgas, Bower, & Krantz, 1984). Mood states have been found to affect self-observations and self-evaluations in a direction corresponding to the valence of the mood (Forgas et al., 1984; & Kavanagh & Bower, 1985). The sad, low-prejudice individuals, therefore, might approach the ASW in a systematic and thoughtful manner. They might think carefully about their responses. The thoughtfulness induced by the mood and its potential self-focus might lead the low-prejudice participants to indicate any negative affect they really do feel in response to gay men. Hence their "would" scores or indications of hypothetical behavior might reflect this negativity. In addition, the low-prejudice participants in the research conducted by Devine et al. (1991) valued their personal standards of egalitarianism and felt that it was important to indicate "should" and "would" responses in line with such standards“ Combining the effects of a sad mood with the already present image and egalitarian concerns of low- prejudice participants might result in an even greater amount of thought given to their internal "should" standards. 60 Therefore, it is predicted that these participants might indicate less negative total "should" scores, in line with egalitarian concerns of appropriate behavior when responding to scenarios containing the gay targets. Consequently, there might be a high discrepancy between sad, low-prejudice participants' total "should" and "would" scores, when responding to scenarios that contain gay targets. This discrepancy might lead to feelings of compunction, replicating the findings of Devine et al. (1991). These participants might feel guilt over the fact that their actual behavior might transgress their personal standards. Therefore, it is predicted that sad, low-prejudice individuals overall will indicate negative "would" scores, positive "should" scores, large discrepancies, and lots of compunction as compared to their other low-prejudice counterparts. Potential Effects of Ange: Anger is associated with less thoughtful and less systematic strategies generally, and with a greater reliance upon stereotypes in forming social judgments specifically (Bodenhausen, Sheppard, Kramer, 1994). Therefore, the responses of high-prejudice people when angry and responding to scenarios containing gay targets are expected to reflect not only negativity but also a lesser degree of systematic thought. Participants under the influence of the mood might simply report responses that are not based on a careful consideration of the issues at hand. These responses would be 61 heuristic and perhaps based upon social stereotypes. Another reason ‘that. these jparticipants' total "would" and. total "should" scores might be negative is that participants might also utilize their mood as information and the primary force from which. to base a response (Schwarz, 1990). These participants might as a result indicate negative total "should" and total "would" scores. These scores would be reflective of prejudiced standards of acceptable behavior and indications of actual behavior. In other words, these scores might be reflective of a greater degree of prejudice as compared to the highly prejudiced participants responding to nongay targets, the neutral mood high-prejudice participants and the low-prejudice participants responding to both kinds of targets. As a result of these uniformly negative responses, it is predicted that there will be a low discrepancy between these participants' total "should" and total "would" scores. It is interesting to speculate about the responses of these participants on the affect indices. Highly prejudiced participants in previous research indicated that they felt a diffuse discomfort on these measures. This affective response was perceived to be due to a perceived discrepancy between personal standards and own behavior. Given that it is predicted in the present research that these participants will not indicate a discrepancy between their sets of scores, perhaps indications of discomfort will not be found. Therefore the affect measure might indicate simply a negative 62 mood. It is predicted that angry, highly prejudiced participants will not evidence much of a sense of compunction. The thoughtlessness associated with anger might prevent any systematic thinking about or focus upon any discrepancies and what they mean. By the same token, the thoughtlessness induced by the angry mood might predispose these participants to not simply be aware:of discrepancies between their two sets of scores. Given that anger is associated with more cursory cognitive styles (Bodenhausen, Sheppard, & Kramer, 1993) , low- prejudice participants, when angry, might not respond exactly in line 'with. their internal standards of egalitarianism (Devine et al., 1991) if it takes a lot of cognitive effort to access these standards. Devine has postulated that while stereotypes ‘might. be cognitive structures that . are automatically activated in the presence of a stimulus indicative of social group membership, personal beliefs or standards regarding social group membership might not be automatically activated, but might require more cognitive effort to activate or access (1989). Therefore low-prejudice participants, due to their affective state, and akin to their high-prejudice counterparts, might give responses indicative of a low level of cognitive effort. For example, they might simply indicate what is salient to them as they respond to the scenarios, instead of giving thoughtful responses. If the low-prejudice people in this experiment have a truly sincere 63 positive attitude towards gay men, or if their egalitarian belief structure is highly salient to ‘them, then 'their responses might be positive in comparison to those of high- prejudice participants. The emotion of anger, due to mood congruency effects, might render these responses of the low- prejudice respondents to be more negative than the scores of their low-prejudice counterparts in the other mood conditions, but still not as negative as those of the high-prejudice participants overall. Several other reasons exist as to why the responses of the low-prejudice participants might be more negative than what would be expected” One reason is that the effects of the mood might interfere with their being able to focus upon personal egalitarian standards as clearly as their sad or neutral counterparts, if‘ these standards are not highly salient. Angry, low-prejudice participants might strongly be influenced by the negative valence of the mood and or the thoughtlessness associated with the mood. It is also worth noting that anger might pose difficulties for the implementation of the self-censorship activities predicted by the Theory of Aversive Racism (Gaertner & Dovidio, 1986). The responses of angry low-prejudice participants might therefore differ from their neutral mood counterparts in the Devine et al. (1991) experiments by being more negative. Therefore, similar to their high-prejudice counterparts, these participants will indicate negative total "should" and total 64 "would" scores. The resulting discrepancy between the two sets of scores perhaps might be small for the angry, low- prejudice participants relative to the other low-prejudice participants in the experiment. On the affect measure, unlike their sad and neutral counterparts in the present experiment, angry low-prejudice people might not feel a significant sense of compunction for several possible reasons. The thoughtlessness associated with anger might lead these participants to not be aware of any discrepancy indicated between their "should" and "would" responses or between their behavior and.their standards. ‘They might not be aware that their responses transgressed their egalitarian standards. It‘stands to reason that it takes time and effort to access such cognitive structures as personal standards and the emotion of anger might prevent or block the effort needed to gain access to such structures. Overall, therefore, it is predicted that angry low-prejudice participants will demonstrate small discrepancies and little compunction. 'a s ' ss Happiness has been associated with less thoughtfulness and less systematic thinking and a greater reliance upon social stereotypes in judgment making (Bodenhausen, Kramer, & Suesser, 1994). It can be predicted that people with a high level of prejudice, who are induced to feel a happy mood, will render prejudiced responses to gay targets reflecting a low 65 degree of thoughtfulness. As specifically compared to the low-prejudice participants, the high-prejudice participants responding to nongay targets, and the neutral mood, high- prejudice participants, happy, high-prejudice participants might indicate more prejudiced personal standards or negative total ”should" scores. Their indications of actual behavior might also tend to be prejudiced, hence their negative total "would" scores. As a result, there will probably be little discrepancy between these two sets of scores. On the affect measure, due to the thoughtlessness associated with happiness, these participants might not perceive a discrepancy between their two sets of scores or the inherent negativity in their responses. Consequently, they might feel little compunction as well as little discomfort, as has been previously documented with high-prejudice participants (Devine et al., 1991). In the responses of happy participants who are low in prejudice, it is predicted that positive reactions will dominate. The thoughtlessness of happiness will lead to these positive responses because participants may not think carefully about their personal standards and or likely behavior. Instead the participants' responses will correspond to the valence of the mood. Indeed.mood congruency appears to be a simple response strategy that people are likely to use when thinking heuristically (Schwarz & Cflore, 1988). Therefore, these participants may glibly report more positive reacti 09 v 11". 103 part1 66 reactions or give responses indicative of a positivity bias. As a result of the similar level of positivity in both total "should" and total "would" scores, there will not be an appreciable discrepancy between these scores. The effects of happiness should thus prevent feelings of compunction from arising. A summary of these predictions is presented in Table 1. Overall, mood is not.expected.toeaffect.the "should", "would", discrepancy, or compunction scores of high-prejudice individuals, who will show the basic pattern reported by Devine and colleagues. That is, they will indicate prejudiced "should" scores, prejudiced "would" scores, low discrepancy and low compunction scores. Mood will affect the "should", "would", discrepancies, and compunction scores of low-prejudice participants. Sad and neutral participants will show the pattern reported by Devine et al. (1991) for low-prejudice participants; relatively nonprejudiced "should" scores, prejudiced "would" scores, large discrepancy scores, and indications of compunction on the affect indices. Angry low-prejudice participants will show a pattern like that of the high prejudice participants in Devine's research: prejudiced "should" and "would" scores, low discrepancy scores, and low compunction. Happy participants will show a pattern not seen in Devine's work. 67 Table l ... . - 01~ e t 99 S 0 ° W o w- ts e e t 119.95! Sheuld. Emile. flimsy. Wen Angry prejudiced prejudiced low low Happy nonprejudiced nonprejudiced low low Sad nonprejudiced prejudiced larger high Neutral nonprejudiced prejudiced larger high co s ' - ' ' ’ ' a ts W need. Staging. 8.91;; W W Angry prejudiced prejudiced low low Happy prejudiced prejudiced low low Sad prejudiced prejudiced low low Neutral prejudiced prejudiced low low 68 They will indicate nonprejudiced "should" and "would" scores, low discrepancy scores, and a low degree of compunction. u,- - s . .... ... - . ., .-.-,.-, v. a. -,. The Attribution Questionnaire. The inclusion of the attribution questionnaire represents an exploratory investigation into ‘the 'tenets of eaversive racism ‘theory (Gaertner & Dovidio, 1986) and the propositions made by Devine and colleagues (1991) regarding the impression management and self-censorship strategies utilized by people to avoid the appearance of being prejudiced. Specifically this measure taps into the attributions that research participants are willing to endorse for their "would" scores. In general, regardless of mood, impression management strategies might be manifested in the attribution of "would" responses to factors in the scenarios that are not related to the intergroup element of the scenario. Therefore to avoid appearing biased, research participants might attribute their likely behaviors to factors ostensibly not related to the social group membership of the targets in the scenarios. If individuals' primary interest is in maintaining an egalitarian image to the self, as suggested by aversive racism theory, strategies of self-deception might also be manifested in a reliance upon ostensibly non-intergroup-related factors as a basis for a response. Therefore self-deception and impression management concerns might both surface in a similar pattern of results on the attribution questionnaire. 69 One difference between impression management and self- deception strategies that.might affect their manifestation is the degree of thoughtfulness the response necessitates. The question. arises as to *what process is more thoughtful, impression management or self-deception? This question is relevant to the present experiment, in that certain moods might predispose participants to think more or less and therefore prevent these two types of responses. For example, sadness might lead to more nongroup-related attributions if thoughtfulness leads to more impression management. By the same token, sadness might lead to more attributions of actions as being due to the social group membership of the targets in the scenario, if an increased level of thoughtfulness leads to less self-deception. Indeed self-deception might be a process that requires less conscious thoughtfulness than impression management. Although there does not exist a strong basis for predictions, the following section presents some tentative hypotheses of how mood and prejudice level might interact with sexual orientation of the target to influence results on the attribution measure. The emotion of sadness and its correspondent thoughtfulness might lead to different attribution strategies for participants of differing levels of prejudice. It is possible that all sad individuals, in the context of the questionnaire, will think more about their "would" scores than they perhaps did when they indicated them 70 due to the effects of the mood and to the instructions to focus upon the responses. As might be predicted by the Theory of Aversive Racism (Dovidio et al., 1992), in a (thoughtful) effort to make their reactions look more socially acceptable, high-prejudice participants might attribute their endorsed behaviors to other sources than the intergroup context of the scenarios. Consequently, it is possible that their responses on the attribution questionnaire will indicate a use of nonminority justification factors (NMJFs). Thus, high-prejudice people, when sad, will give negative responses, but these negative reactions may be accompanied by a greater reliance upon nonminority-related justification factors (NMJFs) to avoid the disapproval of others. Due to the thoughtfulness and self-direction of sadness, it is likely that on the attribution questionnaire, low- prejudice individuals, in contrast will not make use of NMJFs. It is possible that they will not fall back on excuses for their behavior. They might be more aware of the basis of their responses and if it is the social group membership of the targets in the scenarios, be more realistic about it. Therefore, sad, low-prejudice participants will indicate little use of NMJFs. It is unclear how angry, high-prejudice participants will respond on the attribution indices. It might be predicted that highly prejudiced participants will indicate few attributions for their "would" scores. These individuals night atter 1986) to th their avail indie factc in tr shown egali beco; these "Voul scena secon think “egat 71 might not feel any need to justify their behavior or to attempt to portray a nonprejudiced image (Gaertner & Dovidio, 1986) . Low-prejudice participants, when angry and responding to the attribution questionnaire that asks them to focus upon their endorsed. behaviors.:might, however, utilize various available attribution outlets. In other words these, individuals might tend to attribute their likely behavior to factors other than the social group membership of the targets in the scenario. The research of Devine et al. (1991) has shown that low-prejudice individuals want to follow egalitarian standards of conduct, if they are salient, and become upset if they do not follow such standards. Therefore, these participants might be motivated to attribute their "would" scores to non-minority related elements in the scenarios. This use of NMJFs might serve to quell any secondary egalitarian concerns that arise as a consequence of thinking about their "would" scores and their potential negativity. Finally it is interesting to speculate what attribution strategies will accompany happy moods. Generally speaking, it might be predicted that both low and high-prejudice participants while under the influence of a happy mood, might not feel a need to justify behavior through the use of NMJFs as attribution outlets. This lack of attributions might be due to participants' general feelings of positivity which in turn, might undermine a sense of needing to rationalize 72 previous responses. Therefore it could be predicted that all happy participants will fail to utilize nontarget-related factors as attribution outlets. Eart i“div with METHOD OF EXPERIMENT l E !' . ! i 5. J !' Several hundred undergraduate psychology students completed the seven-item Attitudes Toward Homosexuals Scale (ATHS; Bouton, Gallaher, Garlinghouse, Leal, Rosenstein, & Young, 1987), which assesses negative attitudes towards gay men and lesbians, as part of a large amount of pretests administered early in the semester (see Appendix A for a copy of this scale). Respondents rated each item on a five-point rating scale ranging from disagree strongly (1) to agree strongly (S) . Composite ATHS scores were computed by summing participants' ratings, after reverse scoring when necessary. Thus, ATHS scores can range from 7 (low homophobia) to 35 (high homophobia). The mean homophobia score across the sample was 17.31, with a standard deviation of .55. The median of the sample was 17. Men and women who scored either low or high in homophobia were recruited to participate in the experiments. From the distribution of screening scores, individuals scoring in the upper and lower thirds were designated high and low respectively. Specifically, respondents with a homophobia score of 14 or below were designated as low in prejudice towards gay men. Respondents with a homophobia score of 20 and above were designated as 73 74 high in prejudice towards gay men. Based on these criteria, two hundred and sixty-one Michigan State University students were called back to participate in the experiment through sign-up sheets posted in their classes. Participants were not informed of the connection between the present experiment and the pretest completed at the beginning of the semester. Different experimenters from those who administered the pretests conducted the laboratory sessions, and they were blind to participants' level of homophobia. Additionally, 119 students from Appalachian State University participated in experimental sessions at that university. Therefore, a total of 380 individuals participated in this experiment. Of this total pool of respondents, 45 did not complete the pretest and therefore did not have available homophobia scores. Hence these respondents were excluded from the main analyses. An additional 25 respondents had homophobia scores not within.the upper and lower thirds of the distribution of scores and were thus also excluded from the main analyses. As a result, the main analyses of this experiment were conducted with 310 respondents. We Undergraduates participated in mixed homophobia-level groups ranging in size from 12 to 20 people. The experimental sessions were run by trained undergraduate research assistants. The mood induction procedure closely followed that of Bodenhausen, Sheppard, & Kramer (1994) . Upon showing up at exper were unrel PPEF exper parti Appen then 75 up at the laboratory, participants were greeted by the "first" experimenter and given a brief introduction, during which they were told that they would be participating in two short, unrelated experiments as opposed to one long experiment (see Appendix B for a copy of the experimenter protocol for this experiment). At this point, after answering any questions, participants signed a standard departmental consent form (see Appendix C for a copy of this form). ‘The "first" experimenter then asked participants to read the instructions on their materials along with a tape recorded version of the instructions for the "first" experiment. This "first" experiment was described ostensibly as concerning the emotional experiences of college students. The researchers were said to be collecting a variety of samples of such experiences. These instructions resembled those used by other researchers (e.g., Bodenhausen, 1993). In actuality, this task aimed to induce the desired emotion and consisted of what Bodenhausen and his colleagues have labeled "The Life Experiences Inventory" (LEI). ‘The instructions for the experimental groups were the same substituting the appropriate emotion: anger, sadness, happiness (See Appendix D for a copy of this measure). In the control condition, the instructions asked the participants to write about the small, everyday events of their lives, in particular the events that happened to them on the previous day. Participants were then given twelve minutes to write about the particular emotional 76 experience. After the twelve minute writing period, the ”first" experimenter asked participants to stop writing and collected their materials. The "first" experimenter then introduced the "second" experimenter and left the laboratory. The "second" experimenter took over and introduced the "second" experiment, which contained the primary dependent variables. Participants were first told that the experiment concerned peoples' reactions to typical, normal, occurrences that arose:due to the increasing urban.congestion.and cultural diversity of today's society. Then, to underscore the two- study ruse, they were then asked to sign a second standard departmental consent form (printed in a different font; see Appendix E for a copy of this form), and received a booklet. This booklet first asked participants to indicate some of their personal demographics on a participant profile that contained a series of self-ratings of current affective state. After the demographics, the booklet contained a sheet presenting the cover story for the experiment, namely the investigation of the effects of urban congestion and cultural diversity. The rest of the booklet contained the adaptation of Devine et al.'s (1991) "Should-Would" measure, and the attribution measure (see Appendix F for a copy of this entire booklet). After they had indicated their demographics, the "second" experimenter asked participants to listen to and read along with a tape-recorded version of the instructions on the second 77 page of the booklet they had just received (i.e., the cover story regarding urban congestion and cultural diversity; see Appendix G for a copy of the experimenter protocol for this study and these instructions). After listening to the instructions, they were then allowed to work through the booklet at their own pace. When they had finished, they came up to the experimenter, who stamped their experiment card and initiated the debriefing procedure. In both of the experiments, a delayed debriefing was utilized to avoid the true purpose of the experiments becoming common knowledge within the Introductory Psychology Research Participant pool. At the completion of their participation in either experiment, participants were given a participant information sheet to read that contained a general debriefing and asked that participants wishing to know more about the nature of the research complete the address form on the sheet and leave it with the experimenter (see Appendices H and I for copies of these debriefings). Upon the conclusion of the experiments, all participants who turned in the forms were mailed a complete explanation of the experiments. 92:19:: The design of the first experiment was a 4 (mood: happiness, sadness, anger, control) X 2 (prejudice level: low or high) X 2 (Target: Gay, Nongay) between-participants factorial design. Other secondary independent variables examined were gender, and verb order of questionnaire. STATISTICAL ANALYSES FOR EXPERIMENT I Egaliminazy Aaalysas ' u e s The effectiveness of the mood manipulation procedure was assessed by comparing individuals' self-ratings on various indices in the "participant profile" or demographic questionnaire. This profile asked participants to provide demographic information as well as asking them to indicate their current affective state in a series of rating scales (see Appendix F for a copy of this questionnaire). Participants completed this profile measure after the mood manipulation procedure and before completing the primary dependent measures. The self-ratings examined were ones indicative of happiness, sadness, and anger. Planned contrasts were conducted to examine the effects of mood on individuals' adjective ratings. The contrasts were conducted between the mean of the mood groups versus the mean of the neutral group on the appropriate self-rating. The only significant contrast to emerge from these tests was for self- ratings of sadness. Participants in the sad mood condition rated themselves as feeling more sad than the participants in the neutral mood group. The other adjectives did not show statistically reliable evidence of the expected pattern. '78 79 Therefore, in further exploration of the mood data, post hoc comparisons were conducted. The results of these more exploratory post hoc analyses revealed occasional differences between the mood groups in self-ratings. First, a series of oneway analyses of variance (ANOVAs) with rating as the dependent variable and. mood as the independent variable were run on the following adjective scales: happy, pleased, irritated, sad, and depressed. Significant main effects of mood were found on the following adjective rating scales: Happy, £(3, 379) = 3.55, p < .02; Pleased, £(3, 379) = 2.87, p < .04; Irritated, [(3, 379) = 3.33, p < .02; and Sad, {(3, 379) = 4.58, p < .01. Table 2 presents a summary of the means and the significant differences among them. In support of the positive mood induction procedure, participants who had written about a positive life event rated themselves as feeling happier and more pleased than the participants who wrote about an angry life event and those who wrote about a sad life event. Therefore recalling a positive life event appears to lead to more feelings of happiness and being pleased than the recall of a sad or angry life event. Some evidence also emerged in support of the anger eliciting procedure. Participants who had written about an angry life event rated themselves as feeling more irritated than the participants who wrote about a happy life events The recall of an anger producing event appears to render Tab tr (0 cu l 80 Table 2 - s o ced d tve ' Ange! Ham Sag Metal Happy 5.28a 6.14b 5.57a 6.08ab Pleased 4.86a 5.64b 5.12a 5.40ab Irritated 3.38b 2.63a 3.38ab 3.15ab Depressed 2.91ab 2.70a 3.43b 3.01ab Sad 3.02ab 2.51a 3.57b 2.68a Hota. Means in the same row with different subscripts differ significantly at p < .05 by the Tukey HSD comparison. 81 participants more irritated than the recall of a happy event. Finally, in support of the sadness induction procedure, participants who had written about a sad life event rated themselves as feeling more depressed than the participants who wrote about a happy life event, as feeling more sad than the participants who wrote about neutral life events, and as more sad than those participants who wrote about a happy life event. Therefore, it appears that the recall of a sad experience leads.participants to feel.more sad than the recall of neutral events and more depressed than the recall of a happy experience. However, it is problematic that the participants in the sad condition did not significantly differ in their self-ratings of sadness and depression from the participants in the angry condition. Previous research would suggest that this recall procedure would be successful in leading to significant differences between "angry" and "sad" participants (cf . Bodenhausen, Sheppard, 8 Kramer, 1994). Therefore the failure to find such.a difference in the present research is puzzling. These data indicate that the happy mood induction did differ from the angry and sad one in intended ways. However, the various mood inductions generally did not differ significantly from the neutral condition, nor did the anger and sadness conditions differ from each other. 82 W Because data were collected in two different geographic regions, preliminary analyses were conducted to determine whether this factor had any impact on the findings. Specifically, analyses of variance were conducted on all of the dependent variables, with participants' school location, target in the scenario, mood, and prejudice level as the independent variables. Examination of the results of these analyses did not reveal any reliable effects due to the school of the respondent. Therefore, the data were collapsed across the school variable in all subsequent analyses. Eerh_Qrder Verb order of questionnaire refers to whether individuals indicated their "should" responses before their "would" responses or vice versa. The first test of possible verb order effects was an ANOVA performed on total "should" scores with verb order, target in the scenarios, mood, and prejudice level as the independent variables. The results of this analysis did not reveal any significant effects due to verb order. Therefore it appears that the personal standards that people express do not fluctuate due to order of the questionnaire. These standards are roughly equivalent whether they are expressed before or after indicating hypothetical behaviors. This lack of fluctuation in personal standards suggests that personal standards are somewhat stable. The second test of possible verb order effects was an 83 ANOVA performed on total "would" scores, with verb order, target in the scenarios, mood, and prejudice level as the independent variables. The results of this analysis revealed a significant three-way interaction between verb order, mood, and level of prejudice, {(3, 309) = 2.90, p < .04. The pattern of means for this interaction is contained in Table 3. Looking first at the neutral mood conditions, it seems that reports of "would" behavior in the scenarios become more negative if they follow the "should" responses, particularly among high-prejudice participants. Interestingly, this effect is eliminated among happy and angry participants, and it even appears to reverse among sad, high-prejudice participants. The reasons for this reversal are not immediately obvious, and it should probably be treated cautiously pending replication. Based upon the significant three-way interaction found on total "would" scores, it was decided to submit all of the dependent variables to an analysis investigating the effects of verb order. These ANOVAs were conducted with mood, gender, prejudice level, target, and verb order as the independent variables. The only other dependent variable to reveal systematic effects of verb order was participants’ indication of their perception of the match between their "should" and ”would" scores. A main effect for verb order was found, £(1, 307) = 5.00, p < .03. Tabi pea; M “WC Lou- Hiq C0: Table 3 v‘-_! ' O I ._“.. 0.1. ‘ ‘0 YEID_QZ§§I "Should" First Low Prejudice High Prejudice "Would" First Low Prejudice High Prejudice N953. Means with different subscripts, columns, differ significantly at p < comparison. Angry 2.64ab 3.95d 3.00b 3.80de 84 HaPPY 2.45ab 3.84d 2.66b 3.790d Sad 3.17ab 3.65ad 2.59b 4.52cd Neutral 3.13ab 4.44d 2.48b 3.42c within rows and .05 by the Tukey HSD Res; of app: fir; per con the sex for set int- int and par and meat res: tha' fee. Sta: Con: eff. 85 Respondents perceived a greater match between their two sets of scores when asked to give their personal standards of appropriate behavior or "should" responses first ("Should" first: M = 6.51; "Would" first: 3 = 5.82). Indicating one's personal standards first lead. participants to feel ‘more confident that their two sets of scores corresponded or that their "would scores" closely matched their "should" responses. A significant two-way interaction was also found between sex of respondent and verb order, {(1, 307) = 5.75, p < .02 for participants' perceptions of the match between their two sets of scores. The results of this significant two-way interaction were qualified by a significant three-way interaction between sex of respondent, target in the scenario, and verb order of questionnaire that was found for participants’ perception of the match between their "should" and "would" scores, [(1, 307) = 3.97, p < .05. The pattern of means for this interaction is presented in Table 4. With respect to the effects of verb order, the findings indicate that for’men, reporting "should" standards first increases the feeling that their behavior corresponds with their personal standards, and this was particularly true for those considering gay targets. Women's responses, however, were not affected by verb order. Given that research has usually found that men, more so than women, harbor negative feelings towards gay men (Qualls, Cox, & Schehr, 1992), perhaps men experience more conflict 86 Table 4 9’1! H . - . ;- ..,.-, s’ °e -9 . - _;- y; , z- ‘6! w n ,. .n . ,. «w. .u . - - , ., . v . o g‘ . o_- ...a' - a .- , ,- _ - . '. ... -,.- . W W REEL 921191;: 9.8.1): 119119;! "Should" First Men 7.31b 6.33abc Women 7.03b 5.63c "Would" First Men 5.30a 5.60ab Women 6.63b 5.86bc Mafia. Means with different subscripts, in columns and rows, differ significantly at p < .05 by the Tukey HSD comparison. 87 about how to behave toward gay people. Perhaps allowing men to activate or think about their personal standards bafaza deciding how they would behave leads them to have more confidence that their behavior is consistent with their personal standards. These findings suggest that the procedure of Devine and her colleagues of assessing "should" responses before "would" responses may introduce some previously unsuspected influences on participants' "would" responses. On the other hand, "should" responses seem to be relatively unaffected by the order in which they are assessed. Clearly more research is needed to fully understand the implications of this methodological issue. Analysis of Primary Dependent Variables a " o d" n "Wo " ' s For each participant, a composite "should" and "would" score was computed. Higher scores were indicative of greater negativity towards the targets in the expressed personal standards regarding appropriate behavior and indications of hypothetical behavior. These scores were computed by averaging participants' ratings on the "should" and "would" ratings across the four scenarios as specified by Devine et al., 1991; Cronbach's alpha = .78). These scores were submitted to an analysis of variance with sex, prejudice level, target sexual orientation, and mood as the independent variables. 88 Sex exerted a significant main effect on total "should" score, £(1, 307) - 5.78, p < .02. Men indicated more negative total "should" scores than women (Men: M = 2.88; Women: M = 2.33). Also revealed. was the expected main effect of prejudice level on average "should" score, [(1, 307) = 35.13, p < .001. High-prejudice participants reported more negative personal standards than low-prejudice participants (Low: M = 2.05; High: M = 2.98). High-prejudice male participants, across all moods and both targets, express the most negative personal standards. The results of the analysis of variance performed on total "would" scores revealed two significant main effects; one for target, £(1, 307) = 7.66, p < .01 (Gay: M = 3.16; Nongay: M = 3.52) and the other for prejudice level of respondent on total "would" scores, {(1, 307) = 76.04, p < .001 (Low: M = 2.74; High: M = 3.96). These main effects are qualified by a significant two-way interaction between target (i.e., sexual orientation of target in the scenarios) and level of prejudice found for both total "should" scores, £(1, 307) a 26.16, p < .001, and total "would" scores, £(1, 307) = 51.69, p < .001. The means for these interactions are presented in Tables 5 and 6. The same pattern of significant differences among means was found by post hoc comparison tests for both total "should" scores and total "would" scores. For both "would" and "should" measures, prejudice level affected reactions to the gay, but not to the nongay, target. Note 89 Table 5 1‘1! 0 a " 90 d" c- -- .s a _ . . r ~°uoc - [V‘ .f WWW Target 'u 'c ve sax 112nm Low 1.69a 2.39b High 3.42d 2.53bc Magg. Means with different subscripts, in rows and columns, differ significantly at p < .05 in the Tukey HSD comparison. Table hean 1 lees! I E- 90 Table 6 g l o N o a" o S i_ o o a v a 0 e e 1319:: WI flax Nausea: Low 2.01a 3.41b High 4.30d 3.63bc Mate. Means with different subscripts, in rows and columns, differ significantly at p < .05 by the Tukey HSD comparison. HOIEO‘ to g partie targe‘ peopl diffe: eXpec reSpo for IESpc behat appea expre evid. low- dESj ind; 91 Moreover, low-prejudice participants reacted more positively to gay than to nongay targets, while high-prejudice participants reacted more negatively to gay than to nongay targets. Therefore it appears that hypotheses specifying that people of differing levels of prejudice would indicate different responses hold merit. Although highly prejudiced participants, as would be expected, indicate more negative standards and behavior in response to gay versus nongay targets, the opposite is true for the low-prejudice respondents. The low-prejudice respondents actually indicate more positive standards and behavior in response to a gay versus nongay target. It would appear, therefore, that low-prejudice people avoid the expression of prejudice, perhaps overcompensating and thus evidencing a positivity bias towards the gay targets. Indeed low-prejudice people might be more sensitive to the social desirability of their responses. Contrary to what was predicted, mood did not appear to influence "should" scores or the indications of personal standards of appropriate behavior. Mood did appear to play a role in participants' "would" responses. The analyses revealed a significant three-way interaction between mood, sex of respondent, and prejudice level of respondent, £(3, 307) = 2.75, p < .04, for total "would" scores“ The pattern of means is presented in Table 7. 92 Table 7 v.'_! 9 -. "1'- " 0 S - 1 01 0 92‘ I'll o v s o e Gender. Mgaa 222125112: Angry Hana); Sag Ismael Men Low Prej 3.31a 2.67a 3.09a 3.95a High Prej 4.31ab 4.02b 3.92ab 3.72ab Women Low Prej 2.71ac 2.40ac 2.82ac 2.51c High Prej 3.55bd 3.70bd 4.04bd 4.59d M9_t_e,. Means with different subscripts, within columns and rows, differ significantly at p < .05 by the Tukey HSD comparison. 93 Post hoc comparisons conducted. among' the 'means of this interaction revealed the following significant differences between the mood groups. For women, more negative reactions (to both gay and nongay targets) were found among high-prejudice participants compared to those low in prejudice, regardless of mood. In contrast, high-prejudice men were significantly more negative than their low-prejudice counterparts only when they were happy; This effect seems.to»be due primarily to the fact that happy, low-prejudice men are less negative than their other male, low-prejudice counterparts. Thus there seems to be a positive mood congruency effect occurring, but only among low-prejudice male participants. In the previous experiments conducted by Devine et al. (1991), comparisons were not made between participants' responses to scenarios containing gay versus nongay targets. Such a comparison appears warranted, in that the pattern of results of the present study does not always hinge upon.target sexual orientation in the scenarios (e.g., Table 7). Instead, it appears that prejudice level can have effects that are not contingent upon target sexual orientation. Prejudice may be a marker for more general patterns of responding, as well as indicating specific reactions to the stigmatized group per se. In other words, one’s level of prejudice might reflect a more general pattern of responding regardless of to whom one is responding. High-prejudice people might be predisposed to 94 respond more negatively than low-prejudice people. Low- prejudice people might be predisposed to respond to people positively, expecially people who are members of socially stigmatized groups. Therefore, hypotheses predicting differences in responses as a function of levels of prejudice or target of response, might not.be as straightforward as once assumed. W A discrepancy (D) score was computed between each participant's total "would" and total "should" scores, as was done by Devine et al. (1991), by subtracting participants' "should" ratings from their "would" ratings for each situation and summing the discrepancies across the four scenarios, (Cronbach's alpha = .67) . Previous researchers found a Cronbach's alpha = .52 for their four summed discrepancies (Devine et al., 1991). Discrepancy scores could range from 0, meaning no discrepancy between the two sets of scores, to 24, or a maximum discrepancy score, due to the fact that the response scales ranged from 1, "Strongly Disagree", to 7, "Strongly Agree". Higher discrepancy scores were indicative of "would" responses or behavioral expressions that were more negative than participants' expressed personal standards or ”should" responses. After these discrepancy scores were computed for each participant, they were analyzed in an analysis of variance procedure with gender, prejudice level, target sexual orientation, and mood as the independent 95 variables. This analysis revealed a significant main effect of target, {(1, 307) I: 11.79, p < .001. There was a greater discrepancy between respondents' two sets of scores when the target was n9; gay (Gay: M = 2.40; Nongay: M = 4.24). In other words, for all participants, expressed behaviors corresponded more to indicated personal standards when the target person was a member of a stigmatized group. There was a lesser correspondence between expressed personal standards and behavior when the target was portrayed as a neutral person. It would appear that people might be motivated to keep their behavior more in line with their standards when interacting with member(s) of a stigmatized group than a neutral person. It is possible that all people, regardless of prejudice level, when interacting with a member of a stigmatized group pay closer attention to their actions than when interacting with a member of the majority group. One result of the current zeitgeist in this society of opposing prejudice and discrimination is that most people, regardless of prejudice level, are working to keep their overt behavior politically correct. Indeed, as modern theories of racism attest, prejudice has perhaps gone underground such that even those people who possess a great deal of prejudice now carefully mask this side of themselves. It also bears mentioning that this significant main effect could also be indicative of a social desirability bias operating on the part of th regar night apprc best preji were disci High. Part; nen c Score that acre CO: I- 96 of the participants in their reactions and or self-reports regarding gay but nongay targets. Indeed the participants might simply rendering what they consider to be the appropriate response or a response that presents them in the best possible light. Also significant was a main effect for respondents' prejudice level, {(1, 307) = 6.33, p < .012. Respondents who were more prejudiced towards gay people had a greater discrepancy between their two sets of scores (Low: M = 2.74; High: M = 3.94). Therefore, contrary to what was predicted, participants possessing a high degree of prejudice towards gay men did.not have a lower discrepancy between their two sets of scores than the low-prejudice participants. It was presumed that highly prejudiced people would evidence negative scores across the board regardless of target, hence a high correspondence between their standards and behavior. This expected negative congruent reaction was not found, in that their "should" scores were more positive than their "would" scores even in response to nongay targets. It is possible that even the personal standards of high-prejudice people have Jbeen affected by concerns regarding political correctness and social desirability. Therefore, in general, their personal standards are more egalitarian than their endorsed behaviors. However, their endorsed behaviors are not egalitarian as those endorsed by the low-prejudice participants. Even high- prejudice people may have egalitarian standards in the 97 abstract, they might just not be enacted upon overtly to the same degree as with low-prejudice people. Finally, contrary to what was predicted participants’ mood state did not appear to have any systematic effects on the discrepancy between participants' two sets of scores. ;-,—_-,-.-.-1 ’ °-r --t'o . g- ... , ,— ”a ",!-_-"-.1d "W d" es Participants were asked to give an indication of how well they thought that their "should" scores matched their "would" scores. This rating was analyzed via an ANOVA with sex, mood, target, and prejudice level as the independent variables and the rating as the dependent variable. Results revealed a significant main effect for prejudice level, {(1, 307) = 8.61, p < .004. Low-prejudice participants felt that there was a smaller discrepancy or a greater match between their two sets of scores than high-prejudice participants (Low: M = 6.61; High: M = 5.82). Results also revealed a significant main effect for Target, {(1, 307) = 12.78, p < .001. People responding to a gay target felt that there was a smaller discrepancy between their two sets of scores or a greater match between them than people responding to a nongay target (Gay: M = 6.74; Nongay: M = 5.72) . These main effects suggest that low-prejudice people feel that their behavior matches their standards more so than high-prejudice people. This result would have been expected given the findings of Devine et al. (1991). These researchers found that low-prejudice 98 people had personal standards that were more internalized and felt that it was more important to act in accordance with one's standards than high-prejudice people. By the same token, it is also possible that low-prejudice people admit more readily admit to any discrepancy between their standards and behavior. Their readiness to admit such a discrepancy might stem from the fact that perhaps such.a discrepancy makes them feel more guilty as suggested by the research of Devine et al. (1991). The fact that participants perceive a greater match between their scores when responding to gay targets can be interpreted on one hand as suggesting a true avoidance of discrimination. or that, people ‘work. hard. to keep their behavior in line with their personal. standards when interacting with members of socially stigmatized groups. However, the likelihood still exists that this response is driven by social desirability or political correctness concerns. It may be that in response to members of socially stigmatized groups, people are motivated to {apart that their behavior corresponds with their standards more so than when responding to a member(s) of the majority group. This analysis also revealed a complex three-way interaction between prejudice level, target, and mood, {(3, 307) = 2.83, p < .04. The pattern of means is represented in Table 8. Post hoc comparison tests revealed the following significant differences among means. 99 Table 8 i‘-9°!!'l —' e‘el " ‘0 0!— ° 1‘ e-t 1 3‘.W‘Tl re. ” 99 2" :12 "W3. 2" 0 ‘- as a, l e '0! 9 i °‘ 1 1e ~11 os -u ‘2 2°92 and ' -' q - ,rv- . 'e condent IAIQQL H229 Prejudise Angry Heeex Sad Neutral Gay Target Low Prej 7.37a 7.91a 6.84a 7.47a High Prej 5.72b 6.48ab 6.26ab 5.67ab Nongay Target Low Prej 4.85c 6.17c 5.87ac 6.53ac High Prej 6.95b 5.61bd 4.12d 5.50bd M5233. Means with different subscripts, within columns and rows, differ at p < .05 by the Tukey HSD comparison. 100 When in a neutral mood, the trend appears to be for a greater sense of a match between scores among low-prejudice than high- prejudice participants for both targets, but especially for the gay target. Anger seems to decrease the sense of match of low-prejudice people considering nongay targets, but somewhat increase the sense of match of high- prejudice participants considering those targets. In other words, in all conditions, except in the angry mood condition and responding to nongay targets, low-prejudice participants perceive a greater correspondence between their scores than do high-prejudice participants. This one flip in the general pattern does not appear to make much theoretical sense and should not be accorded. much. importance as it could. be spurious. Moreover, the anger mood manipulation procedure did not emerge as effective as would be desired in the manipulation checks assessing mood state. Wises Participants' ratings on the 35 affect items were submitted to a principal-axis factor analysis with varimax rotation. A factor loading criterion of .40 or higher was adopted (cf. Devine et al., 1991). As in Devine's earlier studies, a six-factor solution emerged. This factor solution accounted for 63.4% of the total variance compared to 60.4% of the variance accounted for by the previous researchers. The first five factors that emerged were analogous to the five factors that emerged in previous research i.e. , Devine et al. 101 (1991) . The sixth factor that emerged from the present research did not correspond to its counterpart in the previous research. These factor labels, their loadings, their contents, and.the percent of variance for which.they accounted were as follows: A "negative feelings toward self" factor accounted for 35.6% of the variance and contained the following items: rangry' at :myself (.58), guilty (.61), embarrassed (.69), annoyed at myself (.76), regretful (.55), disappointed with myself (.81), disgusted with myself (.79), shame (.82), and self-critical (.62), (Cronbach's alpha = .91). The "discomfort" factor accounted for 10% of the variance and contained the following items: frustrated (.47), tense (.71), distressed (.68), anxious (.59), bothered (.61), uneasy (.65), and uncomfortable (.62), (Cronbach's alpha = .87). The "positive" factor accounted for 7% of the variance and contained the following items: friendly (.73), happy (.86), energetic (.80), optimistic (.75), content (.67), and good (.79), (Cronbach's alpha = .88). The "negative towards others" factor accounted for 4% of the variance and contained the following items: angry at others (.86), irritated with others (.85), and disgusted with others (.81), (Cronbach’s alpha = .87). The "threatened" factor accounted for 3.6% of the variance and contained the items of threatened (.60) and fearful (.48), (Cronbach's alpha = .70). And the last factor accounted for 3.3% of the variance and contained the items of angry at myself (.48), negative (.46), and concerned (.71). 102 This factor did not correspond to the sixth factor found by Devine et al. (1991), which contained the items of depressed and sad. Similar to Devine et al. (1991), separate affect indices for each of the first five factors identified in the factor solution were constructed by averaging the items that loaded on each factor. This procedure was undertaken to produce more reliable affect measures and produced five emotion scores for each individual. These scores were indicative respectively of the negativity felt towards the self, discomfort, degree of positive affect felt, negative feelings towards other people, and finally a sense of feeling threatened. Participants' emotion scores were considered in an analysis of variance procedure with gender, prejudice level, target, and mood as the independent variables. The ANOVA performed on the emotion score representative of negative affect felt towards the self revealed only a significant main effect of gender, {(1, 307) = 5.86, p < .02. Men reported on the whole less negative affect felt towards the self than women (Men: M = 1.85; Women: M = 2.18). Contrary to predictions, there were no differences in compunction as a function of mood state, nor did the results replicate Devine et al. (1991), who found that low-prejudice people reported more self-directed negative affect than did high-prejudice people. The analysis of the emotion score indicative of 103 discomfort revealed a significant main effect of mood, {(3, 307) c 2.74, p < .04 (Angry: MB 2.75; Happy: M = 2.39; Sad: M = 2.86; Neutral: M = 2.37). Post hoc comparisons among the means revealed that participants induced to feel sad reported more feelings of discomfort than participants induced to feel a neutral mood, {(3, 378) = 3.72, p < .01, or a happy mood, {(3, 378) = 3.72, p < .01. It appears that the participants in the sad condition felt more discomfort than those in the happy and neutral mood conditions. The participants in the angry condition felt roughly the same amount of discomfort as those in the sad mood condition. It is possible that the angry and sad moods do lead to some mood congruent affective reactions. The angry and sad mood manipulation procedures perhaps result in residual negative affect that continues to be expressed throughout the experimental session. Such a pattern was predicted for low- prejudice participants; namely that sadness and neutral moods would lead to more compunction and by inference discomfort than the mood happiness. However, in regards to high- prejudice participants, across all moods, a low degree of negative affect was predicted. Contrary to the results found by Devine et al. (1991), this analysis did not indicate that the high-prejudice participants feel more discomfort than the low-prejudice respondents. 104 The analysis of the emotion score indicative of the positive affect experienced revealed a significant two—way interaction between prejudice level of respondent and mood, {(3, 306) = 3.09, p < .03. The means of this interaction are contained in Table 9. Post hoc comparisons among the means of this interaction indicated that within each prejudice level, there were no differences in positive affect experienced as a function of mood condition. Only the low-prejudice participants, in the happy mood condition, reported experiencing more positive affect than their highly prejudiced counterparts (Low: M = 4.84; High: M = 4.05). Consequently, it appears from this analysis that an induced positive mood might accentuate differences in positive affect experienced by low versus high- prejudice individuals as a result of completing the ASW. The analysis conducted on the negative feelings participants reported feeling towards others, revealed a lack of significant effects. The analysis regarding the emotion score indicative of the degree to which participants felt threatened first revealed a significant main effect of prejudice level of respondent, {(1, 307) = 3.99, p < .05. People who report a high degree of prejudice towards gay men indicated a greater degree of feelings of being threatened than their low-prejudice counterparts (Low: M_= 1.66; High M = 1.88). 105 Table 9 ,, 1 -- V C \ ~ 1.: 01 - 1° e e900 :- ev 0 MM {:ajadige Leval ABEL! EQEEX £99 Neutral Low 4.02ab 4.84a 4.11ab 4.54ab High 4.29ab 4.05b 4.52ab 4.28ab Moge. Means with different subscripts, within columns and rows, differ significantly at p < .05 by the Tukey HSD comparison. 106 Secondly, a main effect of mood was found for reported feelings of being threatened, {(3, 307) -= 2.95, p < .03 (Angry: M_= 1.64; Happy: M = 1.66; Sad: M = 2.07; Neutral: M = 1.66). Post hoc comparisons among means revealed that respondents induced to feel sad reported that they felt more threatened than respondents induced to feel angry, happy, and neutral moods. Sad participants may have felt more threatened due to the nature of what they wrote about to produce the sad mood. It appears intuitive that sadness is often associated with a loss of some sort. The idea that one could experience a loss could render one to feel threatened or fearful. By the same token, the thoughtfulness induced by the sad mood might have led to participants to ruminate about issues regarding prejudice. These thoughts about such a socially sensitive and divisive issue could lead to thoughts of threat. Therefore these reported feelings of threat could be a byproduct of the sad mood manipulation procedure. W The aforementioned ANOVAs did not strongly support the predictions made regarding the compunction that participants representative of the various mood conditions and prejudice levels would report. Nor did these analyses replicate the findings from previous research regarding the compunction felt by people representative of differing prejudice levels (e. g. , Devine et al. , 1991) . A series of Pearson product-moment correlations was run in an attempt both to provide more 107 support for the hypotheses of the present as well as replicate the findings of previous researchers regarding the affective consequences of completing the ASW. The correlations computed were between participants' perception of the match between their two sets of scores, as well as the actual discrepancy between their two sets of scores, and the five emotion scores. Also of interest was the degree of correspondence between Perce ived and actual discrepancies. This set of correlations were run for responses to both types of scenarios i.e. Conta ining a gay target or a nongay target and for each level °f Prejudice. The significant correlations that emerged are Presented in Tables 10 and 11. Jan inspection of these correlations reveals that across '30“! targets and prejudice levels, as the perception of a match between the two sets of scores increases, the lesser the amount of negative feelings towards the self reported. Therefore a perception of a discrepancy between the two sets of 8figures is associated with reports of negative feelings dire(zted at the self. However, only for low-prejudice participants and for all participants responding to gay targets did reports of discomfort increase as the perception of a match between the two sets of scores decreased. Across b°th targets and prejudice levels, the greater the perception or the match between the two sets of scores, the more 1:) . a1:- tJ-cipants reported feeling positive and the less the Pa rticipants reported that they felt threatened. 108 Table 10 '1. 1 . e us --twen rau- . re~ ,. can 5’ e c--_to . e y c; --w--; . l "W- .H ,. .t " go o" . es d we 9: ot'o O 5 1° 1 A ns ‘0-1C- - ° t'c- . _. . We; Variables Target Gay Nongay SMATCH-NEGATIVE FEELINGS TOWARD SELF -.25** -.2o** SMATCH -DISCOMFORT - . 2 1*... ---- SMATCI-I-POSITIVE FEELINGS .20" .20" SMATCH-NEG FEELINGS TOWARDS OTHER ---------- SMATCH-FEELINGS OF BEING THREATENED -.25** -.l6* DISCREP-NEGATIVE FEELINGS TOWARD SELF .15* .15* DISCREP-DISCOMFORT ---------- DISCREP-POSITIVE FEELINGS ---------- DISCREP-NEG FEELINGS TOWARDS OTHER ---------- DISCREP-FEELINGS OF BEING THREATENED ---------- SMA'I'CH-DISCREF - . 31M - . 51H n=152 M=158 We; SMATCH = participants' perception of the match between t . hell‘ "would" scores and "should" scores; DISCREP -= the actual d iscrepancy between participants' two sets of scores. ‘- E < .05; **p< .01 109 Table 11 ’ n' a t'ons w ot'o c s d -: ' “-t' e -- . . he .- - we , ct: "Won d" nd ota "S on d" c d twee ot o co es -. d tr. c a t s ---.e :- w n ° ”ats' co 5 . ud ' ev Va r i ables Prejudice Level Low High SMA'I‘CH-NEGATIVE FEELINGS TOWARD SELF -.29** -.25** SMATCH-DISCOMFORT - . 2 1H ---- SMATcH-POSITIVE FEELINGS .2o* .16* SMATcn-NEG FEELINGS TOWARDS OTHER ---------- SMATcn-FEELINGS OF BEING THREATENED -.21** -.21* DI SCREP-NEGATIVE FEELINGS TOWARD SELF .29“ ----- DI SCREP-DISCOMFORT . 18* ----- DI SCREP—POSITIVE FEELINGS -.17* ----- DISCREP-NEG FEELINGS TOWARDS OTHER ---------- DI SCREP-FEELINGS OF BEING THREATENED .17* ----- SMATcu-DISCREF -.35** -.54** M=156 n=154 % SMATCH = participants’ perception of the match between t . he 1r "would" scores and "should" scores; DISCREP = the actual Q - lgcrepancy between participants' two sets of scores. ‘— p< .05; **p< .01 110 In response to both targets, and particularly in the responses of low-prejudice participants, the greater the actual discrepancy between the two sets of scores the greater degree of reported negative feelings towards the self. Only for the low—prejudice participants did a larger actual discrepancy between the two sets of scores lead to reports of less pos itivity, more discomfort and more feelings of being threatened. Finally, for responses to both targets and across both prejudice levels, the greater the perception of the match between the two sets of scores, the less the actual discrepancy tended to be. Analyses of Secondary Dependent Measures ’ 'o n Respondents' answers to this questionnaire were cthOIidated to render two composite measures for participants respc>nding to scenarios containing gay targets and one composite measure for participants responding to scenarios containing nongay targets. For participants responding to seer‘larios containing gay targets, the first measure was the degree to which participants attributed their "would" responses or hypothetical behavior to the targets being gay. A second composite score, computed for all participants, Pet lected the degree to which participants attributed their ”Q“ 1d responses to other non-target "justification" factors in th _ g scenarios. Both of these two composite scores were chn . . . . buted by summ1ng participants' responses to the pertinent 111 questions across scenarios. For participants responding to scenarios containing gay targets, a higher composite score across the target questions was indicative of a greater attribution of the "would" responses as being due to the targets' social group membership (Cronbach's alpha = .80). For all participants, a higher composite score across the other attribution questions ind icated a tendency to attribute "would" scores to non- minority related elements in the scenarios (Cronbach’s alpha = - 66). The composite scores of target attributions (only Participants responding to gay targets) were analyzed via an ana lysis of variance with gender, mood, and prejudice level as the independent variables. The composite scores indicative of attributions to elements not related to social group mel'ltlkoership (computed for all participants) were analyzed via an analysis of variance with gender, mood, target, and pr‘ejmdice level as the independent variables. The first of these ANOVAs concerned the degree to which resE>c>ndents attributed their likely behavior to the targets in the scenario, if responding to scenarios containing gay targets. This analysis revealed firstly a significant main ef“Sect of prejudice level, {(1, 149) = 94.49, p < .001. l>e°E>le who were more prejudiced towards gay people indicated that their endorsed behavior was due more to the tar t ' th ge In e §Q§l‘nario than did low-prejudice people (Low: M = 2.42; High: Q 7‘ 4.98). As was predicted, low-prejudice participants did 112 not attribute their ”would" responses to the targets in the scenarios to the same degree as did the high-prejudice participants. As specified by aversive racism theory (e.g. , Gaertner & Dovidio, 1986) , maintaining an egalitarian image to sel f and others is important to aversive racists. Therefore, they will not be likely to attribute their behavior to the soc ial group membership of targets to which they respond. It is possible that the low-prejudice participants in the present experiment are analogous to the conceptualization of aversive racists. Research has documented that people scoring on Paper-pencil tests as low in prejudice often share Characteristics with aversive racists, such as the presence of s‘tex‘eotypes, the endorsement of egalitarian personal St~Elr1c3ards, and a general decrying of prejudice (Devine, 1989; Devine et al. , 1991) . Therefore if low-prejudice people share some similarity with aversive racists, then they might resist Inaltirag attributions based on social group membership and perhaps for the same reasons. These reasons would be the deg ire to avoid discrimination and or the appearance of being prerdiced. By the same token, low-prejudice participants might also resist making these attributions so as to enact the it personal beliefs regarding the avoidance of behavior based upon social group membership (of. Devine et al. , 1991) . £>Q:t‘l‘Iaps this maintenance of a nonprejudiced standards and SQ l f—image is less important or relevant to people posseSSing Q. l'ligh degree of prejudice. Therefore highly prejudiced 113 people would not resist attributing their behavior to the social group membership of the target. A significant two-way interaction was also found between prejudice level and mood, [(3, 149) = 3.07, p < .03, for attributions made to the targets in the scenarios. The pattern of means for this interaction is presented in Table 12 . Post hoc comparisons among the means of this interaction indicated that in all four of the mood conditions, low- Prejudice participants indicated to a lesser degree than did hi gh-prejudice participants that their behavior was due to the target in the scenario (Angry: £(1, 35) = 13.78, p < .001; Happy: m1, 43) = 77.00, p_ < .001; Sad: 3(1, 37) = 11.31, p < - 002; Neutral: {(1, 32) = 47.74, p < .001'). An inspection of the means in Table 12 reveals that the magnitude of the 55- f ference between high and low-prejudice participants in the number of attributions made to the targets in the scenario is g“Tealtest in the neutral and happy mood conditions. The low- prejudice participants attribute their behavior to the targets in the scenario the least in the happy condition. High- pI‘ejudice participants appear to make the most target attributions in the happy condition. The thoughtlessness usua 11y associated with happiness might lead to mood- cc>159ruent, positive responses on the part of the low-prejudice pat‘t-icipants. The high-prejudice participants might be simply th°ughtlessly expressing their prejudice. 114 Table 12 ,- - 1 mbe_ -_ ; 'u 118 280- O 1‘ 0' l r '1: ,0 o "W-u " 0 '~ 4: : 1 '01- -u ‘0 290-. :12 ' 7... W921); EQQQ 212514252122 ABEL! H3221 sad neutral L931 2.66a 1.79a 2.76a 2.643 mg; 4.59:: 5.05b 4.61b 5.65b 3%. Means with different subscripts, within columns and rows, differ significantly at p < .05 by the Tukey HSD comparison . 115 The magnitude of the difference between the two groups of participants is the lowest in the sad condition. Perhaps the thoughtfulness usually associated with the emotion sadness, attenuates the difference between people representative of the two prejudice levels. The low-prejudice participants might be thoughtfully and honestly admitting to the basis of their endorsed behavior, thus making more such attributions. The high-prejudice participants in a thoughful attempt to look nonprejudiced might be avoiding such attributions or falling Victim to impression management or social desirability Concerns. The second analysis, performed across all participants, concerned the degree to which respondents attributed their "won 1d" scores to factors in the scenario other than the taI‘get. An analysis of variance was conducted on these composite scores with gender, prejudice level, target, and mood as the independent variables. The results revealed a significant main effect of prejudice level, {(1, 303) = 6.77, p < .01. High-prejudice Part icipants more so than low—prejudice participants indicated that their endorsed behaviors were due to factors other than the target in the scenario (Low: 11 = 4.06; High: M = 4.42). Also found was a significant main effect of target, £(1, 303) ‘ 24.99, p < .001. Participants attributed their endorsed ehaviors more to nontarget-related factors when the scenarios <2 or“‘tained a nongay target (Gay: M = 3.88; Nongay: I! = 4-57) - 116 These findings are contrary to expectations. It was predicted that the low-prejudice participants would be more concerned with maintaining an egalitarian image both to self and others then high—prejudice participants and as a result would be more like 1y to attribute their endorsed behavior to elements in the scenarios not related to social group membership. Instead what was found was that high-prejudice participants indicated more such attributions than did low-prejudice participants and moreover indicated more of these attributions in response to nOrigay targets than to a gay targets. A significant two-way interaction between prejudice level and gender was also found, [(1, 303) = 7.40, p < .01, for the number of attributions made to nontarget-related factors. Results also indicated a significant two-way interaction between gender of respondent and target, £(1, 303) -= 7.75, p < - 006 for number of attributions. These effects are qua l :ified by a significant three-way interaction between prejladice level, gender of respondent, and targets in the ac"’eiI‘lario on the number of attributions for "would" scores to fac'tors in the scenarios other than the social group men"lbership of the target, {(1, 303) = 5.10, p < .03. The pa"rtern of means for this interaction is presented in Table 13 - When responding to scenarios containing gay targets, low- prgjudice women indicate the least number of attributions to n Ontarget-related factors and significantly fewer than Table 13 y ' 2,! u- 0 i '-u '01- na Jrnde d e ' ev SESELJEIEQS 212129.192 Hen (3&3r3g' Low Prejudice 4.51abd High Prejudice 4.07a Nongay Low Prejudice 4.30bd High Prejudice 4.41b 117 .oe a on - nde t Gender r-e -'e,e '1 - 0 S EQEEB 3.38bc 4.19a 4.50d 4.90bd h1S21:;__. Means with different subscripts within, columns and I?<>‘h?=s, differ at p < .05 by the Tukey HSD comparison. 118 high-prejudice women. Low-prejudice women, responding to gay targets, also endorse fewer attributions than low-prejudice women responding to scenarios containing nongay targets. In addition, high-prejudice women responding to gay targets endorse fewer nontarget-related attributions than do high- pre 3’ udice women responding to nongay targets. It appears that low—prejudice women responding to gay targets are less likely to attribute their behavior to nontarget-related factors as compared to high-prejudice women or low-prejudice women responding to neutral targets. 1.: a 'n ' s ' e It was predicted that regarding total "should" scores, 1°"errejudice participants would indicate nonprejudiced or 90$ itive responses in all mood conditions except anger. It "as predicted that the high-prejudice participants would ind icate prejudiced or negative personal standards in all mood (”and itions. It was predicted that indications of hypothetical behavior or total "would" scores, across participants with a low degree of prejudice, would reflect prejudice or be negative in all mood conditions except the happy mood. High- Dre 3 udice participants’ total "would" scores were predicted to be I‘legative or reflective of prejudice in all mood conditions. The results of the analyses of these two measures only Part ially supports these predictions. On both total "should" and total "would" scores a main effect of prejudice level i . hdlcated that high-prejudice participants did endorse more 119 negative standards and behaviors than the low-prejudice participants. Overall, prejudice level affected responses to the gay but not to the nongay target. In addition, the low— prejudice participants reacted more positively to the gay tenacrqget than to the nongay target, whereas the opposite was true for the high-prejudice participants. Contrary to predictions, mood state did not interact significantly with Prejudice level to influence indications of standards and behaviors. This lack of mood effects will be more thoroughly addressed in the following discussion section. Mood was also predicted to interact with prejudice level to affect the actual discrepancy between participants "should" and "would" scores. Low-prejudice participants were predicted t1<> «evidence a smaller discrepancy between their scores when ang:l:'y and happy and larger discrepancies when in a sad or neutral mood state. The high-prejudice participants were 3j-Itlply predicted to have a low discrepancy between sets of ScOx-es across all moods. Results indicated the opposite of the se predictions. Mood did not systematically interact with pIrejudice level to alter discrepancies. Furthermore, high- pI~ejudice participants, across all moods, evidenced greater discrepancies between their two sets of scores than low- Dre j udice participants . Regarding participants' perception of a match between ‘zti‘aflir two sets of scores, no specific predictions were made. Results indicated that low-prejudice participants perceived a 120 greater match between their two sets of scores than the high- prejudice participants. Moreover, all participants, regardless of prejudice level, perceived a greater match between their two sets of scores when responding to scenarios containing gay as opposed to nongay targets. Finally, regarding the affect indices, current predictions were not born out nor were the findings of Previous research replicated. Low-prejudice participants did not report a greater sense of negative self-affect as the disorepancies between their two sets of scores rose, nor did they indicate more of this affect than high-prejudice Participants. The only significant finding indicated that wc>iltlen reported more of this emotion than men. Regarding feel ings of discomfort, contrary to previous research (e.g., De":’Lne et al., 1991), high-prejudice participants did not report these feelings as a result of discrepancies between their two sets of scores, nor did they report more of this emc’t ion than low-prejudice participants. From the present research, it seems that participants induced to feel a sad mood , experience more discomfort than participants induced to fee 1 happy and neutral moods. DISCUSSION OF EXPERIMENT I This experiment intended to demonstrate that various incidental emotional states would impact upon respondents’ indications of their behaviors and standards in response to targets representative of socially stigmatized groups. Moreover, these incidental mood states were also examined as they related to any affective correlates experienced by the research participants as a result of completing the exPerimental measures. It was generally predicted that the reSponses of high-prejudice respondents would be negative and ind icative of a general prejudice towards gay men. Incidental mood states were not predicted to cause much variation in the gerleral negativity expected to be seen in the responses of the high-prejudice respondents. Low-prejudice participants were F’t‘edicted to evidence in their responses both the presence and the lack thereof of prejudice as a function of mood state. The angry and happy mood induction procedures were predicted tQ be associated with congruent responses on the part of low- ht‘Ejudice participants either negative or positive bQSpectively. The sad and neutral moods were predicted to lgad to nonprejudiced indications of personal standards and m§re negative endorsed hypothetical behaviors. ”a. 1.22 ’ de A Although the incidental affect mood induction procedure did not have the impact intended, the results do indicate some support for the hypotheses as well as pointing to some general trends associated with the various moods. Moreover, these trends in the mood effects were generally consistent with the expectations formulated at the outset of this research, a:l-‘i':l‘iough not as statistically powerful as would have been Preferred. It is surprising that the mood induction procedure was not associated with statistically powerful trends given its robust and impactful nature in other experiments (e.g., strack et al., 1983; Bodenhausen, Kramer, & Suesser, 1994; BC>denhausen, Sheppard, & Kramer, 1994). In hindsight, perhaps some feature of the "should-would" task serves to undermine incidental affective states. It is pqssible that certain tasks might not be affected by incidental mood states or by the same token, that certain tg sks might undermine the effects of these mood states. Given that previous research has indicated that research ha~t~ticipants can overcome the cognitive consequences of incidental mood states, if so motivated (Schwarz et al., 1 S 91) , perhaps the should-would task motivates participants to ignore the effects of their current mood state. On the part of low-prejudice participants, this prerimental task might activate their concerns of maintenance at an egalitarian self-concept both to self and to others as 12; well as the determination to respond in line with personal egalitarian standards (Devine et al., 1991; Gaertner £- Dovidio, 1986). Therefore the low-prejudice participants' incidental mood states might be easily dissipated in the face of a task that incites so many issues important to them. Indeed the complexity and interestingness of a task has been shown, in previous research to be linked to more systematic and careful cognitive processing (Isen, 1993). It can be argued that, for people personally dedicated to avoiding prejudiced responses, i.e. low-prejudice participants, a Cc>r1‘l.:act situation with a member from a socially stigmatized 9): clip constitutes an involving situation that might facilitate caI‘eful, monitored responses. One manifestation of this monitoring might be the lack of incidental moods to e ignificantly influence responses. It is a matter of speculation as to why the majority of t‘QSponses of the high-prejudice participants did not qanonstrate statistically robust effects due to mood state and th this might relate to the nature of the experimental task. h.thaps, in general, their responses related to prejudice are % QInewhat stable and not prone to fluctuate due to factors like ‘h incidental mood state. The responses of high-prejudice bQople might simply be generally negative, especially to It"ambers of socially stigmatized groups. There is some Q\’:i.dence to suggest that a sad emotional state somewhat §~t‘tenuates the negativity of the responses of the high- 1.25. prejudice participants. Therefore, in general, future research will be needed to more definitively answer the question of how incidental affective states influence responses on the "Should-Would" measure. Regarding more specific responses, Table 14 presents the mean scores across the primary dependent variables for all Participants. An inspection of the first column of Table 14 r e\reals that across all mood conditions, high-prejudice Participants express more negative personal standards than do low—prejudice participants. This basic difference in personal s'tal'ldards as a function of prejudice level has been documented by previous research (i.e., Devine et al., 1991). These e)“75>eriments indicated that the personal standards of high- prejudice people allowed for a greater expression of Iaegativity or prejudice towards outgroup members than the beIrsonal standards of people possessing a low degree of prejudice. These results are perhaps not surprising given the SQtzial support for prejudice towards gay men. The overall small amount of fluctuation in these scores Que to mood could be a by product of the measure itself as prosed to a lack of effectiveness of the induction procedure. E’QI‘haps personal standards governing appropriate intergroup §'bhtact are somewhat stable cognitive features. These §'tandards may only be prone to shift significantly when the bQrson is experiencing a more intense mood than an incidental °he . Table 14 es de v 'ab 5 s or W 0 es ow- c t nt u Had mm. _ul_Wo d W W Egry 1.61 1.89 1.16 2.03 Happy 1.72 1.92 0.83 1.48 Sad 1.59 2.25 2.63 2.11 Neutral 1.87 2.00 0.53 1.82 $9 5 Ell'l-E .1. I !'i ! M shou 1d Would W Qompunct i on \ Ahgry 3.13 4.31 4.72 2.00 I‘Iappy 3.79 4.06 1.10 2.31 Sad 3.28 4.24 3.84 1.84 “Qutral 3.46 4.64 4.72 2.11 §tg Lower numbers indicate less mean negativity. 1.2.5 In general, it could also perhaps be speculated that manipulations of incidental affect impact the most upon social judgments that are the least self-relevant or self-involving. Indeed most other experiments assessing the cognitive effects of incidental affective states have utilized social judgments tasks that are not as self-relevant as indicating one's Personal standards of conduct. Typical tasks in these experiments involve persuasion or social judgments rendered al:><>1:|t target people (e.g., target peoples' guilt; c.f., Bodenhausen, Sheppard, & Kramer, 1994). Therefore, it is POSsible that the nature of the tasks presented to participants, enables these previous experiments to document robust effects of the mood manipulation procedure. Whereas on a task as self-relevant as the A"Should-Would" measure, these mood manipulations are not as effective and do not produce as large an impact on participants' responses. An alternative e)(planation to the one given above is that participants baaponses are indicative of strong social desirability biases that inhibit them from expressing prejudice. Finally, the analyses involving gender somewhat qualify the previously discussed prejudice effects. With women, those who were highly prejudiced expressed more negative personal Qtandards across all mood conditions and in response to both types of targets. For men, a different trend emerged. Interestingly, high-prejudice men were only significantly more r.‘agative than their low-prejudice counterparts when in a happy mood nega pers expe al., nega wel] res; reac to s in ‘ drix the Per: the rele exan towe will more “1% Deep °Vn 121 mood. Without this mood state, men in general responded negatively to gay targets, irrespective of their level of personal prejudice. It bears mentioning at this point that previous experiments utilizing the "Should-Would" measure (Devine et al., 1991) also found that men appeared to respond more negatively than women to gay targets. In this experiment as well, gender appears to be associated with participants' responses. Women appear to be more positive in their reactions to gay male targets than do men. It is interesting to speculate on how these results might change if the targets in the scenarios were portrayed as lesbians. Perhaps one driving component of homophobia is the correspondence between the gender of the target gay person(s) and the gender of the person evaluator. This correspondence in gender might render the target person more threatening as well as more self- relevant. In response to lesbian targets women might, for example, exhibit a similar pattern of negativity as men do towards gay male targets. Therefore it is possible that women will respond negatively to gay female targets perhaps even more negatively than their male counterparts. Future research might want to investigate more fully the question of whether people are more negative in response to gay people from their own gender as opposed to gay people from the other gender grouping. 12§ W As predicted, the prejudice level of the research participant played a significant role in the majority of participant' responses. In addition, the sexual orientation of the target(s) in the scenarios interacted with prejudice to influence responses, independently of mood. It appears that standards governing appropriate intergroup behavior, when in a mixed social group context, are more negative for men and those participants possessing a high degree of prejudice. The hypothetical behaviors ‘that subjects. endorsed followed. a similar pattern to their standards. Level of prejudice of the participant affected reactions to the gay targets, but not to the nongay targets. Low-prejudice respondents indicated more positive behaviors in response to the gay than the nongay targets. The high-prejudice participants were generally negative towards both targets. The "should" and "would" scores of the high-prejudice participants would not be predicted on the basis of previous research investigating prejudice (i.e., Devine et al., 1991). High-prejudice participants would be expected to react more negatively to members of socially stigmatized groups than to neutral people. Moreover, they would be expected to react more negatively to these outgroup members than low-prejudice people. The results from this experiment, however, support the latter proposition, but not the former. The results from the analysis of the scores of low- 122 prejudice participants was also not as expected. Although their responses to both types of targets was expected to be roughly equivalent, low-prejudice respondents actually showed a positivity bias towards the gay targets: their standards and behavior are more positive in response to the gay targets than to the nongay targets. Perhaps low-prejudice people are overmotivated, so to speak, to avoid acting in a negative or prejudiced manner in response to members of socially stigmatized groups. As a result, low-prejudice people actually demonstrate a positivity bias towards these outgroup members. Lay experience and stereotypes often document that minority group members note that some people, who consider themselves to be low-prejudiced, actually demonstrate an obvious "overdoing it" or blatant positive prejudice towards the outgroup members. These self-assumed low-prejudice people often come off as phony or trying to hard to fit in with minority'group'membersw However, it might also be likely that low-prejudice people have truly renounced or moved beyond any negativity towards social outgroups. Therefore the positivity they are expressing towards the gay targets would be genuine and real. However the potential likelihood of this true positivity still does not adequately explain why the low- prejudice participants were more positive to a gay target than to a neutral target. Additional research should investigate more fully this positivity bias and the mechanisms behind it. It might also be informative to see if this extreme positive 1.3.9 responding is found among low-prejudice people when evaluating other socially stigmatized groups. Regarding the discrepancies between participants scores, an inspection of Table 14 again reveals several trends across the mood conditions that, although not statistically significant, support some of the predictions made for the results of the experiment. It was predicted that the actual discrepancies between participants' expressed standards and behaviors would be low for high-prejudice participants across all mood conditions. They were assumed to express both personal standards and endorsed behaviors that were correspondingly negative. However, an inspection of Table 14 reveals that high-prejudice participants evidence a large discrepancy between scores in all of the mood conditions except the happy one. Therefore the high-prejudice participants express differential negativity in their personal standards versus behaviors rather consistently except when in a happy mood. It is possible that high-prejudice people generally allow for a greater expression of negativity in their behaviors than is specified in their abstract personal standards. Angry and neutral moods might intensify the negativity expressed behaviorally; whereas sadness and especially happiness attenuates it. The greater degree of correspondence between the scores in the happy condition might be due to the positive valence of the mood in conjunction with its usual associated thoughtlessness. A positive, carefree 121 mind set might prompt high-prejudice people to engage in heuristic, but uniform responding, hence a greater correspondence between scores. By the same token the thoughtfulness usually associated with sadness might enable high-prejudice participants to maintain a correspondence between their scores regardless of any presence or lack thereof of a motivation to maintain this correspondence. For example these participants, due to their thoughtful state, might be able to recall their previous responses and continue to respond consistently. The discrepancies between the scores of the low-prejudice participants associated with the various mood conditions approximated the pattern predicted. Although the differences were not significant, angry and happy moods appeared to be associated with smaller discrepancies and sad moods with larger discrepancies between expressed standards and behavior. Interestingly, the smallest discrepancy between scores evidenced by the low-prejudice participants was in the neutral mood condition. The minimal interference with cognitive processing associated with sad and neutral moods might presumably be associated with low-prejudice participants honestly expressing any negativity that might take place in their behaviors; even if it contradicts their personal standards. Therefore larger discrepancies would be seen while under the influence of these moods. The angry and happy moods might lead to smaller 1.3.2. discrepancies between scores due to their strength in inducing unifome mood congruent responses. In other words, both "should" and "would" scores are perhaps following the valence of the low-prejudice participant's prevailing incidental mood state with both being positive or negative depending on the mood. Interestingly enough, just as in several of the other analyses, the type of target portrayed in the scenarios significantly affected the discrepancies between participants' expressed personal standards and hypothetical behaviors. There was a greater expressed discrepancy between standards and behaviors for all participants when the targets in the scenarios were portrayed as neutral as opposed to gay. It is possible that all of the respondents were monitoring their answers to a greater degree when responding to targets who were members of socially stigmatized groups as opposed to those who were portrayed as neutral. This pattern of response would be predicted based upon the prevalent notion in today’s society, and perhaps especially among college students, that prejudice is wrong and not to be condoned (Crosby, Bromely, & Saxe, 1980). However, despite any potential monitoring of responses, and contrary to what was predicted, overall, high-prejudice participants evidenced a greater discrepancy between their sets of scores than did low-prejudice participants. It is possible that highly prejudiced people may not care about 121 maintaining a correspondence between their standards and behavior. This correspondence may not be as important to them as it is to low-prejudice people. Indeed previous research has shown that low-prejudice people have internal standards regarding egalitarian behavior that are more internalized than the equivalent standards for high-prejudice people (e.g., Devine et al., 1991). In addition, high-prejudice participants may not possess as strong a motivation to avoid acting or appearing in a prejudiced manner as low-prejudice participants, as might be suggested by aversive racism theory (Gaertner & Dovidio, 1986). As a result of this lack of motivation, high-prejudice participants would not devote as much cognitive effort to insuring that their behavior upheld their personal standards as low-prejudice people. Therefore discrepancies between standards and behavior would be greater for those possessing a high level of prejudice. In terms of perceived discrepancies, low-prejudice people report a greater perception of a correspondence between their standards and behavior than high-prejudice people. The investment that low-prejudice participants have in acting nonprejudiced could be leading them to actually indicate more correspondent scores. This presumption is supported in the finding that low-prejudice participants do indeed evidence a smaller discrepancy between their two sets of scores than the high-prejudice participants. It is also plausible that low- prejudice participants are motivated to assume a 1&1 correspondence between their standards and behavior. Perhaps, due to social desirability concerns, they are motivated to perceive that their behaviors follow their personal standards, particularly when responding to targets representative of socially stigmatized groups. As was discussed earlier, the correspondence between behaviors and standards might not be that important to high-prejudice people, hence their general lack of a confidence in their scores’ correspondence, as compared to low-prejudice participants. In addition, participants felt that their scores had a greater correspondence when responding to gay as opposed to nongay targets. The current "politically correct" mentality present in today's society might result in at least a desire on the part of people representative of both levels of prejudice to exhibit a vigilance to match behavior to standards when interacting with members of socially stigmatized groups. As a result, all participants report a greater correspondence between their scores when responding to gay as opposed to neutral targets. In addition, participants' perception of the match between their should and would scores increased as the actual discrepancy between the two sets of scores decreased. Therefore, it appears that participants are generally cognizant of the correspondence of their two sets of scores. High-prejudice people attributed their hypothetical behaviors, to a greater degree, more both to target-related 115 and nontarget-related factors in the scenarios than did low- prejudice participants. Low-prejudice women appear to make such attributions for their hypothetical behaviors to the least degree. Given that high-prejudice participants attributed their responses to a greater degree to the social group membership of the targets in the scenarios than did low- prejudice participants there is some support for the contention that aversive racists are analogous to low- prejudice people. Aversive racism theory (i.e., Gaertner & Dovidio, 1986) proposes that aversive racists are motivated to censor their responses to avoid giving the impression to self and others that they are prejudiced. As a consequence, aversive racists, and perhaps low-prejudice people, would be motivated to avoid attributing their responses to anything relevant to social group membership or to prejudice. The maintenance of a nonprejudiced self-image is perhaps less important to high-prejudice people. Therefore, they would not hesitate to attribute their behavior to issues revolving around social group membership. Moreover, it is a theoretical question, as to how people classified as high in prejudice, in the current study, would be classified according to aversive racism theory. B 1. II E i v' ! J {1551} Analyses of participants’ affective states subsequent to indicating "should" and “would" scores, revealed, that the sad mood induction procedure was associated with reports of more 1.3.6 feelings of discomfort and threat than the other mood procedures. This affect could be residual due to the mood induction procedure. It is possible that the nature of the sad experiences that participants write about lead to feelings of discomfort and threat. Experiences that induce sadness perhaps involve a loss of some sort. Rumination about a personal loss might produce as a byproduct, feelings of threat as one contemplates potential vulnerability. Such rumination would be possible given the thoughtfulness that sadness often induces. Another possible explanation for this affect can also be suggested, given that sadness has been linked to a greater propensity to engage in effortful cognitive strategies, (e.g. Bodenhausen, Sheppard, & Kramer, 1994). Perhaps the mood-induced thoughtfulness leads to a rumination upon previously indicated responses. As they think about these previous responses, participants might also think of the possible discrepancies between them and or their negativity and thus feel discomfort and threat as a result of these worries. Mixed evidence emerged in support of the results documented by previous researchers regarding differential reports of feelings of compunction contingent upon personal prejudice level (e.g. Devine et al., 1991). For example, it was not the low-prejudice participants who felt the most negative self-directed affect per se, as in previous research. Instead, women felt more compunction than did the men. Overa] scores and t1 threa1 low-p1 disco: actual colle; discrc with part Part1. f9611: betwel preju. 990131 their Stand 111 Overall, the more participants felt that their two sets of scores corresponded, the more positive affect they reported, and the less negative self-directed affect and feelings of threat were reported. Finally the correlations indicated that low-prejudice participants reported more feelings of discomfort, threat, and negativity towards others, as the actual discrepancy between their two sets of scores increased. Therefore, as the previous 'work of Devine and. her colleagues (e.g., 1991) has suggested, perceptions of discrepancies between standards and behaviors are associated with more negative affective correlates, especially on the part of low-prejudice people. Only for low-prejudice participants and participants responding to gay targets, did feelings of discomfort increase as a perception of a match between standards and behaviors decreased. Therefore low- prejudice people may be more predisposed than high-prejudice people to experience negative affect or feel discomfort if their' behavioral. responses ‘transgress ‘their' personal standards. It bears discussing why the basic findings of Devine et al., (1991) regarding prejudice with and without compunction were not strongly replicated in this experiment. The overall lack of replication of previous findings is puzzling at best. The only ostensible major difference between the measures used in the two studies involves the presence or lack thereof of a cover story to the "Should-Would" measure. The present 12E experiment, unlike the previous researchers introduced the "Should-Would" measure via a cover story that asserted that the measure concerned respondents' reactions to scenarios involving the issues of urban congestion and cultural diversity. The previous researchers explicitly told respondents that the measure assessed reactions to members of socially stigmatized outgroups. They rationalized this approach on the assumption that people could be motivated to respond honestly. In order to motivate their participants they assured them of the confidentiality of their responses, encouraged them to be honest, stressed the importance of the data and the issues that it concerned. Perhaps the compunction and discomfort documented by Devine and colleagues stems from demand characteristics that arise due to the instructions given to respondents regarding the "Should-Would" measure. Knowing that they are responding to a measure assessing prejudice might naturally lead high-prejudice people to feel uncomfortable and perhaps the most salient response for low-prejudice people is guilt. By the same token, it is also possible, that once clued in to the purpose of the measure, respondents indicate responses in line with their social desirability perceptions. The affective responses documented by the Devine laboratory might arise therefore as a function of asking participants to consciously think about and report their standards and behaviors regarding prejudice. Low-prejudice participants assume perhaps that the "correct" 1.22 affective response is compunction; whereas the high-prejudice participants assume that it is discomfort. In addition to not replicating the affective reactions of Devine et al., (1991), the present experiment yielded preliminary results to suggest that the usual order of the ”Should-Would" measure i.e. "should" responses first, could be problematic. Indicating "should" responses first appeared to result in participants' being more confident of the correspondence between their indicated standards and behavior as well as resulting in more negative subsequent "would" scores. These verb order effects appear to be especially pervasive for men responding to gay targets. Women's responses do not appear to be affected by verb order to the same degree as the responses of men. Given that men, in general, appear to respond more negatively towards gay men than women, perhaps men are more confident that their behavior follows their standards, when those standards have been conceptualized prior to indicating behavior. The order of the questionnaire could also be leading to demand characteristics that are noticed by all research participants. Having just been asked to indicate their personal standards, when asked to indicate hypothetical behaviors, perhaps participants perceive that what is expected of them is to give responses discrepant from the previously indicated standards. When asked to indicate hypothetical behaviors first, these demand characteristics may not be as 1.4.9 salient, for participants might not be as easily able to discern the purpose of the study. Despite any present demand characteristics, it is interesting that participants, on the whole, did not appear to heed these characteristics and indicate discrepant scores. If anything, they resisted these demands and responded with scores that pretty much corresponded to each other. In addition the "would" scores given second tend to be more negative. If social desirability or impression management concerns are important to the participants, they would resist endorsing behaviors more negative than their personal standards. These preliminary results suggest that the procedure of Devine and her colleagues of always assessing "should" scores before "would" scores could bias the results obtained with the measure. More explicit testing of these order effects needs to be done, particularly as they relate to a demand characteristics or a general social desirability bias. c s' n The results of this first study suggest several areas of further inquiry. Firstly, future research might want to continue to investigate the effects on the "Should-Would" measure of incidental affective states utilizing other mood induction procedures. Regarding the previous research conducted by Devine et al., (1991), and in light of the present research, the results obtained using the "Should- Would" measure are qualified by both the ordering of the questi questi questi inhere the In seriot well i Also findi: corre Devin repli 15.1 questionnaire and the types of targets portrayed in the questionnaire. It appears that the "Should-Would" questionnaire warrants thorough testing to explore any inherent order or demand effects that might be a byproduct of the measure. These effects of question order could pose serious threats to the interpretation of previous research as well as limiting the use of this measure in future research. Also meriting further exploration is the stability of the findings of previous researchers regarding the affective correlates of perceived discrepancies in responses (e.g. Devine et al. 1991), as these effects were not strongly replicated by the present research. pro and aft Wit ex; aft din int im bet in EXPERIMENT TWO Introduction As was previously discussed, Bodenhausen (1993) has proposed a distinction between two types of affect, "integral" and "incidental" that are relevant to intergroup contexts. Incidental affect is described as an emotion(s) that arises due to factors that are irrelevant to an intergroup context in which the current social perceiver is imbedded. "Integral" affect, on the other hand, is an affective state that is related to the intergroup context of the social perceiver. In other words, this type of affect arises specifically as a response to members of various social groups. This type of affect is associated with the usual conditions and contexts with which the group is associated and encountered. The first experiment contained herein examined the effects of incidental affect upon a social judgment task. This second experiment aims to be a exploratory investigation of the effects of integral affect upon the same social judgment task. Several compelling reasons can be noted that warrant this investigation of integral affect and the empirical comparison between the two types of affect. Firstly there exists a gap in the affect and cognition literature. Previous research has not systematically investigated the effects of integral affect 142 143 upon cognitive processing or specifically upon the use of stereotypes in social judgment. Most of the research has examined the effects of :mild affective states that are independent of the intergroup context (Hamilton, Stroessner, & Mackie, 1993). This second experiment.might serve simply as an extension of previous work, demonstrating the generalizability of the effects of mood state regardless of the its incidental or integral quality. However, the results of this second experiment could raise issues for future research as well as questions about the generalizability of most of the research findings of the affect and cognition literature. Additionally, this experiment is noteworthy in that one compelling reason to propose that incidental and integral affect might lead. to» different effects upon the social judgment maker lies in the motivational concerns associated with each type of affects It is possible that integral affect introduces different motivational concerns from incidental affect. For example, because integral affect arises due to factors that are intergroup related, there might be motivation to handle this emotion differently from incidental emotion so as to maintain a certain quality of intergroup relations. It is also possible that that when making judgments about social groups, affect arising from a vicarious source would affect these judgments differently than affect arising from the same source about which social judgments are to be made (i.e., the gr01 how the aim« inf L int: imp; COUI Equ ange Fir: as ' inVe Foe: harc Sec: WOu] inve is Fina more cenc 144 group in question). It is an empirical question, therefore, how these two types of affect differ, if indeed they do, in the responses they produce in individuals. This experiment aimed to address the question of how integral affect influences social judgments rendered about gay men. The following discussion concerns specifically how integral positive and negative affect might have a different impact upon cognitive processing than their incidental counterparts (cf. Bodenhausen, 1993). This exploratory experiment focused upon the effects of integral happiness and anger. These two emotions were chosen for several reasons. Firstly, they lent themselves nicely to the present paradigm as they are basic emotions commonly elicited in laboratory investigations (e.g., Bodenhausen, Sheppard, Kramer, 1994). Whereas pretesting indicated that sadness was particularly hard to induce via an integral mood induction procedure. Secondly, the choice of these emotions seemed natural as it would allow for one positive and one negative emotion to be investigated. In addition, the emotion of anger was included as it may often be associated with intergroup relations. Finally, the emotion of happiness was included as there exists more literature regarding positive affect from which to draw conclusions. W It might be expected, that people would become cognitively lazy or experience a reduction in cognitive cc Wh gr 192 out the. or 1 slst 145 capacity as a result of happy integral moods, similar to incidental ones (e.g., Worth 8 Mackie, 1987). Therefore a happy mood arising from an intergroup context might lead people to think shallowly and heuristically about the group at hand (e.g., Bodenhausen, Kramer, & Suesser, 1994). Such a cognitive strategy might involve the use of stereotypes. Therefore, expressions of prejudice might be found in the responses of integrally happy participants due to the general thoughtless state and the resultant use of stereotypes induced by the mood. The results of the second study might reveal, however, that integral positive affect has different effects upon the social judgment maker than incidental affect“ Integral affect might introduce additional motivational concerns not present in situations where perceivers are experiencing incidental affect. Given that happy incidental affect has been shown to lead to more heuristic, cursory processing (Bodenhausen, 1993), it stands to reason that people might be unwilling to correct and/or compensate for this lack of cognitive capacity when the outcome of the present task environment is of no great personal consequence (Bodenhausen, 1993, p. 31; Forgas, 1989). However, in situations of direct intergroup contact, outcomes can have a greater relevance to personal interests, therefore positive integral affect might lead to an exertion or motivation on the part of happy individuals to think more systematically, presumably to ensure positive outcomes for the 146 self. In other words, in situations eliciting integral affect, due to the on-line potential ramifications for the perceiver, there might be an increased motivation to overcome any processing constraints due to mood and process the available information more carefully. Indeed, research has shown that participants' level of motivation can interact with the effects of an emotional state. When, subsequent to an incidental mood induction, research participants are informed that they will be held accountable for their responses, they do not render the stereotypic judgments usually found with incidentally happy individuals. This research suggests that happy people are capable of avoiding or attenuating the effects of a happy mood if motivated to do so (Bodenhausen, Kramer, & Suesser, 1994, Exp. 4). The personally involving nature of situations eliciting integral affect might suffice to heighten social perceivers’ motivation (Petty & Cacioppo, 1986) . As a result, people who are integrally happy may not evidence the heuristic and less thoughtful cognitive processing of individuals who are incidentally happy. Other research supports the proposition that situations eliciting integral affect might elicit systematic cognitive processing. An intergroup-contact situation is perhaps intuitively interesting to the social perceiver. This type of situation is distinctive and might have ramifications for the perceiver. In addition, the intergroup situation could in ha 1e 51 ar of be! rec 191 res pre abc rec gen Th1 inc tee 197. Eat emo. 147 involve a cognitive context that is rather complex. Research has shown that both a social perceiver's level of interest in a task and the level of complexity of the cognitive context lead to more systematic thinking on the part of individuals in a positive mood (Isen, 1993). Additional motivational issues may also come into play in situations producing integral affect. Integral emotions may arise due to factors related to intergroup dynamics. Patterns of interaction among ingroup and outgroup members that have been documented by research include phenomena such as reciprocity and retaliation (Donnerstein, Donnerstein, Simon, & Ditrichs, 1972; Dollard, Doob, Miller, Mowrer, 8 Sears, 1939). For instance, individuals induced to feel happy, in response to a member of a particular social group, as in the present experiment, when subsequently asked to form judgments about that group, might produce judgments indicative of a reciprocity or benevolence effect. The happy mood might generalize from its integral stimulus to the group as a‘whole. This generalization effect is one basis for the "contact hypothesis," a prejudice reduction technique that involves increasing positive contact among different groups. This technique has been refined (e.g. Wilder, 1984) and proven to be somewhat successful (e.g. Aronson, Bridgeman, 8 Geffner, 1978). Therefore, positive integral affect might lead to the motivation to return like affect or to reciprocate such an emotion to other members of the group involved. st af 151 th 11' nc st ir CC hc re he fr he CC Ce re 19 in 0f Th 148 HegeSixe_Intsgral_Affe2t Little research exists that would enable one to formulate strong predictions about the effects of negative integral affect, in particular anger. The research that does exist finds that the effects of incidental anger tend to mirror those found with incidental happiness (Bodenhausen, Sheppard, & Kramer, 1994). In other words, participants in whom incidental anger was induced tended to engage in heuristic or nonsystematic cognitive processing, as well as utilizing stereotypes in social judgments. Given that the effects of incidental happiness and anger appear somewhat parallel, one could use the research involving incidental positive affect as a springboard for predictions. The valence of the mood, however, must still be taken into account. First, in accordance with the research previously reviewed, the effects of integral anger might mirror those of negative incidental affect. Therefore an angry mood arising from an intergroup context might lead people to prefer more heuristic cognitive strategies. People might become cognitively lazy or experience a reduction in cognitive capacity as a result of angry integral moods, similar to the results found with incidental anger (e.g., Worth & Mackie, 1987). Social judgments might.manifest this laziness, as with incidental affect (Bodenhausen, 1993), by following a pattern of reliance upon stereotypes when making social judgments. This enhanced use of stereotypes might lead to responses 149 indicative of negativity and prejudice, as many stereotypes are negative. In addition, negativity might dominate the responses of integrally angry participants due to the negative valence of the mood itself. In other words, a simple mood congruency effect might be observed. This mood congruency effect could be distinguished from a stereotyping effect. A stereotyping effect would be target-specific (i.e. negativity directed at only gay targets), while a mood congruency effect would generalize to nongay targets as well. The prediction of a stereotyping effect is perhaps warranted given that the affect commonly associated with stereotyped groups is negative (e.g. , Jackson & Sullivan, 1988). In a situation involving incidental affect, people who become angry as a result of some factor that is not related to an intergroup context may when asked to make social judgments about outgroup members, make negative judgments about them. This negativity might serve a "scape-goating" function. In other words, people might vent or displace their anger upon the minority group. Research has documented such "scapegoating" effects of negative affect (Dollard, Doob, Miller, Mowrer, & Sears, 1939). On the other hand, integrally angry peoples' negative emotion arises specifically due to issues involving outgroup members, so the subsequent reactions of these angry people when placed in an intergroup context might not reveal 150 "scapegoating" but actual retaliation. Depending upon how the integral affect is induced, it might become directed specifically at outgroup members. Research that examines interracial aggression offers support for the prediction that integrally angry individuals might adopt a retaliatory strategy towards the outgroup involved in the affect elicitation. Therefore these integrally angry people might be prone to render negative social judgments about the outgroup involved. In the experiments concerning retaliation (e.g., Donnerstein, Donnerstein, Simon, & Ditrichs, 1972), the researchers found that white college research participants expected more aggression from black than white targets and delivered more aggression to blacks than to whites when retaliation was unlikely. Therefore negative integral affect, as elicited in the present context, could lead to negative judgments on the part of the participants. These individuals will have just experienced a negative interaction with an outgroup member, and in completing the dependent measures, they will have the opportunity to retaliate against the same outgroup, with no possible return retaliation on the part of the outgroup member. The results of this exploratory study might reveal, on the other hand, that integral anger leads.toldifferent.effects than those found with incidental anger. As was discussed earlier, situations of direct intergroup contact can involve outcomes of a greater relevance to personal interests (e.g., 151 Schwarz, 1990); therefore, integral anger might lead to a exertion on the part of angry individuals to think more systematically, presumably to ensure positive outcomes for the self. In other words, these participants will be motivated to overcome the influence of their negative affect, so as to ensure a desirable outcome of the situation. These people might be motivated to resolve the situation positively and avoid retaliation to the self as suggested by Schwarz (1990). This degree of systematic thinking would be greater than that seen with negative incidental affect. Therefore, one might observe that the integrally angry participants do not adopt cursory cognitive strategies (i.e. , stereotyping, Bodenhausen, 1993) of their incidentally angry counterparts. Integral anger might also induce more systematicity in thinking due to the information it presents to individuals. In other words, the emotion of anger might signal to the individual that something is wrong in the environment that necessitates attention (Schwarz, 1990). Individuals might then direct attention not only to the antecedents of the anger, but also to factors that might alleviate it. Therefore, participants would approach any task systematically in an attempt to handle and dispel the emotion. Consequently, these participants might be prone to examine a situation thoroughly and not fall back on stereotypes or other forms of simplified information as a basis for a judgment. Finally, integral angry affect might lead to more VI 91' St im rm Elva. rese 152 systematic cognitive processing due to the interestingness and complexity of the situation that elicits it (Isen, 1993). These factors might stimulate the attention and effort both necessary and conducive to systematic thinking. Again, participants would be motivated to exert more cognitive effort to the situation at hand and avoid such simplification strategies as stereotyping. t e o There exists evidence to suggest that integral affect will differ in its effects from incidental affect (e.g., Schwarz, 1990), as well as evidence to suggest that the two types of affect will have similar effects (e.g., Wilder & Shapiro, 1989). This experiment was designed to be an exploratory study of how integral affect influences individuals' social judgments and in particular, their expressions of prejudice. In this experiment, participants were given an experience designed to elicit an integral emotion, specifically anger or happiness, or a neutral mood state. This manipulation was accomplished by having the participants witness, and indirectly participate in, a staged interaction between the experimenter and two confederates. This interaction was charged with a particular emotion. For example, in the "happy" condition, the experimenter presented a dilemma to the participants that the two confederates resolved by doing the actual participants a favor. Previous research demonstrates that such small tokens as finding a dime 153 or having a favor performed for one can induce a positive mood state (Isen, 1993), so it*was anticipated that the favor would also be an effective mood elicitor. In the integral angry condition, when presented with the dilemma, the confederates staged an interaction in which one confederate presented his solution to the dilemma as an ultimatum and one that must be followed to the possible disadvantage of the actual participants. The pretesting of this scenario to blind raters resulted in most viewers perceiving the situation as one that would upset and irritate the participants. This interaction constituted a manipulation of integral affect for in some conditions, the confederate primarily responsible for the positive or negative outcome of the interaction presented evidence to strongly suggest that he was a member of the target outgroup, gay men. This affect induction procedure satisfies Bodenhausen's definition of integral affect as that "which.is elicited by the social group itself"; or by inference a member of the social group. In the control conditions, this main confederate presented a neutral persona. Therefore the present study examined the responses of people who were high and low in prejudice towards gay men, while under the influence of integral positive, negative, or neutral affect on the "should-would" dependent measures utilized in the previous study. It would also be possible to compare the two studies and thus responses rendered while under the influence of the two types of affect. Method of Experiment II 225.191: The design of the present experiment was a 2 (prejudice level: low/high) x 3 (mood: anger, happiness, neutral) x 2 (confederate: gay/nongay) x 2 (scenario target: gay/nongay) between-participants factorial design. Other secondary independent variables examined were sex of respondent and verb order of the questionnaire. E !° . ! 1 5 J C!' Several hundred undergraduate psychology students completed the seven-item Attitudes Toward Homosexuals Scale (ATHS; Bouton et al., 1987) as part of a variety of pretests administered early in the semester. Composite ATHS scores were computed as in Experiment I. Female and male participants who scored either low or high in homophobia were recruited to participate in the experiment. Approximately 191 people participated in the experiment by signing up for experimental sessions on. sign-up sheets posted in their classrooms. Twenty-five participants for whom no pretest scores were available were dropped from analyses, leaving 166 total participants. A median split was used to designate participants as either high or low in homophobia. The median homophobia score was 17 with a standard deviation of .56. 154 155 Participants were not informed of the connection between the present experiment and the pretest completed at the beginning of the semester. Different experimenters from those who administered the pretests conducted the experimental sessions and they were be blind to participants' level of homophobia. Erocedugg and Materials Undergraduates participated. in. mixed. prejudice-level groups ranging from two to six people. Integral affect was induced in this experiment by having participants witness an interaction between the experimenter and two confederates, one of whom could be identified as a member of the social group in question (e.g., gay men). The emotional context of this interaction was intended to provoke either a happy or angry reaction in participants. Specifically, as participants showed up at the lab, they were directed to take a seat in the front of the room by the experimenter. The experimenter explained that he/she wanted the front of the room to fill up first. Already seated in the room was a confederate, posing as a research participant. This confederate is henceforth labeled Confederate #1. At no later than five minutes past the official starting time of the experiment, or when sure that all of the actual participants had shown up, a white male confederate (Confederate #2) entered the room. (Two research assistants always enacted the role of the second confederate. These two confederates were similar in build and physical attractiveness as determined by an independent panel of 156 raters.) This second. confederate was directed, by’ the experimenter, to shut the door and to take a seat in the available seats in the back of the room. In the conditions where the second confederate was portraying an ostensibly gay image, he presented three signs to indicate that he could possibly be a gay man. Firstly, he wore a shirt containing a pro-gay statement. On the front of this shirt was the following statement enclosed in a large pink triangle: "Don't assume I'm straight." His other clothing was nondescript and held constant across confederates, specifically consisting of pressed blue jeans and loafer type shoes. Secondly, he was wearing an earring in his right ear. A single earring worn in the right ear is recognized in the United States to be a signal of being a gay man. Finally, the ostensibly gay male confederate wore a men's fragrance that pretesting had established to be strong and to waft easily over the lab as the confederate walked by to his seat. These manipulations were to confirm common stereotypes about gay'men (see Herek, 1988, for a review). In the conditions where the second confederate projected a neutral or nongay image, he wore a monotone, short-sleeved t- shirt and the same pants and shoes used in the other conditions, but did not wear the earring and cologne. After the second confederate entered the room he was directed to take a seat among those available in the back of the room. He then walked past the other participants, with his shirt thus 157 in full view (in the "gay" conditions therefore, research participants would be easily able to read the slogan on the t- shirt as established via pretesting of the interactions) and took a seat. After the confederate had been seated, participants were greeted and given a brief introduction to the experiment (i.e., they would indicate their reactions to scenarios involving urban congestion), and role was called. Then the experimenter explained to participants that there was a bit of a problem. It seemed that the researchers were currently running two versions of the same experiment. Participants were told that one version took almost the full experimental hour allotted, and the second version took just a little over a half hour. Both versions of the experiment were to be worth the same amount of experimental credit. Participants were told that each of them was supposed to have been randomly assigned to a particular version of the experiment, with at least two people participating in the longer version, down the hall in another research laboratory, under the direction of another experimenter who was currently waiting for the two research participants. However, this assignment had not been done. The experimenter then said the he/she had been trying to think of fair way to assign each of them, at which point the confederates interrupted with their dialogue. This interaction between the confederates and the experimenter constituted the mood induction procedure. In the 11 in in ex] 0t} Poi 158 "happy" condition, immediately after the experimenter explained the problem, and before the actual research participants could react, both the confederates volunteered to do the longer version of the experiment. Therefore, in the happy condition, the "gay man" in essence did a favor for the participants by volunteering to do the long version of the experiment. In the angry condition, immediately after the problem was explained by the experimenter, the "gay male" confederate protested. any“ random (assignment of the participants to the experiments and asserted that he must partake in the shorter version due to his lack of time. Therefore in this condition the confederate was in essence attempting to "put one over" on the participants by demanding that he participate in the shorter version of the experiment, leaving the actual research participants no choice but to partake of the longer version. of the experiment. The experimenter resolved the dilemma by saying: "I'll just give all of you the shorter version, you two just come with me to the other laboratory (pointing to the confederates) . I’ll work it out with my supervisor later." In the neutral mood conditions, no emotionally charged interaction took place. The confederates entered the lab, as in the other' mood. conditions. 'The experimenter simply explained the problem to the research participants, as in the other conditions. Then the experimenter appeared to randomly point to the two confederates and then asked them to go to the 159 other lab where another experimenter waited to administer a longer version of the experiment to them. In the control conditions, participants witnessed the same mood induction procedures described above (i.e., angry, happy, neutral), but with the second confederate projecting an ostensibly nongay image (i.e., no shirt slogan, earring, or cologne). In all conditions therefore, the confederates were led out of the room, under the pretense that another experimenter was waiting to administer the longer version or in the case of the angry condition a short version of the experiment to them in another laboratory. The experimenter walked the confederates to the door, gave them some directions, ostensibly to the other lab, and then returned to the laboratory and shut the door. (See Appendix G for a complete transcript of the dialogues for these interactions.) The actual research participants were then asked to fill out their consent forms and the "Participant Profile" on the first page of their booklet. Then participants were told to read the second page of their booklet along with the tape recorder. The instructions were the same used in Experiment I to introduce the "Should-Would" measure. The participants were then given the cover story and.dependent.measures used in the "Should-Would" portion of the first experiment. Upon completion of the materials they were debriefed and dismissed as in the first experiment. In Statistical Analyses for Experiment II Preliminary Analyses W The effectiveness of the mood manipulation procedure was assessed by comparing participants’ self-ratings on various indices in the same demographic questionnaire used in the first experiment (see Appendix F for a copy of this questionnaire) . Similar to the first experiment, participants completed this questionnaire subsequent to the mood induction procedure and before completing any of the dependent variables. Oneway ANOVAs with self-rating as the dependent variable and mood as the independent variable were run on the following adjective scales: happy, pleased, and irritated. The adjective anxious was also included in these analyses, as it was deemed possible that anxiety might be a byproduct of the interaction used to induce:the affective states. .Although the results of these ANOVAs did not reveal many significant 3 values, an inspection of the means indicated that some of them were in the predicted direction. Further analyses also offered some support for the mood induction procedure. The oneway ANOVA performed to test for the effectiveness of the positive mood induction procedure did not reveal any significant differences among the means of the mood groups for 160 t1“. ef pr va wa mi of e1 SC re no Or an CO 84 Dr. th. 9f} dic 161 the self-rating of happiness (Angry: u = 6.18; Happy: M = 5.99; Neutral: 11 8 6.12); or feeling pleased (Angry: M = 5.26; Happy: u a 5.40; Neutral: M = 5.38). It was decided to conduct further ANOVAs to test for effects of the mood induction procedure within each level of prejudice, with confederate and mood as the independent variables and the self-ratings as the dependent variables. It was speculated that the integral mood induction procedure might have different effects upon participants as a function of their personal level of prejudice, given that the central element to this mood induction is contact with a member of a socially stigmatized group. The analysis of the self-rating of happiness did not reveal any significant main effects nor interactions between mood and the persona of the confederate (either ostensibly gay or nongay) for participants of either level of prejudice. The analysis of self-ratings of feeling pleased, however, revealed a significant main effect due to the persona of the confederate for participants scoring low in prejudice, £(2, 84) -= 4.09, p < .02. Across all mood conditions, low- prejudice participants reported more positive feelings when the confederate portrayed an ostensibly gay image (Gay: M = 5.48; Nongay: M = 4.66). The initial oneway ANOVA performed to test for the effectiveness of the negative mood induction procedure also did not reveal any significant differences among the means of 162 the mood groups on the relevant self-ratings. The participants in the mood groups did not rate themselves as feeling significantly different in terms of irritation (Angry: u = 3.18; Happy: 5 = 2.52; Neutral: M = 3.07); or anxiety (Angry: M = 4.51; Happy: M = 4.07; Neutral: M = 3.68). Some limited evidence did emerge in support of the effectiveness of the negative mood induction procedure in specific further analyses. One-tailed t-tests were computed among the means of the mood groups on the self-ratings of irritability and anxiety. These t-tests revealed that participants in the "angry" mood group rated themselves as feeling significantly more irritated than those in the "happy" mood group, 3;(133) = 1.78, p < .04 (Angry: M = 3.18; Happy: 1! = 2.52; Neutral: H = 3.07) . In addition, participants in the "angry" group rated themselves as feeling significantly more anxious that those in the neutral mood group, t(2, 122) = 1.81, p < .04 (Angry: M = 4.51; Happy: 1;! = 4.07; Neutral: 1! = 3.68). In addition, ANOVAs were conducted separately for the participants of high and low prejudice, on these negative self-ratings, with confederate and mood as the independent variables. A significant main effect for mood emerged for participants with a high level of prejudice on self-ratings of irritability, £(2, 84) = 4.09, p < .02. Post hoc comparisons conducted among the means indicated that the neutral mood group rated themselves as feeling significantly more irritated 163 than the "happy" group (Angry: u = 3.17; Happy: 3 2.06; Neutral: 3 = 3.29). It appears that, although pretesting indicated that the current procedures would be successful, based upon these analyses, they were not as effective as would be desired. The relevant evidence regarding the positive mood induction procedure suggests that low-prejudice participants reported feeling more positive in the presence of an ostensibly gay confederate. In addition, participants in the "happy" mood condition reported themselves to be less irritated than those in the "angry" mood condition. The results in support of the negative mood induction procedure indicate that participants in the "angry" condition report more anxiety than their neutral mood counterparts. Furthermore, high-prejudice participants in the neutral mood condition report more feelings of irritation than their high-prejudice counterparts in the "happy" mood condition. As a result, some evidence does exist to suggest that the anger manipulation renders a mood not akin to happiness or neutrality. The issue of designing and implementing effective integral affect induction procedures will be addressed further in the discussion section regarding this experiment. Sex_£ffe2t§ The preliminary analyses revealed several effects of gender. The general trend that emerged from these analyses was that women responded more favorably'to‘the‘gay target than 164 did men. In response to the nongay target, the sexes largely did not differ. Due to the lack of interactions between the variables of mood and confederate sexual orientation with gender, and due to the fact that the sample size is not adequate to examine fourth-order and higher interactions, gender was not included in the main analyses. Supplementary analyses of gender effects are presented in Appendix J. 2222422519.: Verb order of questionnaire refers to whether individuals indicated "should" responses first or instead indicated "would" responses first in the questionnaire. ANOVAs were run on all the dependent variables, with prejudice level, mood, and verb order as the independent variables.'.Also included in these analyses were two other independent variables. The first, henceforth referred to simply as "confederate," refers to the type of persona presented by the main confederate (either neutral or ostensibly "gay"). The last independent variable included is labeled "target," and refers to the way targets were portrayed in the questionnaire scenario (either as gay or neutral). Although there were numerous effects of verb order across the dependent variables, they did not bear on the key theoretical issues. Because they are of secondary interest, the details of these analyses will be presented in Appendix K. A few general trends that emerged from these analyses, however, can be briefly noted here. Just as in the first 9f Dr 165 experiment, indicating ”should" scores first was associated with participants' reports of a greater perception of a correspondence between scores. Unlike in Experiment I, in this second study, indicating "should" scores first was also associated with a smaller actual discrepancy between participants' two sets of scores. In addition, in Experiment II, low-prejudice participants perceived a greater match between their two sets of scores when responding to scenarios containing nongay targets, if they answer "should" questions first. Similar to the first experiment however, the opposite was true for high-prejudice participants. They perceived their scores to correspond to a greater degree when responding to gay targets and "should" questions were asked first. Analysis of Primary Dependent Variables " o d" nd "Wo " t'n s For each individual, a composite "should" and "would" score was computed, as in the first experiment (Cronbach's alpha = .81). Higher scores were indicative of greater negativity towards the targets in the expressed standards and indications of hypothetical behavior. Total "should" and total "would" scores were submitted to an analysis of variance procedure with prejudice level, mood, target, and confederate as the independent variables. Although there has been limited empirical testing of the effects of supposed integral affect, various possible predictions about the results of this experiment were th- af re Th re pa th co be) prc mo: re] po: uni mo] ac] Pa] Egg 61. Dr: stl wot the the 166 theoretically suggested. One possibility is that integral affect.might elicit the same effects as incidental affect” .As a result, the integral happy and angry moods might lead research participants to think shallowly and heuristically. Therefore the total "should" and total "would" scores might be reflective of stereotyping and roughly equivalent, for participants representative of both levels of prejudice, in the "angry" and "happy" conditions as compared to the neutral condition. In the neutral mood conditions, participants would be expected to respond in line with their level of personal prejudice, with high-prejudice participants indicating the most negative scores, particularly towards the targets representative of stigmatized social groups. It is also possible that integral moods lead to patterns of effects unlike those seen with incidental moods, perhaps encouraging more thoughtfulness. Specifically, it might be predicted that across the different mood conditions, low-prejudice participants would respond positively, in line with their egalitarian personal beliefs and standards (e.g., Devine et al., 1991). Although the moods might not predispose high- prejudice participants to utilize heuristic cognitive strategies, it might still be predicted that their responses would remain negative and indicative of prejudice, both due to their prejudice regarding the social group in question and their potential lack of internalized personal standards decrying prejudice based upon social group membership. 167 The analyses of both total "should" and total "would" scores revealed, just as in the first experiment, a significant main effect of target in the scenario for total ”would" scores, [(1, 165) = 14.26, p < .001. Overall, it would appear that participants express more positivity in their behaviors in response to gay targets than in response to neutral targets (Gay: H -= 2.88; Nongay: u = 3.52). Also replicated in the second experiment was a significant main effect of prejudice level of participant ("Should": {(1, 165) = 37.43, p < .001; "Would": {(1, 165) = 51.95, p < .001). High-prejudice participants indicated more negative total "should" and total "would" scores than low-prejudice participants (Total "Should": Low: 1! = 1.66; High: )1 = 2.89; Total "Would": Low: H = 2.43; High: M = 3.74). Therefore, analogous to previous research (i.e. Devine et a1, 1991), high-prejudice people express more negative personal standards than low-prejudice people. In this second experiment, high- prejudice participants also indicated more negative hypothetical behaviors than the low-prejudice participants. These main effects are qualified by a significant two-way interaction between prejudice level of respondent and the targets in the scenarios for both total "should" and total "would" scores (Total "Should": [(1, 165) = 6.37, p < .013; Total ”Would": {(1, 165) = 21.67, p < .001). Post hoc comparisons were conducted with the means for both of these interactions. These comparisons revealed the same pattern of t: SC be 51'] 3p; in 168 significant differences among means for both total "should" and total ”would" scores as presented in Tables 15 and 16. Furthermore, this same pattern of significant differences was seen in the data from Experiment I. Low-prejudice participants expressed less negativity in their personal standards and hypothetical behaviors in response to a gay as opposed to a nongay target. In response to scenarios containing gay targets, low-prejudice participants expressed less negativity in their personal standards and endorsed more positive behavior than did high- prejudice participants. Indeed, it appears that low-prejudice people might be motivated to act in a nonprejudiced manner, perhaps even to the point of exhibiting a positivity bias towards members of socially stigmatized groups. High- prejudice participants, across all mood groups, responded in a more negative fashion than low-prejudice participants, especially to gay targets. Contrary to what was predicted, neither mood nor persona of the confederate appeared to influence total "should" and total "would" scores. Prejudice level of respondent and the social group membership of the targets evaluated appeared to be having the largest effect on these scores. Therefore, similar to Experiment I, the mood induction procedure did not appear to have systematic effects upon participants' indications of personal standards or hypothetical behaviors. 169 Table 15 ' " i-a t’l‘ ..-...1, ‘ rve 9 112922 21212Q122_L§!§l Gay EQDQQX Low 1.32a (50) 2.22b (31) High 3.02b (58) 2.62b (27) Ngtg. Means with differing subscripts, in rows and columns, differ at.p < .05 in the Tukey HSD comparison. Cell sizes are indicated in parentheses. 170 Table 16 ,-._, . ._ "W0- 0" . -- .M. i , '., . ' — _o* - l’V‘. . WWW 12:39.3; W Gay. 1191193): Low 1.78a (50) 3.48b (31) High 3.83b (58) 3.56b (27) Note. Means with differing subscripts, in rows and columns, differ at p < .05 in the Tukey HSD comparison. Cell sizes are indicated in parentheses. 171 This paucity of mood effects will be addressed in subsequent discussion sections of this paper. W A discrepancy (D) score was computed between each participant's total "would" and total "should" scores, as was done in the first experiment (Cronbach's alpha = .66). It bears mentioning that the reliabilities computed for the two discrepancy indexes of the experiments contained herein are higher than those reported for the discrepancy index used in previous research (i.e. , Devine et al. , 1991; Cronbach's alpha 2 .52). Discrepancy scores could range from 0, meaning no discrepancy between the two sets of scores, to 24, or a maximum discrepancy score. Higher discrepancy scores meant that the participant's endorsed hypothetical behaviors reflected more negativity than the expressed personal standards. These discrepancy scores were computed for each participant and then analyzed in an analysis of variance procedure with prejudice level, mood, target, and confederate as the independent variables. Specific predictions were not made at the outset of this research regarding the size of the discrepancies between total "should" and total "would" scores as a function of the independent variables. If the effects of integral affect mirror those of incidental affect, the general thoughtlessness engendered by the mood conditions (e.g., Bodenhausen, Kramer, 8 Suesser, 1994; Bodenhausen, Sheppard, & Kramer, 1994) might 172 predispose participants to fail to carefully monitor their responses. This lack of monitoring in conjunction with the fact that previous research demonstrates that, in general, most respondents tend to demonstrate some degree of discrepancy between their scores (Devine et. al, 1991), leads to the prediction that, regardless of level of personal prejudice, the "angry" and "happy" participants would exhibit a large discrepancy between their total "should" and total "would" responses. The discrepancies of the high-prejudice participants might generally being higher than those of low- prejudice participants due to their level of personal prejudice and lack of personal stake in egalitarian values (e.g., Devine et al., 1991). If integral affect does not lead to the lazy cognitive strategies found with incidental affective states, then perhaps, people representative of both levels of prejudice will approach the social judgment task in a systematic manner. They might monitor more closely the correspondence between their responses. Therefore the discrepancy between total "should" and total "would" scores would be small for all participants, regardless of mood state and prejudice level. These discrepancies might even be smaller in the mood conditions than in the neutral condition, as it might be the moods that stimulate the thoughtfulness that leads to a greater correspondence of scores. Mirroring the results of Experiment 1, the analysis of discrepancy scores firstly revealed a significant.main effect pa Cc 173 of targets in the scenarios, [(1, 165) = 5.88, p < .02. There was a greater discrepancy between participants' scores in response to a nongay target as opposed to a gay target (Gay: 3 = 2.58; Nongay: u = 4.45). This significant main effect, found in both experiments, suggests that all participants, regardless of mood or prejudice level, monitor their responses more closely in response to gay as opposed to nongay targets. Similar to the analysis of total "should" and "would" scores, a significant two-way interaction between prejudice level of respondent and the targets in the scenarios emerged for discrepancy scores, F(1,165) = 3.97, p < .05. Post hoc comparisons were conducted among the means of this interaction and revealed only one significant difference as presented in Table 17. Low-prejudice participants exhibited a significantly lower discrepancy between their two sets of scores when responding to scenarios containing a gay target as opposed to a nongay target. High—prejudice participants demonstrated roughly the same amount of discrepancy between their two sets of scores no matter what type of target was presented in the scenarios. It appears that the way that the targets are portrayed.in the scenarios affects.the discrepancy between the total "should" and total "would" scores for low-prejudice participants more so than for their high-prejudice counterparts. 174 Table 17 .-.. .uo_. . I':c -9.;c a; --. . g ” 1o- 9" ..e . "To o." . - 1 n . . i - x 99-“. 9 W 12292: W1 flex Henge! Low 1.84a (50) 5.06b (31) High 3.228 (58) 3.74ab (27) Ngtg. Means with differing subscripts, in columns and row, differ at.p < .05 in the Tukey HSD comparison. Cell sizes are indicated in parentheses. 175 Moreover, it would appear that low-prejudice people might be monitoring their answers more in response to gay targets than to neutral or nongay targets. Unlike the analogous analysis in Experiment I, this analysis did not reveal a significant main effect for prejudice level, whereby high-prejudice participants evidenced a greater discrepancy between their two sets of scores than low-prejudice participants. A significant three-way interaction between prejudice level, target, and mood was also indicated by the analysis of discrepancy scores of Experiment II, £(1,165) = 3.73, p,< .03. Post hoc comparisons made among the means of this interaction revealed the significant differences presented in Table 18. Regarding low-prejudice participants, when in a neutral mood, there is a significantly greater discrepancy between scores when responding to scenarios containing nongay targets as opposed to gay targets. It appears that the neutral mood might induce low-prejudice participants to monitor their responses more closely in response to gay as opposed to nongay targets. By the same token, low-prejudice participants indicate behavior that corresponds to their endorsed standards to a greater degree in response to the gay targets as opposed to the neutral targets. ‘The "angry" and "happy" mood induction procedures perhaps attenuate this correspondence between scores by attenuating the presumed effortful cognitive strategy of monitoring the correspondence one's scores. 176 Table 18 .-.1 Lu0_l ., sis ‘9. 3' w-tn .11 "ago- a" .12 . " -_ ." oes ..s . u c 01- ° - r-v- . ;'-99!2‘1 Ereiudise_nexel need IQIQEE 529:! 32222 Neutral Low Gay 0.38a (l6) 2.45a (22) 2.67a (12) Nongay 3.77ab (13) 5.83ab (6) 6.08b (12) High Gay 2.84a (19) 2.12a (26) 6.003 (13) Nongay 2.64a (11) 9.00b (5) 2.45a (11) Note. Means with differing subscripts, in columns and rows, differ at.p < .05 in the Tukey HSD comparison. Cell sizes are indicated in parentheses. 177 Such a response would be predicted if integral affects lead to the same cursory cognitive styles as has been documented with incidental affective states (i.e., Bodenhausen, Kramer, & Suesser, 1994). The significant differences between means that emerge involving the high-prejudice participants indicate that they demonstrate a significantly greater discrepancy between their two sets of scores when in the "happy" mood condition as opposed to the "angry" or neutral mood conditions. Furthermore, the high-prejudice participants, in the "happy" mood condition, also evidence a greater discrepancy between their two sets of scores in response to a nongay target as opposed tOva gay target. It appears that being in the "happy" condition renders high-prejudice participants similar to the neutral mood, low-prejudice participants. From these results it seems possible that, just as ‘with the low-prejudice participants, integral affect, especially positive, might be leading to the same deficits seen with incidental moods in the high-prejudice participants. For instance the "happy" mood induction might be leading to their scores exhibiting a greater discrepancy supposedly due to a lack of monitoring scores, particularly in response to nongay targets. "—9-n-‘1 :' '- c-- 01 . ,- .a , o-tw-eg " ;o__q" and 392W Participants were asked to give an indication of how well they thought that their "should" scores matched their "would" 178 scores. This rating was analyzed via an ANOVA with mood, confederate, target, and prejudice level as the independent variables and the perception rating as the dependent variable. If the effects of integral affect are similar to those of incidental affect, then it might be predicted that the two mood conditions would lead to a general thoughtlessness. It is hard to predict, however, how this lack of systematic thinking could manifest itself. Just as in the first experiment, a significant main effect of the targets in the scenario emerged from the analysis of discrepancy scores, 2(1, 165) = 5.74, p < .018. Participants, across moods and prejudice levels, perceived a greater correspondence between their two sets of scores when responding to scenarios containing gay targets as opposed to nongay targets (Gay: :1 = 7.31; Nongay: u = 6.41) . Given the fact that the actual discrepancy between sets of scores was greater when responding to nongay targets for all participants, perhaps people representative of both prejudice levels are motivated to censor their responses towards members of socially stigmatized groups to a certain degree. Therefore, people in general, might both monitor their responses more closely and are more aware of the similarity among their responses enacted towards members of socially stigmatized groups. Similar to total ”should", total "would", and discrepancy scores, in both experiments, a significant two-way interaction 1): Pa De 179 between prejudice level of participant and the targets in the scenarios was found for participants' perceptions of the match between their sets of scores, £(1, 165) = 4.39, p < .04. Post hoc comparisons revealed only one significant difference among the means. Low-prejudice participants perceived a significantly greater match between their two sets of scores when responding to scenarios that contained gay targets as opposed to scenarios containing nongay targets (Low-Prejudice: Gay Target: M = 7.80 (N = 50); Nongay Target: M = 6.23 (N=31); High-Prejudice: Gay Target: M = 6.90 (N = 58); Nongay Target: M = 6.63 (N = 27) . Therefore, for participants with a low degree of prejudice, the sexual orientation of the targets in the scenarios especially affects their report of their perception of the match between their two sets of scores. Low-prejudice participants report more confidence that their endorsed behaviors correspond to their expressed personal standards if responding to situations that contain gay targets. It is important to point out that low- prejudice participants might be motivated to overestimate the correspondence between their sets of scores, particularly when the targets is representative of a socially stigmatized group. Again this motivation would stem from the low-prejudice participants' personal stake in looking nonprejudiced (i.e., Devine et al., 1991). This analysis, just as in the first experiment, revealed a significant three-way interaction between prejudice level of th ar Ex si th co 10 an at pa ha 1a of be CO De th "h So 180 participant, mood, and the targets in the scenarios, £(2, 165) s 3.25, p < .04. Post hoc comparisons were conducted among the means of this interaction and the significant differences are presented in Table 19. These means show that, similar to Experiment I, low-prejudice participants perceived a significantly greater should-would match toward gay targets than did high-prejudice participants under neutral mood conditions. In this second experiment, the difference between low and high-prejudiced participants vanished under both angry and happy mood conditions. This pattern was largely attributable to the fact that perceived match of responses to a gay target was especially low among high-prejudice participants under neutral conditions, but not in the angry or happy conditions. It seems that low-prejudice participants largely assume a good correspondence between their indications of personal standards and behavior, regardless of mood. High- prejudice participants' confidence in the correspondence between their sets of scores increased in both of the mood conditions“ It is interesting to speculate whether or not the perceptions of the high-prejudice participants are due to thoughtless overestimation of accuracy, especially in the ”happy" condition given the large discrepancy between their scores as seen in Table 18. Table 19 .1 - ... I y- n,,. .N ,. "w. o '-7000‘1 I..- Prejudice. 11299: Low Gay Nongay High Gay Nongay 181 o ,_01 e O S —- a o ’00. 1° ABEL! 7.75ab (16) 6.54a (13) 7.47ad (19) 6.82a (11) 1e Iis . w -1 -t n tf-n . ' e'-d' ev - 1- e '-5 H229 33221 3:22:11 7.45ab (22) 8.50b(12) 5.50a (6) 7.35a (26) 5.60a (5) 6.25ac (12) 5.15b (13) 6.91abc (11) Mote. Means with differing subscripts, in rows and columns, differ at.p < .05 in the Tukey HSD comparison. indicated in parentheses. Cell sizes are 182 W The same emotion scores computed in Experiment I were computed from the data from Experiment II. Specifically, a "negative feelings toward self" score was computed and contained the following items: angry' at. myself, guilty, embarrassed, annoyed at myself, regretful, disappointed with myself, disgusted with. myself, shame, and self-critical (Cronbach's alpha - .90). The "discomfort" emotion score was based on responses to the items: frustrated, tense, distressed, anxious, bothered, uneasy, and uncomfortable (Cronbach's alpha = .86). The ”positive" emotion score was based upon indications of: friendliness, happiness, energy, optimistic feelings, and feeling content, and good (Cronbach's alpha = .89). The "negative towards others" included scores on the following items: angry at others, irritated with others, and disgusted with others (Cronbach's alpha - .94). Finally, the "threatened" factor accounted for responses to the items of threatened and fearful (Cronbach's alpha - .75). In general, in line with the results of previous research (i.e. Devine et al., 1991), it would be predicted that in the neutral mood conditions, low-prejudice participants would report more negative feelings directed at the self than high- prejudice participants and high-prejudice participants would report feelings of discomfort. It is interesting to surmise how the integral mood induction procedures might undermine the affective correlates documented in previous research. If 183 integral affect mirrors incidental, thus leading to more lazy and heuristic cognitive strategies, these affective responses might not be seen in the present experiment in the two mood conditions. One reason. being that the relatively' more thoughtless state induced by the mood might attenuate these negative feelings“ By ‘the same token, these affective correlates might.be amplified if integral affect leads to more systematic as opposed to cursory cognitive strategies. Participants, due to their enhanced tendency to think about their previous responses, might be more prone to experiencing these affective consequences. The analysis of the emotion scores first rendered several significant main effects across the emotion scores. Regarding the amount of negative feelings directed towards the self, unlike in the first experiment, a significant main effect for target in the scenario was found, [(1, 165) = 4.33, p < .04. All participants reported more self-directed negative affect when responding to scenarios containing nongay targets (Gay: M = 1.75; Nongay: M:= 2.07). It would appear that responding to members of socially stigmatized groups does not necessarily lead to the reporting by low-prejudice participants of negative feelings towards the self. Significant main effects due to the persona presented by the confederate in the experimental sessions were significant for' the emotion. scores of "discomfort" and "positivity" (Discomfort: {(1, 165) = 6.67, p < .01; Positivity: 3(1, 184 165) = 4.12, p < .04). Both of these main effects indicated that more positive reactions were indicated in response to a confederate projecting an ostensibly gay image. Participants reported more discomfort when the confederate projected a nongay image (Gay: M = 2.07; Nongay: M = 2.53). By the same token, participants also reported more positive affect when the confederate projected a gay image (Gay: M = 4.79; Nongay: M = 4.36). A significant main effect, not seen in the first experiment, due to the prejudice level of the subject emerged for the "discomfort" emotion score, £(1, 165) = 3.91, p < .05. This main effect mirrored results seen in the research of Devine and colleagues (1991), presumably with neutral mood participants. In general, high-prejudice participants indicated that they felt more discomfort than did the low- prejudice participants (Low-Prejudice: M = 2.10; High- Prejudice: M = 2.43). It is worth noting that the incidental mood of sadness in the first experiment was associated with more self-reports of discomfort, compared with the neutral mood. Perhaps mood states that provoke thoughtfulness (i.e., sadness) or those that are integral in nature are prone to provoke discomfort, especially on the part of highly prejudiced people. In summary, the significant main effects emerging from the analysis of the emotion scores appear to indicate that in general, high-prejudice participants report that they 185 experienced more discomfort in the experimental sessions. All participants reported more positive feelings after exposure to both questionnaire targets portrayed as gay and experimental confederates portraying an ostensibly gay image. The analyses of the emotion scores regarding positivity and negative feelings towards others revealed a pattern seen in the results of Experiment I and earlier analyses reported for this second experiment. Specifically, a significant two- way interaction between prejudice level of respondent and targets in the scenarios emerged for participants’ positivity scores and their scores representative of negative feelings towards others; (Positivity: £(1, 165) = 11.28, p < .001; Negative Feelings Towards Others: {(1, 165) = 5.85, p < .02). Post hoc comparisons were conducted between the means of both these interactions. The means from these interactions and the significant differences between them are presented in Tables 20-21. First, in looking at Table 20, in terms of positivity, two significant differences among the means emerged. Inw- prejudice participants reported significantly more positive feelings in response to scenarios containing gay targets than to scenarios containing nongay targets (Gay: M = 4.97; Nongay: M = 3.94). Moreover, in response to scenarios containing nongay targets, highly prejudiced participants indicated significantly more positive feelings than the low- prejudice participants (Low-Prejudice: M = 3.94; High- Prejudice: M = 4.75). 186 Table 20 .7. 1.-ug - -. 't ' . 1 1 t:-: - ° - -' - ev-_ . Bespgndggt 99d Tagget in the Scenarios e e ve 11132:. M 11195 Gay 4.97ac (50) 4.55a (58) Nongay 3.94b (29) 4.75a (27) Mg_g. Means with differing subscripts, in rows and columns, differ at p,< .05 in the Tukey HSD comparison. Cell sizes are indicated in parentheses. 187 Table 21 .-.; 11o 1 - 1--. 'v- -- ',. -o 1 ._ o ,- — ._ : 1.. '., . - -' .' - ;v- . ~-;.. .7,t 1,. . .- 1 ,- — - “.5 E . 1° I 1 Target. L519 11138 Gay 1.593 (50) 2.30b (58) Nongay 2.34c (31) 1.95bc (27) Mote. Means with differing subscripts, in rows and columns, differ at p,< .05 in the Tukey HSD comparison. Cell sizes are indicated in parentheses. re Vl‘ [33' be 0t 188 This interaction suggests, as did a substantial number of other analyses, that the low-prejudice participants indicate that they respond more positively to scenarios containing gay targets than nongay targets. In :response to scenarios containing nongay targets, many previous analyses showed that the low-prejudice participants responded to these neutral targets more negatively than to the gay targets. The results obtained from the high-prejudice participants with this analysis of positivity scores are surprising. It appears that they are reporting similar reactions to both types of targets, instead of the predicted more negative reactions to the gay as opposed to nongay targets. Secondly, it is interesting that this analysis replicates a consistent pattern found in the present experiments namely that the low- prejudice participants do not react to the gay and neutral targets the same, but are actually reporting more positive affective reactions to the gay targets. Regarding the negative feelings that participants felt towards others, two significant differences among the means also emerged as presented in Table 21. These differences were not seen in Experiment I. Low—prejudice jparticipants reported significantly more negative feelings towards others when responding to scenarios containing nongay targets (Gay: M =-- 1.59; Nongay: M = 2.34). In addition, high-prejudice participants felt significantly more negative feelings towards others as compared to the low-prejudice participants when 189 responding to scenarios containing gay targets (Low-Prejudice: M B 1.59; High-Prejudice: M = 2.30). Similar to earlier analyses, this interaction suggests that low-prejudice participants indicate more positive reactions towards gay as opposed to nongay targets. Regarding negativity directed towards others, like earlier analyses, it appears that the high-prejudice participants' reactions are roughly equivalent across both types of targets. It appears that fluctuations in responses, as a function of the type of target portrayed in the scenarios, are primarily reflected in the scores of the low-prejudice respondents. A significant interaction between induced mood and persona of the confederate also emerged from the analysis of the positivity emotion scores, £(1, 165) = 3.97, p < .02. Post hoc comparisons conducted. among' the ‘means of this interaction revealed only one significant difference among the means. When in the "angry" mood condition, participants report more positivity in response to a confederate who projects a ostensibly gay image (Angry: Gay: M -= 5.07, Nongay: M = 4.08; Happy: Gay: M = 4.53, Nongay: M = 4.74; Neutral: Gay: M = 4.73, Nongay: M = 4.11). Given that the significant main effects indicated that all participants reported more positive reactions to confederates portraying an ostensibly gay image, this interaction suggests that this positivity towards the "gay" confederate was reported especially when participants were in the "angry" mood C: it 61 CO Sc; thr rev “he Part 190 condition. The analysis of positivity scores also revealed a significant interaction between confederate, target in the scenarios, and induced.mood, [(1, 165) = 3.18, p < .05. IPost hoc comparisons were computed. among the ‘means for' this interaction and the results of these tests are contained in Table 22. Two significant differences among the means emerged. In the "happy" mood condition, when the confederate presented a neutral or nongay image, participants reported more positive feelings when responding to scenarios containing gay as opposed to nongay targets (Gay: M = 4.96; Nongay: M - 3.67). In addition, in the neutral mood condition, and.when responding to scenarios that contain gay targets, participants reported that they felt significantly more positive when the confederate presents an ostensibly "gay" image as opposed to a nongay image (Gay Confederate: M = 5.07; Nongay Confederate: M = 3.65). These effects of mood are hard to interpret, but perhaps suggest, as did earlier analyses, that all participants report more positive reactions to the confederate who projects an ostensibly gay image. Finally, a significant interaction between targets in the scenarios and confederate» emerged for feelings of being threatened, £(1, 165) = 4.37, p < .04. Post hoc comparisons revealed only one significant difference among the means. When the targets in the scenarios were presented as gay, participants reported that they felt significantly more 191 Table 22 . 1 no 1 . . s a , o1 . WW QQDEQQEEQLE M229 1129518892! 39.921: Gay Gay 5.26ab (21) 4.59ab (24) Nongay 4.76a (13) 4.28a (6) Nongay Gay 4.12a (14) 4.96a (24) Nongay 4.02ab (10) 3.67b (5) ut a 5.07bc (17) 4.29a (13) 3.65a (8) 4.52ab (9) Mote. Means with differing subscripts, in rows and columns, differ at,p < .05 in the Tukey HSD comparison. indicated in parentheses. Cell sizes are 192 threatened if the confederate projected a nongay as opposed to an ostensibly gay image (Gay Targets: Gay Confederate: M = 1.35; Nongay Confederate: M = 1.79; Nongay Targets: Gay Confederate: M = 1.55; Nongay Confederate: M = 1.50). In other words, when responding to scenarios containing gay targets, participants indicated that they felt less threatened if the confederate they had seen involved in an interaction projected an ostensibly gay image as opposed to a neutral image. In the first experiment, high-prejudice participants and those participants in the sad mood condition indicated the most feelings of threat. In this second experiment, feelings of being threatened do not appear to be related to prejudice level or mood. It is possible that this result, in Experiment II, could be due to the fact that the previous exposure to a person who could be gay, took the edge off, so to speak, of later having to respond to situations containing gay men. Therefore indicating responses to the scenarios was less threatening to participants who had just had an experience of interacting with gay people. By the same token, this result could be due to demand characteristics. The timing of the fact that some of the participants had just experienced an interaction with a possible gay male, and then were subsequently asked to respond to situations containing gay 'targets, could have clued participants into what was expected of them. Once participants suspected the timing between the m 193 nature of the interaction and the topic of the questionnaire, the questions perhaps were less threatening. The interaction staged with a neutral confederate would have not perhaps raised the same demand characteristics. Participants, in these conditions, even if they suspected that the interaction was staged or related to the questionnaire, would still perhaps not have a clear a hunch to operate on as to the purpose of the study as in the conditions where the confederate projected a gay image and then they were subsequently given questionnaire involving gay people. It is important to point out that this analysis of the emotion scores, similar to that in Experiment I, did not strongly replicate the findings of previous research (e.g., Devine et al., 1991) regarding differences in affect as a function of prejudice level. The low-prejudice respondents are not reporting substantial negative self-affect; nor are the low-prejudice participants reporting more negative self- affect than the high-prejudice participants. However, this experiment does support the finding of earlier research that high-prejudice participants often report more feelings of (discomfort as function of completing the "Should-Would" measure than low-prejudice participants. W A series of Pearson.product-moment correlations was run, as in the first experiment, between participants' perception of the match between their two sets of scores, as well as the 194 actual discrepancy between their two sets of scores, and the five emotion scores. Also of interest was the degree of correspondence between perceived and actual discrepancies. This set of correlations was run for responses to both types of scenarios (i.e., containing a gay target or a nongay target), for both levels of prejudice, and for both types of confederate (gay or neutral). The resulting values of the correlations are presented in Tables 23-25. An inspection of these correlations reveals that only when the confederate projected an ostensibly gay image is a greater perception of a match between the two sets of scores significantly associated with a lesser degree of reported negative feelings felt towards the self. Therefore, only when in the experimental situation, with a hypothetically gay man, did a sense of endorsing behavior consistent with personal standards result in less reported self-directed negative affect. It appears that a greater perception of a discrepancy between the two sets of scores does not necessarily result in more reported negative self-affect by the low-prejudice participants as would be predicted by earlier research (e.g. , Devine et al., 1991). Moreover, the result from the first experiment, namely that greater perceptions of a correspondence between scores were associated with reports of lesser amounts of negative feelings towards the self, was not replicated. H 195 Table 23 :1- 1‘ 10-: '9 ‘01! .0; e Variables Target SMATCH-NEGATIVE FEELINGS TOWARD SELF ---------- SMATCH-DISCOMFORT .......... SMATCH-POSITIVE FEELINGS .37** .36** SMATCH-NEGATIVE FEELINGS TOWARDS OTHER ---------- SMATCH-FEELINGS OF BEING THREATENED ---------- DISCREP-NEGATIVE FEELINGS TOWARD SELF ---------- DISCREP-DISCOMFORT ---------- DISCREP-POSITIVE FEELINGS ---------- DISCREP-NEGATIVE FEELINGS TOWARDS OTHER ---------- DISCREP-FEELINGS OF BEING THREATENED ----- -.24* SMATCH-DISCREP - . 37** -.38** Mote. SMATCH = participants' perception of the match between their "would" scores and "should" scores; DISCREP’= the actual discrepancy between participants' two sets of scores. * p < .05 ** p < .01 196 Table 24 '01. .11 - ‘ o. 01‘ 31‘ WT‘ 3m- 0 9: S' " ’0 'O o 1‘ 9°. C! -'-‘ WT‘ ”Woulo" 1,. . a " 1-u_2" cores ._. :e w-- :10. ‘ ; _-., 1's, e-- _=‘W"1 '- ° .0: Variables ° 8 -‘.1d , - . en .60 es 0 es 0 SMATCH-NEGATIVE FEELINGS TOWARD SELF SMATCH-DI SCOMFORT SMATCH-POSITIVE FEELINGS SMATCH-NEGATIVE FEELINGS TOWARDS OTHER SMATCH-FEELINGS OF BEING THREATENED DISCREP-NEGATIVE FEELINGS TOWARD SELF DI SCREP-DI SCOMFORT DISCREP-POSITIVE FEELINGS DISCREP-NEGATIVE FEELINGS TOWARDS OTHER ----- DISCREP-FEELINGS OF BEING THREATENED SMATCH-DISCREP - . 29** 119112- High -. 46** SMATCH = participants' perception of the match between their "would" scores and "should" scores; DISCREP = the actual (discrepancy between participants' two sets of scores. * p.< .05 1r* p < .01 197 Table 25 ., A, . ..- 21 w- , muo ., . _ ,. o g - ’ o o g o g y , . l , . fl 9 ." 12 o " ,. e" . . . ‘w , mug ., . _ 1. , 0 - . , . e- o 1 s' . . W Variables Confederate Gay Nongay SMATCH-NEGATIVE FEELINGS TOWARD SELF -.24* ----- SMATCH-DISCOMFORT .......... SWATCH-POSITIVE FEELINGS .36** .39** SMATCH-NEGATIVE FEELINGS TOWARDS OTHER ---------- SMATCH-FEELINGS OF BEING THREATENED ---------- DISCREP-NEGATIVE FEELINGS TOWARD SELF ---------- DISCREP-DISCOMFORT ---------- DISCREP-POSITIVE FEELINGS ----- -.25* DISCREP-NEGATIVE FEELINGS TOWARDS OTHER ---------- DISCREP-FEELINGS OF BEING THREATENED ---------- SMATCH-DISCREP - . 34 ** -.45** Mote. SMATCH 8 participants' perception of the match between their "would" scores and "should" scores; DISCREP’= the actual discrepancy between participants' two sets of scores. * p < .05 **p< .Ol 198 In addition, the correlations, from the data of the second experiment, between participants’ perception of the match between their scores and reported feelings of discomfort, do not replicate the findings of Experiment I or those of previous research. The data from Experiment II revealed no significant correlations with discomfort scores. In Experiment I, however, in response to gay targets overall, and for low-prejudice participants, the greater the perception of the match between scores, the lesser amount of reported feelings of discomfort. Furthermore, unlike the results of the research of Devine and her colleagues, in Experiment II, high-prejudice participants' perception of the match between their two sets of scores did not appear to be related to the amount of discomfort experienced. In both experiments, across all conditions, the greater the perception of the match between the two sets of scores, the greater amount of positive affect reported. In other words, generally speaking, a sense of a correspondence between "should" and "would" scores was correlated with reports of feelings of positive affect for all participants. No significant correlations emerged, in either of the experiments, between the perception of the match between the two sets of scores and reported negative feeling directed towards others. Unlike the data from the first experiment, no correlations emerged, in this second data set, between perception of the match between scores and feelings of being 199 threatened. In the first experiment, across both targets and prejudice levels, participants reported less feelings of being threatened the more they perceived their two sets of scores to correspond. The next set of correlations concerned the actual discrepancy between participants' scores. No significant correlations emerged between the actual discrepancy between participants scores and the amount of reported self-directed negative affect, other-directed negative affect, or discomfort experienced. This lack of significant correlations suggests that participants' potential negative affective states may not necessarily be related or contingent upon the actual discrepancy between the two sets of scores. This conclusion is qualified by the correlations that emerged from the first experiment between the emotion scores and the actual discrepancy between sets of scores. Generally, in response to both gay and nongay targets, and for low- prejudice participants, the greater the actual discrepancy between sets of scores, the greater amount of reported negative feelings directed at the self. In addition, the correlations from the first experiment indicate that low- prejudice participants report more feelings of discomfort and less positive feelings as the discrepancy between their scores increases. The first experiment did not reveal any significant correlations between discrepancy and positivity scores. In 200 the second experiment, this correlation was significant in the condition where the confederate projected a neutral (or nongay) persona. In conditions with the neutral confederate, as the actual discrepancy' between 'the scores increased, reported feelings of positivity decreased. Therefore, it appears, that in general, and perhaps unlike the negative affect. ratings. participants' indications. of feelings. of positivity are somewhat contingent upon the actual discrepancy between the two sets of scores. Regarding the relationship between the actual discrepancy between participants' two sets of scores and their reported feelings of being threatened, the only significant correlation to emerge, in the first experiment, indicated that for low- prejudice participants, as the discrepancy increased so did their feelings of being threatened. In the second experiment the only significant correlation between discrepancy scores and feelings of being threatened emerged for participants responding' to scenarios. containing' nongay targets. For participants whose questionnaire concerned reactions to neutral targets, as the actual discrepancy between their two sets of scores increased, their reports of feelings of being threatened decreased. This correlation is counterintuitive and hard.to explain. It1will not be assigned much explanatory power until further replication. Finally, in both experiments, across all conditions, participants' perception of the match between their two sets 201 of scores was negatively correlated with the actual discrepancy between the scores. It appears that participants' subjective sense of "should-would" correspondence is generally sensitive to the magnitude of the actual discrepancy. These correlations provide some support for the findings of Devine et al. (1991). Generally speaking, especially in the first experiment, low-prejudice participants appear to be reporting more affective correlates as a result of the measures as compared to high-prejudice participants. Analyses of Secondary Dependent Measures aeeri be; i on geest i onneize Respondents' answers to the questions contained in this measure were consolidated to render the same two composite measures as in Experiment I. One measure, computed for participants who responded to situations containing gay targets, was the degree to which participants attributed their "would" responses or endorsed hypothetical behaviors to the targets in the scenarios being gay. The, second composite score, computed for all participants, indicated the degree to which they attributed their "would" responses to other factors in the scenarios, specifically non-minority related factors. Higher scores on the target questions indicated a greater attribution of the "would" responses to the group membership of the targets, and thus a higher composite score across these questions indicated the same (Cronbach's alpha - .94). A higher composite score across the other questions indicated a 202 greater amount of nontarget-related attributions (Gay Target: Cronbach's alpha = .85; Nongay Target: Cronbach's alpha -- .64). This attribution measure was included as an exploratory investigation into whether or not participants would engage in the cognitive strategies postulated by Aversive Racism Theory (i.e. , Gaertner e Dovidio, 1986) . Low-prejudice participants might be predicted to be more motivated to attribute their hypothetical behaviors to nontarget-related factors than high- prejudice participants. Correspondingly, it could be predicted that low-prejudice participants, unlike their high- prejudice counterparts, would resist attributing their hypothetical behaviors to the social group membership of the targets about which they were asked. The composite score regarding attributions to the target was analyzed via an analysis of variance with mood, prejudice level, and confederate as the independent variables. Just as in the first experiment, a significant main effect of prejudice level of participant was found, £(1, 106) = 42.99, p < .001. High-prejudice participants endorsed more target- related attributions than low-prejudice participants as explanations for the behaviors they indicated (Low: M = 2.15; High: M = 4.39) . In other words, high-prejudice participants indicated that their hypothetical behaviors or ”would" scores were due to the gay targets' social group membership to a greater degree than the low-prejudice participants. If the 203 low-prejudice participants, in the present study, are analogous to the conceptualization of aversive racists, in the Theory of Aversive Racism, then this result would be expected. Aversive racists, according to this theory (e.g., Gaertner & Dovidio, 1986) would feel motivated to censor their attributions. According to this theory, maintaining an egalitarian self-image to self and others is important to people who harbor aversive racism. Therefore these people, and by extension low-prejudice participants would hesitate to attribute their actions to a person’s social group membership. By the same token, Low-prejudice participants could be truly less biased towards gay targets than high-prejudice participants. A significant main effect of the persona the confederate projected in the mood induction procedure also emerged from this analysis, {(1, 106) = 5.33, p < .023. Participants made more target-related attributions after exposure to a nongay confederate than to a gay confederate (Gay: M = 3.05; Nongay: M = 3.74) . This finding is somewhat counterintuitive and hard to explain unless one again falls back on demand characteristics. Participants perhaps formulated some explanation of the purpose of the study due to the timing of the interaction with the ostensibly gay man and the questionnaire concerning reactions to gays. The research participants might ascertain that the study concerns reactions to gay men. Participants might then be especially sensitized [it De at to 204 to put forth a nonprejudiced image and would, as a result, avoid attributing their responses to the social group membership of the targets. Like in the first experiment, mood state did not appear to systematically affect the degree to which participant attributed their hypothetical behaviors to the targets social membership. The other analysis of responses on the attribution questionnaire involved the number of attributions made to non- minority related elements in the scenarios in explanation for hypothetical behavior. A higher composite score across these questions indicated a greater degree of attributions of "would" scores to nontarget-related factors in the scenarios (Gay Target: Cronbach's Alpha = .85; Nongay Target: Cronbach's Alpha = .64). This composite score was analyzed via an analysis of variance with mood, prejudice level, targets, and confederate, as the independent variables. A significant main effect of the targets in the scenario emerged from this analysis, {(1, 164) = 6.11, p < .015. Participants made more nontarget-related attributions in response to a nongay confederate than an ostensibly gay confederate (Gay: [1: 4.05; Nongay: M: 4.53). Therefore, in general, it appears high-prejudice participants are willing to endorse a greater number of attributions for their endorsed hypothetical behaviors than low-prejudice participants. It seems, in addition, contrary to what would be expected, that the nongay confederates and 205 nongay targets engender more attributions than their gay counterparts. Perhaps these attributions are made to counteract the ambiguity of the experimental session. When the mood induction procedure involves a neutral confederate, there may not exist the demand characteristics that clue participants in to a possible purpose of the study. By the same token, the questions measuring response to neutral targets may have been ambiguous to participants as the purpose of the measures might not be clear. DISCUSSION OF EXPERIMENT II This experiment intended to explore how various integral emotional states would impact upon respondents' indications of their behaviors and standards in response to targets representative of socially stigmatized groups. Moreover, these integral mood states were also examined as they related to any affective correlates experienced by the research participants as a result of completing the experimental measures. e ect oo 3 The results overall did not demonstrate the interactions between prejudice level and mood that were tentatively predicted. Although as predicted, the prejudice level of the research participant played a major role in determining responses, sexual orientation of the targets in the scenarios interacted with prejudice to influence responses, more so than mood per se. Generally speaking, participants in the mood conditions did not rate themselves as feeling significantly different from participants in the neutral mood condition. This lack of evidence in support of the effectiveness of the mood induction procedure was surprising and unpredicted, given the pretesing that attested to its effectiveness. It is possible that mood induction procedures aimed at 206 207 inducing integral affect will have to be stronger and produce more impact than those aimed at eliciting incidental affect in order to assure their effectiveness. Emotions that arise due to interactions with others may be difficult to induce in order to overcome general social norms that call for social interactions between people to remain on a fairly even keel. Indeed we may all receive social training so to speak, to interact, especially with strangers in a fairly polite, neutral manner. One of the first experiments testing the association between attitudes and actions found that contrary to any prejudice expressed in their attitudes, in an on-line social situation involving members from different social groups, norms of politeness and courtesy prevailed (cf. LaPiere, 1934) . In other words, in this experiment, although the restaurant and hotel owners responded by mail that their establishments discriminated against Chinese people, when actually interacting with Chinese people, representative employees from these establishments did not act in a biased manner towards the Chinese. By the same token, manipulation checks to assess the effectiveness of integral mood induction procedures might also have to be more sensitive to changes in mood. It might be advantageous to employ a greater variety of self-ratings as well as other more specific assessment measures to detect the effects of an integral mood. One type of integral mood induction that might be more effective would be one that utilized fewer manipulations to 208 establish the social group membership of the mood eliciting person(s). In the present experiment, the confederate had to convey both the possibility that he could be gay as well as induce a particular emotion in the participants through his interaction with the other confederate. Future studies might want to use confederates to induce moods whose social group status is more explicit and does not necessitate complicated manipulations. For example, confederates representative of certain ethnic groups might be used to elicit moods. Specifically, African-American.or Hispanic confederates could be used to induce the moods. Even though the confederate in the present experiment did not elicit the moods to the degree that would be desired, his presence did produce a variety of reactions in the subjects as described in earlier sections of this paper. Regardless, it will take further research to establish and substantiate effective integral mood manipulations. Regardless of whether the mood manipulation checks were sensitive enough to detect mood differences among the research participants, another reason was proposed, in this paper, as to why responses to the "should-would" task did not reflect the effects of the mood induction procedures. It was proposed in the discussion.of the results of the first experiment that perhaps some feature of the "should-would" task served to tindermine the incidental affective states. Hence, there was a lack of systematic effects on the dependent variables due to 209 mood. However, in this second experiment, such an undermining of the affective state by the questionnaire would perhaps intuitively not be expected. Integral moods might be expected to be intense enough to withstand any inherent problems with the questionnaire. These moods would perhaps be fairly intense for the average person because they arise not only from interaction with other people but from people who are members of different social groups as well. It seems intuitive to presume that, on the average, emotions stemming from another person would be more intense than emotions arising from other incidental aspects of the environment, such as the weather. The social group status of the person eliciting the emotion perhaps only serves. to heighten the emotion experienced. Finally, integral affect could be presumed to be rather intense due to its inherent online ramifications for the social judgment maker. The social judgment maker is in an emotionally charged situation with a target person from another social group. The resolution of this situation will afford to the person the immediacy of several outcomes. Therefore, integral affect presents a difficult task to the social perceiver as it mandates the handling of both an emotion and a potentially problematic interpersonal situation that is ongoing and could have an immediate impact upon its participants. It is also surprising that the ASW scores of the high- prejudice participants did not demonstrate systematic effects 210 due to mood state. It would perhaps be predicted that high- prejudice participants would definitely be affected by the mood induction procedure due to the very nature of its elicitation. Indeed, it would seem that contact with an ostensible member of an outgroup towards which one holds negative attitudes would produce affective reactions. The responses of the low-prejudice participants, on the other hand, due to their well-internalized personal standards (e.g. , Devine et. al, 1991) may be somewhat stable and not prone to fluctuate due to an integral mood state. It is, therefore, an issue for future research as to how integral affective states might influence responses on the "Should-Would" measure. 3 J E. 3' s It appears that just as in the first experiment, personal standards governing appropriate intergroup behavior are more negative for men and those participants possessing a high degree of prejudice. This negativity in the personal standards of high-prejudice people replicates previous research that documented that the personal standards of high- prejudice people allow for a greater expression of prejudice than those of low-prejudice people (Devine et al., 1991). Also, it is a general finding in the literature that men are generally more negative than are women towards gay men (Qualls et al., 1992). The hypothetical behaviors that participants endorsed followed a similar pattern to their standards, again mirroring 211 the results of the first experiment. Low-prejudice respondents indicated more positive behaviors in response to the gay than the nongay targets, and more positive behaviors, in response to the gay targets, than the high-prejudice participants. Overall, the "would" scores of the high- prejudice participants were also more negative than those of the low-prejudice participants. Moreover, for all participants, more positive indications of hypothetical behavior were indicated in response to the gay as opposed to the nongay targets. The results of the analysis of high-prejudice participants' "would" scores, would be predicted on the basis of a previous research (i.e., Devine et al., 1991). As was stated earlier, high-prejudice participants would be expected to react more negatively to members of socially stigmatized groups than to neutral people. Moreover, they would be expected to react more negatively to these outgroup members than low-prejudice people. The results from the analysis of the "would" scores of low-prejudice participants were not exactly as expected. Interestingly, low-prejudice respondents, similar to the first experiment, appear to exhibit a positivity bias towards the gay targets, in that their standards and behavior are more positive in response to the gay targets than to the nongay targets. This positivity bias could be due to the overmotivation of low-prejudice participants, to act in a 212 nonprejudiced manner in response to members of socially stigmatized groups. As a result, low-prejudice people actually demonstrate a positivity bias towards these outgroup members. It is also a possibility that low-prejudice people have truly positive attitudes towards members of various socially stigmatized groups and by the same token overcome any personal prejudices. However, it also bears mentioning that perhaps low-prejudice people have not necessarily overcome any effects of prejudice but simply respond positively to people who are different. Just as with the first experiment, the analyses involving gender somewhat qualify the previous discussion of effects due to prejudice. Women generally appear to respond more positively to the gay targets than do men. The sexes largely did not differ in their responses to the nongay targets. These gender effects replicate the findings of the first experiment as well as the previous experiments utilizing the "Should-Would" measure (Devine et al. , 1991) , whereby men react more negatively to gay men than do women. Even men who score low in homophobia, in this second experiment, exhibit less of a correspondence between their "should" and "would" scores than their female low-prejudice counterparts. In addition, the results of this second experiment indicate that when responding to gay targets, women reported a significantly lesser amount of feelings of being threatened than men responding to the same targets. The results of these two 213 studies suggest, as do a number of others, that gender appears to be a moderator of participants' responses to gay men. The mood induction procedure also did not appear to have a significant effect upon discrepancies in participants’ responses. In both experiments, the type of target portrayed in the scenarios appears to play a big role in the discrepancies expressed between personal standards and hypothetical behaviors. There was a greater expressed discrepancy between standards and behaviors for all participants when the targets in the scenarios were portrayed as neutral as opposed to gay. It is possible that all of the respondents were monitoring their answers, to a greater degree, when responding to targets who were members of socially stigmatized groups as opposed to those who were portrayed as neutral. This pattern of response could be due to the prevalent notion in today’s society that prejudice is wrong and not to be condoned (Crosby, Bromely, & Saxe, 1980). Unlike in Experiment I, high-prejudice participants did not evidence a greater discrepancy between their sets of scores than did low-prejudice participants. However, low- prejudice participants demonstrated a larger discrepancy between their indicated standards and behaviors in response to the neutral as opposed to the nongay targets. Indeed it appears that low-prejudice people may work hard to maintain a correspondence between their standards and behavior, in response to members of socially stigmatized groups. 214 Maintaining such a correspondence may not be as important to high-prejudice people. Hence the discrepancies between the scores of the high-prejudice participants may be roughly the same for all targets. These results would be easily predicted based upon previous research since studies have shown that low-prejudice people have standards regarding egalitarian behavior that are more internalized than they are for high-prejudice people (e.g., Devine et al., 1991). As a result, low-prejudice participants would be motivated to maintain a correspondence between their personal standards and behaviors, especially, in response to outgroup members. Perceptions of the correspondence between scores also varied among the participants depending upon their prejudice level and the social group membership of the targets to which they were reacting. In the first experiment, low-prejudice people reported feeling more confident that their standards and behavior corresponded than high-prejudice people. This pattern was not seen in the data from the second experiment. Instead, low-prejudice participants felt more confident of the correspondence between their two sets of scores when the targets to which they are responding were gay as opposed to neutral. In fact, across all participants in this second experiment, a greater perception of the match between the two sets of scores was reported in response to gay as opposed to nongay targets. The "politically correct" mentality present 215 in today's society might lead to at least a perception, on the part of both high and low-prejudice participants, of exerting effort to match behavior to standards when interacting with members of socially stigmatized groups. Finally, participants' perception of the match between their should and would scores increased as the actual discrepancy between the two sets of scores decreased. Therefore, it appears as in Experiment I, that participants may be somewhat accurate in their perception of the correspondence of their two sets of scores. Analyses of participants' affective states subsequent to indicating "should" and "would" scores revealed, contrary to predictions, few systematic effects cf mood state. Furthermore, scant evidence emerged in support of the results documented by previous researchers regarding differential affective responses contingent upon personal prejudice level (Devine et al., 1991). For example, it was not the low- prejudice participants who felt the most negative self— directed affect as a result of perceived discrepancies per se, as in previous research. Instead, participants exposed to a mood induction procedure involving the ostensibly gay man indicated more feelings of compunction than participants exposed to the neutral confederate as their perception of the match between their sets of scores decreased. Overall, as seen in the first experiment, the more participants felt that their two sets of scores corresponded, 216 the more positive affect they reported. In addition high- prejudice participants report less feelings of threat as their perception of the match between their scores increased. Regarding correlations between emotion scores and the actual discrepancy between participants' sets of scores, fewer significant effects emerged than from the analyses of Experiment I data. In response to mood induction procedures involving nongay confederates, as the actual discrepancy between participants' scores increased, the amount of positive feelings reported decreased. Only one other significant correlation between emotion scores and the actual discrepancy between participants’ "should" and "would" scores emerged from the data of the second experiment. This correlation indicated that in response to scenarios containing nongay targets, as the actual discrepancy between participants' scores increased, the amount of reported feelings of being threatened decreased. These last two correlations are hard to interpret and should be treated cautiously pending replication. Finally, In the second experiment, unlike the first, low- prejudice participants did not report more feelings of discomfort, threat, and negativity towards others, as the actual discrepancy between their two sets of scores increased. Therefore, the second experiment generally did not replicate the findings of Devine and her colleagues (1991) . Perceptions of discrepancies between standards and behaviors were not associated with more negative affective consequences, 217 especially on the part of low-prejudice people. The attributions that participants offered in explanation for their hypothetical behaviors followed a consistent pattern depending on the prejudice level of the participant. High- prejudice people attributed their would scores more to the social group membership of the targets in the scenarios than did low-prejudice participants. Participants in this second study also made more target-related and nontarget-related attributions when the confederate in the mood induction procedure projected a neutral as opposed to ostensibly gay image. These results can be explained perhaps, as in the first experiment, according to aversive racism theory (i.e., Gaertner & Dovidio, 1986) . Aversive racists and perhaps low- prejudice people are motivated to avoid attributing their responses to social group membership or any factor relevant to prejudice so as to avoid projecting an image to self and others of being prejudiced. The maintenance of a nonprejudiced self-image would perhaps be less important to high-prejudice people. Therefore, they would not hesitate to attribute their behavior to the issue of social group membership. Finally, similar to Experiment I, there were no effects of mood state upon attributions. GENERAL DISCUSSION 5 'ce Recent research postulates that prejudice still exists, although somewhat devious in its manifestation (Crosby et al. , 1980), and the present research certainly bears out this prediction. Clear predictions regarding the interaction of prejudice effects with those of other variables can be made but might be hard to evaluate in the laboratory setting. One's level of prejudice might not be as neat a coat hanger upon which to base predictions as it once was when more old- fashioned forms of racism and prejudice were prevalent. Indeed, modern racism might often lead to conflicting and ambiguous responses. Future research will want to address the direct effects of prejudice more carefully as it fluctuates due to variables like gender, current mood state, and target of response. This research indicated that high-prejudice individuals tend to respond in a generally negative fashion towards gay men despite any transitory mood state. On the other hand, low-prejudice people appear to react more positively to members of socially stigmatized groups than to neutral targets. This positivity might not reflect actual feelings toward outgroup members, but the tendency of low-prejudice 218 219 people to want to strenuously avoid the image both to self and others of being prejudiced (Gaertner & Dovidio, 1986). Low- prejudice people might simply be working too hard to avoid a prejudiced image and therefore end up exhibiting a positivity bias towards members of other social groups. By the same token, the positivity of the low-prejudice participants might reflect true positive attitudes. Low-prejudice people might feel and express more positive attitudes when responding to members of various social groups as opposed to people who are not members of a distinctive social group. It is for future research to distinguish between the image maintenance, social desirability, and true attitudes of low-prejudice people. Future research might also want to address the question of why low-prejudice people might actually feel more positive towards people from socially stigmatized groups. In addition, future research might want to address the issue of what is really driving the responses of people low and high in prejudice respectively. Does being high in prejudice translate into a general negativity towards people who are members of socially stigmatized groups? By the same token, does being low-prejudice translate into a general positivity or unidirectional positivity towards members of other distinct social groups? Perhaps people who score high in prejudice are simply generally more negative in their responses as compared to low-prejudice people. Therefore a general dispositional factor could be playing a role in high- 220 prejudice participants' responses as opposed to outright prejudice against the group in question. Another issue that this research highlights is the question of why women generally respond more positively to gay men than do men themselves. Future research might want to attempt to tease apart the reasons behind gender differences in homophobia. It is possible that these differences are due to a methodological flaw in most homophobia scales. In general, many of these scales appear to assess attitudes towards gay men or gays in general rather than assessing attitudes towards both gay men and lesbians. It could be important to assess reactions to both gay men and lesbians as a function of other variables, such as gender, in order to fully understand the typology and scope of homophobia (Bouton et al., 1987). It might also be worthwhile for additional research to explore other gender differences in prejudice. In other words, how do men and women differ in their responses to members of different socially stigmatized groups. It may be that men tend to more prejudiced towards different groups than women. Although Devine et al. , (1991) did not find systematic gender differences in prejudice towards African—American, attitudes towards the obese, for example, might differ as a function of gender of respondent. Finally, future research might want to address the finding, in both studies that participants perceive a greater 221 correspondence between their "should" and ”would" scores, when responding to scenarios containing gay as opposed to nongay targets. Research might want to address whether this perceived correspondence is due to a closer monitoring, by all people, of their responses to members of socially stigmatized groups. If this monitoring occurs, it would lend support to the Theory of Aversive Racism (e.g. , Gaertner 6 Dovidio, 1986) by demonstrating that most people have traits of aversive racists and thus work to avoid appearing biased. Indeed, perhaps prejudice has gone even further underground than previous research would indicate. WW Regarding the effects of the moods' in the present research, unfortunately no clear pattern of results emerged. In the first experiment, this paucity of mood effects was clearly not expected given the ability of previous research to effectively utilize this mood induction procedure to induce the desired affective states (e.g., Bodenhausen, Sheppard, & Kramer, 1994). Researchers might want to continue to investigate the effects on the "Should-Would" measure of incidental affective states. If, as was speculated earlier some component of the "should-would" task undermines the incidental affect induction procedure, perhaps other mood induction procedure would be more successful at eliciting and documenting robust systematic effects due to this type of emotion. Future experiments might utilize an incidental 222 affect induction that is ongoing throughout the experiment so that the desired emotion is maintained throughout the experimental session. For example, the musical mood induction procedures used in other studies investigating the impact of incidental affect (c.f., Bodenhausen, Kramer, & Suesser, 1994) could be modified for the present paradigm. Specifically, music pretested to establish a certain affective state could be played throughout the duration of completing the "Should- Would" measure. This continuous presence of the mood eliciting stimulus might serve to maintain the desired affective state throughout the social judgment task and thus documents the effects of incidental affect the present study did not. The lack of systematic mood effects in the second experiment is not as surprising as it was an exploratory investigation. Several recommendations might be made for future investigations of integral affect. Perhaps it would help their effectiveness to design integral mood induction procedures to be rather short in duration. Therefore experiments would be streamlined with a smooth transition from mood induction to assessment of the dependent variables. This temporal variable might make the induction procedure more successful and might also reduce potential demand characteristics. It bears mentioning that previous successful attempts to elicit positive affect have been as simple as having subjects find a dime (Isen, 1984). Therefore, it is 223 somewhat surprising that the procedures in the second experiment were apparently so ineffective. Perhaps the interaction was not clear and participants did not fully realize the "favor" the confederate was doing for them by volunteering to be in the longer version of the experiment. By the same token, participants might not have grasped the disservice the confederate was doing to them by refusing to participate in the longer version of the experiment. Regardless, the presence and persona of the confederate did produce some affective consequences for the participants, suggesting that effective procedures of this sort could be designed and that it is possible to induce integral affect in a laboratory setting. Future research will hopefully address the challenges of eliciting integral affect as well as assessing its effects upon the social judgment maker. The persona of the confederate, in the second experiment, did affect participants' responses. In general, they reacted more positively to the interactions involving a gay confederate as opposed to a neutral one. It is interesting to speculate how the reactions of the participants might change if the contact experience that elicited the integral affect were heightened a bit. What might participants' reactions be if the affect induction involved a waiting room scenario, where they were placed in closer physical contact with the mood eliciting confederate? In the waiting room situation, participants might demonstrate the opposite pattern seen in 224 the present research or rather more negative reactions to the members of stigmatized groups than to members from majority groups. People might not mind and even respond favorably to members of stigmatized groups as long as they are simply in the area. However, when people interact with members of stigmatized groups in close contact situations, negative reactions might be more likely to be experienced by the majority group members. Such a pattern of responses just described could constitute the old, "I don’t mind the show, but don't step on my toe effect." In other word, members of socially stigmatized groups are considered to be O.K. as long as they stay in their place and among their own. The question still remains, therefore, as to how integral affect influences social judgments. It might also be important for future research to address the robustness of the affective correlates, as a result of perceived discrepancies in responses, on the measure, found in previous research (e.g., Devine et al., 1991) , as these effects were not strongly replicated by the present research. Moreover, still an interesting empirical issue to investigate and or document is the proposed censorship mechanisms of aversive racists or of low-prejudice people as manifested in the attributions they offer for their behavior. Experiments in order to measure this censorship as reflected in attributions, might have to employ more sensitive attribution measures. Respondents perhaps need more leeway in the 225 attributions they are allowed to make. Hence, the use of an open-ended questionnaire might be recommended. e u st' ‘ e Regarding the previous research conducted by Devine et al. , (1991) , and in light of the present research, the results obtained using the "Should-Would" measure are perhaps qualified by the ordering of the questionnaire. The experiments reported here present fairly clear evidence that the order in which participants are asked to indicate their personal standards as opposed to behaviors could be an important factor in determining their responses. Rendering ”should" responses or personal standards first appears to lead to a lesser discrepancy between respondents' two sets of scores, as well as to a greater perception on the respondents’ part of a correspondence between the scores, and more reported positive affect. There are several plausible reasons why these order effects might have taken place. First there might have existed some demand characteristics in the timing of the questionnaire so that participants became aware that what was being investigated was the discrepancy between their two sets of scores. In response to these possible clues, it is interesting that participants do not give the experimenter perhaps what they perceive was wanted, but instead indicate scores with a small discrepancy between them. It appears that even if such demand characteristics exist, participants might ignore them and hesitate to express discrepancies between 226 their standards and behavior. This reluctance to express discrepancies could be due to a social desirability bias that overrides any influence of demand characteristics. Perhaps respondents are more concerned with projecting an image of consistency than following any demand characteristics. Future research utilizing the "Should-Would" measure might want to question respondents after the experiment as to their perceptions of the true purpose of the experiment. This questioning would allow for participants to be removed from the sample if they knew too much about the experiment as well as documenting any of their hypotheses regarding the purpose of the study. The end goal of this research would be the weeding out of any possible demand characteristics inherent to the questionnaire. The verb order effects might also have been due to the fact that the personal standards, of the participants, were in place to guide subsequent indications of behavior. Given the differences found in the personal standards of high and low- prejudice participants, the pattern of results found would be predictable. Low-prejudice participants would respond more positively, in line with their personal standards and the high-prejudice participants would respond negatively, following their standards, to members of socially stigmatized groups (cf. Devine et al., 1991). In order to test verb order effects, future research might want to ask participants to just answer just ”should" or 227 just ”would" questions. It would be interesting to see how "should" and "would" scores would change as a function of prejudice level of respondent, and targets in the scenarios. It might be found, for example that the "should" scores of low-prejudice participants, in response to gay targets, are more negative when given first or alone as opposed to after "would" scores. Finally, there could be some inherent problem with asking participants to indicate their personal standards first. Perhaps asking low-prejudice people this question kicks in their postulated censorship mechanisms (of. Gaertner & Dovidio, 1986) . Therefore the responses of the low-prejudice participants would be predicted to be egalitarian, in line with their beliefs of what is right and wrong. The effects of transitory mood states might also be overridden by the enacting of these censors regarding prejudice, hence the paucity of mood effects upon the main dependent variables. The responses of high-prejudice participants might not be as affected by verb order as those of the low-prejudice participants. These people might just not care to censor their responses despite any demand characteristics as to the purpose of the study. They will continue to respond negatively in line with their personal prejudice. The affect inductions might not have affected the scores of the high- prejudice participants, simply because their expressions of their beliefs about other social groups are stable and immune 228 to fluctuation. The Should—Would questionnaire needs to be thoroughly tested to document the inherent order or demand effects that might be a byproduct of the measure. These potential problems with the measure would pose serious threats to the interpretation of previous research as well as limiting future research with this measure. c s'o d ' s This research points to the ongoing issue that prejudice constitutes for society. While evidence abounds that it exists, there does not always appear to be a clear manifestation of prejudice (Crosby et al., 1980) . Indeed, the results from the present studies suggest that high-prejudice individuals do not without fail respond more negatively to members of stigmatized social outgroups than to their own ingroup members. Moreover, due to the increasing cultural diversity of modern life and the social group interactions it necessitates, delineating the characteristics of prejudice is particularly worth empirical consideration. In order to maintain a harmonious and democratic society, it is important to understand factors that influence social group relations. Specifically, it is worth investigating precursors to both egalitarian and prejudiced behavior. As our understanding of the workings of prejudice increases, society is provided a powerful tool in the eventual demise of prejudice, namely the knowledge of how it works and is perpetuated. APPEN DI CES APPENDIX A Appendix A !’ 9‘ ‘ ° ‘.14: 1 1 -°.‘ ° '1 0. 2911.9‘.1' ~ ‘ (Bouton, Gallaher, Garlinghouse, Leal, Rosenstein, 8 Young, 1987) All responses scored on a f ive-point Likert Scale ranging from "1", strongly disagree to "5", strongly agree. 1. Homosexuals contribute positively to society. 2. Homosexuality is disgusting. 3. Homosexuals are just as moral as heterosexuals. 4. Homosexuals should have equal civil rights. 5. Homosexuals corrupt young people. 6. Homosexuality is a sin. 7. Homosexuality should be against the law. 229 APPENDIX D Appendix B EXR2IiE2nL§£_EIQ£QQQl_EXR£Iim§n£_I INSTRUCTIONS FOR THE "FIRST" EXPERIMENTER Once all of the participants have shown up at the laboratory, please greet them with the following: What we.have for you to‘participate in today is two»short experiments. Because the first experiment is so short, it is being piggy-backed with another experiment. I am conducting the first experiment for Lori Sheppard of the Psychology Department. The first experiment involves having you recall and write about emotional events from your life. Researchers are interested in the emotional experiences of today’s college students and wish to tabulate some of these experiences and how they affect people. The second experiment is being given by the Michigan Department of Urban Affairs in collaboration with various sociologists and psychologists. A second experimenter will be conducting this experiment. It involves having you give your answers to some questions about population diversity and congestion in modern urban environments. Then pass out the consent form and say the following: If you agree to participate today, please sign and date this standard department consent form. 230 231 Then pass out the LEI sheets and say the following: Please read the instructions on the top of your sheet along with the tape recorder. The instructions for the experimental groups, on the tape, will be as follows, substituting the appropriate emotion: anger, sadness, happiness. We are interested in scientifically studying the relationship between memory and emotion. Today we are interested in . To study the relationship between memory and , we would like you to recall a single or event in your life that caused you to feel very at the time it occurred. USe the rest of this sheet to write down the event as you now remember it. In particular, please describe how the event came about-as concretely and vividly as you can. In fact, before you begin writing, take a few minutes to try to reexperience this event as vividly as possible. Then take about ten to fifteen minutes to write your description. Please remember that all of your responses will be kept confidential. When we signal that this first experiment is finished, please place your materials on the table in front of you. Please feel free to be honest and express your emotions. You may now begin. 232 In the control condition the instructions, on the tape, will be as follows: We are interested in scientifically studying the nature of people’s memories about the small, everyday events of their lives. To study this issue, we would like you to recall the events that happened to you yesterday and write them down on the page below. There is no need to go into great detail. Simply describe the events and experiences you had yesterday, based on your current memories. Pdease remember that all of your responses will be kept confidential. When we signal that this first experiment is finished, please place your materials on the table in front of you. Please feel free to be completely honest. You may now begin. Time the participants for 12 minutes, giving them a two minute warning. Once the writing period has passed, ask participants to turn the sheets into you, and collect their sheets. Then say: I want to thank-you for your participation today. I will now turn you over to the second experimenter. THE FOLLOWING INSTRUCTIONS ARE FOR THE "SECOND" EXPERIMENT!!! 1. Introduce yourself and say the following: Hello, I am and I am conducting this experiment for Dr. Galen Bodenhausen of the Psychology Department, in conjunction with various other colleagues. This second experiment is being given by the Michigan 233 Department of Urban Affairs in collaboration with various sociologists and psychologists. Today we will ask you to indicate your responses to questions about population diversity and congestion in modern society. Then pass out the experimental booklets. Ask participants to first sign and date the consent form if they agree to be in the experiment. Then ask participants to fill out the "Participant Profile" on the first page of their experimental booklet. Then, thirdly, ask participants to read the second page along with the tape recorder. The instructions, on the tape will read as follows: A group of researchers is cooperating to investigate two recent trends in American society. These trends are the increasing congestion and the increasing cultural diversity of modern society. These trends impact on daily life in a variety of ways, affecting everyday living conditions and interpersonal situations. The researchers involved in the project have generated a variety of such typical, everyday situations. We intend to sample responses to these scenarios from students taking a wide variety of introductory classes, such as psychology, sociology, anthropology, and communications. Respondents are being presented these situations and asked for their reactions. 234 Some respondents are being asked to consider situations involving a variety of different groups. These groups differ across questionnaires and are randomly selected. Examples of these groups include: the obese, ethnic minorities, religious groups, gays, senior citizens etc. You may or may not be asked about one of these groups. You will be asked to give your .responses to the situations from several standpoints. PM WWW It is inwpor-tant to keep in mind that there are no right or wrong answers. Our' primary interest is in the ‘wide diversity of responses, not the singular responses given by you. Therefore feel free WW. When you finish, you may leave your materials on the table in front of you. You may then go up to the experimenter, who will stamp your credit card. When participants are finished, ask them to leave all their materials the table. Stamp their credit card and give them the debriefing sheet. Ask them to read the debriefing before they leave and to return the sheet to you with their address on it, if they want to be notified of the results. APPENDIX E Appendix C onse t o ' t 'm nt 0 r' t 1. MICHIGAN STATE UNIVERSITY Department of Psychology Research Consent Form I freely consent to participate in the scientific research project being conducted by Lori A. Sheppard entitled "Moods and Memories". The basic nature of the study has been explained to me. I understand that I will be asked to recall and write about potentially emotional events from my life. I further understand that my responses will be made anonymously and that no attempt will ever be made to identify my responses nor to refer to my name in any report of the research findings. I understand that I am free to discontinue my participation in this research at any time without penalty. I understand that participation in this project will require less than 30 minutes to complete. I further understand that participation in this research is not intended to produce any particular benefits to me other than the educational value of learning about psychological research first-hand as well as learning more about certain topics that social psychologists like to study. 235 241 6. I understand that I may address any questions or concerns about this research, including .questions about the results obtained, to Galen Bodenhausen of the Psychology Department. Signed Date APPENDIX F Appendix F -W e PARTICIPANT PROFILE W. Your responses in today's research will be completely anonymous. However, it is helpful to us to know a bit about your background and current psychological characteristics in order to understand the types.of'people*who are participating in today’s study. Therefore, before we begin today’s study, please take a moment to answer each of the questions below. A. Your age: years B. Gender: Male Female (circle one) C. Listed below are several descriptors of psychological qualities. Please indicate how much you feel each term describes you 11m. Do this by circling a number on the scale from 0 (not at all) to 9 (extremely much) that reflects how much each quality applies to you at this time. not at extremely all much depressed 1 2 3 4 5 6 7 8 alert 1 2 3 4 5 6 7 8 anxious 1 2 3 4 5 6 7 8 pleased I 2 3 4 5 6 7 8 irritated 1 2 3 4 5 6 7 8 confused I 2 3 4 5 6 7 8 sad 1 2 3 4 5 6 7 8 242 \OO‘OKD‘DKDKO 243 haPPY distracted agitated conscientious calm 244 INSTRUCTIONS TO THE SOCIAL TRENDS SURVEY A group of researchers is cooperating to investigate two recent trends in American society. These trends are the increasing congestion and the increasing cultural diversity of modern society. These trends impact on daily life in a variety of ways, affecting everyday living conditions and interpersonal situations. The researchers involved in the project have generated a variety of such typical, everyday situations. We intend to sample responses to these scenarios from students taking a wide variety of introductory classes, such as psychology , sociology , anthropology , and communications. Respondents are being presented these situations and asked for their reactions. Respondents are being asked to consider situations involving a variety of different groups. These groups differ across questionnaires and are randomly selected. Examples of these groups include: the obese, ethnic minorities, religious groups, gays, senior citizens etc. You will be asked to give your responses to the situations from several standpoints. e se 0 u'c 'rs t s . It is important to keep in mind that there are no right or wrong answers . Our primary interest is in the wide diversity of responses, not the singular responses given by you. Therefore feel free :9 be as h s as ou . When you finish, you may leave your materials on the table in front of you. You may then go up to the experimenter, who will stamp your credit card. 245 Often times, we set up personal standards or guidelines for evaluating our own behavior in interpersonal situations. We usually phrase these guidelines in terms of how we believe we SHOULD respond or behave in various situations. Based on your own personal standards for how you should respond, consider the following situations. For each situation, you will circleethe number between."1" (strongly disagree) and "7" (strongly agree) that best.reflects.how*you.believe you.SHOULD respond in the situations. SITUATION NUMBER ONE: TRANSPORTATION CONGBSTION Imagine yourself sitting on a bus on a hot summer day. At the next stop a gey person boards the bus and takes the seat next to you. You SHOULD feel bothered that this person is sitting next to you. 1 2 3 4 5 6 strongly disagree strongly agree SITUATION NUMBER TWO: SUBURBAN LIVING You own a home in a comfortable middle-class neighborhood. Until recently, the lot to the right of your home has been wooded. You have just found our that this lot was purchased by a developer. He has a contract from a young gey couple to build a house. You SHOULD NOT feel upset about this couple moving in. 1 2 3 4 5 6 7 strongly disagree strongly agree 246 SITUATION NUMBER THREE: CONNOTING While driving home from work, traffic is backed up due to an unannounced rally for gey people. You SHOULD NOT think to yourself "I wish they would go somewhere else". 1 2 3 4 5 6 strongly disagree strongly agree SITUATION NUMBER POUR: PUBLIC TRANSPORTATION You are in a rush to get to a downtown meeting. You reach a cab as the same time as a gey person who offers to share the cab with you. You SHOULD feel uncomfortable. 1 2 3 4 5 6 strongly disagree strongly agree 247 Although we set up personal standards for how we should respond, our actual responses may or may not be consistent with these standards or’ guidelines. Consider the four situations you responded to previously. You will now be presented with these same situations again. But this time, report on the same 1 to 7 scale how you believe you WOULD actually respond in the situations. Base these responses on your personal thoughts and feelings. Your responses may or may not be the same as the ones you gave earlier. SITUATION NUMBER ONE: TRANSPORTATION CONGESTION Imagine yourself sitting on.a bus on a hot summer day. .At the next stop a gey person boards the bus and takes the seat next to you. You SHOULD feel bothered that this person is sitting next to you. 1 2 3 4 5 6 strongly disagree strongly agree SITUATION NUMBER TWO: SUBURBAN LIVING You own a home in a comfortable middle-class neighborhood. Until recently, the lot to the right of your home has been wooded. You have just found our that this lot was purchased by a developer. He has a contract from a young gey couple to build a house. You SHOULD NOT feel upset about this couple moving in. l 2 3 4 5 6 7 strongly disagree strongly agree 248 SITUATION NUMBER THREE: COMMUTING While driving home from work, traffic is backed up due to an unannounced rally for gey people. You SHOULD NOT think to yourself "I wish they would go somewhere else". 1 2 3 4 5 6 strongly disagree strongly agree SITUATION NUMBER POUR: PUBLIC TRANSPORTATION You are in a rush to get to a downtown meeting. You reach a cab as the same time as a gey person who offers to share the cab with you. You SHOULD feel uncomfortable. l 2 3 4 5 6 strongly disagree strongly agree 249 We would now like you to report how well you think your two sets of responses matched each other. Please rate the overall degree to which your ratings on the "SHOULD" scales matched your ratings on the "WOULD" scales. Ratings did not Ratings matched match at all perfectly 0 1 2 3 4 S 6 7 8 9 10 250 Please now take a moment and think back to the answers you gave earlier regarding you probable responses in the various scenarios. Think about how well your should and would responses matched up. We are interested in how you are feeling right now, about how well your SHOULD responses that you reported earlier, match your WOULD responses that you also indicated. Below are words that can describe different types of feelings. For each. word, please indicate how' much it describes how you are feeling AT THIS MOMENT by circling a number on the scale. "1" means "does not apply at all" and "7" means "applies very much" to how you are feeling at this particular moment. does not applies apply at all very much 1. angry at myself 2. negative 3. friendly 4. guilty 5. concerned 6. happy 7. angry at others 8. threatened 9. depressed 10. embarrassed 11. frustrated 12. energetic 13. irritated at others 14. fearful 15. sad H F4 H >4 H +4 H +4 H Pi H Pd H P4 H N no N no N a: N a; N a: w «a N a: N cit» u Cit» u 811» m cat» u 611» u a J» a as a as e 4s A :s e as e as e Ultn 011» m cu m cu m Ln m L» m L» m ox