r- ‘nu~ . ,(V :5! r a .‘ C 13€H7537 QETAT UNIVERSITVIIB BRARIE lliliiliilliiliill. iili Iii lliiil 3 1293 00606 0424 llBRARY Michigan State University V‘D This is to certify that the thesis entitled Toward a Resolution of Self-Enhancement and Self-Conaiatency Theories presented by Cynthia Marie Kaufman has been accepted towards fulfillment of the requirements for M-A- degree in MP8 alt/W ‘X/ flfl‘d 913/64 Z/Cj/W/L’ Major professor Date (3/37/37 0.7639 MS U is an Affirmative Action/Equal Opportunity Institution -‘a PLACE IN RETURN BOX to remove this checkout from your mood. TO AVOID FINES return on or beta. due due. DATE DUE DATE DUE DATE DUE MSU Is An Alfirmative Action/Equal Oppommily Institution TOWARD A RESOLUTION OF SELF-ENHANCEMENT AND SELF-CONSISTENCY PROCESSES BY Cynthia Marie Kaufman A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1989 (004065 ABSTRACT TOWARD A RESOLUTION OF SELF-ENHANCEMENT AND SELF-CONSISTENCY THEORIES BY Cynthia Marie Kaufman A crucial test was conducted between self-consistency theory, which suggests that individuals are motivated to maintain a stable self-concept, and self-enhancement theory, which asserts that the self-concept may be altered if such change would result in a more positive self-view. Results of a first experiment were inconclusive but suggested that attention to a stimulus may not indicate liking for it. In a second experiment, subjects pretested to be self-perceived dominant or submissive were informed that either dominance or submissiveness is more a desirable trait. Subjects then interacted with a confederate who stated that they appeared dominant or submissive. Subjects altered their behavior and their self-conceptions to be more in line with the desirable trait, regardless of their initial self-conceptions. Results were interpreted as support for the notion that individuals will self-enhance rather than self—verify when given the choice. ACKNOWLEDGEMENTS The author wishes to recognize those persons who made this thesis possible with their assistance and support. First, many thanks to my advisor, Dr. Ann Baumgardner, who taught me the skills I needed to complete this.research as well as provided invaluable assistance with the project itself. Second, I would like to thank the other members of my committee, Dr. Norbert Kerr and Dr. Galen Bodenhausen, who were extremely helpful with their comments and suggestions. Also extremely helpful were the research assistants who helped me conduct the experiments: many thanks to Michele Ceiling, Dave Goran, Jerry Juncker, Daylene Richmond, Jack Novak, Dina Myers, Stephanie Vlahos, Ann Butala, Allen Madison, Stuart Cohen, MaryKay Stakoe, and Marilyn Verdonk. Finally, I would like to thank two people who provided endless amounts of love and support. The first is Dan Mondoux, who has been through the process himself and was therefore highly sympathetic. The second is my mother, Joan White, who taught me always to believe in myself and to reach for the stars. iii TABLE OF CONTENTS Page LIST OF TABLES O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O 0 v LIST OF FIGURES O O O O OOOOOOOOOOOOO O O O O O O O O O O O O O O O O O O O O 0 Vi INTRODUCTION 0 O O O O O O O O O C O O O O O C O O O O O O O O O O O O O O O O O O O O O O O O 1 Preliminary Evidence for Self-Enhancement ....... 2 Evidence for Self-Consistency ................... 4 Self-Enhancement of Low Esteem Persons .......... 6 Toward a Discovery of the Critical Difference ... 14 Self-Consistency and Value-Ambiguous Traits ..... 16 Overview of Experiments ......................... 20 EXPERIMENTI0.0.0.0.........OOOOOOOOOOOOOOOOOO0...... 23 Method ............ ...... .. ...... ................ 23 Subjects ................................... 23 Procedure .................................. 23 Results and Discussion .......................... 25 EXPERIMENT II . ....................................... 31 Method .......................................... 31 Subjects 31 Procedure .................................. 31 Desirability Instructions .................. 32 Dominant-Submissive Feedback ............... 35 The Interaction Opportunity ................ 36 Dependent Measures ......................... 37 Results ......................................... 37 Manipulation Check Items ................... 37 Interrater and Scale Reliabilities ......... 40 Behavioral Dominance ....................... 42 Resistance to Feedback ..................... 47 Changes in Self-Perceived Dominance ........ 50 Discussion ...................................... 55 GENERAL DISCUSSION ........................ ........... 59 Boundaries of Self-Consistency Theory ........... 60 Clinical Implications ................ ....... .... 62 iv APPENDICES ...O....0.0000000000000000...0.00000...O... Appendix Appendix Appendix Appendix Appendix Appendix Appendix Appendix A: B: C: D: E: F: Rosenberg Self-Esteem Inventory .... Self-Perceived Likability Scale .... Manipulation Checks--Experiment I .. Measure of Pre-Dominance ........... Desirability Information Sheet ..... Second Measure of Dominance Self-Ratings ....................... Manipulation Check Items Experiment II ...................... Videotape Coding Form .............. LIST OFREFERENCES ......OOOOOOOOO0.00.00...0.0.000... 65 65 67 71 74 76 77 79 81 83 LIST OF TABLES TABLE Page 1 Ratings of Score Accuracy ......................... 28 vi LIST OF FIGURES FIGURE Page 1 Subjects' Posture Subsequent to Feedback ......... 44 2 Behavioral Dominance Findings of Swann and Hill ... 46 3 Swann and Hill's Findings for Resistance to FeedbaCR .................OOOOOOOOOOOOOOOO00...... 48 4 Resistance to Feedback ................ ............ 49 5 Change in Self-Rating as a Function of Self-Perceived Dominance .............. ............ 51 6 Change in Self-Rating as a Function of Feedback .............................. ............ 53 7 Change in Self-Rating as a Function of Information ..........OOOOOOOOOOOOOOO0.0... ........ 54 vii INTRODUCTION Is the self-concept stable or malleable? One of the primary foci in this area of study has been how information from the social environment is rejected by, transformed by, or incorporated into the self-concept. Although this question is far from resolved, the focus of the present paper goes beyond the mechanisms by which feedback is processed and focuses on the motivational processes underlying self-concept change. At the time of this writing, two popular perspectives on the motivational processes underlying self-concept change exist, namely, self-consistency and self-enhancement theories. They make competing and mutually incompatible predictions about why initial self-view influences the acceptance or rejection of social feedback (Jones, 1973). According to self-consistency theory, individuals are motivated to seek out evaluations, whether positive or negative, that support their self-views, because this brings about a comfortable state of congruency. Self-consistency theory suggests that people will selectively seek out, attend to, and give credence to feedback that supports their private self-conceptions. It therefore predicts that people with low self-esteem (LSE) would selectively choose and be more satisfied with negative feedback since it is consistent 2 with their negative self-image. Self-enhancement theory, however, asserts that people are motivated to have positive self-regard, even if they have generally negative self-concepts. This suggests that people prefer to receive positive rather than negative feedback regardless of their self-views, because such feedback would support or create a more positive self-view. Self-enhancement theorists, therefore, predict that LSE people prefer to hear positive rather than negative feedback in order to improve or enhance their self-esteem. WWW Several studies suggest that people prefer to be liked rather than disliked presumably because the former leads to more positive esteem than does the latter (Jones, 1973; Regan, 1976: Wyer & Frey, 1983). For example, Wyer and Frey (1983) found that subjects given bogus negative feedback about their performance on an intelligence test were highly motivated to refute arguments supporting the validity of the results and of intelligence tests in general. It has been further demonstrated that rejecting negative and incorporating positive social feedback into the self-concept may serve to enhance and protect positive self- regard (Baumgardner & Arkin, 1987, 1988; Baumgardner, Kaufman, and Levy, 1989: Tesser, Campbell, & Smith, 1984). For instance, Baumgardner and Arkin (1988) found that, relative to neutral mood conditions, inducing either 3 positive or negative mood in subjects resulted in an exaggerated self-serving bias. This is where successes are attributed to internal causes (e.g., one's effort or ability) and failures are attributed to external causes (e.g., task difficulty or bad luck). The researchers concluded that causal attribution of success to internal factors can create or maintain positive affect. The self-serving bias is but one example of a self- enhancement strategy (Bradley, 1978). Examples of other such strategies include the overestimation of past successes (Nelson 8 Craighead, 1977), the overestimation of personal control (Baumgardner, Heppner, 8 Arkin, 1986: Taylor 8 Brown, 1988), and unrealistic optimism about future events (Taylor 8 Brown, 1988). These self-enhancing 'illusions' may foster motivation, persistence at tasks, and more effective performance (Taylor 8 Brown, 1988). While prevalent in the general population, self- enhancement strategies seem to be much less pronounced among those of low as opposed to high self-esteem (Baumgardner et a1., 1986; Ickes 8 Layden, 1979), and this has led many researchers to conclude that LSE people may not wish to raise their esteem level. Self-consistency theorists, for example, assert that those with high self—esteem (HSE) use such positively-biased cognitive distortions because they wish to verify their initially positive self-views (e.g. Snyder, 1984: Swann, 1987; Swann and Read, 1981a,b). 4 Conversely, according to this view, those with low self- esteem will actually seek out and prefer negative social feedback because it is consistent with their self-views and therefore produces a sense of well-being. To self- consistency theorists, this explains why such cognitive coping mechanisms are generally not found among LSE persons. W The notion that LSE persons prefer consistent rather than enhancing social feedback is based on previous research suggesting that the self-concept is very stable and resistant to change (Schrauger 8 Schoeneman, 1979: Wylie, 1979). It has been shown, for example, that even years of intensive therapy may not produce a change in self-concept (wylie, 1979). This suggests that individuals may be reluctant to or unable to alter their self-views. Swann (1987) explains this reluctance to change self- view by arguing that people have a fundamental desire to know what to expect from their social environments. To facilitate this, children learn to observe their behaviors and the reactions of others to them. Through this process they develop relatively stable self-conceptions. They then begin to seek out evidence confirming these self-conceptions in order to foster a sense of personal control of, and predictability in, the environment. Therefore, individuals prefer to receive self-consistent social feedback, even if such feedback is negative or unfavorable. 5 Swann and Read (1981a) provide evidence that people prefer feedback consistent with their initial self- conceptions. They argue that people use social interactions as a means to receive self-consistent feedback, and as evidence, demonstrated that subjects deemed "self-likables" or "self-dislikables" spent more time viewing partner evaluations when they expected them to confirm their initial self-conceptions than when they did not. They also reported that self-likables recalled more positive than negative statements, although self-dislikables did not recall significantly more negative statements than positive. Swann and his colleagues continue to provide evidence for the self-consistency hypothesis. Swann and Read (1981b), for example, assessed subjects for self-perceived emotionality and assertiveness, and told them they would have an opportunity to interact with a partner. They were told their partners had read some preliminary information about them and had evaluated the subjects by answering several questions. Subjects were allowed to see the list of questions their partner had answered and could choose the questions for which they most wanted to see answers. The researchers found that subjects preferentially chose to see answers to questions that were consistent with their initial self-views (i.e., the assertives chose questions asking the partner to describe why the subjects seemed assertive rather than unassertive). Extending the findings of Swann and Read (1981a,b), Swann and Hill (1982) found that subjects who perceived themselves as either dominant or submissive disconfirmed self-discrepant feedback (e.g. self-perceived dominants told they had been judged ”submissive" subsequently rejected and refuted this feedback) and did not change self-ratings. Swann and Ely (1984) extended this finding to the realm of confirmation of perceiver expectancies (i.e., the way in which an observer's expectations for behavior influence the actual behavior of a target) by demonstrating that self- perceived extraverts or introverts self-verified rather than confirmed mistaken expectancies (e.g., extraverts expected to be introverted acted gregariously). This finding held in all conditions except when perceivers were certain of their expectancies at the same time that targets were uncertain of their self-conceptions. W In contrast to the work of Swann and his colleagues, the self-enhancement tradition sees the self-concept as malleable and as constantly influenced by social feedback. Individuals are seen as motivated to enhance their self- concepts, and for those with initially low self-regard, this involves an active attempt to shift it upward. This idea has theoretical precedent: Festinger (1954), for example, spoke of the "unidirectional drive upward," 7 i.e., that there is a basic human drive to progressively perform at higher levels and to continuously improve one's abilities. This provides a foundation for the notion that there is motivation for positive self-concept change. Markus and her colleagues (e.g. Markus 8 Kunda, 1986: Markus 8 Nurius, 1986, Markus 8 Wurf, 1987) argue convincingly for a multi-faceted, dynamic self-concept. Such a dynamic self-concept would more likely be influenced by the social environment than would the stable, static self suggested by self-consistency proponents. In fact, Markus and Wurf (1987) suggest that people use social interactions as opportunities for self-enhancement by engaging in biased processing during social perception. They cite evidence that individuals are likely to evaluate others on dimensions that are personally important to themselves, use selectively self-enhancing social comparison strategies, and selectively interpret events so as to support positive self-conceptions (Markus 8 Wurf, 1987). Kunda and Sanitoso (1989) provide direct evidence suggesting that the self-concept may be influenced by the perceived desirability of various traits. A first experiment demonstrated that when informed that either extraversion or introversion was predictive of future success, individuals began to view themselves as possessing more of the desirable attribute. A second experiment demonstrated that individuals exposed to successful and 8 failing individuals came to see themselves as possessing traits more like those exhibited by the successful persons. Social information, then, is capable of influencing the self-concept, at least for an unspecified period of time. According to self-enhancement theory, therefore, individuals are motivated to develop positive self-views and may do so through social interaction. How then do individuals cope with directly negative evaluations? They must somehow avoid incorporating negative information into the self-view, because this would make the self-view less positive. Recall that LSE relative to HSE persons are less likely to show cognitive self-enhancement strategies. In order to assert that even LSE individuals wish to develop positive self-views, the self-enhancement tradition must explain the lack of self-serving cognitive distortions among LSE persons in a manner quite different from self- consistency theorists. We (Baumgardner, Kaufman, and Levy, 1989) have argued that LSE people are unable to use such cognitive coping mechanisms as the self-serving bias because their negative self-views prevent them from cognitively filtering social feedback as do their HSE counterparts. Those who have initially negative or uncertain self-conceptions, however, may actually feel a greater need to make a favorable impression on others than those who are more self-confident or more self-certain (Baumgardner, 1988). According to this 9 view, therefore, LSE persons do wish to self-enhance, and so they must find means other than cognitive distortions by which to accomplish this. Perhaps the most convincing argument in favor of self- enhancement comes from the series of experiments conducted in our laboratory. Baumgardner et al. (1989) tested the hypothesis that LSE persons engage in strategic self- presentations as a means to raise self-affect. In the first three experiments, we presented subjects with personality feedback ostensibly generated by either an interaction partner (Experiment 3) or a computer program (Experiments 1 8 2) and then gave the subjects an opportunity to respond to the feedback and its source either privately or publicly. In the private response condition, subjects were told to put no identifying marks on their questionnaires and that they would place them in a box labeled "Confidential" at the end of the experimental session. In the public response sessions, subjects were told to put their names and student numbers at the tOp of the page and, when finished, to hand them directly to the experimenter, who would "look them over to see what you wrote.” Although HSE persons showed a general self-serving bias, i.e. rating sources of positive feedback as more accurate, intelligent, and likable than sources of negative feedback, LSE persons were likely to compliment a source of positive feedback or derogate a source of negative feedback 10 only when they believed that their reactions would be publicly available to others. When they thought their reactions were private and confidential, however, they did not rate the source of positive feedback as more competent than the source of negative feedback. We did explore an alternative social exchange explanation for the findings: perhaps subjects were merely attempting to reward or punish the feedback source. However, Experiments 1 and 2 demonstrated that LSE subjects publicly self-enhanced when using computers as feedback sources, and it seems unlikely that subjects thought they could either reward or punish a computer. Another possible alternative explanation springs from impression management theory: perhaps LSE participants were attempting to impress the experimenter rather than regulate their self-views. However, the results were unaffected by whether another person, the experimenter, was aware of the initial feedback subjects received. In Experiment 2, we created ”experimenter aware" conditions, in which subjects were instructed to raise their hands immediately after receipt of their computer feedback. The experimenter then walked over and read through their feedback before clearing the screen. In the ”experimenter unaware" conditions, subjects were told to hit return after reading their feedback to clear the computer screen before anyone else could see it. Although 11 manipulation check questions suggested that the aware/unaware manipulation was effective, no difference was found between the two conditions. This made an impression- management explanation less tenable. Baumgardner et al. (1989) further demonstrated in Experiment 4 that when induced to publicly compliment a source of positive feedback or derogate a source of negative feedback, LSE persons reported a greater resultant rise in self-affect than those induced to do so privately. These results suggest that public self-enhancement does indeed function to regulate self-affect. Taken together, the results of these four experiments indicate that LSE persons attempt to discredit negative social feedback, rather than preferentially solicit and attend to it. However, the strategies they employ may be qualitatively different from those cognitive strategies chosen by persons of high self-esteem. Previous studies have suggested other ways in which LSE persons might attempt to convince themselves that they are worthy individuals. For example, LSE persons may be more likely than those of high self-esteem to self-handicap, or acquire impediments to performance when unsure of success so as to provide an excuse in case of failure (Baumgardner, Lake, 8 Arkin, 1985: Arkin 8 Baumgardner, 1985). For example, a tennis player may point out that she is playing with an old, poorly strung racket in order to excuse any 12 poor performance. Low self-esteem persons are also more likely to attempt to lower evaluation standards through strategic failure at the beginning of a task (Baumgardner 8 Brownlee, 1987). For example, a "B" student whose teacher tells him she expects him to get an ”A+" in her class may withdraw effort and get a ”C” on the first quiz in order to lower what he considers to be unrealistic expectations, thus increasing the probability of future success. A recent study by Brown, Collins, and Schmidt (1988) found that LSE persons are more likely than HSE persons to engage in what the researchers call indirect forms of enhancement, i.e., enhancements focusing on one's association with other persons rather than explicitly self- related esteem-enhancing biases. These researchers found that although HSE individuals were more likely to display group favoritism when directly involved in group processes, thereby self-enhancing directly, LSE subjects were more likely to show favoritism when only indirectly associated with the group, thereby avoiding direct self- enhancement. Brown et al. (1988) explained this finding by asserting that everyone strives to self-enhance but that LSE persons are more limited in their ability to do this directly. This seems similar to the findings of increased self- handicapping, strategic failure, and self-enhancing public 13 responses to evaluating others among LSE persons, in that it again represents an inability to directly, cognitively self- enhance among LSE individuals and yet demonstrates that even these persons do attempt self-enhancement. It appears that LSE persons may attempt to bolster self-esteem through either association with other persons or by attempting to regulate their expectations. They seem less direct and more interpersonal than those coping mechanisms, such as self-serving biases and cognitive distortions, chosen by persons of high self-esteem. It may be that LSE persons are unable to cognitively self-enhance, because their negative self-regard prevents them from using cognitive, positive distortions. Instead, they seem to use social interactions as interpersonal enhancement opportunities. These interpersonal strategies also appear to play a role similar to the HSE cognitive buffers in regulating affect; research indicates a link between self-enhancement strategies and a subsequent rise in self-esteem. Experiment 4 of Baumgardner et al. (1989) provided evidence for this link most directly, but additional research also supports this conclusion at least indirectly (Jones, Berglas, Rhodewalt, 8 Skelton, 1981). These researchers found that subjects induced to make either a positive or negative impression on others showed corresponding shifts in esteem. This would suggest that people incorporate public statements 14 about themselves into their self-concepts. Taken together, the results of our experiments, along with findings of self-handicapping, strategic failure, and indirect enhancement, suggest that even persons with negative self-views desire to self-enhance, but that the methods they use to accomplish this are qualitatively different from those used by persons with high self-esteem. These studies appear to run directly counter to findings of self-consistency. Such conflicting perspectives seem difficult to reconcile, particularly since both traditions seemingly have direct evidence supporting their basic tenets. Most theorists have concluded that both approaches have their merits, but that there must be a yet unidentified critical difference that makes one operative over the other in any given situation (e.g., Epstein, 1973). WWW One study that has postulated a possible critical difference was conducted by Swann, Griffin, Predmore, and Gaines (1987). This research suggests that "affective” reactions to feedback may conform to self-enhancement theory while "cognitive" reactions follow self-verification principles. Subjects were given either favorable or unfavorable feedback about their social skills after delivering a speech and then they responded to questionnaires designed to tap their cognitive and affective responses to the feedback. 15 Swann et al. (1987) found that LSE persons found the negative feedback to be more self-descriptive than did HSE persons yet felt depressed after receiving it. Questioning the assumption of 'psychological unity' within the individual, Swann and his colleagues argue for the independence of affective and cognitive systems from each other, stressing that these systems may at times (e.g., when receiving negative social feedback) have conflicting goals for the LSE individual. Persons with low self-esteem, they hypothesize, do cognitively desire negative feedback because it is more consistent with their self-image, but they feel more depressed and hostile than their high self-esteem counterparts upon receiving such feedback. Because the cognitive and affective reactions of LSE persons are incongruent with each other, this could account for evidence for both self-enhancement and self-consistency. The cognitive ”system” appears to desire consistency while the affective "system" appears to desire enhancement. A related alternative, however, is suggested by a review of the self-consistency literature. All but one of the studies finding evidence for this process measured ”value-ambiguous" traits or aspects of the self-concept, such as dominance, emotionality, assertiveness, and extraversion. They are considered value-ambiguous in that there is no clear positive or negative value assigned to 16 either of the polar ends of the given trait. For example, it is not considered clearly more advantageous to be highly emotional as opposed to very unemotional, since both traits represent extremes, and each has a slightly negative connotation. It may be that individuals tend to respond in a self- consistent manner when asked to manifest such value- ambiguous traits because there exists no opportunity to engage in self-enhancement. For example, an individual who considers herself highly emotional may be informed that she seems unemotional. She has no motivation to accept the discrepant view since her self-concept will not be enhanced by doing so. WWW As noted above, the traits used to show self- consistency processes have been almost entirely value- ambiguous. For example, Swann and Hill (1982) gave self- perceived dominants and submissives either self-consistent or self-discrepant feedback (e.g., told self-perceived dominants they seemed dominant or submissive, respectively). They then either did or did not give them an opportunity to refute this feedback before measuring resultant self-rating change. They found that subjects given a refutation opportunity did attempt to refute the discrepant feedback by making verbal protestations and showed minimal self-rating change. 17 However, subjects may have had no opportunity to engage in self-enhancement in this situation, for it is not considered clearly better to be dominant as opposed to submissive. Both may be seen as extremes and may even have a slightly negative connotation. A person who is thought of as dominant may be considered overly aggressive or "pushy,” whereas a submissive person may be thought of as cowardly or as "lacking a mind of his or her own." Other studies by Swann and his colleagues also use such value-ambiguous character traits to provide evidence for self-consistency. Swann and Ely (1984) showed that when perceivers formed expectancies discrepant from targets' self-views as either introverted or extraverted, these were disconfirmed in all conditions except when targets were uncertain of their self-conceptions at the same time that perceivers were certain of theirs. Similarly, Swann and Read (1981b) demonstrated that participants more often verbally requested as well as spent more money to solicit feedback confirming their emotionality and assertiveness self-conceptions than those that disconfirmed them. Characteristics like extraversion/introversion and emotionality are also ambiguous in their social value, however, for an extravert may be thought of as friendly and social whereas an introvert may be considered thoughtful and sincere. Emotionality may also be similarly ambiguous in its social desirability, for although it is not considered 18 positive to be a ”cold fish,” neither does one want to be considered a person who is ”high strung” or who always overreacts. It may even be argued that self-consistency findings may be reinterpreted in terms of self-enhancement. Perhaps individuals given no clear indication of the desirability of various trait attributes tend to believe that the traits they feel they possess are more desirable than the opposite traits (cf. Kunda 8 Sanisoto, 1989: Taylor 8 Brown, 1988). For example, a self-perceived dominant may believe that dominance is better than submissiveness if given no evidence to the contrary, simply because it would be self-enhancing to maintain this belief. This would certainly be consistent with findings of self-serving biases and cognitive distortions that have been so well-documented. Future research might examine this possibility. The one exception to the use of value-ambiguous traits by consistency theorists is a study by Swann and Read (1981b), in which they argued that subjects who disliked themselves preferred unfavorable as opposed to favorable evaluations. This finding is crucial since it is the only published finding clearly supporting self-consistency over self-enhancement theory. There are several issues in this study, however, that suggest a replication is in order. First, the dependent measure of evaluation preference was the amount of time spent scanning received feedback. 19 This is problematic because there are alternative explanations for why a self-dislikable might spend more time scrutinizing negative feedback. Self-dislikables may need more time to discredit the negative feedback or its source than do self-likables because they have less cognitive coping mechanisms available to them (Baumgardner et al., 1989: Fuhrman, 1988). Therefore, it is unclear whether the amount of time spent scanning feedback is an indication of desire for that information. It is also important to note that although self- likables showed better recall of positive statements than negative ones, self-dislikables did not show better recall of negative than positive statements. This latter finding, therefore, lends support to neither self-verification nor self-enhancement. The purpose of the present research was to assess the hypothesis that subjects would prefer to self-enhance rather than self-verify if given both opportunities. The second experiment addressed the limitations of the Swann and Read (1981a) study by examining more than simply looking at positive versus negative information. Instead, subjects were allowed to verbally respond to social feedback indicating they possessed a certain personality trait and were also asked how positive the trait was compared to its opposite. Self-concept change along the dimension of the trait was also measured. 20 W A preliminary experiment was designed to test the hypothesis that subjects will choose to self-enhance rather than self-verify when they receive feedback on a trait with unequivocal desirability. For example, although Swann and his colleagues have found evidence for self-verification when using traits of ambiguous desirability, e.g., dominance or extraversion, they have provided little clear evidence for the same pattern of results with less ambiguous traits such as intelligence or likability. In a first experiment, subjects' levels of self-esteem and self-perceived likability were measured and then bogus likability ratings, ostensibly generated by judges hidden behind a one-way mirror, were delivered. Each subject received five ratings, ranging from 59 to 93 on a scale of 100. Subjects were then informed that they would have an opportunity to meet one of the judges and receive a more detailed explanation of the attractiveness score she had given them. The main dependent measure was which judge they would choose to meet, and it was hypothesized that all subjects, regardless of self-esteem or self-perceived attractiveness, would be more likely to choose to meet a judge who provided a higher score than one who provided a lower score. A questionnaire assessing subjects' motives for wishing to meet a certain judge was also distributed 21 and analyzed in order to assess their reasons for selecting a given judge. The second experiment was designed to provide a more direct crucial test between self-enhancement and self- consistency theories. In this experiment, the method used by Swann and Hill (1982) was partially replicated and extended. In their study, subjects were pretested for self- perceived dominance and were then given feedback by a confederate indicating that they either seemed dominant or submissive. The main dependent measures were the levels of behavioral dominance and the resistance to feedback shown by subjects as well as any changes in reported self-concept. However, the present study added an additional manipulation to their design: the desirability of dominance was systematically varied so that half the subjects were led to believe that dominance is more socially desirable and more predictive of future success than submissiveness, while the other half was led to believe the opposite. It was hoped that manipulating trait desirability in this way would provide subjects with an enhancement opportunity that did not exist in the original study. It was predicted that subjects, given this opportunity, would choose to self-enhance by accepting self-discrepant feedback if it was self-enhancing (i.e., if the discepant trait was described as the more desirable one). Subjects were pretested for self-perceived dominance, 22 and those who scored in the top and bottom thirds (self- perceived dominant and submissive, respectively) were then told that either dominance or submissiveness is socially desirable. Subjects were given feedback from a confederate stating that they appeared either dominant or submissive. The question of interest was whether subjects would subsequently choose to self-verify or to self-enhance in both their behavior and their public self-descriptions. The design of this study was a 2 (self-perceived dominance vs. submissiveness) x 2 (instructions: dominance or submissiveness socially desirable) x 2 (feedback: dominant vs. submissive) between-subjects factorial design. In the first experiment it was predicted that individuals who received feedback on their likability would prefer to self-enhance, i.e., choose to meet a judge who gave them a high score, rather than self-verify, i.e., choose to meet a judge who gave them a low score. It was hoped that the second experiment would provide a more direct test of the hypothesis because subjects were led to believe that either dominance or submissiveness was more desirable than the other. It was hypothesized that subjects would accept self-discrepant feedback if it indicated that they possessed the desirable trait, and that they would even alter their self-concepts in the direction of the desirable trait. EXPERIMENT I Method‘ 5.111212% Thirty undergraduate female students enrolled in an introductory psychology course participated in the study in exchange for partial fulfillment of a course requirement. EIQQQQBIQ Subjects participated in the experimental session individually. Upon arrival at the experimental room, the subject was asked to complete the Rosenberg (1965) Self- Esteem Inventory (Appendix A, p.65) while waiting in a waiting room for the experiment to begin. Then she was asked to sit in front of a one-way mirror and complete a questionnaire in which questions assessing self-perceived likability were embedded (Appendix B, p.67). She was told that the researchers were planning to conduct an experiment to see how likability correlates with the several other factors such as grade-point average and attitudes towards family and friends. In order to do this, the subject was informed, some judges needed to be trained to make accurate ratings of likability. The subject was then told that four female judges were hidden behind the one-way mirror, and that these judges were going to rate the likability of the subject while she answered a few questions posed by the experimenter. The subject would then be allowed to see the ratings, "in order 23 24 to help us judge their accuracy.” She would be asked to rate the accuracy of the scores in order to help train the judges, and might even be "allowed to meet some of the jusges and learn more about the scores, if time permits." It was determined by pilot-testing that the questions posed to subjects were low in difficulty so as to avoid performance-debilitating anxiety. They also allowed the subjects to express enough personal opinions to make the likability ratings seem more plausible. Questions ranged from, "What is your first childhood memory?" to "What is/are your roommate(s) like?" After the subject had responded to all of the questions, the experimenter left the room for a moment, ostensibly to retrieve the scores from the judges. She then returned with four pieces of paper. On each one was written one of the following handwritten scores: 59%, 68%, 84%, and 93%. Before this experiment was begun, these scores had been pretested to determine that two of them were above and two below an independent sample of college females' mean ratings (u = 76) of their own likability. The subject was allowed to examine the scores for about fifteen seconds. The experimenter then informed her that it would be helpful to one of the judges to be allowed to give a more detailed explanation of her score. This would allow the judge an opportunity to give a more detailed explanation of her rating, which would be very helpful in her training. 25 Additionally, this would give the subject an opportunity to learn some more detailed information about one of the scores. The experimenter added that if there were more time, the subject could meet all of the judges, but that the time remaining in this session would only allow for one. Subjects were told that they should choose one judge they would like to meet based on the scores and hear a more detailed explanation of the score. It was emphasized that it was irrelevant which judge they chose to meet, since the experimenters were running enough subjects that eventually all of the judges would have a chance to explain their ratings. After subjects selected a score, they were told that they would meet that judge but that first they should fill out a questionnaire evaluating the accuracy of all of the scores. This questionnaire was actually a Likert-type manipulation check questionnaire assessing their reasons for choosing that particular judge (e.g., I wanted to change his or her mind; I wanted to hear further information about my score, etc.)(Appendix C, p.71). Subjects were then debriefed thoroughly according to standards outlined by Mills (1976) and excused. Results and Discussion Prior to analysis, subjects were divided into low and high levels of both self-esteem and self-perceived likability using median splits. The main dependent 26 measure was which judge's score subjects chose to learn about. However, twenty-one out of thirty subjects chose to meet the judge who gave the lowest score. Of the remaining nine, seven chose to meet the judge who had given the second-lowest score of 68%, and two chose to meet the judge who gave the highest score of 93%. Chi-square analyses revealed no significant differences in self-perceived likability (X2(2)-0.06, p > .97) or self-esteem (X2(2)-2.08,p > .35) between those who chose to meet different judges. This finding was quite surprising, since it appears to support neither self-enhancement nor self-consistency theories. The former predicted that all subjects would choose to meet a judge who gave a high score, whereas the latter predicted that subjects high in self-perceived likability would choose to meet a judge who gave them a high score but subjects low in self-perceived likability would choose to meet a judge who gave them a low score. It was therefore decided to conduct internal analyses of the motivation questions for those subjects who selected the lowest score in order to determine why they did so. Twenty-one subjects were included in these analyses. A two-factor (high vs. low esteem, likability) analysis of variance was conducted to determine if there existed differences between the ratings of how accurate each likability score was between these low and high in self- 27 esteem and self-perceived likability. There was a marginal effect of self-esteem for ratings of the lowest score, £(1,20) - 3.18, p <.09, indicating that LSE persons were somewhat more likely to rate the score of 59% as more accurate (M - 4.27) than were their HSE counterparts (M = 2.80). A similar finding was obtained for ratings of the highest score of 93%. Those who were high in self-perceived likability rated this score as more accurate (M = 6.18) than did those low in self-perceived likability (M - 5.10), £(1,20) - 4.89, p <.05. No other significant differences were found for any other ratings when analyzed either by self-esteem or by self—perceived likability. It is interesting to note, however, that the accuracy ratings are sequentially greater for each higher score, indicating that both those high and low in self-esteem and self-perceived likability rated higher scores as more accurate (see Table 1). This would seem to indicate that although these subjects chose to meet the judge with the lowest score, they regarded this score as the least accurate of them all. Why then did subjects wish to meet this judge? Some insight into their motivation is provided by responses to the seven-point Likert-type manipulation check items (Appendix C). There were no significant differences as a function of either self-esteem or self-perceived likability, and the group means were so similar to the overall mean for 28 Table 1. Ratings of Score Accuracy. Rating Self-Esteem Self-Perceived Likability low high low high 59 4.27 2.80 3.50 3.64 68 4.36 4.20 4.00 4.55 84 5.36 5.50 5.00 5.82 93 5.36 6.00 5.10 6.19 29 each variable that only the latter will be discussed here. Subjects were neutral about how much they expected to like the judge they would meet (M = 4.52), but they agreed somewhat with the statement that the score they had chosen (59%) was the least accurate of all of them (M = 3.14). They stated that they wished to hear more information about this low score (M a 5.90) and that they were curious about the judge who had given it to them (M = 5.81). They reported that they did not choose the score randomly (M a 1.71) and did not like this score better than the others (M = 1.90). They did feel that this judge had put careful thought into the score (M - 5.10). Taken together, these results suggest that subjects chose the lowest score not because they preferred it but simply because they were curious about why it was so low. Content analyses of subjects' responses on a three-point scale to the question, "Why did you choose to meet this judge?” overwhelmingly indicated that they were simply curious (M = 2.04), and there were no differences as a function of self-perceived likability (E (1,20) = .08, p > .75) or self-esteem (2 (1,20) = .80, p > .35). In general, then, this first experiment cannot be interpreted as support for either self-enhancement or self- consistency theory. Subjects overwhelmingly chose to meet the judge who gave them the lowest score, and it appears that they were motivated by simple curiosity. Future 30 research might address this problem by using a more restricted range of scores so that the bottom one does not appear to be so low. These findings underscore the argument that attention to positive or negative feedback does not necessarily imply a motivation to validate or to incorporate that feedback. As discussed earlier, Swann and Read (1981a) found that self-perceived likables and dislikables attended more to self-consistent that self-discrepant feedback. Such attention does not necessarily indicate agreement with the information. An alternative reason for attending to negative information is suggested by research on selective exposure (e.g., Frey, 1986). This literature suggests that individuals may attend to information that is discrepant from their self-view but useful: for example, upon being told that she seems unfriendly, an individual may seek out further information about this judgement in order to learn what behaviors to avoid. As this experiment has demonstrated, curiosity may be another reason for seeking out further evaluative information. Subjects sought more information about a score that they actually liked the least and thought was the least accurate. Therefore, attention to or even seeking out information is not an adequate operationalization of preference. In general, researchers must be careful to be explicit about the assumptions made when interpreting behaviors. EXPERIMENT II Method 5383991.: Two hundred female undergraduate introductory psychology students were pretested for self-perceived dominance/submissiveness using the scale developed by and described in Swann and Hill (1982) (Appendix D, p.74). Those scoring in the upper and lower thirds of the dominance scale were considered self-perceived dominant or submissive, respectively, and were invited by phone to participate in an ostensibly unrelated study, which was actually a second experimental session. Eighty subjects participated in the second experimental session. EIQQEQQLQ Upon arrival at the second experimental session, each subject was told that she was going to participate in a problem-solving task with a partner. As in the experiment by Swann and Hill (1982), she was introduced to a female confederate partner posing as an introductory psychology student, who had supposedly been contacted by phone in the same manner as the subject had been. The subject and confederate were then told that the study involved comparing how people solve problems alone versus with someone else, and that it was expected that social embarrassment may make people use different problem-solving strategies when in a group than when by themselves. 31 32 The experimenter then explained how to perform the experimental task. Following the procedure outlined by Swann and Hill (1982), the subject and confederate were shown a modified version of the game 'Mastermind' and were given instructions about how to play it. The experimenter selected four different colored pegs from a bowl without letting the others see which ones and inserted them side by side on a concealed pegboard. The players were asked to guess the relative position of each of the pegs by selecting any four pegs and placing them in a row on a second pegboard. The experimenter then examined their peg placement and informed them how many pegs were of the right color and how many in the correct position. After each such trial, the players had the opportunity to try a new grouping of pegs until they either discovered the correct order or reached a 10-trial limit. Dfifiizihili§¥_ln§£IB££iQn§ Before beginning to play, the subject and confederate were given one of two desirability instruction sets. They were told that the experimenter had available some information sheets that were actually from another study, but that she thought they might like to read them since they described "characteristics relevant to group problem- solving." The partners were allowed an opportunity to read the information, and all but one subject did so. The subject and confederate then read one of two bogus 33 information sheets (Appendix E, p.76) bearing a Princeton University logo and one of two typewritten messages. The first read as follows: Dear Colleagues, Recently my colleagues and I have conducted a series of remarkable studies in our laboratory examining the personality trait known as dominance. We have uncovered irrefutable evidence suggesting that in group situations, 0 a o a et e - WW- Furthermore. our results prove that W 0.: - O O 2-- they have we and W ten years after graduation than do submissive people. These results, although incontrovertable, were somewhat surprising. We would like to collect students' opinions from across the country about why dominance may be a more desirable trait than submissiveness. Please ask your students to list three reasons why they think dominance may be a better trait than submissiveness, and send these sheets back to us. Thank you for your assistance. The sheet then gave the subject and confederate blanks on which to write any explanations that occurred to them. All but 2 subjects wrote two or more reasons on their sheets. This procedure for inducing belief in the desirability manipulation is similar to that described by Kunda and her colleagues (Kunda 8 Fong, 1988; Kunda 8 Sanitioso, 1989). The second bogus information sheet was identical to the first except that the words 'submissiveness' and 'dominance' were reversed, thereby indicating that submissiveness was more desirable than dominance. The confederate did not know which of the two information sheets the subject had read, in 34 order to avoid biasing the delivery of the feedback (to be discussed below). It is important to note that, throughout the experiment, the experimenter, the confederate, and the audiotape judges were all blind to several conditions. All were blind to the initial self-perceived dominance or submissiveness of the subject. Because of this, they did not know which type of behavior constituted a self-enhancing response for each subject, rendering an experimenter expectancy effect explanation of the results less tenable. Additionally, the experimenter was blind to the feedback delivered by the confederate until after it had been delivered. Finally, as mentioned above, the confederate was blind to the information sheet the subject had read until after the session was completed. While the subject and confederate filled out the sheets, the experimenter went into the next lab and turned on a videocamera, which was facing a one-way mirror looking into the experimental setting. The experimenter then returned and instructed the partners to begin playing the game. He or she informed them how many pegs they had in the correct positions and how many of the correct color until they guessed correctly or completed ten trials. All but two sets of partners were able to guess the pattern within the ten-trial limit. In the Swann and Hill (1982) study, the experimenter 35 then informed the participants that during a second set of trials one person would be a leader, in charge of making the final decisions about peg placement, and the other partner would be an assistant. The partners were told they should decide who would play each role, and the confederate delivered the dominance-submissiveness feedback immediately afterwards. However, this instruction is problematic. It may be that submissive people who are told they are dominant are afraid of having to assume a leadership role. They then reassert their submissiveness in an attempt to regulate the partner's expectations and avoid having to act as the leader. In this case, Swann and Hill's findings might be explained by expectancy-regulation theory (Baumgardner 8 Brownlee, 1987) and may not be evidence for self-consistency theory. Therefore, in the present Experiment 2, the partners were merely told that the experimenter had forgotten a questionnaire in the next lab and that they should just relax while he or she went to get it. We]: Immediately after the experimenter left the room, the confederate delivered dominant or submissive feedback to the subject. In dominant conditions, the confederate stated: Well, that information sheet is pretty wild. I didn't know they studied things like that. You seem to be a more dominant person, at least from what I could tell during 36 the game. Is that how you think of yourself too? In the submissive-feedback conditions, the confederate repeated the same words but substituted 'submissive' for 'dominant.' W The subject then had a one-minute opportunity to react to the feedback statement made by the confederate. During this time, the confederate responded as nondirectively as possible to statements made by the subject in an attempt to avoid biasing the subject's reponses. The confederate simply nodded in response to statements the subject made and responded, "I've never really thought of myself that way," if asked if the confederate thought she herself was dominant or submissive. One minute subsequent to feedback, the experimenter turned off the camera and returned to the room, carrying two questionnaires. The subject and confederate were then given a set of questionnaires in which the questions from the first dominance scale were imbedded (Appendix F, p.77). Also included was a manipulation check survey in order to make sure the desirability instructions were understood and the cover story believed (Appendix G, p.79). After completing these questionnaires, the subject was debriefed thoroughly using principles outlined by Mills (1976), given credit for participation, and then excused. 37 Dependent_uea§ures Several main dependent measures were analyzed. First, the videotape of the game interaction and feedback response was coded by four independent and pretrained judges. The judges rated the dominance of the subjects during the game using both subjective scales and objective measures, including the number of pegs placed by the subjects and the number of verbal suggestions made. Judges also rated the resistance of the subjects to the confederates' feedback with a 4-point scale ranging from 1 ("accepted it willingly”) to 4 ("questioned or refuted it") (Swann and Hill, 1982) (Appendix H, p.81). In addition, change in self-reported dominance was measured by examining the responses to the second dominance scale. These scores were subtracted from the subjects' initial pre-session responses and then compared to the changes of other subjects in different conditions. Results Manipulatien_§hesk_ltems Subjects' responses to several seven-point Likert manipulation check items suggested that the experimental manipulations were perceived and interpreted as intended. Subjects who received dominant feedback from the confederate indicated that their partner thought they were much more dominant (M a 5.41) than did those who received submissive feedback (E = 2.68) (2(1, 74) - 111.58, p < .0001). 38 All but two subjects remembered having read an information sheet about dominance or submissiveness, and of those who remembered, all were able to correctly report whether they had read that dominance is better than submissiveness or that submissiveness is better than dominance. When asked how much they agreed with the information sheet they had read, most subjects indicated that they did at least somewhat (M = 4.81). Those who read that dominance is better reported agreeing with it more (M = 5.43) than did those who read that submissive is better (M = 4.26), F (1,74) = 6.65, p < .01. However, both means are above the midpoint of the seven-point scale, indicating that both groups agreed at least somewhat with the information. When asked if they thought they themselves were more dominant or submissive, subjects in the pre-dominant group were marginally more likely to see themselves as more dominant (M = 4.42) than were those in the pre-submissive group (M = 3.85), E(l,74) = 3.45, p < .07. This provides additional evidence supporting the partitioning of subjects into these groups. In addition, those who read information sheets indicating dominance was better than submissiveness saw themselves as more dominant (M s 4.49) than did those who read the opposite (M a 3.76), F (1,74) = 6.36., p < .01. This gives some indication that subjects had read and were 39 influenced by the bogus information sheets and replicates the findings of Kunda and Sanitioso (1989). Subjects' responses to this question were also influenced by the type of feedback they received. Those who were told they seemed dominant reported themselves to be more dominant (M = 4.57) than did those who were told they seemed submissive (M - 3.48), E (1,74) - 12.47, p < .001. Subjects, therefore, were not only influenced by the information sheets, but were also influenced by social feedback. When subjects were asked whether it was better to be submissive or dominant on a Likert scale ranging from 1 (submissiveness is better) to 7 (dominance is better), three main effects emerged. First, those who had read that dominance is the better trait rated dominance as opposed submissiveness (M = 5.32) higher than did those who read that submissive is better (M - 3.89), F (1,74) = 29.22, p < .0001. This supports the notion that subjects attended to, and were at least momentarily influenced by, the information sheets they read. Second, those who received dominant feedback were more likely to indicate that dominance is better (M = 4.91) than were those who received submissive feedback (M = 4.16) (E (1,74) = 4.99, p < .03). This suggests that individuals tend to see the traits others think they possess as being desirable. ”If I appear dominant/submissive," subjects 40 appear to have reasoned, "then that trait must be better than its opposite.” Third, those who were in the pre-dominant group were more likely to think dominance was better (M s 4.92) than were those in the pre-submissive group (M - 4.31), E (1,74) - 4.67, p < .03. This finding provides even stronger support for the idea that individuals tend to see the traits they possess as more favorable than others. Taken together, the results of the manipulation check analyses indicate that subjects understood and were influenced by the feedback and information manipulations. It also appears that the division into presubmissive and predominant groups before the main experimental sessions was a meaningful one, since predominant persons reported themselves to be more dominant than presubmissive persons at the end of the session. Finally, subjects were more likely to see their own trait as more desirable. WW Interrater reliabilitities were calculated for a number of behaviors videotaped during the experimental sessions. Every fifth tape was viewed by all four judges, resulting in sixteen total subjects coded by all four. Interrater reliabilities were calculated using the Spearman-Browne formula for these sixteen tapes. All remaining tapes were coded by two of the four and final scores were calculated by averaging the scores of raters who had demonstrated 41 acceptable reliabilities. The number of game pegs subjects placed in the board and number of verbal suggestions they made were accurately recorded by all four raters. Interrater reliabilities for these measures were high (I s .99 for the former and I =.96 for the latter). Overall ratings of subjects' dominance were also calculated by having judges make ratings every three minutes during game play. Judges used several criteria for making these judgments, including eye contact, tone of voice, posture, and "taking charge" of the game. One judge showed poor reliabilities for this measure and was eliminated from it. Resulting interrater reliability was good (3 = .81). This correlation indicates a slightly lower yet still acceptable reliability for this measure. Judges were also asked to code subjects' responses to the confederate's feedback. All coders were able to reliably code the responses on a five-point Likert scale that ranged from 1 (agreed strongly) to 5 (argued strongly) as well as the number of responses to the feedback made by the subject. Interrater reliability was good for both (; = .88 for the former and I a .90 for the latter). Finally, judges rated subjects' posture subsequent to receiving feedback on a three-point scale ranging from 1 (leaning back) to 3 (leaning forward). Interrater reliability for this measure was good (I = .92). It appears, then, that the 42 judges were able to reliably code a number of behaviors recorded during the experimental sessions. A final note here concerns the scale reliability of the dominance scale. This scale consisted of five Likert items (see Appendices D, F) and was used to divide the sample into pre-dominant and pre-submissive groups as well as to look for self-rating change. The Cronbach's Alpha for the scale measured at the time of the initial division was .86, indicating an acceptable level of reliability (Crocker 8 Algina, 1986). The distribution of pre-dominance scores ranged from 5 to 24 on a possible range of 5 to 25, and the means for the pre-submissive and pre-dominant groups were 10.5 and 17.9, respectively. Maximum Many differences between conditions in behavioral dominance during gameplay were uncovered. First, subjects who had read that dominance is better than submissiveness made more suggestions (M - 1.17 per trial) than did those who read the opposite (M = 0.77 per trial), E (1,55) = 4.33, p < .05. General ratings of dominance on a five-point Likert scale revealed that those who read that dominance is better were perceived as behaving more dominantly during the game (M - 3.72) than were those who read that submissiveness is better (M - 3.05), 1 (1,54) - 8.30, p < .007. There were no differences between conditions for the number of game pegs placed in each trial. This is probably 43 due to the fact that most subjects appeared to follow a turn-taking norm, whereby they would place two pegs and allow the confederate to place two pegs during each trial. Although confederates attempted to discourage this by placing pegs "out of order,” this practice was extremely difficult to extinguish. Two measures of behavioral dominance were taken subsequent to the confederate's feedback. The first was the subject's posture after hearing that she appeared either dominant or submissive. Subjects' posture was measured on a Likert scale ranging from 1 (leaning back) to 3 (leaning forward). A marginal two—way interaction of information (dominant desirable vs. submissive desirable) and feedback (dominant vs. submissive) was uncovered: those who received feedback that was discrepant from the information sheet (e.g., being told one is dominant after reading that submissiveness is better) were more likely to lean forward than those who were told that they possessed the trait that was better according to the information sheet, {(1,52) = 3.16, p < .08 (see Figure 1). Simple effects analyses revealed that the only statistically significant difference among the means was for those who read that submissiveness is better than dominance. Of these subjects, those who received dominant feedback were more likely to lean forward than those who received (D'IC—o-tnO'D 44 Figure 1. Subjects' Posture Subsequent to Feedback. Leenlng forward 3 1 l l - Oebnleelve better Denise-t better Leaning beck Information +0.».qu dek +0".qu dek 45 submissive feedback, 2 (1,52) - 6.60, p < .025. However, the means for those who read that dominance is better are in the expected direction, and it may be that the subjects generally valued dominance and so were comfortable with dominant feedback regardless of the information manipulation. These subjects, therefore, were not surprised by the feedback, and perhaps this is why they did not lean forward. This finding contrasts with the data of Swann and Hill (1982), who found a two-way interaction for behavioral dominance: only when given self-discrepant feedback did self-perceived dominants behave more dominantly than self- perceived submissives (see Figure 2). In the present data, however, the interaction of the first measure of behavioral dominance and feedback was not significant, E (1,52) = 0.43, NS. The second measure of behavioral dominance taken subsequent to feedback was whether subjects initiated a conversation with the confederate. This variable was measured on a four-point Likert scale. A main effect of information was discovered, such that those who read that dominance was better were more likely to initiate a conversation (M = 3.71) than were those who read that submissiveness was better (M - 3.00), F (1,56) = 5.16, p < .03. No effects of confederate feedback were uncovered for this measure (all Es < 1.00). 46 Figure 2. Behavioral Dominance Findings of Swann & Hill. 50 momma—BOO 25— ............................................................................................................................................................................................. 20. L 1 Feedback +Ielt-Pereelvee De. +IeH-Pereelvee Deb from Swarm 8 Hlll (1982) 47 am For their resistance measure, Swann and Hill (1982) had judges code responses on a scale ranging from 1 (argued strongly) to 5 (agreed strongly) and found a main effect of feedback discrepancy (see Figure 3): subjects showed higher resistance to self-discrepant feedback whether they perceived themselves as dominant or submissive. However, these data did not replicate Swann and Hill's findings. A cross-over interaction of information and feedback was instead discovered, 2 (1,55) = 2.81, p < .10. (see Figure 4). Simple effects analyses revealed that subjects showed a marginally higher level of resistance to submissive feedback if they had been led to believe that dominance is a better trait than if told that submissiveness is better. However, subjects showed a marginally higher level of resistance to feedback indicating they seemed dominant if led to believe submissiveness was more desirable than if led to believe the opposite, [(1,55) - 2.80, p < .10. There was no effect of self-perceived dominance on this measure, and so the consistency or discrepancy of the feedback from the initial self-concept did not affect these results. Although the simple effects were only marginally significant, the important finding here is that the slopes are in opposite directions, such that subjects resisted feedback when it was discrepant from the information sheet they had read, 48 Figure 3. Swarm & Hill’s Findings for Resistance to Feedback. 10 maflflnw-WON Feedback +8elt-Pereetved be. +|eH-Pereetvee Deb from Swann 8 HIII (1982) QOSN—ew—MQN 49 Figure 4. Resistance to Feedback. Oebleleelee better Dealeeet better - Information +sub-mm dek +0".qu mu 50 regardless of their initial self-conceptions. WW Swann and Hill (1982) measured changes in self- perceived dominance subsequent to feedback and found no significant difference in dominance change between those receiving consistent and discrepant feedback as long as subjects were given an opportunity to refute the feedback they received. In the present experiment, all subjects were given an opportunity to refute the feedback, but Swann and Hill's (1982) results were not replicated. Instead, three two-way interactions were uncovered using repeated-measures ANOVA. First, a significant two-way interaction of time and initial dominance level was discovered, E (1,149) - 19.81, p < .0001 (see Figure 5). Simple effects analyses revealed that self-perceived submissives came to see themselves as more dominant at the end of the experiment than before it, £(1,149) = 5.88, p < .025, whereas self-perceived dominants came to see themselves as more submissive, [(1,149) = 7.37, p < .025. This finding is probably best ascribed to regression towards the mean, for no theoretically meaningful explanation comes readily to mind. Simple effects analyses also revealed that self-perceived dominants saw themselves as more dominant than did self-perceived submissives both before, E(l,149) = 108.48, p < .001, and at the end of the experiment, E(1,149) = 27.73, p < .001. This provides an additional manipulation momma—sou -o.o com-army ...—mg) 61 Figure 5. Change in Self-Rating as a Function of Self-Perceived Dominance 20 18 16 I4 )— .......................................................................................................................................................................................... - Before After - Time —'— Submissive —*— Dominant 52 check suggesting that self-perceived dominants and self- perceived submissives were categorized correctly. A second significant interaction was of time and feedback, E (1,149) = 4.29, p < .05 (see Figure 6). Simple effects analyses revealed that those given submissive feedback were marginally more likely to see themselves as more submissive subsequently, E(1,149) - 3.29, p < .08. Those given dominant feedback saw themselves as slightly . more dominant subsequently, although this difference was not significant, E(1,149) = .62, NS. Again, the important finding here is that the slopes of those given submissive versus dominant feedback are opposite in direction, indicating self-concept change in the direction of the feedback. In addition, although there were no differences in self-perceived dominance initially between those given submissive and dominant feedback, those given dominant feedback subsequently saw themselves as more dominant than did those given submissive feedback (see Figure 6). Taken together, these findings indicate that subjects were influenced by the feedback manipulation, and those given submissive feedback even altered their self-conceptions because of it. A final marginal interaction of information valence and time was uncovered, E (1,149) = 3.27, p < .08 (see Figure 7). Simple effects analyses indicated that those who read 00393—300 -O (JD—whim fl—OCD 16 12 11 10 63 Figure 6. Change in Self-Rating as a Function of Feedback. )— .......................................................................................................................................................................................... l 1 - Before After - Time —‘— Submissive dek —+— Dominant dek momma—300 *0 Oil—#913 ”HI-0CD 16 13 12 64 Figure 7. Change in Self-Rating as a Function of Information. - Before After - Time —'— Submissive better —*— Dominant better 55 that submissiveness is better tended to rate themselves as more submissive than they had before, E(1,149) a 2.78, p < .06. Those who read that dominance is better subsequently rated themselves as slightly more dominant than they had before, although the difference was not significant, E(1,149) = 1.29, NS. In addition, no initial difference in self-perceived dominance was found between those who read that dominance is better and those who read the opposite, but subsequent to reading the information, the former rated themselves marginally more dominant than did the latter, E(1,149) = 3.38, p < .08 (see Figure 7). Again, the important finding is that the slopes are in opposite directions for those reading submissive versus dominant information. Discussion Swann and Hill (1982) found that individuals resist discrepant feedback and do not show self—attitude change if allowed to refute discrepant feedback. However, Kunda and Sanitioso (1989) found that individuals who have been led to believe that one trait is more desirable than another may change their self-conceptions to more closely match the desirable trait. Based on these studies, therefore, it was predicted that subjects would resist self-discrepant feedback only when given information indicating that their self-perceived trait was the more desirable one. When led to believe that 56 their self-perceived trait was less desirable than another, subjects were expected to show little resistance to feedback indicating that they possessed the discrepant but desirable trait. However, the present results generally found no effect for self-perceived dominance, indicating that the consistency or discrepancy of the feedback did not appear to affect responses. Instead, subjects altered their behaviors to be more in line with the trait reported to be desirable in the information sheet. This finding in particular seems compelling because although subjects could alter their verbal responses and self-reports in response to demand characteristics, it seems much less plausible that they could similarly alter their non-verbal behaviors. Subjects also resisted feedback indicating they possessed the non-desirable trait, and they even changed their self-ratings to be more in line with the desirable trait. All of these behaviors occurred regardless of the level of self-perceived dominance reported before participation in the experimental session. The results also indicate that individuals may be influenced by social feedback. Those who were told they seemed dominant subsequently rated dominance as better than did those who were told they seemed submissive. In addition, subjects changed their self-perceived dominance ratings in the direction of the feedback they received. 57 This finding supports the notion that the self-concept may be influenced by the social environment, both through social feedback and through information about the desirability of character traits. Individuals may even alter their self- conceptions based on such social information. Although no significant self-rating changes were found for those who read that dominance is better or who were told they seemed dominant, the fact that there were differences in the expected direction (i.e., they did report themselves to be slightly more dominant after readiang the information than before) for these conditions seems compelling. It may be that in our society we have a general bias towards preferring dominance to submissiveness: the present manipulation check results would seem to support this notion, since both self-perceived submissives and dominants rated dominance as slightly better than submissiveness. If this is the case, then perhaps those who read that dominance is better were only reading evidence confirming what they already believed to be true, whereas those who read that subissiveness is better were surprised and therefore more affected by the information. All subjects included in the analyses were unable to guess the purpose of the study or even that it related to the desirability of submissiveness versus dominance Those who were generally suspicious tended to guess that the true purpose of the experiment was to "see how dominant and 58 submissive people get along." Since subjects were unable to guess that the experiment was designed to examine self- conceptions or trait desirability, a demand characteristics explanation for the findings does not seem tenable. The failure to replicate Swann and Hill's findings is somewhat surprising. Perhaps it may be explained by the elimination of their instruction to "choose a leader." As explained earlier, this may have been problematic because submissive individuals may have been attempting to avoid assuming a leadership role by behaving submissively. If this were the case, than the supposedly self-confirming behaviors may actually have represented an attempt to regulate the expectations of both the confederate partner and the experimenter in the subsequent trials. It seems likely that the elimination of this instruction in the present experiment is responsible for the failure to replicate Swann and Hill (1982), since the method of the interaction-opportunity conditions was otherwise replicated. The pattern of the present results provides evidence for the hypothesis that subjects, regardless of their initial self-conceptions, are likely to alter their behaviors and self-conceptions to be more in line with a desirable trait. The findings suggest that self- enhancement opportunities take precedence over concerns with self-consistency when such opportunities are clearly available. GENERAL_DI§§!§§IQH Although the results of the first experiment were inconclusive for reasons discussed above, the results of the second experiment provide clear support for the hypothesis that individuals will self-enhance when given the opportunity to do so. When a character trait that is normally ambiguous in value is suddenly believed to be highly desirable, individuals are more likely to exhibit that trait and to change their self-conceptions in its direction. The present Experiment 2 was consistent with the findings of Kunda and Sanitioso (1989), in that both demonstrated that changing the desirability of a given character trait may alter the way an individual views him- or herself along that dimension. However, the present experiment went further. It demonstrated that social feedback as well as desirability information may be capable of producing changes in the self-concept. It also demonstrated that when given the opportunity to respond either self-enhancingly or self-consistently, individuals chose the former. Preliminary evidence was provided for the notion that individuals tend to believe that the traits they possess are more desirable than alternatives, so perhaps this explains why social feedback may alter the self-concept. Individuals who are informed that they appear to possess a given trait 59 60 may then believe that if they appear to manifest it, it must be desirable. This in turn would strengthen their belief that they possess this trait. W The present experiment provided an attempt at a crucial test between self-enhancement and self-consistency theories. Subjects were given enough information to clearly indicate whether endorsing or refuting a given trait both through responding to feedback and reporting self-ratings would represent a self-enhancing or a self-consistent response. In the present experiment, subjects clearly chose self- enhancing responses, regardless of their self-discrepancy or self-consistency. This suggests that individuals may prefer self- consistent information only when opportunities to engage in self-enhancement are not available. If told that he or she possesses a value-neutral trait (e.g., dominance, emotionality), then an individual may well choose to examine whether the information is consistent with or discrepant from his or her self-concept. However, the present results suggest that if the information is self-enhancing (i.e., the trait is favorable), then an individual is likely to accept that information without comparing it to his or her self- conception. It was expected that the effects in Experiment 2 might be smaller in magnitude for submissive individuals because 61 they may be less verbal by nature. However, very few differences as a function of self-perceived dominance emerged. Submissive individuals responded with strength equal to their more dominant counterparts. The discrepancy of the information they read or the feedback they received appeared to matter little in the presence of the enhancement opportunity that the information sheet provided. In general, then, it appears that under certain conditions the self-concept may indeed by influenced by both social feedback as well as information about the desirability of a given trait. The measures of self-rating change discussed in Experiment 2 indicate that both social feedback and information about what traits are desirable influenced the self-concepts of participants. It is striking that subjects not only attempted to manifest a desirable trait (i.e., dominance or submissiveness) in the presence of others, but actually altered their self- conceptions in the direction of the favored trait. These findings are important because they do not simply add additional support to either self-enhancement or self- consistency theory. Instead, by pitting the two theories against each other, the research subjected them to a crucial test. It was demonstrated that individuals prefer to self- enhance when allowed to do so, even if this response is not consistent with their initial self-conception. The findings strongly support the notion of a malleable as opposed to a 62 rigid self-concept, and suggest that the self may allow for change, especially if such changes are thought to be self- enhancing. It could be argued that subjects did not self-verify because they did not expect to engage in any future interactions with the confederate or the experimenter. This possibility could be examined by conducting a third experiment in which a expectation for future interaction is manipulated. Three expectancy conditions might be added, including one in which a leader is to be chosen (following the method of Swann and Hill, 1982), one in which subjects are simply told they are to play another round of Mastermind with the confederate, and one in which there are no expectations for future interactions. If the present pattern of results were maintained for the second condition, this would provide evidence supporting the notion that subjects self-enhance even when anticipating future interactions. This would also lend additional support to the suggestion that Swann and Hill's (1982) results may have resulted from submissive subjects' fear of assuming a leadership role. Wen: The implications of these findings are multiple. Perhaps the most important is in the area of clinical psychology. If, as has been suggested (Markus 8 Kunda, 1986: Markus 8 Wurf, 1987), it is true that the self-concept 63 may be changed, then this provides hope for those in a therapeutic capacity beacause a major goal in therapy is often to change the negative self-conceptions of the client. The present study provides evidence that such change may occur both through social feedback and the availability of desirability information. Future research might examine clinical populations to determine whether they are similarly able to use enhancement opportunities provided by social feedback and information. If, as might be suspected, they are not able to do so, it could be determined whether they are unable to understand the information, absorb it, or manifest it behaviorally. Understanding where this process "breaks down" might be a useful step towards learning to rebuild it. One limitation of the present study is that only short- term self-concept change was addressed. Although subjects altered their self-ratings at the end of an experimental session, these changes may have been short-lived. It would be useful to examine how long such changes persist in the course of daily life. Future research might address this limitation by increasing the length of the experiment and taking measures of self-perceived dominance across a longer period of time. Future research should also examine the possibility that dominance is not representative of other kinds of personality traits. For example, perhaps one's level of 64 dominance is not highly central to the self-concept and so changing one's self-conception of it is not very difficult. It may be that traits an individual considers highly central and important to his or her identity may be more resistant to change (Markus 8 Nurius, 1986). A wide range of personality traits should therefore be examined to address this possibility. APPENDICES APPENDIX A 65 APPENDIX A Rosenberg (1965) Self-Esteem Inventory 1. I feel that I'm a person of worth, at least on an equal basis with others. 1 2 3 4 5 strongly strongly agree disagree 2. I feel that I have a number of good qualities. 1 2 3 4 5 strongly strongly agree disagree 3. All in all, I am inclined to feel that I am a failure. 1 2 3 4 ' 5 strongly strongly agree disagree 4. I am able to do things as well as most people. 1 2 3 4 5 strongly strongly agree disagree 5. I feel I do not have much to be proud of. l 2 3 4 5 strongly strongly agree disagree 6. I take a positive attitude towards myself. 1 2 3 4 5 strongly strongly agree disagree 7. 0n the whole, I am satisfied with myself. 1 2 3 4 5 strongly strongly agree disagree 8. I wish I could have more respect for myself. 1 2 3 4 5 strongly strongly agree disagree 66 9. I certainly feel useless at times. 1 2 3 4 strongly agree 10. At times I think I am no good at all. 1 2 3 4 strongly agree 5 strongly disagree 5 strongly disagree APPENDIX B 67 APPENDIX B Self-Perceived Likability Scale 1. I often find myself becoming angry in situations I have no control over. 1 2 3 4 5 6 7 not at very all like me much like me 2. Within a social situation, I like to stand out and be seen as a unique person. 1 2 3 4 5 6 7 not at very all like me much like me 3. In activities I find myself pressured for time, not able to relax. I 2 3 4 5 6 7 not at very all like me much i like me 4. I am think I am mostly a likable person. 1 2 3 4 5 6 ‘7 not at very all like me much like me 5. If given the opportunity to do something risky, maybe even a little dangerous, I would consider doing it. 1 2 3 4 5 6 7 not at very all like me much like me 6. I hold back from criticizing people and their ideas. 1 2 3 4 5 6 7 not at very all like me much like me 7. I think I am a friendly person. 1 2 3 4 5 6 7 not at all very much 68 8. I wish I could change my physical appearance, for instance, make myself taller or shorter: change my skin color, etc. 1 2 3 4 5 6 7 hardly very ever often 9. I feel awkward meeting people because I often expect they won't like me. 1 2 3 4 5 6 7 hardly most of ever the time 10. I would rather be part of a classroom that consists of 20-30 people rather than a lecture hall with 200 or more people. 1 2 3 4 5 6 7 not at all very like me much like me 11. I like the thrill of competitive sports such as softball, basketball, etc. 1 2 3 4 5 6 7 not at all very like me much like me 12. Other people usually think I'm friendly. 1 2 3 4 5 6 7 almost almost never always 13. When I meet new people I initiate the conversation. 1 2 3 4 5 6 7 almost almost never always 14. Most people who meet me like me. 1 2 3 4 5 6 7 almost almost never always 69 15. I watch at least three hours of TV a night. 1 2 3 4 5 6 7 almost almost never always 16. I have learned from bitter experience that most people are not trustworthy. 1 2 3 4 5 6 7 not at all very like me much like me 17. I am more of a sociable person than an unsociable one. 1 2 3 4 5 6 7 very unsociable very sociable 18. I would rather participate in a sport than relaxation. 1 2 3 4 5 6 7 almost almost never always 19. I make smart, sarcastic remarks to people if I think they deserve it. 1 2 3 4 5 6 7 not at all very like me much like me 20. I think of myself as more of a cold person than a warm one. 1 2 3 4 5 6 7 very cold very warm 21. I prefer to go home on the weekends rather than stay at MSU. 1 2 3 4 5 6 7 almost almost never always 22. I feel I am more serious about my school work than most of my peers. 1 2 3 4 5 6 7 not at all very much like me like me 70 23. I think other people see me as likable. 1 2 3 4 5 6 7 not at all very much like me like me 24. I read my horoscope at least four times a week. 1 2 3 4 5 6 7 hardly almost ever always 25. I consider myself more guarded than open. 1 2 3 4 5 6 7 very very guarded open APPENDIX C 71 APPENDIX C Manipulation Checks--Experiment I 1. What four ratings did you receive? % % % % 2. How accurate do you think the first score you listed is? Please circle the number that is closest to your answer. 1 2 3 4 5 6 7 not at all extremely accurate accurate 3. How accurate do you think the second score you listed is? Please circle the number that is closest to your answer. 1 2 3 4 5 6 7 not at all extremely accurate accurate 4. How accurate do you think the third score you listed is? Please circle the number that is closest to your answer. 1 2 3 4 5 6 7 not at all extremely accurate accurate 5. How accurate do you think the fourth score you listed is? Please circle the number that is closest to your answer. 1 2 3 4 5 6 7 not at all extremely accurate accurate 6. If there is time to meet a judge, which judge did you choose do meet? Please circle a. b. one and fill in the blank. There is not enough time so I will not meet a judge. I chose to meet the judge who gave me % **If you are not going to meet a judge, please skip the remaining questions. answer them.** 7. If you are going to meet a judge, please Why did you choose to meet this judge over the others? Please turn the page and answer the following questions: 72 We'd like to know why subjects choose the judge they did so that we can make further determinations about each judge's ability: 8. How much do you expect you'll like the judge you chose to meet? 1 2 3 4 5 6 7 not at all very much 9. I thought the score I chose to hear more about was the most accurate of all of them. 1 2 3 4 5 6 7 not at all very much 10. I thought the score I chose to hear more about was the least accurate of all of them. 1 2 3 4 5 6 7 not at all very much 11. How much were each of these reasons a part of your decision to meet this judge instead of the others? Please circle the number that is closest to your answer for each one; you can rate as many reasons as you want to as 'important,' because you may have had lots of reasons. a. I wanted to hear information about this score more than the other scores. 1 2 3 4 5 6 7 Not at all extremely important important b. I wanted to change the judge's mind about my score. 1 2 3 4 5 6 7 not at all extremely important important c. I was curious about the judge who gave me this score. 1 2 3 4 5 6 7 not at all extremely important important 73 d. I chose this score randomly from the others: there was no reason why I picked this particular score. 1 2 3 4 5 6 7 not at all extremely important important e. I would like to meet the judge who gave me this score. 1 2 3 4 5 6 7 not at all extremely important important f. I like this score better than the others. 1 2 3 4 5 6 7 not at all extremely important important 12. How much thought do you think the judge put into giving you this score? 1 2 3 4 5 6 7 no thought careful at all consideration APPENDIX D 77 APPENDIX F Second Measure of Dominance Self-Ratings 1. CREATIVE 1 2 not at all like me 2. INTELLIGENT 1 2 not at all like me 3. DOMINANT 1 2 not at all like me 4. INQUISITIVE 1 2 not at all like me 5. COMMANDING 1 2 not at all like me 6. SHORT TEMPERED 1 2 not at all like me 7. ANXIOUS 1 2 not at all like me 8. SOMEONE WHO TAKES CHARGE 1 2 not at all like me very like very like very like very like very like very like very like very like much me much me much me much me much me much me much me much me 9. JEALOUS 1 not at all like me 10 . ’ WELL ORGANIZED 1 not at all like me 11. DOMINEERING 1 not at all like me 12. A GOOD SENSE OF HUMOR 1 not at all like me 13. QUICK THINKER 1 not at all like me 14. QUIET 1 not at all like me 15. FORCEFUL 1 not at all like me 2 2 2 2 78 very like very like very like very like very like very like very like much me much me much me much me much me much me much me APPENDIX E 76 APPENDIX E Desirability Information Sheet PRiNCETQN UNWERSHTY Department of Psychology Dear Colleagues. Recently my associate: and l have conducted a series of remarkable studies in our laboratory examining the personality trait known as dominance. We have uncovered irrefutable evidence suggesting that in group situations. WW R2221:- Furthermore. our results prove that W WWII“ unfazther “NW and We» in years after graduation than do submissive people. These results. although incontrovertable, were somewhat surprising. We would like to collect students' opinions from across the country about why dominance may a more desirable trait than submissiveness. Please ask your students to list three reasons why they think dominance might be a better trait than submissiveness. and send these sheets back to us. Thank you tor your assistance. Reason .9 l: ............................................... Reason #2: ............................................... Reason #3: ............................................... Sincerely yours. Dr. Aaron Spauldtng. Chair Dr. Susan Grenache Department of Psychology Department of Psychology 79 APPENDIX G Manipulation Check Items--Experiment II 1. Do you think your partner is a more dominant or a more submissive person? 1 2 3 4 5 6 7 very very submissive dominant 2. Does your partner seem to think you are a more dominant or a more submissive person? 1 2 3 4 5 6 7 thinks I'm isn't sure thinks I'm very submissive very dominant 3. Do you think you are a more dominant or a more submissive person? 1 2 3 4 5 6 7 very very submissive dominant 4. Which kind of people, those who are dominant or submissive, generally have more success later in life? 1 2 3 4 5 6 7 submissive neither are dominant more successful more successful more successful than dam. than the other than submiss. 5. a. Did you read an information sheet duscussing the relationship between dominance/submissiveness and future success? Yes No b. If yes, which trait was reported to be the better one? Please mark an X in the appropriate box: dominance better unsure submissiveness better 80 c. How much do you agree with the information sheet? 1 2 3 4 5 6 7 not at all somewhat very much 6. Do you think it's better to be dominant or submissive? 1 2 3 4 5 6 7 submissive neither dominant is better is better is better 7. What do you think the purpose of this study is? 8. Can you think of any additional purposes? APPENDIX H APPENDIX H Videotape Coding Form Subject # Experimenter Confederate W # pegs placed on trial 1 # pegs placed on trial 2 # pegs placed on trial 3 # pegs placed on trial 4 # pegs placed on trial 5 # pegs placed on trial 6 # pegs placed on trial 7 # pegs placed on trial 8 # pegs placed on trial 9 # pegs placed on trial 10 total # trials ratio place/trial # verbal suggestions trial 1 # verbal sugg. trial 2 # verbal suggestions trial 3 # verbal sugg. trial 4 # verbal suggestions trial 5 # verbal sugg. trial 6 # verbal suggestions trial 7 # verbal sugg. trial 8 # verbal suggestions trial 9 # verbal sugg. trial 10 total # trials ratio suggestions/trial W Subject's dominance during the first three minutes: 1 2 3 4 very somewhat neither somewhat very submiss. submiss. or both dominant dominant Subject's dominance during the second three minutes of the game: 1 2 3 4 very somewhat neither somewhat very submiss. submiss. or both dominant dominant Subject's dominance during the third three minutes of the game: 1 2 3 4 very somewhat neither somewhat very submiss. submiss. or both dominant dominant 81 5 5 5 82 Subject's dominance during the fourth three minutes of the game: 1 2 3 4 5 very somewhat neither somewhat very submiss. submiss. or both dominant dominant Subject's dominance during the fifth three minutes of the game: 1 2 3 4 5 very somewhat neither somewhat very submiss. submiss. or both dominant dominant W Subject's overall dominance during the game: total above scores total # minutes scores/minutes Subject's most prevalent posture during the game: 1 2 3 leaning back neither leaning forward Did the subject and confed. solve it within ten tries? Y N Winn What feedback was the subject given? Dominant Submissive How did the subject respond to the feedback? 1 2 3 4 5 agreed ' agreed no argued argued strongly somewhat reaction somewhat strongly How many statements did the subject make in response? What was the subject's posture subsequent to feedback? 1 2 3 leaning back neither leaning forward Did subject initiate a conversation after feedback? Y N LIST OF REFERENCES LIST OF REFERENCES Arkin, R.M., 8 Baumgardner, A.H. (1986). Self- presentation and self-evaluation: Processes of self- control and social control. In R. Baumeister (Ed.), Eublis_§elf_and_£rixate_§elf. (pp- 75-97)- New York: Springer-Verlag. Baumgardner, A.H. (1988). To know oneself is to like oneself: Self-certainty and self-affect. Manuscript under review, Michigan State University. Baumgardner, A.H., 8 Arkin, R.M. (1988). Affective state mediates causal attribution for success and failure. hatixatian_and_8m9tian. 12. 99-111- Baumgardner, A.H., 8 Arkin, R.M. (1987). Coping with the prospect of social disapproval: Strategies and sequelae. In C.R. Snyder 8 C. Ford (Eds.), gliniggl '._!e__ :e . e; 1e ee . e- _._e‘ L; e, .;eg : - gygngg. .New York: Plenum. Baumgardner, A.H., 8 Brownlee, E.A. (1987). Strategic failure in social interaction: Evidence for expectancy disconfirmation processes. quzna1_gfi 2srs2nalitx.and.§ccial_£sxchalesx. 52. 525-535- Baumgardner, A.H., Heppner, P.P., 8 Arkin, R.M. (1986). Role of causal attribution in personal problem solving- Iaurnal_2f_2ersanalitx_and_fiacial_2§xcbalggx. 5g, 636-643. Baumgardner, A.H., Kaufman, C.M., 8 Levy, P.E. (1989). Regulating affect interpersonally: When low esteem leads to greater enhancement. laurnal_2f Earsanalitx_and_§acial_£§xchclegx. 56. 907-921- Baumgardner, A.H., Lake, E., 8 Arkin, R.M. (1985). Claiming mood as a self-handicap: The influence of spoiled and unspoiled public identities. Eggsgn§11;y_§nd Sacial_£§xchglegx_§ulletin. 11. 349-357- 83 84 Bradley, G.W. (1978). Self-serving biases in the attribution process: A reexamination of the fact or fiction question. Iaurnal_af_£ersanalitx_and_fiocial Psychclegx. if. 56-71- Brown, J.D., Collins, R.L., 8 Schmidt, G.W. (1988). Self-esteem and direct versus indirect forms of self- enhancenent- Iaurnal__f_Eer§Qnalitx_and_§ccial Psychal_gx.. 55. 445-453- Crocker, L., 8 Algina, J. (1986). Intrgguggign_tg_ Classical_and_hndern_Test_Thearx- New York: Holt. Rinehart, 8 Winston. Darley, J.M., 8 Fazio, R. (1980). Expectancy confirmation processes arising in the social interaction sequence- American_2§xchalogist. 35. 867-881- Epstein, S. (1973). The self-concept revisited. American_zsxshologi§t..28. 404-416. Festinger, L. (1954). A theory of social comparison processes. Human_3g1agign§, 1, 117-120. Frey, D. (1986). Recent research on selective exposure to information. Adxances.in.£xnerinental.§ocial £§YQDQIQQY..12. 41-80- Fuhrman, R.W. (1989). Effects of information accessibility and trait importance on self-judgements. Unpublishled manuscript, University of Illinois at Urbana-Champaign. Ickes, W., 8 Layden, M.A. (1978). Attributional styles. In J. Harvey, W. Ickes, 8 R. Kidd (Eds.), Mg! ° ' V . Hillsdale, NJ: Erlbaum. Jones, S.C. (1973). Self and interpersonal evaluation: Esteem theories versus consistency theories. Esychglogical Eglletig, 12, 185-199. Jones, E.E., Rhodewalt, F., Berglas, S., 8 Skelton, J.A. (1981). Effects of strategic self-presentations on subsequent self-esteem. 1aurnal_af_2ersonalitx_and Social_£§xcbolch. 11. 407-421- Kunda, 2., 8 Sanitioso, R. (1989). Motivated changes in the self-concept. Inurnal.af_£xnerimental social_£§29halcsx. 25. 272-285. 85 Markus, H., 8 Kunda, z. (1986). Stability and malleability of the self-concept. figurng1_gfi_2g;§gngligy and_§acial_zsxchalosx. 51. 858-866. Markus, H., 8 Nurius, P. (1986). Possible selves. Americsn_zsxchologist..11. 954-969- Markus, H., 8 Wurf, E. (1987). The dynamic self- concept: A social psychological perspective. Amgzigan Bexiex_2f_£§xcholngx. 38. 299-337- Mills, J. (1976). A procedure for explaining experiments involving deception. Egzfigngli;y_§ng_§ggigl Psychologx_nulletin. 2. 3-13- Nelson, R.E., 8 Craighead, W.E. (1977). Selective recall of positive and negative feedback, self-control behaviors, and depression. igurna1_gf_5pngzmgl EEXQDQLQQY: ER. 379-338- Regan, J.W. (1976). Liking for evaluators: Consistency and self-esteem theories. Journal_af Exuerimental_and_§ocial_zsxchnlogx. 12. 159-169- Shrauger, J.S., 8 Schoeneman, T.J. (1979). Symbolic interactionist view of self-concept: Through the looking glass darkly- Psychological_hnlletin. as. 549-573. Snyder, M. (1984). When belief creates reality. e , 13, 247-305. Swann, W.B., Jr. (1987). Identity negotiation: Where two roads meet. I2urnal_af_£erscnalitx_and_ficcial Perchalosx. 53- Swann, W.B., Jr., 8 Ely, R.J. (1984). A battle of wills: Self-verification versus behavioral confirmation. I2urnal_2f.£er§2nalitx_and_ficcial EEYQDQIQQY: 15: 1237-1302- Swann, W.B., Jr., Griffin, J.J., Jr., Predmore, S.C., 8 Gaines, B. (1987). The cognitive-affective crossfire: When self-consistency confronts self- enhancement- 12urnal_af_£er§2nalitx_and_fiacial 2512821292: 52. 381-839- Swann, W.B., Jr., 8 Hill, C.A. (1982). When our identities are mistaken: Reaffirming self-conceptions through social interaction. Social_zsxchalcax. 13. 59-66- 86 Swann, W.B., Jr., 8 Read, S.J. (1981a). Acquiring self-knowledge: The search for feedback that fits. Ianrnal_af_2ersanalitx_nnd_§221al_2sxchcleax. 11. 1119- 1128. swann’ WeBe' Jre' & Read, SeJe (1981b) e self- verification processes: How we sustain our self- conceptions- 12urnal_af_Exnerimental_§ncial_2§xshclogx. 11, 351-372. Taylor, S.E., 8 Brown, J.D. (1988). Illusion and well-being: A social psychological perspective on mental health. Psychological_hulletin. 121. 193-210- Tesser, A., Campbell, J., 8 Smith, M. (1984). Friendship choice and performance: Self-evaluation maintenance in children. i2urnal_af_Ber§2nalitx_and 5991§1_£§¥£h91291. 11. 672-682- Wyer, R.S., Jr., 8 Frey, D. (1983). The effects of feedback about self and others on the recall and judgments of feedback-relevant information. Qggzngl_gfi Exnerimental.§acinl_£sxcholch. 12. 540-559. Wylie, R. (1979). Thg_§g1f_ggngep;. Lincoln: University of Nebraska Press.