“6 1a..) nfflw; vi: : . i . .A $5.... ,. , . V ‘ . uni?” ..mV_:wu:Wmm 5.361%” . , .v , . . .4 L33.....vb31. ”HES"! llllllll/llllllllllllllllllWillllllllllllllllll 015 572 2055 LIBRARY Michigan State University This is to certify that the thesis entitled Emotional Responses to the Success of Out-Group Members presented by Ruth Ellen Fleury has been accepted towards fulfillment of the requirements for Masada—degree in ism mpg/rem Date W 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution PLACE IN RETURN BOX to remove this checkout from your record. :I’O AVOID FINES mum on or before date duo. DATE DUE DATE DUE DATE DUE .‘5 i m MSUIOAn ‘"‘ ' ‘ " 1 ' r, EMOTIONAL RESPONSES TO THE SUCCESS 0F OUT-GROUP MEMBERS By Ruth Ellen Fleury A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1996 ABSTRACT EMOTIONAL RESPONSES TO THE SUCCESS OF OUT-GROUP MEMBERS By Ruth Ellen Fleury Previous research has looked at the relationship between cognition and emotion in prejudice, as well as at personal standards for responding to out-group members (Devine, Monteith, Zuwerink, & Elliot, 1991). The current study examined emotional responses and discrepancies in emotional responses to the successful outcome of an out-group member, when affirmative action was or was not a possible cause of that outcome. It was hypothesized that individual differences in prejudice and different types of affirmative action policies would interact to affect actual emotion and emotion discrepancies. It was found that the presence of an affirmative action policy tended to yield larger discrepancies between how participants felt and how they thought they ought to feel. Additionally, results suggested that participants may associate race more than sex with affirmative action. h-‘tt 3' (I! .IY '.‘. 3.1.. . 3‘30 liia. . ":16 .: 90. e v A v-\ .C .9. i.” r\ r . . .D A. ‘ v ». ..\. l. .L .\ in a. . T l. E to J flu an .0 . r t. .u re . mm. . W ., . m . M .. ”a . .. _.m . 2.... . y -_ ... In t 5A... as t V . , Donn: ACKNOWLEDGMENTS First, and foremost, I am indebted to my advisor, Linda A. Jackson, for all of her support and all of her feedback on the innumerable drafts of this thesis. I would also like to thank my committee members Lawrence Messé and Galen Bodenhausen for all of their valuable suggestions. Only someone who has been there can really understand this process. For this reason, I am so grateful to my companion SPIG students: Carole, Kris, Keith, Susan, Dale, Donna, Mike, Shira, and especially Jen for all of their support throughout this process. TABLE OF CONTENTS List of Tables ........................................................ vii Introduction ........................................................... 1 Group Membership, Ability, and Attributions ............................ 2 Emotional Responses to Out-Group Members ........................... 6 Discrepancies in Responses to Out-Group Members ...................... 8 Affirmative Action ................................................ 10 Individual Differences in Racism and Sexism ........................... 16 The Current Research .............................................. 18 Hypotheses ...................................................... 20 Evaluation Extremity ........................................ 20 Actual Emotional Responses .................................. 21 Emotional Discrepancies ..................................... 21 Method .............................................................. 22 Participants ...................................................... 22 Overview of Design ............................................... 22 Measures ....................................................... 23 Academic Ability ........................................... 23 Emotional Responses and Discrepancies ......................... 23 Attitudes Toward African Americans ........................... 25 Attitudes Toward Women .................................... 26 Demographics ............................................. 26 Familiarity with Afiirmative Action ............................ 26 Materials ....................................................... 27 Procedure ....................................................... 29 Results ............................................................... 29 Evaluation Extremity .............................................. 29 Actual Emotional Responses ........................................ 34 Discrepancies in Emotions .......................................... 39 Undefined Policy Definitions ....................................... 42 Additional Influences on Target Qualifications Ratings ................... 43 Additional Influences on Actual Emotions ............................. 43 Additional Influences on Emotion-Discrepancies ........................ 49 Ought Emotions .................................................. 53 Discussion ............................................................ 55 Qualifications Ratings ............................................. 55 Actual Emotional Responses ........................................ 57 Discrepancies in Emotion .......................................... 59 Directions for Future Research ...................................... 63 Conclusions ..................................................... 64 Appendices ........................................................... 66 Appendix A: Target Qualifications Rating Scale ........................ 66 Appendix B: Actual and Ought Emotions Rating Scales .................. 67 Appendix C: Modern Racism Scale (MRS) ............................. 69 Appendix D: Modern Sexism Scale (MSS) ............................. 70 Appendix E: Demographics Measure ................................. 72 Appendix F: Unfair Advantage Policy Scholarship ....................... 73 Appendix G: Qualifications First Policy Scholarship ..................... 74 Appendix H: Undefined Policy Scholarship ............................ 75 Appendix I: No Policy Scholarship ................................... 77 List of References ...................................................... 78 vi LIST OF TABLES Table 1: Male participants' mean qualifications ratings by policy and target ......... 32 Table 2: Female participants' mean qualifications ratings by policy and target ....... 33 Table 3: Mean actual emotion for African American targets by racism and policy . . . . 36 Table 4: Males' mean actual emotion for female targets by policy and sexism ........ 38 Table 5: Mean emotion discrepancies for African American targets by policy and racism ............................................................... 40 Table 6: Males' mean emotion discrepancies for female targets by policy and sexism . . 41 Table 7: Males' mean emotion scores by policy and racism ...................... 45 Table 8: Females' mean emotion scores by policy and racism .................... 46 Table 9: Mean actual emotion ratings by target race and participant racism .......... 48 Table 10: Participants' mean emotion discrepancies by policy and target ............ 50 Table 11: Mean emotion discrepancies by participant sex and participant sexism ..... 52 Table 12: Mean ought emotion ratings by target race and participant racism ......... 54 Introduction “Is it reasonable to start a race between two runners, one of whose legs are shackled at the beginning of the race and unshackled halfway into the contest and declare a fair ending without adjusting for the handicap?” (Taylor, 1989). In the spring of 1965, President Lyndon Johnson asked this question of the audience at the commencement at Howard University. Later that year, Johnson signed an executive order requiring federal contractors to “take affirmative action to ensure that applicants are employed, and that employees are treated during employment, without regard to their race, color, religion, sex, or national origin” (Holloway, 1989). In the three decades since Johnson issued the executive order, affirmative action policies have been controversial. These policies have been conceptualized both as ways to overcome past discrimination by helping qualified members of disadvantaged groups and as reverse discrimination, or discrimination against Whites and/or males (Crosby & Blanchard, 1989). Misperceptions of affirmative action abound; many people believe that affirmative action policies require quotas and overlook qualifications (Kravitz & Platania, 1993). On the other side, proponents of affirrnative action have accused opponents of both racism and sexism (Cohn, Turque, & Brant, 1995; Fineman, 1995). In the popular press, affirmative action has been portrayed as an emotional issue, involving anger on the parts of both proponents and opponents (Cohn et al., 1995; Fineman, 1995). Researchers in psychology have examined emotional responses to affirmative action both from the perspective of the intended beneficiaries and the majority group members (Arthur, Doverspike, & Fuentes, 1992; Kinder & Sanders, 1990). A great 2 deal of empirical research has focused on emotional processes in prejudice and discrimination more generally, (e.g. Allport, 1954; Devine, Monteith, Zuwerink, & Elliot, 1991; Jackson & Sullivan, 1988; Monteith, Devine, & Zuwerink, 1993). Some of these researchers have also looked at the role of personal standards in responding to out-group members and the emotional consequences of responding in ways which do not match these standards (Allport, 1954; Devine et al., 1991; Monteith et al., 1993). However, the possible situational influences on the magnitude of the discrepancies between actual emotion responses and personal standards for responding have received little attention. The current study investigated emotion responses and discrepancies in emotion responses to the successful outcome of an out-group member, when affirmative action was or was not a possible cause of that outcome. In this study, several factors were expected to influence both actual emotion responses and personal standards for responding to an out-group member’s success. Participants’ perceptions of the target’s ability, of the causes of the target’s success, and individual differences in prejudice, all were expected to affect how individuals felt about the success of an out-group target, and how they thought they ought to feel about that success. Group Membership, Ability, and Attributions Researchers have found that people judge others in part by their group membership, whether or not group membership is relevant to the judgment. Dovidio and Gaertner (1981) found that White participants viewed the cognitive abilities of Whites and African Americans differently. In a study which manipulated perceptions of a 3 partner’s status and ability, they included a relative rating of intelligence. Participants thought that White partners who had been labeled as high-ability were somewhat more intelligent than themselves. However, White participants still rated high-ability African American partners as less intelligent than themselves. Dovidio and Gaertner (1983) found evidence that people also view the abilities of males and females differently. Using the same measure of relative intelligence, they examined how participants rated their own intelligence relative to a female or a male partner. While participants rated themselves as more intelligent, overall, than their partners, they rated themselves as more intelligent than a female partner compared to a male partner. There was no effect for participant sex; both female and male participants demonstrated the same pattern. However, not all researchers have found that women and members of racial minorities are evaluated less favorably than males and/or Whites, respectively. Linville and Jones (1980) found that White participants evaluated a strong law school candidate more favorably when they believed the candidate was African American than when they believed the candidate was White. Consistent with this finding, Jackson, Sullivan, and Hodge (1993) found that African American college applicants with strong credentials were evaluated more favorably than White candidates with the same credentials. Additionally, Linville and Jones found a similar effect for target sex. Among White applicants, White participants rated other-sex individuals as more qualified than same-sex individuals. Linville and Jones (1980) explained their findings in terms of cognitive 4 complexity. People have more complex schema for the in-group than the out-group. Because the schemas for out-group members are less complex, extreme information, either positive or negative, is likely to have a large effect on evaluations. For in-group members, the same information has less of an effect because there is more information available in the schema. In the current study, qualified out-group members are expected to be rated as more qualified than equally qualified in-group members due to differences in cognitive complexity for in-groups and out-groups. Sex and race, or group membership more generally, of a target can also affect the attributions made for the behavior and for the success or failure of the target. According to Weiner (1986), attributions for an event or outcome vary on three dimensions: locus (intemal/extemal), controllability, and stability. The first dimension, locus, varies depending on whether the behavior or outcome is perceived as due to something about the person (internal) or something about the situation (external). Attributions about controllability depend on whether or not the event is seen as under the control of the actor. An attribution to stable causes will be made if the event is seen as something which is stable across time, while an attribution to unstable causes will be made if the event is seen as transient. Islam and Hewstone (1993) found that group membership of both the participant and a target affected attributions for the target’s behavior. In this study, religion, Muslim or Hindu, was the basis for group membership. Participants were asked to imagine an individual of the same or a different religion behaving either positively or negatively toward them. Participants tended to make more in-group than out-group favoring 5 attributions. For instance, participants made more stable attributions when an in-group member behaved positively toward the target than when an in-group member behaved negatively. Additionally, participants tended to rate negative behaviors by in-group members as less controllable than the same negative behaviors by out-group members. In this study, group membership of a target led participants to explain the same outcome differently. Heilman and Herlihy (1984) studied the effects of group membership on attributions for success. They asked participants about their interests in an occupation when an increase in the number of women holding that position was due to the number of qualified women, to preferential treatment given to women, or when no information was given. The results for male, but not female, participants in the no information condition mirrored those for the preferential treatment condition. The authors speculated that "without evidence to the contrary males assumed a non-merit-based placement of women when they appeared in significant numbers" in the occupational description. The male respondents, then, seemed to be making external attributions for the women’s employment. In one survey of employed White men (Heilman, Block, and Lucas, 1992), participants were asked to describe a co-worker who had been hired recently and who was a member of a group that did not typically have that kind of job (e.g. a woman in a typically male position). Data were analyzed only for those participants who named a co- worker who was a member of a group protected under affirmative action. They found that the group membership of the co-worker did affect perceptions of the degree to which 6 qualifications and affirmative action were believed to have influenced the hiring decision. The attributions made about the reasons for the co-worker's employment were related to perceptions of the co-worker's competence. The more a co-worker was perceived as being hired due to affirmative action, the less qualified she or he was thought to be. As the authors pointed out, however, these differences are difficult to interpret due to the correlational nature of the study. Other research has explicitly examined the effects of an affirmative action policy on evaluations of racial minority group members and women. Garcia et al. (1981) found that participants rated the qualifications of minority applicants to a graduate program as lower when the program was described as having an affirmative action policy than when it was not. The mere presence of an affirmative action policy affected evaluations; this research did not say that the minority group member was accepted because of the policy. Group membership of a target, then, can affect how that target is evaluated and the attributions made for positive and negative outcomes of that target. These attributions in turn can affect how individuals feel about the target’s outcome. Emotional Responses to Out-Group Members Attributions for one’s own behavior and the behavior of others can have emotional consequences. According to Weiner (1986) emotions following a success or failure sometimes depend on the attributions made for that outcome. For instance, following a success, an actor will probably feel happiness, regardless of the attributions for that success. The happiness, then, is outcome dependent, but attribution independent. If the actor believes the success was due to effort or ability, the actor may also feel pride. 7 If, however, the actor believes the success was due mainly to luck, the actor is likely to feel surprise. Pride and surprise, in this case, are both outcome dependent and attribution dependent. lslarn and Hewstone (1993) found a strong relationship between causal attribution and the resulting affect for both in-group and out-group actors. Attributions about the cause of an actor's behavior toward the participant predicted either positive or negative emotions of the participant, following positive or negative outcomes respectively. For instance, when the actor was an in-group member, attributing the actor's positive behavior to an internal cause led to greater feelings of happiness and pride than attributing the actor's behavior to an external cause. The way participants attributed an actor's behavior affected their emotional responses to the behavior. Prejudice against a social group also has an emotional component. Gaertner and Dovidio (1986) differentiate dominative racists, or people who openly express and act on their racist beliefs, from aversive racists, who regard themselves as nonprejudiced, but who also possess negative feelings about African Americans. This negative emotion is not hatred, but rather discomfort, disgust, and sometimes fear. Cognitions about a social group and emotions about that group are ofien inconsistent. Jackson and Sullivan (1988) examined the role of cognition and emotion in evaluations of a stereotyped group member (a male homosexual). They found that the negative emotional responses to the stereotyped target better predicted evaluations than did cognitions about the target's group. In related research, Stangor, Sullivan, and Ford (1991) examined both affective and cognitive components of attitudes toward several 8 different social groups, including those with positive stereotypes (e.g. Americans) and negative stereotypes (e.g. homosexuals). They found that affect, both positive and negative, was a more consistent predictor of attitudes toward the target group than were cognitions. In general, then, group membership affects evaluations of and attributions about out-group members. Different attributions about in and out-group members can lead to difi‘erent emotional responses. In addition, people tend to have emotional reactions to in- groups and out-groups. However, how individuals actually respond to out-groups and how they think they ought to respond are not necessarily the same. Discrepancies in Responses to Out-Group Members People ofien experience discrepancies between how they believe they should react and how they think they actually would react in certain situations. Allport (1954) described contradictions within individual attitudes toward out-group members. Often, these conuadictions are between cognitions about a group and emotional responses to the group. People may have cognitions about a group which are less prejudiced than their emotional responses to the group. As Allport phrased it, “Defeated intellectually, prejudice lingers emotionally” (p. 328). These discrepancies also have emotional consequences. In one study of prejudice, White participants were presented with several hypothetical situations involving contact with people of color. They were asked to indicate both how they thought they should react and how they thought they would react (Devine, Monteith, Zuwerink, & Elliot, 1991). Low prejudiced individuals who indicated should-would discrepancies felt 9 compunction. That is, they felt shame and guilt when they thought they might do or feel something prejudiced. High prejudiced participants, on the other hand, felt some global discomfort when they thought they might discriminate against others. They did not, however, feel as much guilt or self-criticism as the low prejudiced participants. Devine et al. also found that those participants in the middle of the prejudice distribution were slightly more likely to experience these discrepancies. That is, both high and low prejudiced participants were more likely to indicate that they would act or feel as they should than moderately prejudiced participants. Participants moderate in prejudice were less likely to indicate that they actually would react in accordance with their personal standards. This research by Devine et al. (1991) used a fairly explicit notion of what is prejudiced; the nonprejudiced response was quite clear. For instance, one situation involved participants imagining feeling uncomfortable that an African American was sitting next to them on a bus. In the real world, not all situations have such clear-cut standards for non prejudiced responses. As other researchers have demonstrated, people may, consciously or not, use situational ambiguity as an "excuse" to discriminate. For instance, in one study Gaertner (1973; cf. Gaertner & Dovidio, 1986) had either African American or White targets call a "wrong" number and ask the respondent to make a phone call because the target had no more change with which to reach the correct number. Liberal respondents were equally likely to make the additional phone call for African American and White callers. However, they were also more likely to hang up on African American callers than White callers before the caller had a chance to 10 explain the problem. The standards for helping behavior in this situation are clear once the situation has been explained; one should make the phone call for the target. However, the standards for responding to a wrong number, that is, how soon one should hang up, are less clear. When the standards for non-prejudiced behavior were less clear, then participants tended to discriminate. The current study used affirmative action as a potential “excuse” to change standards for non-prejudiced responses. Afirmative Action Affirmative action policies have caused a great deal of controversy. The prevalence of these policies, and the controversy and misunderstanding that often accompanies them, provide an opportunity to examine them as a potential source of attribution for the success and failure of a target. Additionally, the social norms for responding to an individual’s success when affirmative action is a potential explanation for the success are not as clear as the norms for responding when the success is likely due to internal factors. Under Title VII of the Civil Rights Act of 1964 it is illegal for an employer to discriminate against people because of their race or sex in hiring, placement, or promotion. Employers who have been found in violation of Title VII may be required to set up affirmative action goals in hiring and promotion. In other words, companies may be required to set up programs to recruit and train more members of a previously disadvantaged group. Some employers, in the absence of violations of Title VII, have set up such programs on their own (Bergmann, 1986). Some universities have also set up affirmative action programs to recruit and retain more students, faculty, and staff from ‘ minority groups (Newman, 1989). According to the manual of the US. Department of Labor's Office of Federal Contract Compliance Programs (OFCCP), affirmative action is "those results oriented actions which a contractor by virtue of its contracts must take to ensure equal employment opportunity. Where appropriate, it includes goals to correct underutilization, correction of problem areas, etc." (Holloway, 1989). Two groups have been identified as target groups, or groups these programs are designed to help: women and members of ethnic minority groups. Specifically, the government defines minority groups as American Indians or Alaskan Natives, Asian or Pacific Islanders, Blacks not of Hispanic origin, and Hispanics. In the 1970's, two additional groups were added: persons with disabilities and Vietnam-era veterans (Holloway, 1989). Affirmative action, then, was originally conceived as a way to guarantee equal opportunity to people who might otherwise face discrimination. One survey (Klugel, 1985) reported that although a majority of Whites believed there was at least some discrimination against African Americans that limit their chances of getting ahead, a majority also believed that African Americans have an average or better than average chance to get ahead. These apparently conflicting perceptions are probably influenced by a perception (also held by more than half the respondents) that there is at least some "preferential treatment for minorities...that improves their chances to get ahead." This preferential treatment many people perceive may sometimes be in the form of affirmative action policies. One common misperception of affirmative action policies is that unqualified 12 members of protected groups are favored over more qualified majority group members (e.g. Kravitz & Platania, 1993). Kravitz and Platania (1993) found that participants rated hiring and promotion of a certain number of people in a group and quota hiring as probable components of an affirmative action plan. In other words, participants expected that aflinnative action plans generally involve benefits based on membership in a certain group (e.g. females, African Americans) even if those group members were not as qualified as majority group members. As the authors point out, however, affirmative action plans cannot legally involve strict quotas or force an organization to hire or promote unqualified individuals. Membership in a protected group may only be considered "after competence is used for an initial screening" (Kravitz & Platania, 1993). A related misperception is that there is no discrimination against people of color and/or women in the first place. As one researcher described this situation, "if there isn't a problem, you don't need a solution" (Klugel, 1985). Affirmative action, then, becomes a policy that at best does nothing and, at worst, hurts both the (White male) people it "discriminates" against (Kravitz & Platania, 1993), as well as intended beneficiary groups (Arthur et al., 1992; Garcia, Erskine, Hawn, & Casmay, 1981; Heilman & Herlihy, 1984). Members of intended beneficiary groups may be seen by others as having a particular position or reward because of group membership rather than objective qualifications. Moreover, members of intended beneficiary groups may also come to see their achievements as a result of group membership rather than ability. That is, both majority and minority groups may come to attribute success to an external rather than an internal source. 13 Given the common view of affirmative action as reward distribution based on group membership rather than objective qualifications and popular support for equal opportunity for all Americans, different sorts of affirmative action policies should produce different emotional responses to an affirmative action decision. Kinder and Sanders (1990) describe "framing" as the way parts of an issue are emphasized or downplayed; in other words, Wing is the context around the issue. Some studies have shown that the way that affirmative action policies are flamed impacts reactions to decisions based on the policies, but these results are not consistent across studies (Fine, 1992; Kinder & Sanders, 1990). Fine (1992) presented participants with frames of afi'rrmative action that emphasized the opposition in terms of either discriminating against Whites or giving African Americans advantages that they had not earned. White participants were more likely to be strongly opposed to affirmative action when it was presented as reverse discrimination than when it was presented as giving African Americans an unfair advantage. Fine suggests that Whites were reacting as potential victims of discrimination in the reverse discrimination condition, but not in the unfair advantage condition; that is, they were reacting against a policy that could hurt them in contrast to a policy that would unfairly help someone else. Kinder and Sanders (1990) also studied reactions to affirmative action when it was framed to emphasize either reverse discrimination (discrimination against Whites) or an unfair advantage that African Americans have not earned. Unlike Fine's (1992) findings, opinions were not significantly afiected by the frames; people opposed it nearly l4 equally no matter how it was described. The reason for these slightly different findings is unclear; both studies used the exact same flames, and both used data flom the National Election Survey (flom 1984 and 1986, respectively). Kinder and Sanders did use a smaller sample size; the pattern of means Kinder and Sanders found is consistent with Fine’s results. In their study of afl'rrmative action flaming, Kinder and Sanders (1990) also measured emotional responses to different affirmative action policies. Framing affirmative action as an unfair advantage that had not been earned yielded negative emotions like anger, disgust, and infirriation more than did flaming it as a policy that discriminates against Whites. They also found that positive emotions like pride, hope, and sympathy were not affected by the flaming. This last finding is not surprising given that in both flames the justification for affirmative action was the same; only the reason against it changed across flames. They examined reactions to the policies themselves, rather than the outcomes of the policies or any individuals involved. The characteristics of the intended beneficiary groups may also affect emotional responses to an affirmative action decision. For instance, Eberhardt and Fiske (1994) argue that affirmative action for Aflican Americans may been seen as less fair than affirmative action for White women because of the differences in stereotypes between these groups. They argue that the stereotypes of White women tend to be more complex than those of African Americans; Whites may see women as less similar to each other than Aflican Americans are to each other. Therefore, fewer White women (only "career" women) than Aflican Americans (all African Americans) may be seen as potential beneficiaries of affirmative action. Clayton (1992) found that affirmative action directed toward people of color was perceived more negatively than affirmative action efforts directed toward women. Participants thought that their chances of being treated fairly by an employer who was trying to recruit more women were higher than their chances of being treated fairly by an employer that was trying to recruit more minorities. An employer described as having a large percentage of minority managers but not described as trying to recruit minorities was perceived as fair, as was an employer actively flying to recruit women. In their survey of employed White men, Heilman et al. (1992) found that the perceived qualifications of a co-worker were inversely related to the extent to which affirmative action was seen as a cause of the co-worker’s employment. They also found that this relationship differed with characteristics of the co-worker. This inverse relationship between affirmative action status and qualifications was stronger for African American women than for African American men. Although Heilman et al. did not measure the emotional reactions to these co-workers, it would not be unreasonable to think that since Aflican American women were perceived differently than African American men, participants may also have felt differently about working with African American women than African American men. In a study of psychology faculty members, Arnirkhan, Betancourt, Graham, Lopez, and Weiner (1995) also found differences in admissibility ratings (to a graduate program) for members of different minority ethnic groups. Academic background was the most important criteria in decision making. However, African Americans received 16 higher admissibility ratings than did Latino Americans and Asian Americans. The authors interpret this difference to mean that the affirmative action may be seen as a way to compensate for past injustice; Latino Americans and Asian Americans may be seen as less in need of compensation for historical discrimination. In contrast to those studies which found differences in attitudes toward affirmative action depending on the recipients' characteristics, Kravitz and Platania (1993) did not find significant differences between attitudes toward affirmative action as a function of sex or race of the recipients. Participants were equally opposed to the programs whether they were designed to help women or to help people of color. Personal standards, then, for emotional responses to affirmative action decisions may or may not be affected by the group membership of the beneficiary. Individual Differences in Racism and Sexism Individual differences in beliefs and attitudes should influence emotional responses to the same affirmative action policy. One of these differences is a person's level of racism. McConahay's (1986) concept of modern racism is different flom more traditional racism in several ways. First, those who endorse modern racism do not consider themselves racists. They believe that discrimination against Aflican Americans no longer exists; African Americans are flee to compete with Whites. Second, modern racists think that African Americans are pushing themselves too far too fast where they are not wanted. They believe that recent gains made by Aflican Americans are not due to merit, so Aflican Americans have gotten more than they deserve. From this definition of modern racism, it is not hard to predict that those who endorse the tenants of modem 3 17 racism would tend to oppose affirmative action policies that benefit African Americans. If Aflican Americans are not discriminated against, then there is no need for programs that help African Americans achieve equality. What is less clear is how those who do not endorse the modern racism items should feel about affirmative action to help African Americans or other minority group members. Even if one is aware of discrimination against Aflican Americans, it does not automatically follow that one would support affirmative action as a means to end discrimination. Gaertner and Dovidio's (1986) concept of aversive racism explains how Whites can discriminate against Aflican Americans and still maintain a nonprejudiced self- image. Where nonracial characteristics are salient, discrimination is often easier. That is, if the discriminatory behavior can be attributed to some situational characteristic besides race, such as an affirmative action policy, then even Whites who consider themselves egalitarians might discriminate. From this, is would be reasonable to expect that if aversive racists can justify their discrimination or their negative reactions to an Aflican American, they might also experience smaller discrepancies between how they do feel and how they think they ought to feel. A second individual difference factor which may influence emotional responses to affirmative action is a person's level of sexism. Swim, Aiken, Hall, and Hunter (1995) proposed that sexism, like racism, has become more subtle over time. Using concepts similar to McConahay's (1986) modem racism, Swim et al. (1995) developed a concept of modern sexism, which differs from old-fashioned sexism. Modern sexism includes a denial of continuing discrimination against women, and antagonism toward the demands 18 of women. Modern sexism also includes resentment of special favors to women. These three components of modem sexism clearly relate to emotional responses to affirmative action for women. People who deny there is discrimination against women, who do not understand the anger of women or women's groups, or who think the government has shown "more concern about the treatment of women than is warranted by women's actual experiences" (Swim et al., 1995) should oppose affirmative action for women. If there is no discrimination or if women are already receiving favors, then affirmative action is, at best, unnecessary or, at worst, harmful to women as well as other groups. If there is no discrimination, and yet an out-group target receives special favors, then that outcome should lead to negative emotions on the part of in-group members. Similar to expectations about racism, if sexists are able to justify their negative reactions to a successful woman, based on an affirmative action policy, then they might also experience smaller discrepancies between how they do feel and how they ought to feel. The Current Research The current research examined the effects of potential attributions for success of an in-group or out-group member on emotional responses to the outcome. This research will examine both actual emotions and discrepancies between how one feels and how one thinks one ought to feel. Where previous research has focused on differences between cognitions and emotions in prejudice (e.g. Allport, 1954) or emotional responses to these discrepancies (Devine et al., 1991), this research will examine discrepancies within emotions. The successful outcome was receiving a scholarship over another candidate, who 19 was always a White male. Different affirmative action policies were included as part of the scholarship description to provide one potential attribution for the outcome. The target individual varied in both race and sex, creating a double in-group (same race and sex) relative to the participant, a double out-group (different race and sex), and two crossed groups (same race, different sex and different race, same sex). Four types of affirmative action policies were used. One afi'rrmative action policy emphasized membership in a disadvantaged group as a consideration in awarding the scholarship, even if that candidate is less qualified than a non-disadvantaged candidate (unfair advantage condition). In other words, less qualified candidates had an advantage because of their group membership. The policy in this condition is consistent with one of the common misperceptions of affirmative action, that unqualified minority candidates will be favored over qualified non-minority candidates (Kravitz & Platania, 1993). A second affirmative action policy emphasized consideration of membership in a target group only if all other qualifications are equal. In this condition, qualifications were more important than group membership, although membership was considered if candidates are equally qualified (qualifications first condition). The policy in this condition is closer to how affirmative action decisions may legally be implemented (Kravitz & Platania, 1993). In a third condition, participants were asked to write their own description of affirmative action (undefined condition). In the control condition, affirmative action was not mentioned. Participants' levels of racism and sexism were also considered. Participants read one of four scholarship descriptions which varied in affirmative 20 action policy. They then read applications flom and evaluated the qualifications of two “finalists” for the scholarship. Next, participants were told which of the candidates received the scholarship. Then, participants filled out measures of their actual emotion in response to the decision and of how they thought they ought to feel. Hypotheses Evaluation Extremity According to previous research, (e.g. Linville & Jones, 1980), well qualified out-group members are rated more positively than similar in-group members. In this study, then, the White participants should rate Afiican American scholarship candidates as more qualified for the scholarship than the White candidates when there is no affirmative action policy. Similarly, both male and female participants should rate other-sex candidates as more qualified than same-sex candidates in the absence of an affirmative action policy. Earlier research (e.g. Garcia et al., 1981; Heilman et al., 1992) has shown that the presence of an affirmative action policy can cause people to discount or minimize a potential beneficiary’s qualifications for a position. In the current study, then, qualifications ratings should be affected by the presence of an affirmative action policy when the out-group being rated is a potential beneficiary of the policy. Ratings of African American candidates by White participants and of female candidates by male participants should be lower when there is an affirmative action policy than when there is not. In effect, then, the presence of an affirmative action policy was expected to eliminate the out-group favorability bias when the policy is applicable to the out-group. In addition, the particular affirmative action policy should make a difference; a policy 21 which emphasizes qualifications over minority status should lead to higher qualifications ratings than a policy which emphasizes minority status over qualifications. Actual Emotional Responses Based on previous research (e.g. Islam & Hewstone, 1993), participants should react more negatively and less positively when an out-group member receives the scholarship than when an in-group member receives the scholarship, in the absence of an affirmative action policy. When there is an affirmative action policy, and that policy is applicable to an out-group target, prejudice and the policy should interact to affect actual emotional responses. Given the prevalence of negative attitudes toward affirmative action (e.g. Fine, 1992; Kinder & Sanders, 1990), participants should react more negatively when an affirmative action policy explicitly states that group membership may be more important than qualifications (an “unfair advantage” policy) as compared to a no policy condition, with a “qualifications first” policy somewhere between these two. This expected difference between policies should be magnified by individual differences in levels of prejudice, that is, participants higher in prejudice should react more negatively when there is an affirmative action policy than participants lower in prejudice. Emotional Discrepancies Consistent with Devine et al. (1991), moderate racists should have larger discrepancies between how they think they ought to feel and how they actually do feel than either low racists or high racists when the scholarship recipient is African American and there is no affirmative action policy. However, when there is an “unfair advantage” policy, the success of an Aflican American candidate and any resulting negative emotion can be attributed to the policy, which should reduce the 22 discrepancies somewhat relative to a “qualifications first” policy and, greatly relative to a no policy condition. Parallel findings are expected for male participants rating female candidates with respect to sexism. Method Participants Participants were 463 undergraduate psychology students at a large Midwestern university. Most were White (83.6%), 231 were male, and 223 were female (9 did not indicate their sex). The average age was 19.36 years old (SD = 1.92). Participants received class credit in exchange for taking part in the study. Eleven participants were eliminated from the analyses due to incomplete data. The small number of students of color and of rrrixed race (14.4%) did not allow for racial comparisons. Because previous research would predict that race differences exist, both in attitudes toward affinnative action (e.g. Kravitz & Platania, 1993) and in evaluations of African American candidates (e.g. Jackson et al., 1987), the analyses were performed on the White sample only. Thirty participants (11 males and 19 females) were dropped flom the analyses either because they indicated directly that they were not familiar with affirmative action, or gave a definition of affirmative action which indicated a very different definition of the concept than the one of interest in this research (e.g. “Taking action up (sic) something”). The final sample, then, consisted of 180 White males and 174 White females. Overview of Design Participants evaluated two scholarship applications. The first of the scholarship 23 candidates was always a qualified White male. The race and sex of the second candidate (the target) varied, as did the affirmative action policy. Therefore, this research used a 2 (sex of target) x 2 (race of target: African American or White) x 4 (affirmative action policy: none, undefined, qualifications first, or unfair advantage) x 2 (sex of participant) between subjects factorial design. The dependent variables were emotional responses to the committee’s decision to award the scholarship to the target and discrepancies between how participants did feel and how they thought they should feel about this decision. Evaluations of the target, attitudes toward Afiican Americans, and attitudes toward women were also examined as other possible influences (besides the manipulated variables) on participants’ emotional responses. Measures Academic Ability To measure perceptions of the targets’ qualifications for the scholarship, participants were asked to rate the applicants on nine characteristics, such as “intelligent” and “creative” using a 5-point semantic differential scale (See Appendix A). Previous research by Garcia et al. (1981) has shown these nine traits to load on one factor, which they called academic ability. In the current study, internal consistency was good, with a Cronbach’s alpha of .83 and corrected item-total correlations of .42 to .66. Emotional Responses and Discrepancies Discrepancies between how participants believed they should feel and how they actually did feel were measured using two identical emotion word checklists. Each nine-item checklist contained both positive emotion words, (e.g., “happy”) and negative emotion words, (e.g., “disgusted”). 24 Participants were asked to indicate, on a five-point semantic differential scale, how they actually did feel about the committee's decision to award the scholarship to the target candidate and, on the same set of emotion words, how they thought they ought to feel about the decision. The order of the ought/would measures was alternated across participants (See Appendix B). Subscales of positive and negative emotion words were created for both the actual emotion and the ought emotion measures. Due to low corrected item-total correlations, two emotion words (“sympathetic” and “surprised”) were dropped flom the scales. The remaining seven items formed two subscales for both the actual emotion measure and the ought emotion measure. In other words, four subscales were created. The three item actual positive emotion measure (Cronbach’s alpha = .80) and the three item ought positive emotion measure (Cronbach’s alpha = .85) used the same three emotion words. Similarly, the four item actual negative emotion measure (Cronbach’s alpha = .90) used the same emotion words as the four item ought negative emotion measure (Cronbach’s alpha = .90). To test the hypotheses regarding participants' actual emotions following the success of the target, an overall actual emotion scale was created. The actual positive emotion subscale and the actual negative emotion subscale were found to be significantly correlated (r (267) = -.42, p < .01). Because these two subscales were related, and because it was hypothesized that actual positive and actual negative emotions would be inversely related, an overall actual emotion measure was created. The difference between the mean actual positive emotion and mean actual negative emotion score was calculated 25 for each participant as an index of overall actual emotion. Positive scores reflect positive emotion overall, negative scores reflect negative emotion overall, and scores of or near zero reflect ambivalence or neutrality. In order to calculate emotion discrepancy scores for each participant, first an index of overall ought emotion was calculated. The positive ought emotion scores and the negative ought emotion scores were correlated (r (267) = -.42), and it was hypothesized that positive and negative emotions would be inversely related. An overall index of ought emotion was created the same way as the overall index of actual emotion: by taking the difference between the positive emotion mean and the negative emotion mean. Positive scores indicate that, overall, a participant thought she or he ought to feel positive about the outcome, while negative scores indicate the participant thought she or he ought to feel negative. Scores near zero indicate that participants thought that they ought to feel ambivalent or neutral. Discrepancies between how participants felt and how participants thought they ought to feel were calculated by taking the difference between each participant's overall ought emotion and overall actual emotion. Positive overall discrepancy scores indicate that participants thought they ought to be more positive than they actually were; negative discrepancies indicate that they thought they ought to be more negative than they actually were. Discrepancies of zero indicate that participants thought they ought to feel the same as they actually did feel. Attitudes Toward African Americans The Modern Racism Scale (McConahay, 1986) is a seven-item scale designed to measure prejudice toward Aflican Americans. 26 Unlike measures of old-fashioned racism, the modern racism scale is designed to measure more subtle negative attitudes toward African Americans. Participants were asked to indicate their agreement with items on a five point Likert type scale (1 = "strongly disagree" and 5 = "strongly agree") (See Appendix C). Afier reverse scoring one of the items, higher scores on this scale indicate higher levels of modern racism. Previous research has established the reliability and validity of this scale (e.g. McConahay, 1986). In the current study, internal consistency was good, with a Cronbach’s alpha of .88 and corrected item-total correlations ranging flom .55 to .74. Attitudes Toward Women The Modern Sexism Scale (Swim, Aiken, Hall, & Hunter, 1995) is an eight item scale designed to measure prejudice toward women. The authors argue that discrimination against women has become more subtle over time; this scale is designed to measure this newer, subtle sexism. Participants were asked to indicate their agreement with statements on a five point Likert-type scale (1 = "strongly disagree" and 5 = "strongly agree"). After reverse coding three items, higher scores indicate more sexist attitudes (see Appendix D). Previous research (Swim et al., 1995) has demonstrated both the validity and the reliability of this measure. In this study, internal consistency was .83 with corrected item-total correlations flom .31 to .70. Demographics Participants were asked to indicate their year in school, grade point average, age, race and ethnic background, sex, and social class (see Appendix E). Familiarity with Affirmative Action After running several experimental sessions, it was noticed that a few participants in the undefined policy condition were providing definitions of affirmative action that showed the participant was not familiar 27 with the concept as it was used in this study. Therefore, partway through the study, an additional question at the end of the demographics questionnaire asking participants to define affirmative action in their own words was added, in order to make sure that participants in all conditions were at least slightly familiar with affirmative action. Materials Two packets were prepared: a scholarship packet and a social issues packet. In the scholarship packet, the first page was a brief description of a (nonexistent) scholarship offered by a different university than the participants’ to one undergraduate for his or her senior year. The affumative action manipulation was achieved by using different descriptions of the affirmative action policy associated with the scholarship. In the unfair advantage condition of the affnmative action manipulation the affirmative action policy was described as having been implemented "to compensate for paSt or present discrimination... Although this scholarship is open to all undergraduates at this university, a candidate's minority status may be considered, even if that candidate is not as qualified as other, non-minority candidates” (see Appendix F). In the qualifications first condition, the affirmative action policy was described identically, except for the last sentence, which read “However, a candidate’s minority status will only be considered if that candidate is as qualified as all other top candidates for the scholarship” (see Appendix G). In the undefined condition, an affirmative action policy was not described. The scholarship description simply concluded with the sentence “This scholarship is awarded in accordance with the university’s affirmative action policy.” In this condition, on the 28 following page, participants were asked to describe, in their own words, what affirmative action meant to them (see Appendix H). In the control policy condition, the description of the scholarship did not contain any mention of affirmative action (see Appendix 1). Pilot testing was used to assure that the two applications used in the study would be equivalent in terms of qualifications for the scholarship. An additional 48 undergraduate students ( 19 males, 27 females, 2 did not indicate their sex) read a brief description of a scholarship which made no mention of affirmative action, followed by four applications for that scholarship. The applications varied slightly in college major, grade point, and honors received. The order of the four scholarship applications was varied across participants. No indication of race or sex of the applicant appeared on any of the applications in the pretesting. These pretest participants were asked to evaluate each candidate on the academic ability scale (Garcia et al., 1981; see Appendix A). Internal consistency ranged from .77 to .86. Two equivalent applications were selected for use in the final study. The next pages of the scholarship packet contained these two applications and the qualifications measures for each candidate. To highlight the fact that this was a competition between a White male and a target, the White male's application materials always appeared first. The race and sex manipulation was achieved by including this information along with other demographics information (year of high school graduation, age) in both applications. The two applications selected flom the pretesting were alternated such that half of the target applicants and half of the White male applicants had each of the applications. 29 Additionally, in one condition two White male candidates were compared as a check that the two sets of qualifications were equal. Following the applications, there was a page indicating that the target candidate received the scholarship, followed by the actual and ought emotion word measures. The order of these two measures was alternated. The second packet, the social issues packet, contained the racism measure and the sexism measure. The order of these two measures was alternated. The last page of the second packet was the demographics information questionnaire, and the familiarity with affirmative action check. Procedure Participants completed the experiment individually or in groups of two to fifieen. They were told that they were participating in a study of the factors which influence evaluations of scholarship applicants. They then read and completed the first packet. The experimenter then collected the first packet flom each participant. After everyone had finished the first packet, the participants were told that because the experiment had only taken about half an hour, and they were receiving credit for an hour, they were asked if they would mind filling out a brief survey on some social issues as part of another study for the experimenter’s advisor. Participants then completed the social issues questionnaire packet. By collecting the second packet in the same order as the first, the experimenter was able to match the two packets for each participant. Results Evaluation Extremity It was hypothesized that Linville and Jones’ (1980) findings that highly qualified 30 out-group targets were rated more favorably than similarly qualified in-group targets would be replicated when there was no affirmative action policy. Inspection of the means revealed that there was no difference in ratings between Afiican American targets and White targets (M = 4.25, SD = .47 M = 4.25, SD = .48). Additionally, contrary to the hypothesis, the same-sex candidates were rated slightly more favorably than the other-sex candidates (M = 4.25; SD = .46; M = 4.24; SD = .49). As such, the out-group extremity hypothesis was not supported when there was no affirmative action policy. Means and standard deviations are listed in Table 1 for male participants and in Table 2 for female participants. It was also hypothesized that the presence of the unfair advantage affirmative action policy would result in much lower qualifications ratings of African American targets by the White participants compared to the no policy condition and in somewhat lower qualifications ratings compared to the qualifications first policy. The hypothesized main effect for policy among participants rating Aflican American targets was not found (E (2, 240) = .05, NS) (See Tables 1 and 2). Similar to the hypothesis for ratings of African American targets, it was hypothesized that males’ ratings of female targets’ qualifications would be lowest in the unfair advantage policy condition and highest in the no policy condition, with the qualifications first policy condition in between. However, the mean qualifications ratings were not in the predicted direction (No policy: M = 4.15, SD = .42; Qualifications first: M = 4.14, SD = .39; Unfair advantage: M = 4.20, SD = .35). Female targets were 31 actually rated slightly more positively by male participants when there was an unfair advantage policy (see Table 1). 32 Table 1: Male participants' mean qualifications ratings by policy and target Target race African American White Target sex: Male Female Male Female Policy None M 4.26 4.23 4.20 4.07 SD .35 .44 .62 .41 n 10 11 ll 12 Qualifications first M 4.28 4.16 4.09 4.11 SD .35 .37 .50 .43 n 1 1 13 10 10 Unfair advantage M 4.15 4.26 4.18 4.15 SD .35 .38 .58 .33 33 Table 2: Female participants' mean qualifications ratings by policy and target Target race Afiican American White Target sex: Male Female Male Female Policy None M 4.20 4.29 4.45 4.25 SD .62 .47 .42 .44 n 12 12 l 1 ll Qualifications first M 4.38 4.15 4.34 4.31 SD .49 .41 .42 .48 n 12 ll 10 10 Unfair advantage M 4.35 4.09 4.19 4.38 SD .35 .31 .44 .45 34 Actual Emotional Responses It was hypothesized that participants would feel more positive when the target was an in-group rather than an out-group member, in the absence of an affirmative action policy. Inspection of the means revealed that contrary to the hypothesis, the emotion scores following the success of an African American target were more positive than those following the success of a White target when there was no affirmative action policy (M = 1.25, SD = 1.42; M = 1.08, SD= 1.81). To test the effect of target gender, a planned comparison was performed within the flamework of a 2 (sex of target: same or different from participant) by 2 (sex of participant) by 2 (race of target) by 3 (affirmative action policy: none, qualifications first, or unfair advantage) analysis of variance on participants’ overall actual emotion scores.I It was hypothesized that participants would be more positive when the target was of the same sex than of the other sex. This hypothesized effect for target sex was not found (1 (243) = .84, NS). Individual differences in racism and the affirmative action policy were expected to interact to affect emotion responses to African American targets. To test this, the modern racism scale scores of participants were trichotimized into low, medium, and high levels of racism. It was expected that, if the target was African American, participants who were higher in racism would react more negatively when there was an affirmative action policy than participants lower in prejudice. Inspection of the means for participants who ‘An additional factor, the order of the actual and ought emotion questionnaires, was also tested in this and subsequent analyses on actual emotion and emotion discrepancies. This factor was not significant in any of the analyses, and was not considered further. 35 evaluated an Aflican American target revealed that this predicted interaction pattern was not found (see Table 3). For instance, participants moderate in prejudice indicated more positive emotion when there was a qualifications first policy than did participants low in prejudice. 36 Table 3: Mean actual emotion for African American targets by racism and policy Racism Low Medium High Policy None M 1 .53 1.69 .70 SD 1 .36 1 . 1 1 l .58 n 12 15 18 Qualifications first M .93 1 .20 -.52 SD 1 .53 1.33 1 .82 n 9 20 18 Unfair advantage M 1.40 1 .10 -1 .08 SD 1.76 2.02 2.43 37 Male participants’ level of sexism and the affirmative action policy were expected to interact when rating female targets. A preliminary t-test revealed that male participants had significantly more sexist attitudes than female participants (1 (264) = 6.49, p < .01; M = 2.59, SD = .74; M = 2.07, SD = .57). To compensate for this difference, the sexism scores for male participants and female participants were trichotimized into low, medium, and high levels of sexism separately. Then, each level of sexism for male participants was combined with the corresponding level for female participants to create an overall trichotirnization of sexism. The decision was made to analyze sexism data flom male and female participants combined for two reasons. First, by combining data flom males and females, the sample size for each test is higher, thus increasing the power of the test. Second, if the data were analyzed separately, the number of statistical tests used to explore the data would increase, thus increasing the type I error rate. Male participants were expected to react more negatively to a female target when there was an affirmative action policy than when there was no policy. In addition, the policies were expected to have a greater impact on male participants higher in sexism. Inspection of the means for male participants who rated female targets revealed that this predicted interaction pattern was not found (see Table 4). For instance, when there was no policy, males high in sexism were more positive than males moderate in sexism. Additionally, when there was a qualifications first policy, male participants moderate in sexism were more positive than those high or low in sexism. 38 Table 4: Males' mean actual emotion for female targets by policy and sexism Sexism Low Medium High Policy None M 1 .76 08 58 SD 1.05 1 .87 l .43 n 7 8 8 Qualifications first M .29 1.41 -2.00 SD 1.82 1 .23 1 .23 n 8 9 6 Unfair advantage M .27 -.30 1 .00 SD 2.80 2.50 1 .93 39 Discrepancies in Emotions Based on Devine et al. (1991), it was hypothesized that in the absence of an affirmative action policy, participants moderate in prejudice would have larger discrepancies between how they felt and how they thought they ought to feel than participants who were either low or high in prejudice when the target was an out-group member. Inspection of the means for participants who read about an African American target revealed that moderate racists actually had smaller discrepancies than participants either low or high in racism (see Table 5). Similarly, inspection of the mean emotion discrepancy scores for male participants who evaluated a female target revealed that those moderate in sexism had discrepancy scores closer to zero than either low or high sexists (see Table 6). This hypothesis, then, was not supported either for White participants responding to Afiican American targets or for male participants responding to female targets. 40 Table 5: Mean emotion discrepancies for African American targets by policy and racism Racism Low Medium High Policy None M .75 .49 .72 SD 1 .44 1 .06 1 .89 n 12 15 1 8 Qualifications first M 1.33 .58 .98 SD 1 .56 l .91 l .74 n 9 20 18 Unfair advantage M .80 .83 1.29 SD 1.63 l .30 2.42 n 10 20 16 41 Table 6: Males' mean emotion discrepancies for female targets by policy and sexism Sexism Low Medium High Policy None M -.43 .04 2.13 SD .85 1.87 1.36 n 7 8 8 Qualifications first ' M .79 -.07 1.56 SD 1 .21 .91 l .81 11 8 9 6 Unfair advantage M .73 .67 -.20 SD 1 .12 1 .67 l .60 42 Undefined Policy Definitions An additional condition was included in which participants provided their own definition of affirmative action. This condition allows for the examination of participants' own conceptions of affirmative action, as well as their responses on the other measures given their own definitions of affirmative action. Participants did provide a wide range of definitions of affirmative action. Nearly half (49.4%; n = 43) of the participants in this condition defined affirmative action positively, as a way to overcome discrimination. For instance, one participant wrote “Affirmative action was developed to help the minority populations have equal opportunities in jobs, schooling, etc.” A number of participants (29.9%; n = 26) defined affirmative action as something negative or unfair. For instance, one participant wrote “Affirmative action to me means that minorities are favored over White men! Which is wrong!” The remaining 20.7% (n = 19) of participants gave mixed definitions of affirmative action, such as “I feel that affirmative action means including Aflican Americans, and other minorities, and even sometimes choosing them over another candidate simply because of their racial background. I believe that this policy is both positive and negative.” Nearly all mentioned racial groups or “minorities” as potential beneficiaries. Over a third (36.3%; n = 32) specifically included women as potential beneficiaries as well. Some participants’ definitions also included quotas (16.1%; n = 13) and/or overlooking qualifications to favor particular groups of people (20.7%; n = 19). A number of participants expanded their definition of affirmative action to include people 43 who are members of groups which are not included in most legal definitions of affirmative action, such as religious groups or gays, lesbians, and bisexuals (12.6%; n == 11). Additional Influences on Target Qualifications Ratings To test for any unhypothesized influences on target qualifications ratings, a 2 (participant sex) by 2 (target sex: same or different flom participant) by 2 (target race) by 3 (affirmative action policy: none, qualifications first, or unfair advantage) analysis of variance was performed on the target qualifications ratings scale. No significant effects were found. Additional Influences on Actual Emotions To test for any unhypothesized influences on actual emotion ratings, a 2 (participant sex) by 2 (target sex: same or different flom participant) by 2 (target race) by 3 (affirmative action policy: none, qualifications first, or unfair advantage) by 3 (racism: low, medium, or high) analysis of variance was performed on the actual emotion measure. Several effects were found. A three way interaction between participant sex, participant racism, and affirmative action policy was found (13 (4, 196) = 3.89, p < .01). Among male participants, a main effect of racism was found (E (2, 196) = 5.81, p < .01). A Tukey's HSD test showed that male participants low in racism were more positive than participants who were high in racism (g (3, 196) = 4.66 p < .01). Means and standard deviations are listed in Table 7. Among female participants, a borderline significant interaction was found between racism and affirmative action policy (E (4, 196) = 2.27, p 44 < .07). Simple simple effects tests showed that among female participants medium or high in racism, actual emotion was not affected by the affirmative action policy (E (2, 196) = .27, NS; E (2, 196) = 1.23, NS). Female participants low in racism, however, were more positive when there was no affirmative action policy than when there was a qualifications first policy (g (3, 196) = 3.94, p < .05). Means and standard deviations are listed in Table 8. 45 Table 7: Males' mean emotion scores by policy and racism Racism Low Medium High Policy None M 1.06 1.17 1.02 SD 1.50 2.14 1.42 n 1 l 12 21 Qualifications first M 1.70 .83 -.27 SD 1.24 1.37 1.96 n 9 18 17 Unfair advantage M 1.10 .88 -.51 SD 2.14 1.67 2.31 n 10 l9 l7 46 Table 8: Females' mean emotion scores by policy and racism Racism Low Medium High Policy None M 1 .87 1.18 .63 SD 1.38 .99 2.08 n 18 13 16 Qualifications first M .17 1.67 .64 SD 1 .78 1.28 1.79 n 13 13 16 Unfair advantage M 1 .47 1 .39 -.21 SD 1 .13 1 .93 2.33 47 This 2 (participant sex) by 2 (target sex: same or different flom participant) by 2 (target race) by 3 (affirmative action policy: none, qualifications first, or unfair advantage) by 3 (racism: low, medium, or high) analysis of variance also yielded a main effect of target sex, which was qualified by a target sex by target race interaction (E (1, 196) = 6.25, p < .05). Simple effects tests revealed that when the target was African American, the target’s sex did not affect participant emotion. When the target was White, however, the (White) participants were more positive when the target was of the same sex than ofthe other sex (E (1, 196) = 8.55, p < .01; M = 1.32, SD = 1.48; M = .56, SD = 1.86). A participant sex by target race interaction was also found (E (l , 196) = 4.24, p < .05). Simple effects tests showed that participant sex did not affect emotional responses to African American targets (E (1 , 196) = .69, NS). However, when the target was White, female participants were more positive than male participants (E (1 , 196) = 5.81, p < .05; M =1.25, SD = 1.63; M = .62, SD =1.76). Another interaction involving target race was found; target race interacted with participant racism (E (2, 196) = 7.18, p < .01). Simple effects tests showed, not surprisingly, that when the target was White, participant racism did not significantly affect actual emotion. When the target was Aflican American, racism did affect actual emotion (E (2, 196) = 13.53, p < .01). Tukey’s HSD tests revealed that participants high in racism were less positive than participants either low or moderate in racism (5; (3, 196) = 6.13, p < .01; g (3, 196) = 6.33; p < .01). Means and standard deviations are listed in Table 9. 48 Table 9: Mean actual emotion ratings by target race and participant racism Target race African American White Racism Low M l .30 l .17 SD 1 .61 1 .61 n 46 51 Medium M 1 .20 .96 SD 1.55 1 .62 n 66 55 High M -.09 .66 SD 1.95 l .89 49 In order to test for any unexpected effects of participant sexism on actual emotions, a 2 (participant sex) by 2 (target sex: same or different flom participant) by 2 (target race) by 3 (affirmative action policy: none, qualifications first, or unfair advantage) by 3 (sexism: low, medium, or high) analysis of variance was performed on the actual emotion measure. This analysis did not yield any significant effects. Additional Influences on Emotion Discrepancies To test for any unhypothesized influences involving participant racism on discrepancies between how participants thought they should feel and how they actually did feel, a 2 (participant sex) by 2 (target sex: same or different flom participant) by 2 (target race) by 3 (affirmative action policy: none, qualifications first, or unfair advantage) by 3 (racism: low, medium, or high) analysis of variance was performed on the emotion discrepancy measure. This analysis yielded" a significant target race by affirmative action policy interaction (E (1, 196) = 3.06, p < .05). When there was no affirmative action policy target race did not affect emotion discrepancies. When there was a qualifications first policy, a main effect of target race was found (E (1, 196) = 4.45, p < .05). Participants experienced larger positive emotion discrepancies when the target was Afiican American than when the target was White. When there was an unfair advantage policy, participants also experienced larger positive emotion discrepancies when the target was African American than when the target was White (E (l, 196) = 5.67, p < .05). Means and standard deviations are listed in Table 10. 50 Table 10: Participants' mean emotion discrepancies by policy and target Target race Afiican American White Target sex: Same Different Same Different Policy None M .91 .41 .14 .83 SD 1.48 1.52 1.54 1.53 n 22 23 23 23 Qualifications first M .47 1.24 .10 .10 SD 1.69 1.80 1.13 1.10 n 22 25 20 20 Unfair advantage M 1.12 .86 -.18 .47 SD 1.59 2.00 1.55 1.95 a 22 24 19 24 51 To test for any unhypothesized influences of participant sexism on emotion discrepancies, a 2 (participant sex) by 2 (target sex: same or different flom participant) by 2 (target race) by 3 (affirmative action policy: none, qualifications first, or unfair advantage) by 3 (sexism: low, medium, or high) analysis of variance was performed on the emotion discrepancy measure. A participant sex by sexism interaction was found (E (1, 195) = 3.34, p < .05). Sexism did not affect the emotion discrepancies of male participants (E (2, 195) = 1.22, NS). Sexism marginally affected the emotion discrepancies of female participants (E (2, 195) = 3.00, p < .07). A Tukey’s HSD test revealed that female participants low in sexism had larger positive emotion discrepancies than those high in sexism (g (3, 196) = 3.46, p < .05). Means and standard deviations are listed in Table 11. 52 Table 11: Mean emotion discrepancies by participant sex and participant sexism Participant sex Male Female Sexism Low M .44 93 SD 1 .08 1 .65 n 41 56 Medium M .25 .51 SD 1.56 1 .81 n 69 49 High M .68 .21 SD 1 .64 1 .73 53 Ought Emotions Although no specific hypotheses were formulated regarding what participants would think they ought to feel, possible influences on participants’ ought emotions were examined. A 2 (participant sex) by 2 (target sex: same or different flom participant) by 2 (target race) by 3 (affirmative action policy: none, qualifications first, or unfair advantage) by 3 (racism: low, medium, or high) analysis of variance was performed on participant’s ought emotion means. A main effect of target sex was found (E (1, 196) = 4.04, p < .05). Participants thought they ought to be more positive than they were for both same-sex and other-sex targets. However, this effect was stronger for same-sex than for other-sex targets (M = 1.57, SD = 1.74; M = 1.26, SD = 1.82). Additionally, a main effect for target race was found, which was qualified by a target race by participant racism interaction (E (2, 196) = 4.23, p < .05). When the target was White, participant racism did not affect how they thought they ought to feel (E (2, 196) = .60, NS). However, when the target was Aflican American, racism did affect how participants thought they ought to feel (E (2, 196) = 7.29, p < .01). Overall, participants thought they ought to be more positive than they were. A Tukey’s HSD test revealed that participants who were moderate or low in racism thought they ought to feel more positive than did participants who were high in racism (5; (3, 196) = 3.90, p < .05; g (3, 196) = 5.07, p < .01). Means and standard deviations are listed in Table 12. 54 Table 12: Mean ought emotion ratings by target race and participant racism Target race Aflican American White Racism Low M 1.99 1.41 SD 1.64 1.79 n 46 51 Medium M 1.64 1.27 SD 1.85 1.50 n 66 55 High M .82 1.06 SD 2.08 1.96 n 70 64 55 To test for any effects of sexism on ought emotions, a 2 (participant sex) by 2 (target sex: same or different flom participant) by 2 (target race) by 3 (affirmative action policy: none, qualifications first, or unfair advantage) by 3 (sexism: low, medium, or high) analysis of variance was performed on participant’s ought emotion means. No effects were found. Discussion This research examined emotional responses and discrepancies in emotional responses to an out-group member’s success when affirmative action was or was not a potential explanation for that success. The qualifications of the target, target characteristics, the particular affirmative action policy, and individual differences in prejudice all influenced actual emotion and/or emotion discrepancies to some degree. However, the relationship between these variables appears quite complex. Qualifications Ratings Based on Linville and Jones (1980), it was hypothesized that participants would rate out-group targets as more qualified for the scholarship than in-group targets when there was no affirmative action policy. The findings did not show this expected out- group extremity effect; the differences between ratings of out-group targets and ratings of in—group targets were quite small in this condition. It was also expected that the affirmative action policy would lead to a discounting of an out-group target’s qualifications, if the policy was applicable to the out-group. This hypothesis was not supported either. This lack of difference between qualifications ratings of in-group and out-group 56 targets is, however, consistent with Biemat, Manis, and Nelson’s (1991) work on shifiing standards in evaluations of different groups. Their research found evidence that when individuals are asked to rate members of different groups on stereotyped traits, anchors for subjective, but not objective, judgements may shift. For instance, a woman who is five-foot-nine may be rated as “tall,” but a man of the same height may be rated as “average” in height. The objective height has not changed, but the standard for rating male and female height is different. In the current research, participants were told that the two applicants were the finalists for the scholarship before participants were asked to fill out the qualifications and emotions measures. It is still possible that, rather than rating both applicants with regard to their qualifications for this particular scholarship, the standards of judgement shified, depending on group membership of the target. For instance, an Aflican American target may have been rated in comparison with Afiican Americans in general, rather than in comparison with the other scholarship finalist, the White male. Another possible explanation for the null effect of the target’s group membership on qualifications ratings is a ceiling effect. All of the means for target qualifications ratings were greater than four on a five point scale (see Tables 1 and 2). In the no policy condition, there may not have been any room on the scale for participants to rate cut- group members as more qualified. In the conditions with an affirmative action policy, which was expected to lead to a discounting of qualifications, the targets may have been too qualified for participants to be able to discount. That is, they may have been so obviously qualified for this scholarship that the presence of an affirmative action policy 57 was not enough to lead participants to think that this out-group target was less qualified. Using two scholarship applications that were qualified, but perhaps not so highly qualified may have led to support for the predicted extrerrrity effect. Actual Emotional Responses It was hypothesized that without an affirmative action policy, participants would be more positive when the target was an in-group member than when the target was an out-group member. This hypothesis was not supported for either same race or same sex groups. The (White) participants were actually slightly more positive when the target was African American than when the target was White. Though the emotion means for sex in-group and sex out-group members were in the predicted direction, the difference was not significant. Some evidence was found, though, for an in-group bias in emotional responses. If the target was White, participants were more positive when the target was also of the same sex as the participant. This finding suggests that both race and sex contribute to the perception of in-group membership. In this study, neither race nor sex alone was enough to create an in-group bias; both seemed to be necessary. In addition, it was found that if the target was White, female participants were more positive than male participants. It was also hypothesized that individual levels of prejudice would interact with the affirmative action policies if the policy was applicable to an out-group member. The presence of an affirmative action policy was expected to have a greater impact on the actual emotional responses of individuals higher in prejudice. This hypothesis was not supported for either the (White) participants rating African American targets or for male 58 participants rating female targets. Some interesting, though not hypothesized, effects were found for this measure. For instance, racism affected actual emotion reactions to Aflican American targets. Participants high in racism had slightly negative reactions to Aflican American targets while participants moderate or low in racism had somewhat positive reactions to the Aflican American targets across the policies. This effect is hardly surprising. However, male participants low in racism were more positive than male participants high in racism across policies and across targets as well. The majority of males high in racism (45 of 55) read about an affirmative action policy, an Aflican American target, or both. Participants high in racism may be reacting negatively to an affirmative action policy, even when it doesn’t apply, and to African American targets, even when there is no policy. For high racists, the mere presence of one of these circumstances may be enough to yield a negative reaction. A rather unusual finding was that female participants low in racism were more positive when there was no policy than when there was a qualifications first policy. Perhaps, similar to the finding for high racist males, these females’ negative reactions to a policy are enough to affect emotional responses to any target. However, why this occurred for low racists, rather than high racists, remains unclear. Participants, in general, thought they ought to be at least somewhat positive about the outcome. A main effect for the sex of the target was found; participants thought they ought to be more positive when the target was of the same sex as the participant than when they were different sexes. Additionally, when the target was African American, 59 participants low and moderate in racism thought they ought to be more positive than did participants high in racism. These findings are consistent with the findings for actual emotion. When the target was African American, participants high in racism reacted with slight negative emotion, while participants low and moderate in racism were somewhat positive. The relationship between ought emotion, actual emotion, and racism is a frmction of what it means to be racist: peOple high in racism react negatively to African Americans, and believe they ought to react less positively than do people lower in racism. In this study, as in Devine et al. (1991), individuals with different levels of racism also had different standards for responding to Aflican Americans. Discrepancies in Emotion Consistent with Devine et al. (1991), it was hypothesized that participants who were moderate in prejudice would have larger discrepancies between how they thought they ought to feel and how they actually did feel than participants who were either low or high in prejudice when the target was an out-group member. This hypothesis was not supported; participants moderate in prejudice actually had slightly (though not significantly) smaller discrepancies when the target was an out-group member. It was also hypothesized that the presence of an affirmative action policy would minimize these discrepancies between how participants felt and how they thought they ought to feel, particularly if the participant was moderate in prejudice. Evidence was not found to support this hypothesis. Exploratory analyses did yield some interesting effects for this measure. Generally, participants thought they ought to feel more positive than they actually did. 60 When there was an affirmative action policy, participants had larger positive discrepancies if the target was Aflican American than if the target was White. This race effect is logical; affirmative action should not affect discrepancies in emotional reactions to White male targets because they are not intended beneficiaries of the policy. Affirmative action may also be more closely associated in participants’ minds with African Americans than with White women. If this is the case, then affirmative action might not have much of an impact on discrepancies when the target is a White woman either. Contrary to the hypotheses, however, the presence of an affirmative action policy led to larger, rather than smaller, discrepancies. Rather than providing an explanation for experiencing less positive or negative emotion, as expected, resulting in smaller discrepancies, the affirmative action policy seems to have led to larger discrepancies. This increase could be due to ambiguity of the reason for the target’s success. All of the targets were highly qualified for the scholarship, so participants may not have known whether to attribute the target’s success to ability or to the affirmative action policy. Perhaps participants thought they ought to be quite positive because the target was highly qualified, but also could not help thinking that the affirmative action policy, and not the target’s qualifications, led to the target’s success. The potential to attribute the target’s success to the affirmative action policy may have led to less positive actual emotional responses, leading to larger emotion discrepancies. Additionally, it was found that, among female participants, discrepancies in emotion were larger among females low in sexism than among those high in sexism. 61 This effect is particularly interesting, because it occurred across target race, target sex, and affirmative action policy. Individuals low in sexism may have internalized standards for responding to all people that they feel they are not living up to, leading to these larger discrepancies across conditions. However, a corresponding pattern was not found for ought emotion responses, making interpretation of this effect more tentative. Contrary to the hypotheses, the findings regarding an affirmative action policy were not clear cut; affirmative action did not lead to a simple decrease in positive emotion and a decrease of a different magnitude among individuals with different levels of prejudice. The affirmative action policy did not simply allow participants to justify any negative emotion, leading to a decrease in discrepancies either. Rather, some of the conditions where an affirmative action policy was applicable led to larger emotion discrepancies; participants thought they should feel better than they did. This finding may also be due to the ambiguity surrounding the reason that the target received the scholarship. Believing the target was indeed qualified, but attributing the target’s success to affirmative action could lead to these larger emotion discrepancies. Further research should clarify the attributions participants made for the target’s success, perhaps by simply asking participants to explain why they thought the target was successful. In the real world, individuals may also not be sure whether a co-worker or fellow student is there because of affirmative action, because of qualifications, or both. This uncertainty could lead to emotion discrepancies which may have behavioral consequences. Previous research (e. g. Devine et al., 1991) has shown that discrepancies between personal standards and actual behavior have consequences for emotion. People 62 may feel shame, guilt, or global discomfort. These unpleasant feelings could in turn lead to negative behaviors toward the “source” of the emotion: the minority group member. In the workplace, then, individuals who are unsure whether to attribute the success of a co-worker to affirmative action or qualifications could experience emotion discrepancies and unpleasant emotion resulting flom these discrepancies. This resulting unpleasant emotion could, for instance, lead workers to avoid a co-worker, which could result in lower productivity. Making sure workers realize that individuals are hired because they are qualified and not solely on the basis of group membership could reduce tension in the workplace. Contrary to what was hypothesized, the pattern of results for Afiican American targets rated by White participants, and the results for female targets rated by male participants were not identical. Racism and affirmative action policy and target race appear to be more closely related to emotion responses and emotion discrepancies than sexism and target sex. In the open-ended definition condition, more participants indicated that affirmative action was aimed at minority racial groups than at women. Despite the legal definition of affirmative action, which generally apply equally to sex and race, participants seem to more closely associate affirmative action with race. In order to achieve and maintain diversity in a workplace or academic environment, these differences in responding to different discriminated-against groups must be considered. What is effective for minimizing discrimination against one group may not minimize discrimination against another group. 63 Directions for Future Research The current research examined emotions and discrepancies in emotions when the success of an out-group target could be attributed to an affirmative action policy. Future research is needed to address a few limitations of the current research, as well as to further examine influences on discrepancies in emotional responses to out—group members. The current research did not have a manipulation check to make sure participants noticed the race and sex of the scholarship candidates, leaving a possibility that some participants did not notice the applicants’ group membership. Additionally, participants were not probed for suspicion regarding the actual purpose of the study. Certainly, future research needs to correct these omissions. The applicants in this study were also well- qualified for the scholarship. Had either or both been less qualified, participants may have had different emotion reactions and different standards for responding. Participants in the current research were all White college students. A more racially diverse sample and a sample of participants who have been in the workforce and might have had more personal experience with affirmative action might have produced different results. The current research asked participants to indicate how they felt and“ how they thought they ought to feel about the committee’s decision to award the scholarship to the target. The results might have been different had they been asked about their emotional responses to the targets themselves. For instance, participants might have perceived the decision to award the scholarship as being constrained by the affirmative action policy in the first place. In other words, the committee may not have been flee to award the 64 scholarship to the most academically talented student. If this were the case, participants might have been more negative toward the target than toward the outcome. On the other hand, participants could hold the committee completely responsible for the outcome and not exhibit any negative reaction toward the target. To clarify this, future research should directly measure the degree to which the committee, the target, and the affrrrnative action policy are seen as causes of the outcome. Future research should also examine individuals’ behavior toward out-group members following a success which could be attributed to an external cause, such as affirmative action. For instance, the amount of time spent working with a target to solve a problem, or the desire to work with a target again could be measured. Devine and colleagues (e.g., Devine et al., 1991; Monteith et al., 1993) have demonstrated that discrepancies between personal standards for responding and actual responses to out- group members have emotional consequences. Affirmative action policies, in this study, were associated with larger discrepancies in emotion response. These discrepancies in turn should affect how these individuals actually behave toward out-group members, and how the out-group members respond. Conclusions The results support the notion that target characteristics, prejudice, and affirmative action do indeed influence the way individuals feel, and the discrepancies between how they feel and how they think they ought to feel about a target’s success. However, the patterns of interactions among these factors are quite complex. Participants seemed to associate race with affirmative action more closely than sex with affirmative action. 65 Additionally, the affirmative action policies tended to lead to larger positive emotion discrepancies. Rather than giving individuals an excuse to discriminate, the policies may have confounded the reasons for the target’s success. Mass media portrayals of the “angry White man” (e.g., Cohn et al., 1995; Fineman, 1995) appear too simplistic to capture actual reactions. Many variables combine to create actual emotional responses and emotion discrepancies following an out-group members’ success. APPENDICES APPENDIX A Appendix A Target Qualifications Ratings Scale Based on the information you have read, please circle the number which best represents your Opinion of Michelle B. on each of the dimensions. Feel flee to look again at the application and the scholarship description. Not at all intelligent 1 2 3 4 5 Very intelligent Not at all competent 1 2 3 4 5 Very competent Not at all qualified 1 2 3 4 5 Very qualified Not at all resourceful 1 2 3 4 5 Very resourceful Not at all disciplined 1 2 3 4 5 Very disciplined Not at all responsible 1 2 3 4 5 Very responsible Not at all creative 1 2 3 4 5 Very creative Not at all conscientious 1 2 3 4 5 Very conscientious Not at all trustworthy l 2 3 4 5 Very trustworthy 66 APPENDIX B Appendix B Actual and Ought Emotion Rating Scales Please circle the number which best indicates how you actually do feel about the committee's decision to award the scholarship to Joseph D. There are no right or wrong answers. Not at all happy 1 2 3 4 5 Very happy Not at all disgusted l 2 3 4 5 Very disgusted Not at all proud 1 2 3 4 5 Very proud Not at all angry 1 2 3 4 5 Very angry Not at all hopeful 1 2 3 4 5 Very hopeful Not at all surprised I 2 3 4 5 Very surprised Not at all upset I 2 3 4 5 Very upset Net at all sympathetic 1 2 3 4 5 Very sympathetic Not at all sad 1 2 3 4 5 Very sad 67 Often, we have personal standards for how we think we should feel or react in various situations. Based on your personal standards, please circle the number which best indicates how you think you W about the scholarship committee's decision to award the scholarship to this individual. These answers may or may not be the same as the answers you gave earlier. There are no right or wrong answers. Not at all happy Not at all disgusted Not at all proud Not at all angry Not at all hopeful Not at all surprised Not at all upset Not at all sympathetic Not at all sad 1 1 2 2 3 3 4 4 5 5 Very happy Very disgusted Very proud Very angry Very hopeful Very surprised Very upset Very sympathetic Very sad APPENDIX C Appendix C Modern Racism Scale (MRS) Read each of the following statements carefully and, for each statement, circle the number which best represents your own opinion. There are no right or wrong answers for any of the statements. 1. Over the past few years, the government and news media have shown more respect to Blacks than they deserve. Strongly disagree 1 2 3 4 5 Strongly agree It is easy to understand the anger of Black people in America. Strongly disagree 1 2 3 4 5 Strongly agree Discrimination against Blacks is no longer a problem in the United States. Strongly disagree 1 2 3 4 5 Strongly agree Over the past few years, Blacks have gotten more economically than they deserve. Strongly disagree 1 2 3 4 5 Strongly agree Blacks have more influence over school desegregation plans than they deserve. Strongly disagree 1 2 3 4 5 Strongly agree Blacks are getting too demanding in their push for equal rights. Strongly disagree 1 2 3 4 5 Strongly agree Blacks should not push themselves where they are not wanted. Strongly disagree 1 2 3 4 5 Strongly agree 69 APPENDIX D Appendix D Modern Sexism Scale (MSS) Read each of the following statements carefully and, for each statement, circle the number which best represents your own opinion. There are no right or wrong answers for any of the statements. 1. Discrimination against women is no longer a problem in the United States. Strongly disagree 1 2 3 4 5 Strongly agree Women often miss out on good jobs due to sexual discrimination. Strongly disagree 1 2 3 4 5 Strongly agree It is rare to see women treated in a sexist manner on television. Strongly disagree 1 2 3 4 5 Strongly agree On average, people in our society treat husbands and wives equally. Strongly disagree 1 2 3 4 5 Strongly agree Society has reached the point where women and men have equal opportunities for achievement. Strongly disagree 1 2 3 4 5 Strongly agree It is easy to understand the anger of women's groups in America. Strongly disagree 1 2 3 4 5 Strongly agree It is easy to understand why women's groups are still concerned about societal limitations of women's opportunities. Strongly disagree 1 2 3 4 5 Strongly agree 70 71 Over the past few years, the government and news media have been showing more concern about the treatment of women than is warranted by women's actual experiences. Strongly disagree 1 2 3 4 5 Strongly agree APPENDIX E Appendix E Demographics Measure Please let us know a bit about yourself. Your answers will remain confidential. What is your sex? Male Female What is your age? years What is your racial/ethnic background? Please mark all that apply. _ Afiican American or Black not of Hispanic origin _ Asian American, Asian, or Pacific Islander _ Caucasian or White not of Hispanic origin _ Hispanic American or Latino _ Native American or Native Alaskan _ Other not listed above What is your overall college GPA? _ _ Not applicable because this is my first semester. Check the category that best describes the social class of your current household/family, or the household/family you lived with most of your life. _ Lower class/poor _ Lower middle class _ Middle class _ Upper middle class _ Upper class/wealthy 72 APPENDIX F Appendix F Unfair Advantage Policy Scholarship The following scholarship is awarded each year at another university (NOT Michigan State University). Identifying information about the university has been removed. The H. C. Monroe Memorial Scholarship is awarded each year to an outstanding student for the final year of undergraduate study. The recipient of this scholarship must demonstrate strong academic ability. The scholarship will cover the costs of tuition and fees for a student's final year. This scholarship is awarded in accordance with the university's affirmative action policy. The university policy aims to compensate for past or present discrimination. The award of this scholarship may be based in part on being a member of a rrrinority group (such as women, Aflican Americans, and Latino Americans). Although this scholarship is open to all undergraduates at this university, a candidate's rrrinority status may be considered, even if that candidate is not as qualified as other, non-minority candidates. On the next two pages you will find part of the applications flom two of the finalists for last year’s H.C. Monroe Memorial Scholarship. Identifying information has been deleted. Please read each candidate's application carefully. 73 APPENDIX G l-I Appendix G Qualifications First Policy Scholarship The following scholarship is awarded each year at another university (NOT Michigan State University). Identifying information about the university has been removed. The H. C. Monroe Memorial Scholarship is awarded each year to an outstanding student for the final year of undergraduate study. The recipient of this scholarship must demonstrate strong academic ability. The scholarship will cover the costs of tuition and fees for a student’s final year. This scholarship is awarded in accordance with the University's affirmative action policy. The University policy aims to compensate for past or present discrimination. The award of this scholarship may be based in part on being a member of a minority group (such as women, Aflican Americans, and Latino Americans). However, a candidate's minority status will only be considered if that candidate is as qualified as all other top candidates for the scholarship. On the next two pages you will find part of the applications flom two of the finalists for last year's H.C. Monroe Memorial Scholarship. Identifying information has been deleted. Please read each candidate's application carefully. 74 APPENDIX H Appendix H Undefined Policy Scholarship The following scholarship is awarded each year at another university (NOT Michigan State University). Identifying information about the university has been removed. The H. C. Monroe Memorial Scholarship is awarded each year to an outstanding student for the final year of undergraduate study. The recipient of this scholarship must demonstrate strong academic ability. The scholarship will cover the costs of tuition and fees for a student's final year. This scholarship is awarded in accordance with the University's affirmative action policy. On the next two pages you will find part of the applications from two of the finalists for last year's H.C. Monroe Memorial Scholarship. Identifying information has been deleted. Please read each candidate's application carefully. 75 76 In the space provided, please explain in your own words what “affirmative action” means to you. We are interested in what you think of when you hear this phrase and not a dictionary definition. APPENDIX I Appendix I No Policy Scholarship The following scholarship is awarded each year at another university (NOT Michigan State University). Identifying information about the university has been removed. The H. C. Monroe Memorial Scholarship is awarded each year to an outstanding student for the final year of undergraduate study. The recipient of this scholarship must demonstrate strong academic ability. The scholarship will cover the costs of tuition and fees for a student's final year. On the next two pages you will find part of the applications flom two of the finalists for last year's H.C. Monroe Memorial Scholarship. Identifying information has been deleted. Please read each candidate's application carefully. 77 LIST OF REFERENCES List of References Allport, G. W. (1954). W Reading, MA: Addison-Wesley Publishing Co. Amirkhan, J., Betancourt,H., Graham, 8., Lopez, S. R. & Weiner, B. (1995). Reflections on affirmative action goals in psychology admissions. Emhglggigal Streamed. 140-148 Arthur, W., Doverspike, D., & Fuentes, R. (1992). Recipients' affective responses to affirmative action interventions. A cross-cultural perspective. Dehaxigral summmm 229-243 Bergman. B. R. (1986). WWW Basic Books, Inc. Biemat, M., Manis, M, &Nelson, T. E. (1991). Stereotypes and standards of judgement mummmmmmmam Clayton, S. D. (1992). Remedies for discrimination: Race, sex, and affirmative action Behaxmmlficrcnrasandrhemm 245-257 Cohn, B., Turque, B., & Brant, M. (1995, March 27). What about women? W25. 22-25. Devine, P. G. Monteith, M. J. Zuwerink, J. R. & Elliot, A. J. (1991). Prejudice with and without compunction Icumalrrfficrscnalinaandicnallmhclchfl 817- 830. Dovidio, J. F. & Gaertner, S. L. (1983). The effects of sex, status, and ability on helping behavior NW 191-205 Dovidio, J. F. & Gaertner, S. L. (1981). The effects of race, status, and ability on helping behavior. W 192-203. Eberhardt, J. L., & Fiske, S. T. (1994). Affirmative action in theory and practice: Issues of power, ambiguity, and gender versus race. W W 201-220 78 79 Fine, T. S. (1992). The impact of issue flaming on public opinion: Toward affirmative action programs. W 323- 334. Fineman, H. (1995, April 3). Race and rage. W 23—25. Gaertner, S. L, & Dovidio, J. F. (1986). The aversive form of racism. In J. F. Dovidio and S. L. Gaertner (Eds) Wm San Diego, CA. Academic Press. Garcia, L. T., Erskine, N., Hawn, K., & Casmay, S. R. (1981). The effect of affirmative action of attributions about minority group members. W31. 42, 427-437. Heilman, M. E., Block, C. J. & Lucas, J. A. (1992). Presumed incompetent? Stigmatization and affirmative action efforts. W 536- 544. Heilman, M. E., & Herlihy, J. (1984). Affirmative action, negative reactions? Some moderating conditions. . . . 204-213. Islam, M. R. & Hewstone, M. (1993). Intergroup attributions and affective consequences in majority and minority groups. W91 MW 936-950 Jackson, L. A. & Sullivan, L. A. (1988). Cognition and affect in evaluations of StereotYPed group members W 65 9-672 Jackson, L. A., Sullivan, L. A., & Hodge, C. N. (in press). Stereotype effects on attributions, predictions, and evaluations: No two social judgments are quite alike. I le l' lS'lE 11 Johnson, S. D. (1980). Reverse discrimination and aggressive behavior. Ille W495. 11-19. Kinder, D. R. & Sanders, L. M. (1990). Mimicking political debate with survey questions: The case of White opinion on aff'umative action for Blacks. Smialgqgnifign, S. 73- 103. Klugel, J. R. (1985). "If there isn't a problem, you don't need a solution": The bases of contemporary affrrrnative action attitudes. WM 761 -784. Kravitz, D. A., & Platania, J. (1993). Attitudes and beliefs about affirmative 80 action: Effects of target and respondent sex and ethnicity. .ImrmalgflApplied WM 928-938. Linville, P. W. & Jones, E. E. (1980). Polarized appraisals of out-group members W689 703 McConahay, J. G. (1986). Modern racism, ambivalence, and the modern racism scale InJ G Dovidio&S L Gaertner(EdS) Prejudicediscriminaticnandracrsm. San Diego, CA: Academic Press. Newman, J. D. (1989). Affirmative action and the courts. In F. A. Blanchard & F J Crosby (Ede) Aflinnatixeacticninnerspsctixc. New York Springer-Verbs Stangor, C., Sullivan, L. A., & Ford, T. E. (1991). Affective and cognitive determinants of prejudice. W 359-3 80. "llllllllllililllli