Reflections on affirmative action goals in psychology admissions

PSYCHOLOGICAL SCIENCE General Article REFLECTIONS ON AFFIRMATIVE ACTION GOALS IN PSYCHOLOGY ADMISSIONS By James Amirkhan,^ Hector Betancourt,^ Sandra Graham,^ Steven Regeser Lopez/ and Bernard Weiner"^ 'Department of Psychology, California State University, Long Beach; ^Department of Psychology, Loma Linda University: ^Department of Education, University of California, Los Angeles; and ''Department of Psychology, University of California, Los Angeles Four goals of affirmative action in higher education are de- irrespective of race or gender. On the other hand, the 1970s scribed as they relate to psychology admissions. Broadly con- and 1980s legitimated a full range of policies and practices, ceived, these goals are compensating for past injustice, cor- collectively known as affirmative action, that partially recting present inequity, promoting intellectual diversity, and defined individual merit in terms of ethnicity and gender. enhancing the presence of role models. It is argued thai the Because of this tension, few issues have generated as four goals differ in their underlying assumptions about the pur- much controversy or lack of consensus as affirmative poses of affirmative action and that these differences can result in disparate admission decisions. Data from three experiments action in higher education. Proponents view some form on decision making in graduate psychology admissions are pre- of preferential treatment based on gender and ethnicity as sented to illustrate the analysis. In these studies, academic necessary at this time to ensure full participation in the psychologists rated the admissibitity of hypothetical graduate academic enterprise of heretofore excluded groups. Opstudent applicants who varied on a number of characteristics ponents dismiss such policies as unfair to meritorious (e.g., ethnicity, social class, interest in minority research) per- nontargeted groups and ultimately detrimental to the tartinent to affirmative action. A consistent pattern of ethnic geted beneficiaries themselves. At issue are fundamental group differences in admissibility ratings was documented, il- questions such as these: What kinds of affirmative action lustrating that compensation for past injustice can be inter- policies are truly effective in redressing past wrongs? Do preted as a salient affirmative action goal in graduate admis- the methods have unintended counterproductive outsions decisions. Implications of the analysis for clarifying comes? What kinds of burdens do the policies impose on admissions decisions guided by affirmative action goals are other people, and are such policies fair? discussed. We offer no defmitive answers to these questions, for this is not an article about the pros and cons of affirmaA moral and political tension exists in contemporary tive action in higher education. Other authors have articAmerica between the ideal of meritocracy, wherein ulated the opposing viewpoints far more knowledgeably achievement is based on individual effort, and the reality and passionately than we possibly could (see, e,g,. of societal injustice, wherein whole groups of individuals Carter, 1991; Francis, 1993; Glasser, 1988; Sowell, 1989), have been denied an equal opportunity to pursue the mer- Rather, we hope to both clarify and concretize the debate itocratic ideal. Fifty years ago, Myrdal (1944) recognized by engaging our readers in some reflection on the relation this strife as our "American dilemma," It persists today between their graduate admissions decisions and affirmalargely because our social agenda contraposes two seem- tive action goals. ingly incompatible goals. On the one hand, the civil rights Federal legislation has mandated that academic instimovement of the 1950s and 1960s defined a national pri- tutions adhere to some form of affirmative action guideority to become a society that rewards individual merit. lines in their student admission policies. Although most academics probably agree that steps should be taken to Address correspondence to Sandra Graham, Department of Educa- ensure that admissions procedures conform to established guidelines, they may be unaware of the assumption, University of California, Los Angeles, CA 90095-1521. 140 Copyright © 1995 American Psychological Society VOL. 6, NO. 3, MAY 1995 PSYCHOLOGICAL SCIENCE J, Amirkhan et al. tions that shape their adherence to these guidelines— assumptions as to the very goals of affirmative action. We argue that such goals, whether explicit or implicit, are not always completely compatible with one another, and that different goals may sometimes result in quite disparate admissions decisions. For example, suppose that a psychology department at a large research university has committed itself to increasing its ethnic minority enrollment by 15% in the next academic year, A pool of eligible candidates is identified, and the arduous process of selection begins. What criteria are to be judged most important? Should aSl nonwhite groups, including African Americans, Latinos, Native Americans, and Asian Americans, be equally represented? Should economic disadvantage be considered a relevant criterion in addition to (or in lieu of) ethnicity? Should preference be afforded to applicants who express a desire to study minority issues over applicants with more traditional interests? One's answer to the first question, which focuses on ethnic differences, depends in part on whether one believes that some groups have stronger claims to preferential treatment than others based on historical victimization. One's position on the second question, which calls attention to social status, is shaped by assumptions about the inclusiveness of affirmative action guidelines. And a response to the third question, concemed with research directions, is partly determined by one's stance on the importance of a diversity of intellectual viewpoints. Because issues such as these capture much of the controversy surrounding affirmative action, the process as a whole stands to benefit from conscious efforts to better articulate what affirmative action policies are intended to accomplish. In the following sections, we begin by describing four prevalent goals of affirmative action, focusing on how each goal might lead to a differential weighting of various criteria for student admissions to graduate programs in psychology. Broadly conceived, these goals are defined as affirmative action to compensate for past injustices, correct present inequities, promote intellectual diversity, and enhance the presence of role models. Next, we attempt to shed light on how these goals underlie graduate psychology admissions decisions by turning to preliminary data gathered in three experiments designed to simulate the admissions process. We conclude the article by considering how our analysis might lead to a better understanding of the graduate admissions process when affirmative action remedies are desired, THE GOALS OF AFFIRMATIVE ACTION Compensating for Past Injustice One ofthe major goals of affirmative action is based on the notion of compensation for past injustice (e,g,, FranVOL. 6, NO. 3, MAY 1995 cis, 1993): Because ethnic minorities have been victimized by a protracted history of prejudice and racial discrimination, affirmative action procedures are necessary to balance the moral scales—that is, to position targeted individuals as much as possible in the situation they would have expierienced had the injustice not occurred. In authorizing Executive Order 11246, which was the 1965 legislation that mandated affirmative action. President Johnson articulated a compelling case for the compensatory argument when he stated: You do not wipe away the scars of centuries by saying: Now you are free to go where you want, do as you desire, and choose the leaders you please. You do not take a man who has been hobbled by chains, liberate him, bring him to the starting line of a race, saying, "you are free to compete with all the others," and still justly believe that you have been completely fair. Thus it is not enough to open the gates of opportunity, (cited in Glasser, 1988, p, 348) Affirmative action as compensation implies that individual victims of past discrimination can be identified and that consensus can be achieved regarding who should be compensated and why. Executive Order 11246 defined targeted minority groups as "Blacks, Spanish-surnamed Americans, American Indians, and Orientals." But if the defining feature of a compensatory argument is the cumulative effect of a history of victimization, then the case may be made that these four groups are not equally entitled to special consideration under affirmative action guideUnes, Without minimizing the plight of any of these groups, many of our readers might agree that the legacy of slavery endured by African Americans and the disenfranchisement suffered by Native Americans are unique to these groups and are not matched by the experiences of more recent and voluntary immigrant groups, such as Asians and most Latinos, Furthermore, although Asian Americans are a designated target group under Executive Order 11246, and the injustices toward Chinese immigrants in the 19th century and toward Japanese Americans during World War II are well known, many members of the academic community and much of the lay public do not perceive Asians as a disadvantaged minority. Even if consensus could be reached as to which minority groups have stronger cases for redress, affirmative action as compensation forces the issue of which individuals within a targeted minority group are the rightful beneficiaries of preferential treatment. Is group membership alone sufficient? For example, should any qualified black applicant to graduate school be considered under affirmative action guidelines, or must a case be made that an individual applicant has been a victim of past discrimination? Although no serious scholar of affirmative action would claim that case summaries documenting individual 141 PSYCHOLOGICAL SCIENCE Affirmative Action in Psychology histories of discrimination should become parts of admissions folders, it has been argued that socioeconomic status (SES) is at least a gross marker of individual victimization and should therefore become a relevant criterion for affirmative action, perhaps even superseding that of ethnicity (e.g,, Simon, 1993), Affirmative action scholars sympathetic to this view, including some African American social scientists, have lamented that affirmative action programs have never been constructed to address the problems of the economically disadvantaged (e,g,, Sowell, 1989; Wilson, 1987). Rather, such programs have always favored minority group members who were from advantaged backgrounds and were in a position to compete for preferred positions such as high-paying jobs and admission to college. If these arguments have merit, and a generic category of "disadvantaged" should be used as a marker of past victimization, then a host of additional complexities arises in the application of affirmative action guidelines. Most obviously, a white applicant from an impoverished background might have a legitimate claim to preferential treatment over an affluent minority candidate. And how is one to evaluate minority subgroups, such as Latinos of South American descent, who have (relatively) little history of oppression in this country? It is common practice for Spanish-surname applicants to receive affirmative action consideration, as directed by Executive Order 11246, in the absence of information about their country of origin, A strict interpretation of affirmative action as compensation for past injustice might question the fairness of preferential treatment for members of minority groups that have not suffered economic disadvantage and have little or no history of injustice in this country. In sum, affirmative action as compensation for past injustice draws attention to issues of the severity of historic wrongs, group versus individual victimization, and characteristics of individuals targets (i.e., their particular ethnicity, SES, and country of origin). An admissions pattem that favors African Americans, the economically disadvantaged, and native-born over foreign-born minority applicants might be taken as evidence for the salience of a compensatory goal. Correcting Present Inequities characteristics such as race, the analysis does not rely on the historical bases of these racial inequalities. How, then, does one determine the existence of present inequities without appealing to the historical argument? A straightforward argument, although not without controversy, is that of proportional representation (see Wolf-Devine, 1993), Ethnic minorities ought to be represented in the workforce and in academia to the same extent that they are represented in the general population. If they are not, there are inequities that need correction. By all available indicators, ethnic minorities are markedly underrepresented in psychology. According to data recently reported in the APA Monitor, minorities constitute about 5% of all doctoral-level psychologists and about 15% of current psychology graduate and professional school students (Tomes, 1993). Ethnic minorities make up 25% ofthe total U.S. population. Furthermore, no one ethnic group has fared much better in psychology than any other: As of 1986, 1,9% of doctoral psychologists were Asians, compared with 2,2% for African Americans and 1,4% for Latinos (Kohout & Pion, 1990). Because the disparities between population statistics and representation in the discipline are so great, affirmative action goals based on correction are largely directed toward increasing opportunities for all ethnic minority groups, independent of particular group membership or characteristics of targeted individuals. In this sense, the goals of correction and compensation at times may seem incompatible, as when choices have to be made between two "equally qualified" members of different ethnic groups. On the other hand, the distinction between compensation and correction may become blurred when SES is a relevant characteristic, inasmuch as economic disadvantage can also be linked to correcting present inequity. For example, consider the choice between a poor Asian American whose parents are Cambodian immigrants and an affluent first-generation Japanese American. In neither case is there a protracted history of victimization in this country. Yet affirmative action goals are likely to favor the disadvantaged Cambodian because present economic disparities, as well as the spectre of growing discrimination against more recent Asian immigrant groups, no doubt will be salient. Thus, we cannot fully ascertain whether preferential treatment on the basis of SES "counts" as compensation or correction, or where, in the broader scheme of affirmative action goals, compensation for past injustice leaves off and correction for present inequity begins. In contrast to the historical basis of affirmative action as compensation for past injustice, a second goal of affirmative action is grounded in the notion of correction and is more ahistorical in nature (see Francis, 1993). According to this argument, discriminatory practices have allowed some groups to receive a greater share of scarce, Promoting Intellectual Diversity prized, or competitive rewards than other groups, and The diversity argument is seemingly the one goat of this inequity must be corrected. Other than acknowledging that such inequality in outcomes is attributable to affirmative action that is specific to academia. Education 142 VOL. 6, NO. 3, MAY 1995 PSYCHOLOGICAL SCIENCE J, Amirkhan et al. is an enterprise in which students leam by being exposed to different perspectives on the world. Minority (and women) scholars have a special role in this endeavor because they are likely to be pursuing original work, both substantive and methodological, in areas not studied by mainstream scholars. Contributions in the areas of Afrocentricity and feminist theory are contemporary cases in point. Minorities in academia also have unique experiences with prejudice and discrimination that shape their thinking and research agendas in ways not possible for people not encountering these experiences directly. Even when such work generates controversy or is greeted with skepticism, it can still be perceived as important because it sharpens intellectual dialogue and provides a forum to challenge prevailing views. As in the case of the two goals already discussed, some implicit assumptions guide admissions decisions when the affirmative action goal is intellectual diversity. Consider two equally qualified ethnic minority applicants to a doctoral program in personality. Applicant A writes in her personal statement that she wants to study the role of ethnic identity in the formation of the self-concept, whereas Applicant B indicates an interest in structural theories of emotion. Applicant A links her interest to personal experiences with role confiict, whereas Applicant B refers exclusively to the need for theory-driven models of emotion with no suggestion of how these theories might have some particular applicability to "socially relevant" issues. Which applicant should receive higher priority? If Applicant A is preferred over Applicant B, there is an implicit assumption that intellectual diversity is best fostered when minority scholars pursue racerelevant scholarship. Of course, there is by no means universal agreement about this interpretation of diversity. For example, some ethnic scholars have argued that it is demeaning to expect members of minority groups to hold particular viewpoints or to think "differently" from whites, and that recruitment on either of these bases promotes intellectual conformity rather than diversity (e,g,. Carter, 1991), Furthermore, there may be areas in psychology that are more amenable than others to intellectual diversity based on ethnic group membership. Surety an Afrocentric view is more important in the study of interpersonal relations than in the study of neuropsychotogy. In the latter case, as in most natural sciences, intellectual diversity would appropriately be achieved by allowing for the representation of different viewpoints and methodologies, but using criteria that have nothing to do with gender or ethnicity. These complexities aside, for our purposes, we argue that adhering to the affirmative action goal of promoting intellectual diversity in psychology would lead to admissions decisions that give priority to applicants' stated research interests. VOL, 6, NO, 3, MAY 1995 Enhancing the Presence of Role Models A fourth goal of affirmative action is to increase the representation of minorities in academia so that successive cohorts of minority students can identify with minority scholars who are successful (Brooks, 1990). For all minority students, but particularly those on predominantly white campuses, it is considered essential to have the opportunity to observe faculty of color teaching classes, participating in hiring and student admissions decisions, sitting on committees, and otherwise visibly engaging in university business. This goal might also embrace other consequences of being in the public eye, such as exposing nonminority students to an ethnically diverse faculty. But even this seemingly noncontroversial goal might generate disagreement in admissions decisions depending on how one chooses to define the characteristics of a "good" role model. For example, does the Latino faculty member who came from an impoverished background and worked two jobs to put himself through his local college make a better role model than his Harvardeducated counterpart with professional parents who supported him throughout? In other words, must one be perceived as having struggled and having overcome seemingly insurmountable odds to be considered a faculty member students of color will readily identify with? A strong proponent of the role model argument might answer this question in the affirmative and therefore prefer the applicant from an economically disadvantaged background. Here, then, personal history rather than mere ethnic group membership does emerge as relevant. Thus, some ofthe same complex issues about individual suffering that surfaced in the discussion of affirmative action as compensation and correction are also evident in the role model argument. The role model goal might also be either compatible with or in conflict with the intellectual diversity goal. As indicated, a strict interpretation of the diversity argument suggests that candidates who study minority-relevant topics should receive preferential treatment over candidates who do not. At the same time, those who create the intellectually diverse environment are also available to serve as role models, so in this sense, these two goals are complementary. However, in a strict interpretation ofthe role model argument, being a role model should not depend on one's research perspective. That is, for example, the black neuropsychologist and Latino biostatistician serve the same role model functions as do their counterpeirts studying topics more sensitive to race and ethnicity issues. To summarize the discussion thus far, four goals of affirmative action have been articulated as they relate to a variety of complex issues in the graduate student ad143 PSYCHOLOGICAL SCIENCE Affirmative Action in Psychology missions process. In some cases, the goals are compatible and overlapping, leading to general agreement about the criteria for student selection under affirmative action guidelines, ln other instances, the goals may be contentious and divergent, resulting in potential lack of consensus about admissions policies. We now turn from these abstract ideas and speculations to a set of experiments that we conducted to apply our analysis to the decision-making process in graduate admissions. We wanted to ascertain whether academic psychologists' weighting of various admissions criteria conformed to our thinking about affirmative action goals. The general method involved presenting respondents with a set of hypothetical student applicants to their department's graduate program in psychology. The applicants systematically varied on a number of factors pertinent to admissions in general (e.g., grades; Graduate Record Examination, GRE, scores) and to affirmative action goals in particular (e.g,, ethnicity, SES). Respondents rated the likelihood that each candidate would be admitted to their program, METHOD Stimulus Materials In designing the stimuli, we considered a number of factors that might infiuence admissions decisions in some ofthe ways articulated in the previous section. The list of "indispensable" factors was held to five: ethnicity (black. Latino, Asian, or white), academic achievement (excellent or respectable), family SES (working class or professional), country of origin (American born or foreign born), and interest in minority research (yes or no). Native Americans were not included in the ethnicity variable in order to keep the number of manipulated factors at a manageable level. We felt that all these variables should be presented in as ecologically valid a format as possible. That is, rather than having the respondent simply rate the perceived importance of each applicant characteristic, we thought it best to cross these factors systematically in a series of lifelike application packets, A full design crossing all of the variables would have yielded 64 cases for review, an enormous burden for the respondent. Thus, our preference to capture all combinations of variables had to be weighed against the potential loss of subjects. We therefore decided to decompose the investigation into three separate experiments, each of which would examine four of the five factors in a crossed design. Specifically, Study 1 varied ethnicity (four levels), country of origin, SES, and academic achievement. Thus, one "applicant" was a black student who was bom in the Caribbean, came from a working144 class family, and had an excellent academic record (defined as a grade point average, GPA, greater than 3.6 and a combined Quantitative and Verbal GRE score between 1301 and 1400). Study 2 crossed ethnicity (four levels), SES, academic achievement, and interest in minority research (rather than country of origin). Thus, one hypothetical student in this study was an Asian American (all applicants were U.S. bom) who came from a middleclass background, had a respectable academic record (defined as a GPA between 3.0 and 3.2 and a combined GRE score between 1000 and 1150), and was interested in researching ethnic minority issues. For Studies 1 and 2, there were 32 hypothetical applications (4 X 2 X 2 X 2) to be evaluated. To preserve as much ecological validity as possible, stimulus materials were presented in the format of case summaries, A standard form was prepared, mimicking applicant summary forms used at the authors' universities, with spaces for ethnicity, GRE scores, and the rest ofthe information. Manipulated data for each case were written in by hand. The 32 case summaries were presented in a different random order to each respondent. The respondent was instructed simply to make a decision regarding admission to his or her own department based on the data presented, given that these data were admittedly incomplete. Admissions decisions were made on 9-point scales anchored at the extremes with definitely accept and definitely reject and at the midpoint with undecided. Study 3 differed from the first two in that it forced the use of affirmative action criteria in the decision-making process. Thus, respondents were advised that the students to be considered had not been accepted in the general admissions process, but were eligible for consideration under the affirmative action guidelines of the respondents' departments. To be consistent with the cover story that the hypothetical applicants had not been selected in first-round decisions, their academic records were held constant at a respectable level (i.e., GPA between 3.0 and 3.2, GRE between 1000 and 1150). Furthermore, because all candidates were ostensibly eligible for consideration under affirmative action guidelines, the ethnicity variable included only the three minority groups (i.e., black. Latino, and Asian). The other three factors manipulated in Study 3 (SES, country of origin, and interest in minority research) were the same as in Studies 1 and 2. Thus, respondents in Study 3 evaluated 24 cases (3 X 2 X 2 X 2), presented in random orders. Admissions decisions were elicited on the same 9-point scales used in the first two studies. Selection of Respondents A quasi-random selection procedure was used to procure names from the American Psychological Society VOL, 6, NO, 3, MAY 5995 PSYCHOLOGICAL SCIENCE J. Amirkhan et al. (APS) directory, excluding people who were listed by a home or business address rather than an academic institution, affiliated with institutions that did not grant doctoral degrees, and classified as retired or of emeritus status. Six hundred potential respondents were recruited in this manner, with 175 of them randomly assigned to Study 1, 175 to Study 2, and 250 to Study 3. We oversampled Study 3 because it alone specifically asked subjects to consider affirmative action goals. Materials (including the appropriate case summaries, a demographic survey, and a stamped return envelope) were sent to the 600 potential subjects. A cover letter assured recipients that there was no hidden political agenda to this research, and that their responses would be kept anonymous. Reminder postcards were sent to all 600 potential respondents approximately 3 weeks after the initial mailing. Within 3 months ofthe initial mailing, 231 questionnaires, or 39%, had been returned. Of these, 62 either were incomplete or expressed the respondent's unwillingness to participate. Thus, the rate of return for usable data was 28%, a percentage almost identical to that documented by Nelson, Rosenthal, and Rosnow (1986) in a mailed survey with a similar population. Although low, our response rate was not unexpected, given the prior results and other researchers' warning that academic psychologists are not likely to invest much time in anonymous surveys (Zuckerman, Hodgins, Zuckerman, & Rosenthal, 1993). Ofthe 169 respondents with usable data, there were 50 subjects in Study 1, 52 in Study 2, and 67 in Study 3; all these numbers were sufficient for our research purposes. Seventy-four percent were men and 26% were women, with an average age of 46,2 years. The vast majority of respondents were white (95%), with small numbers of Latinos (2.4%), Asians (1.2%), and African Americans (0,6%). Thus, our sample was predominantly white and male, which refiects the ethnic and gender composition of current doctoral psychologists (Kohout & Pion, 1990). These respondents had spent an average of 17,7 years in academia, the majority being full professors (56%), although associate (24%) and assistant (19%) professors were reasonably represented. Most respondents indicated that they typically (62%) or often (16%) participate in graduate admissions decisions. Random assignment procedures apparently were successful, inasmuch as chisquare tests revealed no significant differences in any of these respondent characteristics across the three studies. Before presenting our findings, we wish to acknowledge the limitations of our methodology. Clearly, simulation studies of this type have many drawbacks, not the least of which are the artificiality of the stimuli and the potential for experimental demand. In what follows, we have (by necessity) greatly oversimplified how admisVOL. 6, NO, 3, MAY 1995 sions decisions are actually made. We interpret the data as illustrative of some of the complexity and assumptions underlying affirmative action decision making, rather than as capturing the admissions process as a whole. RESULTS In the analysis, our strategy was to look for pattems of main effects and interactions across the three studies that might suggest adherence to one or more of the articulated affirmative action goals. Thus, documenting consistent main effects of ethnicity, with some minority groups given higher admissions ratings than others (with "all else being equal") could be interpreted as supporting compensation for historical injustice as a goal of affirmative action. The absence of main effects of ethnicity involving the three minority groups would be more consistent with the correction (e.g., representation) and role model goals. Similarly, main effects of research interest, with applicants who reported a commitment to study minority issues being favored over those who did not, would be indicative of support for the intellectual diversity goal. Studies 1 and 2 We turn first to Studies 1 and 2, which included white applicants and manipulated academic record. Applicant ratings in Study 1 were analyzed i n a 4 x 2 x 2 x 2 (Ethnicity x SES x Academic Record x Country of Origin) repeated measures analysis of variance (ANOVA), The ANOVA for Study 2 ratings was based on the same four-factor design, except that the last factor was applicant's interest in minority research (yes or no), rather than country of origin. Table 1 shows the F values only for the significant main effects and interactions in each study. In both studies, the major variance was accounted for by academic record (see Table I and Fig. 1), Figure 1 shows that applicants with a strong academic record were overwhelmingly preferred; this variable accounted for 72% and 71% of the variance in applicant ratings for the two studies, respectively. Thus, respondents were most influenced by traditional "quality" criteria in their decisions about which students to admit to their graduate psychology programs. But respondents also attended to affirmative action concerns, as indicated by the main effects of ethnicity. Figure 1 reveals that in both studies, higher ratings were given to blacks and Latinos, whose ratings did not differ, than to Asians and whites, who also were seen as equivalent. There also was an Ethnicity x Academic Record interaction in each study, which captured the finding that the importance of ethnicity was more evident for students 145 PSYCHOLOGICAL SCIENCE Affirmative Action in Psychology Table 1. F values for significant maineffects and interactions in the three studies Study Effect Academic record Ethnicity SES Country of origin Academic record x ethnicity Academic record x SES Academic record x country of origin Ethnicity x country of origin Research x academic record Research x SES Research x country of origin 1 2 428,12*** 20.55*** 5.82* 5,97* 3,63* 9,22** 6.24* 6.76*** N/A N/A N/A 300,25*** 7,37*** n.s. N/A 5,43** n.s. N/A N/A 4,66* n,s. n,s. 3 N/A 28.82*** 23.84*** 13,26*** N/A N/A N/A n.s. N/A 9,07** 5,74* Note, SES = socioeconomic status, N/A = effect not examined in that study n,s, = nonsignificant effect. In Study 1 (n = 50),factors were academic record, ethnicity, SES, and country of origin, ln Study 2 (n = 52) , factors were academic record, ethnicity. SES, and interest in minority research. In Study 3 (fl = 67), factors were ethnicity, SES, country of origin, and interest in minority research. *p < ,05, **p < ,01, ***p < ,001, with acceptable records than for students with excellent applicants received higher ratings than foreign-born aprecords. The preference for blacks and Latinos (and the plicants in Study 1 (Ms = 5,8 vs, 5,6, p < ,05), By maperception of Asians as a nontargeted minority group) nipulating interest in minority research. Study 2 examsuggests that the compensatory goal of affirmative action ined the intellectual diversity goal, The absence of a main was operative in these admissions decisions. Even effect of research interest and the one relatively weak though the magnitudes of these differences are not large, interaction effect suggest that respondents gave little they may capture the real-life situation, in which small weight to this goal. differences in graduate applicant ratings often determine acceptance or rejection. Study 3 We tum next to the other factors manipulated in these In Study 3, we held academic record constant at a studies to better ascertain which particular goals of affirmative action were being considered. In Study 1, there respectable level, thereby providing an opportunity to were significant effects of SES, both as a main effect and assess independently the effects of the other factors. In in interaction with academic record (see Table 1), Low- addition, we instructed respondents to attend particularly SES applicants were favored over high-SES applicants, to affirmative action goals. Participants rated the admisparticularly when their academic records were more sibility of 24 adequately qualified black. Latino, and modest (Ms = 3,6 vs, 3.4, p < ,05). Also, native-bom Asian hypothetical applicants who varied in SES, country of origin, and interest in minority research in a 3 x 2 X 2 X 2 design. Table 1 shows only the significant main effects and interactions, and Figure 2 depicts the main STUDY 2 findings graphically. Consistent with the first two studies, there was a main effect of ethnicity (see Table 1). Figure 2 shows that this effect was explained by the fact that blacks (M = 5.6) were rated higher on admissibility than Latinos (M = 5.2), who, in tum, were rated higher than Asians (M = 4.8). The difference between each pair of these groups Whitfi was significant (ps < .05). In addition, low-SES candidates (M = 5.2) were preferred over high-SES applicants Fig. t. Mean admissibility rating as a function of ethnicity and (M = 4.9), as were American-bom (M = 5.5) over foreign-bom (M = 4.9) applicants; both findings are consisacademic record: Studies 1 and 2. 146 VOL, 6, NO, 3, MAY 1995 PSYCHOLOGICAL SCIENCE J, Amirkhan et al. Black Latino Ethnic Group Asian Fig. 2. Mean admissibility rating as a futiction of ethnicity and socioeconomic status (SES): Study 3. tent with results from Study 1, The preference for lowSES and American-born students was equally evident in the three ethnic groups (all ps < ,001). As in Study 2, there was no main effect of research interest, and neither of the interactions involving this variable was readily interpretable. DISCUSSION Examining the pattern of findings across the three studies, how might we interpret our respondents' admissions decisions in terms of affirmative action goals? The consistency of the ethnicity main effects across the three studies leads us to believe that compensation for past injustice is a salient goal in graduate psychology admissions. In Studies 1 and 2, white and Asian applicants were rated lower in admissibility than blacks and Latinos, and in Study 3, with whites excluded from the analysis, the rank ordering was blacks, followed by Latinos, followed by Asians. We find it particularly noteworthy that Asians apparently were not perceived as a targeted minority group, despite the fact that they, too, are greatly underrepresented in the field of psychology. In addition, both Studies 1 and 3 showed effects of country of origin, such that native-born applicants were preferred to those who were foreign born. This pattem is also consistent with a compensatory argument inasmuch as the aim is to provide preferential treatment to people who have histories of discrimination or victimization within this country. Interest in minority research was not a central determinant of admissions decisions in either Study 2 or Study 3. This may mean that promoting intellectual diversity is not a Scilient affirmative action goal for these respondents. Or it may mean that admitting students on the basis of research interests must satisfy goals in addition to affirmative action, such as intellectual compatibility VOL, 6, NO, 3, MAY 1995 between faculty and student. As one respondent wrote, "Interest in the study of minority issues does not generally affect admission decisions unless such interests dovetail with those of a faculty member." If facultystudent fit is an important requirement in admissions decisions, but few faculty members are engaged in ethnic research (because so few faculty members are minorities), this state of affairs could undermine intellectual diversity as an affirmative action goal. Results for SES show that the economically disadvantaged applicant was favored over the advantaged applicant in Studies 1 and 3, As indicated in our discussion of goals, this main effect of SES can be interpreted as supportive of either compensatory (historical injustice) or correction (present inequities) goals. It could even connote adherence to the role model argument, depending on how one conceptualizes a "good" role model. Had there been Ethnicity x SES interactions such that, for example, the African American applicant from an impoverished background was particularly favored, a stronger case for the compensation goal might have been made. There probably is more ambivalence and far greater disagreement among respondents concerning the extent to which socioeconomic background, compared with ethnicity, should enter into the affirmative action decision process. If SES carries independent weight as a marker for preferential treatment, among whites as well as ethnic minorities, then the very meaning of affirmative action has undergone a conceptual shift. Not only does evidence of past and present discrimination count as a criterion for preferential treatment, so does the overcoming of personal barriers, as in the case of the poor white from Appalachia who made it to the university. This raises a whole host of related questions about affirmative action decisions. Just as underrepresentation is evidence of discriminatory barriers faced by ethnic minorities, other groups, such as the physically disabled, also face opportunity barriers. Should they receive preferential treatment in admissions decisions? After all, handicapped individuals are also underrepresented in psychology, and surely there is the need for role models, ln other words, would it be prudent to reconceptualize affirmative action as a remedy for past and present barriers to opportunity, be they ethnic, gender related, social status related, or determined by other pertinent historical, cultural, or biological factors? Such a rethinking would certainly broaden the inclusiveness of groups targeted for affirmative action; it would also heighten the competition among targeted groups for increasingly scarce resources. We offer no resolution to these complex philosophical and policy issues. We mainly want to raise the question of what is affirmed (i.e., declared of value) in affirmative action. We want to acknowledge again the limitations of the 147 PSYCHOLOGICAL SCIENCE Affirmative Action in Psychology three experiments, which we conducted to illustrate the main themes articulated here. By manipulating a pertinent set of hypothetical applicants' characteristics, we wanted to focus respondents' attention on what we believe are the important factors underlying affirmative action decisions. The studies were done to provide a forum for discussion, rather than to examine how academic psychologists actually make graduate admissions decisions. In the spirit of this kind of discussion, we conclude the article on a more concrete note with suggestions for how our analysis might aid the general affirmative action process. First, our analysis underscores the need for regular evaluation within academic departments to determine how effectively affirmative action goals are being achieved. For example, if enhancing the presence of role models in academia is one such goal, then departments need to document whether minority graduates are, indeed, competing for and accepting faculty positions in colleges and universities. This need is especially important inasmuch as employment trends suggest that successful minority Ph,D,'s often do not choose jobs in academia, given the lure of private industry, as well as the perceived need to maintain close ties with the community (e,g,, Simon, 1993). Second, evaluation of goal effectiveness also entails knowing when to revise, adjust, or even abandon particular goals in response to changing needs or shifts in the political or legal climate. Several important Supreme Court decisions of the late 1980s involved the rights of unprotected groups (e.g., white mates) in employment discrimination cases and are likely to make it more difficult for universities to sustain aggressive affirmative action goals (Francis, 1993; Lee, 1989; Robinson, Allen, & Abraham, 1992). For example, the Croson decision of 1989 requires that a "compelling case" be made for affirmative action policies. Disallowed as insufficiently compelling under this ruling are general arguments for redressing historical discrimination (i.e., the goal of compensation). Similarly, the Wards Cove case, also decided in 1989, decreed that plaintiffs cannot rely solely on statistics to establish a prima facie case of discrimination (i.e., the goal of correction); a particular discriminatory practice has to be identified first, thus shifting the burden of proof from the employer to the plaintiff. The political climate fostered by these landmark cases may necessitate a shift away from the affirmative action goals of compensation and correction, which generate the most controversy, in favor ofthe ideologically safer diversity and role model goals. 148 Ensuring affirmative action in academia is complex work. We suspect that many academic psychologists will admit to some degree of ambivalence when they weigh their preferred affirmative action strategies against their own fundamental beliefs about equity and faimess. We hope that this article will stimulate our readers to think further about these issues and to discuss them with colleagues who hold divergent viewpoints. Even if there is no consensus on the proper goals of affirmative action, there can at least be greater clarity about the underlying meaning of affirmative action as articulated in these goals. Acknowledgments—The authors are listed alphabetically. Preparation ofthe manuscript was facilitated by Grant No, DBS-9211982 to Sandra Graham and Bernard Weiner from the National Science Foundation, We thank Wendy Berry for her assistance with data entry and analysis. REFERENCES Brooks, R, (1990), Rethinking the American race prohtem, Berkeley; University of California Press. Carter, S, (1991), Reflections of an affirmative action baby. New York: Basic Books, Francis. L, (1993). In defense of affirmative action, ln S. Cahn (Ed,), Affirmative action and the university: A philosophical inquiry (pp, 9—47), Philadelphia: Temple University Press, Glasser, I, (1988). Affirmative action and the legacy of racial injustice. In P, Katz & D, Taylor (Eds,), Eliminating racism (pp, 341-357), New York: Plenum Press, Kohou(, J,, & Pion, G, (1990). Participation of ethnic minorities in psychology: Where do we stand today? In G, Strieker, E, Davis-Russell, E, Bourg, E, Duran, W, Hammond, J, McHolland, K, Polite, & B, Vaughn (Eds,), Toward ethnic diversification in psychology education and training (pp, 153165), Washington, DC: American Psychological Association, Lee, B, (1989, June 28), Recent Supreme Court rulings could disrupt or halt affirmative-action recruiting and hiring in academe. Chronicle of Higher Education, 35. Bl, B3, Myrdal, G, (1944), An American dilemma: The Negro problem and modern democracy. New York: Harper & Brothers, Nelson, N,, i*osenthal, R,, & Rosnow, R, (1986). Interpretation of significance levels and effect sizes by psychological researchers, American Psychologist, 44, 1276-1284, Robinson, R,, Allen, B,, & Abraham, Y, (1992). AfTirmative action plans in the 1990s: A double-edged sword? Public Personnel Management, 21. IM-lll. Simon, R, (1993), Affirmative action and the university: Faculty appointment and preferential treatment, ln S, Cahn (Ed,). Affirmative action and the university: A philosophical inquiry (pp, 48-92), Philadelphia: Temple University Press, Soweil, T, (1989), 'Affirmative action": A world-wide disaster. Commentary, 88, 21-41, Tomes, H, (1993. October), More efforts needed to close diversity gap, APA Monitor, p, 46, Wilson, W, (1987), The truly disadvantaged: The inner city, the underclass, and public policy, Chicago: University of Chicago Press, Wolf-Devine, C, (1993), Proportional representation of women and minorities. In S, Cahn (Ed,), Affirmative action and the university: A philosophical inquiry (pp, 223-232), Philadelphia: Temple University Press, Zuckerman, M,, Hodgins, H,, Zuckerman, A,, & Rosenthal, R, (1993), Contemporary issues in the analysis of data: A survey of 551 psychologists. Psychological Science. 4. 49-53, (RECEIVED 7/7/94; ACCEPTED iO/26/94) VOL, 6, NO, 3, MAY 1995