the stigma of affirmative action: a stereotyping-...

THE STIGMA OF AFFIRMATIVE ACTION: A STEREOTYPING-BASED THEORY AND META-ANALYTIC TEST OF THE

CONSEQUENCES FOR PERFORMANCE

LISA M. LESLIENew York University

DAVID M. MAYERUniversity of Michigan

DAVID A. KRAVITZGeorge Mason University

Affirmative action plans (AAPs) are designed to facilitate workplace success formembers of the groups they target (e.g., women, ethnic minorities), yet may have theironic effect of stigmatizing AAP targets and, in turn, decreasing their performanceoutcomes. Prior work has focused on the stigma of incompetence as the primarymechanism that links AAPs to performance; however, the broader social psychologicalliterature suggests that additional mechanisms may also play a role. We use stereo-typing theories to develop a more comprehensive model of the pathways throughwhich AAPs limit targets’ performance outcomes. Drawing from the stereotype contentmodel, we propose that the negative effect of AAPs on others’ evaluations of targets’performance is driven by perceptions of incompetence and low warmth. Drawing fromstereotype threat theory, we propose that the negative effect of AAPs on targets’self-evaluated and objective performance is driven by perceptions of low self-compe-tence, negative state affect, and perceived stereotyping by others. Meta-analytic pathanalyses support our hypotheses. Our theory and findings demonstrate that multiplemechanisms explain the negative consequences of AAPs for targets’ performanceoutcomes, highlight differences in reactions to AAP targets by others versus the self,and provide insight into preventing the unintended negative effects of AAPs.

Despite improvements in recent decades, womenand ethnic minorities continue to face employmentdiscrimination and other barriers to advancement(e.g., Bertrand & Mullainathan, 2004; Blau & Kahn,2007; Leslie, King, Bradley, & Hebl, 2008). To levelthe playing field, organizations around the globehave implemented affirmative action plans (AAPs),which are policies designed to improve work out-comes for underrepresented groups by providingthem with extra help in the employment process

(Jain, Sloane, & Horwitz, 2003; Yang, D’Souza,Bapat, & Colarelli, 2006). AAPs increase the num-ber of women and ethnic minorities in managerialpositions (Holzer & Neumark, 2000; Kalev, Dobbin,& Kelly, 2006; Leonard, 1984), and thus help pro-mote organizational diversity and redress societalinjustice.1

Nevertheless, AAPs are not without drawbacks.They can stimulate backlash among non-beneficia-ries who may feel unfairly disadvantaged by thesepolicies (Harrison, Kravitz, Mayer, Leslie, & Lev-Arey, 2006; Lynch, 1992; Shteynberg, Leslie,Knight, & Mayer, 2011). In addition, AAPs cancause the very employees they are intended to ben-

This research was conducted with support from aGrant-in-Aid from the Society for the PsychologicalStudy of Social Issues. Portions of the results were pre-sented at the 2007 annual conference of the Society forIndustrial and Organizational Psychology and the 2011annual conference of the International Association forConflict Management. We are grateful to Yeonka Kim forher research assistance. We also thank Ray Sparrowe andthree anonymous reviewers for their feedback.

1 For the sake of parsimony, we use the term “ethnic-ity” to refer to the confluence of race (e.g., Asian, Black,Native American, White) and ethnicity (Hispanic, non-Hispanic) (cf. Leslie, Snyder, & Glomb, 2013; Phin-ney, 1996).

964

� Academy of Management Journal2014, Vol. 57, No. 4, 964–989.http://dx.doi.org/10.5465/amj.2011.0940

Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s expresswritten permission. Users may print, download, or email articles for individual use only.

efit to be stigmatized as incompetent by both othersand the self (e.g., Heilman, 1994). Existing theoryused to explain the stigma of incompetence hasbeen grounded primarily in discounting and re-lated notions of self-doubt. Specifically, the pres-ence of an AAP raises the possibility that membersof the groups the AAP targets were hired or pro-moted due to their demographics, not their qualifi-cations. Scholars have theorized that others there-fore discount the possibility that AAP targets arecompetent (e.g., Garcia, Erskine, Hawn, & Casmay,1981; Heilman, Block, & Lucas, 1992) and, in aparallel fashion, that AAPs and the associated pos-sibility that demographics played a role in selec-tion cause AAP targets to doubt their self-compe-tence (e.g., Heilman, Simon, & Repper, 1987;Niemann & Dovidio, 2005). Perceived incompe-tence is likely to result in poor performance out-comes (e.g., Heilman & Alcott, 2001). To the extentthat AAPs and the associated stigma of incompe-tence limit targets’ performance, AAPs may havethe opposite of their intended impact.

Understanding the mechanisms that drive theunintended consequences of AAPs is essential forensuring that AAPs facilitate, rather than constrain,targets’ work outcomes. Existing theory groundedin discounting and self-doubt has focused on thestigma of incompetence, yet the broader social psy-chological literature suggests that additional mech-anisms contribute to the effects of AAPs on perfor-mance. For example, AAPs increase the ability oftarget groups to compete for jobs and other work-place resources. Several theories of stereotypingand intergroup relations suggest that a targetgroup’s competitiveness threatens others and leadsto perceptions of the competing group as cold andunlikable (e.g., Bobo, 2000; Fiske, Cuddy, Glick, &Xu, 2002), which negatively affects others’ evalua-tions of the competing group’s performance (Cardy& Dobbins, 1986; Nisbett & Wilson, 1977). Thus, thestigmatization of AAP targets by others may bemore complex than prior theory suggests, such thatothers perceive AAP targets as lacking both compe-tence and warmth, each of which may contribute tothe negative effect of AAPs on others’ evaluationsof targets’ performance.

Similarly, in the case of targets’ own reactions toAAPs, mechanisms other than the stigma of incom-petence may link AAPs to performance. Prior worksubstantiates that AAPs negatively affect targets’perceived self-competence (e.g., Heilman et al.,1987), and are thus likely to threaten targets’ desireto view the self favorably. A number of theories in

the broader literature on stereotyping and stigmasuggest that threats to a positive self-image triggernegative emotions, such as stress and anxiety (e.g.,Major & O’Brien, 2005; Steele, Spencer, & Aronson,2002), which limit performance (Kaplan, Bradley,Luchman, & Haynes, 2009; Schmader, Johns, &Forbes, 2008). Thus, AAP targets are likely to ex-perience negative state affect, in addition to lowself-competence, both of which may affect theirperformance negatively.

In summary, prior theory on why AAP targets arestigmatized by others and by the self is groundedprimarily in discounting and self-doubt, respec-tively, and focuses on the stigma of incompetence;however, the broader social psychological litera-ture suggests that additional other-driven (e.g., per-ceived warmth) and self-driven (e.g., state affect)mechanisms contribute to understanding whyAAPs negatively affect targets’ performance out-comes. An alternative theoretical framework maytherefore be needed to gain a fuller understandingof the stigma of AAPs and the associated conse-quences for performance. To this end, we drawfrom stereotyping theories and build a more com-prehensive model in which the stigma of incompe-tence is only one of several mechanisms throughwhich AAPs negatively affect targets’ performanceoutcomes.

In using the stereotyping literature to advancetheory regarding the unintended negative effects ofAAPs, we rely on two specific theories: the stereo-type content model (SCM) and stereotype threattheory (STT). In the literature on stereotyping andsocial cognition more broadly, self–other discrep-ancies—or differences in individuals’ perceptionsof others in a given situation versus their percep-tions of the self when placed in the same situa-tion—are well documented (Jones & Nisbett, 1972;Malle & Knobe, 1997; Miller & Ross, 1975; Pronin,Lin, & Ross, 2002). It is therefore not surprising thatscholars have developed distinct theories that de-tail how others react to negatively stereotyped in-dividuals versus how individuals react to negativestereotypes about the self. Specifically, the SCMand STT provide comprehensive accounts of howindividuals are stereotyped by others and by theself, respectively, and also detail how negative ste-reotypes affect a number of outcomes, includingperformance (Cuddy, Glick, & Beninger, 2011;Fiske et al., 2002; Schmader et al., 2008; Steele etal., 2002). Consistent with the two fundamentalstereotype content dimensions identified in theSCM, we theorize that others perceive AAP targets

2014 965Leslie, Mayer, and Kravitz

as lacking competence and warmth, both of whichnegatively affect others’ evaluations of targets’ per-formance. Consistent with key mechanisms identi-fied in STT, we theorize that AAP targets experi-ence low self-competence, negative state affect, andperceived stereotyping by others, each of whichnegatively affects their self-evaluated and objectiveperformance.

We use path-analytic meta-analyses to test ourhypotheses. Prior work has focused primarily onthe effects of AAPs on competence and perfor-mance. Researchers, however, have also exploredthe effects of AAPs on other outcomes, includingwarmth, state affect, and perceived stereotyping(e.g., Brutus & Ryan, 1998; Heilman et al., 1992;Taylor, 1994). We are therefore able to test ourtheory with meta-analysis. Although our analysesare based on existing studies, we depart from priorwork by testing a model in which competence,warmth, affect, and perceived stereotyping eachcontribute to the effect of AAPs on performance.

Our stereotyping-based theory and meta-analytictest advance knowledge by demonstrating that ste-reotyping theories provide a more comprehensiveaccount of the unintended negative effects of AAPs.Consistent with prior theory, we propose that AAPsimpede targets’ performance because targets areperceived as incompetent by both others and theself. At the same time, we build new theory byproposing that additional mechanisms contributeto the effect of AAPs on performance and that thereare some differences in others’ reactions to AAPtargets (e.g., perceived low warmth) versus theself’s own reactions to being an AAP target (e.g.,negative state affect). A fuller understanding of thepathways through which AAPs impede perfor-mance is essential for preventing AAPs from hav-ing the opposite of their intended impact. Thus,from a practical standpoint, our research suggestsnovel strategies for ensuring that AAPs do not havethe ironic effect of limiting the workplace outcomesof their intended beneficiaries.

THEORY DEVELOPMENT

AAPs are policies designed to improve the em-ployment outcomes of members of underrepre-sented social groups by providing them with extrahelp in the employment process (cf. Kovach, Krav-itz, & Hughes, 2004). AAPs originated with Execu-tive Order 11246, implemented by President John-son in 1965. Notably, AAPs differ from equalemployment opportunity (EEO) policies, which

originated with the Civil Rights Act of 1964. Spe-cifically, EEO policies mandate non-discrimina-tion, but AAPs go further by allowing positive ac-tions aimed at helping target groups (Kovach et al.,2004). Moreover, EEO policies are identity-blindpolicies that focus on equal treatment regardless ofgroup membership, whereas AAPs are identity-conscious policies that target specific groups(James, Brief, Dietz, & Cohen, 2001; Konrad & Lin-nehan, 1995).

The positive actions provided by AAPs take dif-ferent forms that vary in strength (Harrison et al.,2006; Kovach et al., 2004). For example, opportu-nity enhancement (OE) AAPs—the weakest AAPtype—provide extra resources to AAP targets, butdo not give demographics any weight in employ-ment decisions (e.g., targeted recruitment). Weakpreferential treatment (WPT) AAPs give targetspreference in employment decisions if and only iftheir qualifications are equivalent to those of non-targets. Finally, strong preferential treatment (STP)AAPs give demographics a positive weight in em-ployment decisions, in addition to qualifications.For example, an AAP target may be hired over amore qualified non-target, assuming the targetmeets the minimum qualifications for the job. De-spite this variability, all AAP types share the com-mon goal of providing extra help to members ofunderrepresented groups to enable them to achievemore favorable outcomes.

Paradoxically, AAPs can stigmatize the very em-ployees they are intended to benefit (e.g., Garcia etal., 1981; Heilman, 1994). Current theory focuseson the stigma of incompetence and is groundedprimarily in discounting and self-doubt. Specifi-cally, qualifications are the typical causal factor inemployment decisions, which leads to perceptionsthat job incumbents are competent. AAPs raise thepossibility that demographics also influence em-ployment decisions. In the case of stigmatization byothers, scholars have drawn from the discountingprinciple, which states that the perceived impor-tance of a given cause is diminished when plausi-ble alternatives exist (Kelley, 1972) and have theo-rized that the assumption that demographicsinfluence employment decisions causes others todiscount the possibility that AAP targets are com-petent (e.g., Heilman et al., 1992; Resendez, 2002).In the case of stigmatization by the self, scholarshave not drawn explicitly from discounting, buthave similarly theorized that AAPs and the associ-ated assumption that demographics influence em-ployment decisions cause AAP targets to doubt

966 AugustAcademy of Management Journal

their self-competence (e.g., Brutus & Ryan, 1998;Heilman et al., 1987). Others’ perceptions that AAPtargets are incompetent are likely to lead others toevaluate targets’ performance negatively, and AAPtargets’ self-perceptions of incompetence are likelyto lead to low self-evaluated and objective perfor-mance. Consistent with theory, evidence indicatesthat AAPs negatively affect the perceived compe-tence and performance outcomes of AAP targets(e.g., Evans, 2003; Heilman & Blader, 2001; Hei-lman et al., 1987; Resendez, 2002).

Existing theory provides a compelling account ofwhy AAP targets are stigmatized as incompetent.At the same time, the stereotyping literature sug-gests that additional mechanisms may contribute tounderstanding why AAPs impede targets’ perfor-mance outcomes. We therefore use theories of stereo-typing to build a more comprehensive model of themechanisms that link AAPs to performance. Notably,distinct stereotyping theories have emerged that de-tail how others react to negatively stereotyped in-dividuals versus how individuals react to negativestereotypes about the self. Thus, we use differenttheories to advance understanding of the other-driven (SCM) and self-driven (STT) processes thatexplain the negative effects on AAPs on targets’performance outcomes.

Stereotype Content and Others’ Reactions to AAPTargets

The SCM (Fiske et al., 2002) is a useful frame-work for building theory regarding why and howothers react to AAP targets, given that this theoryprovides a comprehensive account of the dimen-sions along which individuals are negatively ste-reotyped by others. Although any number of morespecific stereotypes may exist, a core tenet of theSCM is that the two overarching dimensions ofstereotype content are competence and warmth.The basis for the two dimensions of stereotypecontent stems from individuals’ basic need to un-derstand whether others have the capacity and in-tent to do them harm. Individuals are thereforemotivated to evaluate others in terms of their com-petence (i.e., capacity) and their warmth (i.e., in-tent) (Cuddy, Fiske, & Glick, 2008; Fiske et al.,2002; Fiske, Cuddy, & Glick, 2007). Consistent withtheory, research indicates that competence andwarmth are the primary dimensions of person per-ception and explain more than 80% of individuals’global evaluations of others (Asch, 1946; Woj-ciszke, Bazinska, & Jaworski, 1998). In addition to

identifying competence and warmth as the primarydimensions of stereotype content, the SCM alsospecifies the conditions under which individualsform competence- and warmth-based stereotypesabout others, as well the consequences of thesestereotypes (e.g., Cuddy, Fiske, & Glick, 2007;Cuddy et al., 2011; Fiske et al., 2002). We thereforedraw from the SCM and build theory regarding whyothers are likely to perceive AAP targets as lackingboth competence and warmth, as well as why bothof these stereotypes cause others to evaluate AAPtargets’ performance negatively.

According to the SCM, status is the key anteced-ent of competence-based stereotypes and lack ofcompetition is the key antecedent of warmth-basedstereotypes (Cuddy et al., 2008; Fiske et al., 2002).More specifically, the status–competence linkageis explained by just-world beliefs; namely, the ba-sic human motivation to believe that individualsget what they deserve (Cuddy et al., 2008). Individ-uals assume that members of low status groupsdeserve their low status because they lack compe-tence, whereas members of high status groups haveearned their status through competence and hardwork. The connection between lack of competitionand warmth is explained by the basic human mo-tivation to enhance self-interest (Cuddy et al.,2008). Individuals seek to maximize their own out-comes and therefore view those with whom theymust compete for valuable resources as a source offrustration. Moreover, individuals assume that oth-ers’ competitive behaviors stem from malice andintent to do harm, and groups who are able tocompete for resources are therefore perceived aslacking in warmth. Consistent with theory, evi-dence indicates that a group’s status positively af-fects the perceived competence of group membersand that a group’s ability to compete for resourcesnegatively affects the perceived warmth of groupmembers (Caprariello, Cuddy, & Fiske, 2009; Fiskeet al., 2002).

AAPs simultaneously signal that AAP targetshave low status and increase targets’ ability to com-pete for resources. Drawing from the SCM, wetherefore propose that others stigmatize AAP tar-gets as lacking in both competence and warmth.AAPs, by definition, provide extra help to targetgroup members to improve their employment out-comes. The notion that certain groups require extrahelp is likely to create perceptions that the grouphas poor outcomes and thus occupies a position oflow status (e.g., Maio & Esses, 1998; Steele, 1990).Given that AAPs signal that targets have low status,


and status is the key antecedent of competence,others are likely to perceive members of a givengroup as less competent when the group is targetedby an AAP than when it is not. AAPs are similarlylikely to affect perceived warmth. The goal of theextra help provided by AAPs is to enable targetgroups to compete for workplace resources (e.g.,jobs, promotions). Given that AAPs increase tar-gets’ competitiveness, and lack of competition isthe key antecedent of warmth, others are likely toperceive members of a given group as less warmwhen the group is targeted by an AAP than when itis not (e.g., Aquino, Stewart, & Reed, 2005).

Importantly, stereotypes are context dependent(Crocker, Major, & Steele, 1998) and the presence ofan AAP in a given setting is therefore likely to affectthe perceived competence and warmth of anygroup it targets in that setting, regardless ofwhether the group is chronically negatively stereo-typed in the broader society. For example, ethnicminorities are generally stereotyped as lower incompetence than Whites, yet both groups are seenas incompetent if portrayed as poor and as compe-tent if portrayed as professionals (e.g., Fiske et al.,2002). Thus, AAPs are likely to affect the perceivedcompetence and warmth of any group they target.

The tendency to view AAP targets as lacking inboth competence and warmth is consequential,given that stereotypes of incompetence and lowwarmth negatively affect how others view and treatstereotype targets. Research on the consequences ofcompetence and warmth stereotypes has focusedlargely on emotions (e.g., contempt) and behaviors(e.g., harassment) (Cuddy et al., 2007). More re-cently, however, SCM scholars have suggested thatcompetence and warmth judgments also affectworkplace evaluations (Cuddy et al., 2011). Specif-ically, others’ evaluations of a target’s performanceare vulnerable to subjective influences, such thatindividuals perceive others in stereotype-consis-tent ways; for example, by seeking out and attend-ing to stereotype-confirming, but not stereotype-disconfirming, information (e.g., Dunning &Sherman, 1997; Feldman, 1981). If others perceivea target as competent, and thus have high perfor-mance expectations for that target, they are moreattentive to that target’s successes than failures,resulting in favorable impressions of performance,whereas the reverse is true if others perceive atarget as incompetent. Thus, the perceived compe-tence of a target is likely to be positively related toevaluations of that target’s performance.

Perceived warmth is also likely to shape perfor-mance evaluations. Due to preferences for consis-tency, evaluators are subject to a halo effect. Spe-cifically, if a target is positively stereotyped on agiven dimension, evaluators’ subsequent judg-ments—even on unrelated dimensions—tend to beconsistent with that positive impression (Feldman,1981; Nisbett & Wilson, 1977). For example, evi-dence documents that positive perceptions of a tar-get’s warmth lead to positive judgments of thattarget’s physical attractiveness (Nisbett & Wilson,1977). As a result, when others evaluate the perfor-mance of a target they perceive as warm, they seekout and attend to information that confirms theirpositive impression, whereas the reverse is true fora target they perceive as cold. Thus, others’ percep-tions of a target’s warmth, similar to perceptions ofcompetence, are likely to be positively related toothers’ evaluations of the target’s performance.

The above arguments suggest that perceptions ofincompetence and low warmth are both mecha-nisms through which AAPs negatively affect oth-ers’ evaluations of AAP targets’ performance (seeFigure 1A). Notably, the predicted effect of AAPson perceived competence is consistent with priorwork grounded in discounting, which has focusedon the stigma of incompetence. Our SCM-basedtheory advances prior work, however, by suggest-ing that a stigma of low warmth is a second mech-anism that links AAPs to evaluations of targets’performance and thus provides a more comprehen-sive account of the stigmatizing effects of AAPs.

Hypothesis 1. The presence of an AAP has anindirect negative effect on others’ evaluationsof AAP targets’ performance through perceivedcompetence.

Hypothesis 2. The presence of an AAP has anindirect negative effect on others’ evaluationsof AAP targets’ performance through perceivedwarmth.

Stereotype Threat and Self-Reactions amongAAP Targets

The SCM provides a useful framework for build-ing theory regarding how others react to AAP tar-gets, but is less relevant to the self’s own reactionsto being an AAP target. For example, we draw fromthe SCM and propose that, because AAPs increasetargets’ ability to compete for resources andthreaten non-targets’ work outcomes, others per-ceive AAP targets as low in warmth. Yet this logic


does not apply to targets’ self-reactions becauseAAPs increase, rather than decrease, targets’ likeli-hood of receiving resources, and are thus unlikelyto lead to self-perceptions of low warmth. It istherefore not surprising that distinct theorieshave emerged that explain reactions to negativestereotypes about the self. In particular, STT(Steele et al., 2002) provides a detailed account ofhow individuals react to being negatively stereo-typed, as well as the consequences for their per-formance. We therefore draw from STT to buildtheory regarding the self-driven process that linkAAPs to performance.

Stereotype threat—defined as the risk of con-firming that a negative stereotype about one’s groupapplies to the self—is a psychological state with thepotential to inhibit performance (Steele, 1997;Steele & Aronson, 1995; Steele et al., 2002). Ac-cording to STT, negative cognitions (e.g., low self-competence) and negative state affect (e.g., anxiety,fear, etc.) are two key mechanisms that play a rolein explaining why negative stereotypes limit per-formance (Cadinu, Maass, Rosabianca, & Kiesner,

2005; Schmader et al., 2008). Notably, stereotypethreat is not a unitary phenomenon, but takes dif-ferent forms depending on the source of the threat(Shapiro & Neuberg, 2007). Specifically, self-as-source stereotype threat is driven by one’s ownconcern that a group stereotype may be true of theself, whereas other-as-source stereotype threat isdriven by a concern that others may believe agroup stereotype is true of the self. Self-as-sourcestereotype threat therefore suggests a direct effectof negative stereotypes on self-competence andstate affect, whereas other-as-source stereotypethreat suggests an indirect effect through per-ceived stereotyping by others. Drawing from STT,we build theory regarding the negative effects ofAAPs and propose that self-competence, state af-fect, and perceived stereotyping serve as mecha-nisms through which AAPs negatively affect tar-gets’ performance.

According to self-as-source STT, negative stereo-types breed negative cognitions, including self-doubt (e.g., Cadinu et al., 2005), and negative affect,including anxiety and fear (e.g., Keller & Dauen-

FIGURE 1Hypothesized Mediated Models

A) Others’ Reactions to AAP Targets; B) Self-Reactions among AAP Targets

Hypothesis 1:Hypothesis 2:

AAP � Competence � PerformanceAAP � Warmth � Performance

B)

Hypotheses 3a–3b:Hypotheses 4a–4b:

AAP � Self-Competence � PerformanceAAP � State Affect � Performance

Hypotheses 5a–5b: AAP � Perceived Stereotyping � Self-Competence � PerformanceHypotheses 6a–6b: AAP � Perceived Stereotyping � State Affect � Performance

Affirmative Action Plan

(AAP)

Perceived Target

Competence

Perceived Target

Warmth

Evaluationsof Target

Performance

– +

– +

Affirmative Action Plan

(AAP)

Perceived Self-

Competence

State Affect

Perceived Stereotyping by

Others

PerformanceSelf-evaluated•

• Objective

+

+

+

–

–

–

–

A)


heimer, 2003; Osborne, 2001). If a group stereotypeof incompetence exists, group members questionwhether the stereotype applies to them personallyand thus doubt their own competence. Negativestereotypes also increase negative affect; stereo-types are a source of anxiety because they are in-compatible with a self-interested desire to view theself favorably and lead to a fear of performingpoorly and confirming that a stereotype is true ofthe self (Schmader et al., 2008; Spencer, Steele, &Quinn, 1999; Steele et al., 2002). For example, con-sistent with theory, the presence of negative stereo-types about women’s math ability causes women toreport low self-competence and negative state af-fect, and negative stereotypes about ethnic minori-ties’ academic ability have corresponding effects onethnic minority individuals (Cadinu et al., 2005;Keller & Dauenheimer, 2003; Osborne, 2001; Spen-cer et al., 1999; Steele & Aronson, 1995).

Stereotype threat research has focused on stereo-types about women’s math ability and ethnic mi-norities’ academic ability that are activated, for ex-ample, by indicating that group-based performancedifferences exist on a given test (e.g., Cadinu et al.,2005; Keller & Dauenheimer, 2003; Spencer et al.,1999). AAPs also are likely to activate a negativestereotype and thus affect targets’ self-competenceand state affect. Specifically, AAPs provide targetgroups with extra help in the employment process,signaling that they cannot achieve favorable out-comes on their own and creating the stereotype thatAAP targets lack competence. To the extent thatAAPs create a negative stereotype, STT suggeststhat individuals experience lower self-competenceand higher negative state affect when their group isthe target of an AAP than when it is not.

Notably, like the SCM, STT suggests that stereo-type threat is a context-dependent phenomenonthat can affect any group in certain settings, even ifthe group is not chronically negatively stereotypedin the broader society (Aronson, Lustina, Good,Keough, Steele, & Brown, 1999; Steele et al., 2002).For example, Whites are generally stereotyped ashigh in competence, yet fall victim to stereotypethreat in contexts where their performance is com-pared to that of a higher ability group, such asAsians (Aronson et al., 1999). Thus, STT suggeststhat the presence of an AAP in a given settingcreates a negative stereotype about any group ittargets in that setting and, in turn, has detrimentalconsequences for the perceived self-competenceand state affect of AAP targets.

The negative cognitions (i.e., low perceived self-competence) and state affect (e.g., anxiety, fear)triggered by negative stereotypes inhibit self-eval-uated and objective performance. STT holds thatindividuals engage in self-regulatory processes,such that they seek to suppress unwanted negativecognitions and negative affect in an attempt to pre-vent them from impeding performance (Schmaderet al., 2008). Yet attempts at suppression requireeffort and diminish individuals’ cognitive capacityand ability to concentrate (e.g., Gross, 2002; Mu-raven & Baumeister, 2000). Cognitive resources arecritical to performance, and such attempts at sup-pression therefore have the paradoxical effect ofinhibiting performance (Schmader et al., 2008;Schmader & Johns, 2003). Consistent with theory,evidence indicates that negative stereotypes impairthe performance of members of the groups theytarget and that this effect is driven by both lowself-competence and negative state affect (e.g.,Cadinu et al., 2005; Keller & Dauenheimer, 2003;Osborne, 2001). It follows that the low self-compe-tence and negative state affect triggered by AAPsare likely to inhibit AAP targets’ self-evaluated andobjective performance.

The above arguments, derived from self-as-sourceSTT, suggest that AAPs negatively affect the self-evaluated and objective performance of AAP targetsthrough two pathways. Specifically, the negative ef-fect of the presence of an AAP on targets’ perfor-mance outcomes is explained by both low self-com-petence and negative state affect (see Figure 1B).

Hypothesis 3. The presence of an AAP has anindirect negative effect on AAP targets’ (a) self-evaluated and (b) objective performancethrough perceived self-competence.

Hypothesis 4. The presence of an AAP has anindirect negative effect on AAP targets’ (a) self-evaluated and (b) objective performancethrough state affect.

In addition to the two hypotheses derived fromself-as-source STT, other-as-source STT suggeststhat AAPs affect performance through two otherpathways. Negative group stereotypes have not only adirect effect on self-competence and state affect, butalso indirect effects through perceived stereotypingby others. When a negative group stereotype exists,members of that group are aware that others maybelieve the stereotype is true of them personally (e.g.,Steele & Aronson, 1995; Steele et al., 2002). To theextent that AAPs provide target groups with extra


help, and thus create a stereotype of incompetence,individuals are more likely to perceive that othersbelieve a stereotype of incompetence applies to thempersonally when their group is targeted by an AAPthan when it is not.

Perceived stereotyping by others, in turn, islikely to negatively affect self-competence and stateaffect. In addition to the direct effect of negativestereotypes on self-competence, targets’ percep-tions that others believe a stereotype of incompe-tence applies to them personally reinforces the pos-sibility that the stereotype is true and causes targetsto further doubt their self-competence (Gunderson,Ramirez, Levine, & Beilock, 2012; Jacobs, 1991; Par-sons, Adler, & Kaczala, 1982). Perceived stereotyp-ing by others also increases negative affect by cre-ating anxiety and fear of performing poorly andconfirming that the stereotype is true of the self, notonly in one’s own mind, but also in the eyes ofothers (Steele & Aronson, 1995; Steele et al., 2002).As previously noted, low self-competence and neg-ative state affect inhibit performance. Thus, draw-ing from other-as-source STT, we propose thatAAPs increase perceived stereotyping by others,which inhibits self-competence, state affect, and,ultimately, targets’ self-evaluated and objectiveperformance (see Figure 1B).

Hypothesis 5. The presence of an AAP has anindirect negative effect on the (a) self-evalu-ated and (b) objective performance of AAP tar-gets through perceived stereotyping by othersand perceived self-competence.

Hypothesis 6. The presence of an AAP has anindirect negative effect on the (a) self-evalu-ated and (b) objective performance of AAP tar-gets through perceived stereotyping by othersand state affect.

In summary, we draw from STT and theorize thatAAPs negatively affect targets’ self-evaluated andobjective performance through four pathways (i.e.,Hypotheses 3–6). Notably, consistent with priortheory grounded in self-doubt, we propose that lowperceived self-competence explains why AAPsnegatively affect targets’ performance outcomes, atleast in part. At the same time, our hypothesesdiffer from past work by providing a more compre-hensive account of the self-driven mechanisms thatexplain the negative effects of AAPs on targets’performance. In particular, we propose that AAPslead to negative state affect and perceived stereo-typing by others, in addition to low perceived self-

competence, and that the pathways through whichAAPs have indirect negative effects on targets’ per-formance include self-competence, negative affect,perceived stereotyping and self-competence, andperceived stereotyping and negative affect.

METHOD

We used a meta-analytic approach to test ourhypotheses. To identify relevant studies, wesearched databases—including ABI-Inform, Busi-ness Source Premier, ERIC, and PsycINFO—as wellas the programs for the annual meetings of theAcademy of Management and Society for Industrialand Organizational Psychology using Boolean com-binations of relevant search terms (“affirmative ac-tion,” “preferential selection,” and “employmentdiscrimination” with “stigma”, “competence,”“evaluation,” “judgment,” “perception,” “qualifi-cation,” “warmth,” “affect,” “attitude,” “perfor-mance,” and “career”). No start date was specifiedand our search included papers available throughthe end of 2011. We also manually searched 19high-quality management journals, identified byGomez-Mejía and Balkin (1992), and six high-qual-ity social psychology journals that often publishaffirmative action research (from 1980 through2011).2 Finally, we requested unpublished work byposting on discussion groups and contacting affir-mative action scholars.

Studies were included in the meta-analysis ifthey investigated the effects of AAPs on one of thedependent variables of interest and reported thestatistics necessary to calculate effect sizes (e.g.,t-statistics, correlations, etc.). We conducted an ini-tial inspection of each paper identified through thesearch procedures, which resulted in 85 potentiallyrelevant manuscripts. The 85 manuscripts werecoded in greater detail by one of the authors or aresearch assistant. The first author then indepen-dently coded all studies initially coded by someoneelse (40% of studies). Agreement in the coding wasalmost perfect (kappa � .87; Landis & Koch, 1977)and all coding discrepancies were resolvedthrough discussion. The coding revealed that 35manuscripts met our inclusion criteria. These

2 The social psychology journals were Basic and Ap-plied Social Psychology, the Journal of Applied SocialPsychology, the Journal of Experimental Social Psychol-ogy, the Journal of Personality, the Journal of Personalityand Social Psychology, and the Personality and SocialPsychology Bulletin.


manuscripts included data from 6,432 individu-als, across 45 independent samples. Tables 1 and2 summarize the studies included in the sample.

Independent and Dependent Variables

The independent variable was the extent towhich an AAP was present in the study setting. Thestudies in the sample assessed the presence of anAAP in one of two ways. Researchers either con-ducted a field study and asked participants to re-port the extent to which an AAP was present intheir organization or conducted a laboratory studyand manipulated the extent to which an AAP waspresent—for example, by providing participantswith a rationale for why they were chosen for atask. Most laboratory studies manipulated the pres-ence versus absence of an AAP, but some manipu-lated the presence of a stronger versus weaker AAP.

We conducted separate analyses for others’ reac-tions to AAP targets and AAP targets’ own reac-tions. The variables included in the other-drivenmodel were perceptions of targets’ competence,warmth, and performance. For competence, weincluded both ability-based competence (e.g.,competence, ability, efficacy, qualifications) andagency-based competence (e.g., ambition, po-tency, activity). For warmth, participants ratedtargets’ interpersonal characteristics (e.g.,warmth, sincerity, likability). Performance evalua-tions included perceptions of targets’ actual andexpected performance either on a specific task or intheir career in general.

The variables included in the self-driven modelwere perceived stereotyping by others, self-compe-tence, state affect, and self-evaluated and objectiveperformance. Perceived stereotyping captured par-ticipants’ perceptions regarding the extent to whichthey expected to be evaluated as incompetent byothers. Perceived self-competence included bothability-based and agency-based competence. Stateaffect captured participants’ ratings of their generalaffect at a given point in time (e.g., nervous–calm,tense–relaxed), coded so that high ratings indicatepositive affect and low ratings indicate negativeaffect.3 Performance evaluations included partici-

pants’ perceptions of their actual performance andexpectations for future performance. Objective per-formance included objective measures of partici-pants’ task performance.

Meta-Analytic Procedures

We used Hunter and Schmidt’s (2004) randomeffects methodology to calculate meta-analytic ef-fect sizes. The analyses were conducted in SAS,using the programs provided by Arthur, Bennett,and Huffcutt (2001). We first transformed the effectsize reported in each study into a correlation andcorrected the correlations for unreliability. If thereliability of a variable was not reported in a givenstudy, we used the average reliability reported inother studies for the same variable. We assumedperfect reliability for manipulations of the presenceof an AAP, which is a conservative approach.When a sample contributed more than one effectsize for a given dependent variable, we transformedthe correlations to z-scores, averaged the z-scores,and used the back-transformed average correlationin the analyses to maintain independence acrosseffect sizes (Hunter & Schmidt, 2004). Followingprior work, we calculated a meta-analytic effectsize for a given dependent variable if at least threestudies were available (e.g., Bhaskar-Shrinivas,Harrison, Shaffer, & Luk, 2005; Dalton, Certo, &Roengpitya, 2003).

We report the uncorrected sample-weightedmean effect sizes (r), the variance of the uncor-rected effect sizes (var. r), the corrected sample-weighted mean effect sizes (�), and the variance ofthe corrected effect sizes (var. �). We also report95% confidence intervals (CI95) and conclude thatan effect is significant if the CI95 excludes 0. Wereport the Q-statistic and the credibility intervalaround �; significant Q-values and/or credibilityintervals that are large or include 0 indicate thatthere is substantial variability across studies andsuggest that moderators may be present. Finally, toaddress the file drawer problem (i.e., the possibilitythat non-significant studies are unlikely to be pub-lished and, therefore, unlikely to be included inmeta-analyses), we calculated the fail-safe k statis-tic (Dalton, Aguinis, Dalton, Bosco, & Pierce, 2012;Orwin, 1983; Rosenthal, 1979). Fail-safe k (kFS) in-dicates the number of additional studies, each pro-ducing an effect size of 0, that would be needed tonullify a significant effect.

To identify potential outliers, we calculated thesample-adjusted meta-analytic deviancy (SAMD)

3 There is a debate regarding whether positive and neg-ative state affect are distinct constructs or bipolar ends of asingle dimension (e.g., Brief & Weiss, 2002). Consistentwith the approach taken in the studies in our sample, wetreat state affect as a unidimensional construct.


TA

BL

E1

Stu

die

sof

Oth

ers’

Rea

ctio

ns

toA

AP

Tar

gets

Stu

dy

AA

PT

arge

tG

rou

p(s

)A

AP

Typ

e(s)

aR

esea

rch

Set

tin

gP

erce

ived

Com

pet

ence

bP

erce

ived

War

mth

bP

erfo

rman

ceE

valu

atio

nsb

Aqu

ino

etal

.(2

005)

,S

tud

y1

Eth

nic

min

orit

ies

SP

TL

abor

ator

y—

—N

SA

quin

oet

al.

(200

5),

Stu

dy

2E

thn

icm

inor

itie

sS

PT

Lab

orat

ory

NS

NS

NS

DeM

atte

o,D

obbi

ns,

Mye

rs,

and

Fac

teau

(199

6)W

omen

SP

TL

abor

ator

yS

IG—

MIX

Eva

ns

(200

3)E

thn

icm

inor

itie

sW

PT

,S

PT

Fie

ldM

IX—

MIX

Gil

bert

and

Ste

ad(1

999)

,S

tud

y1

Wom

enG

EN

Lab

orat

ory

SIG

——

Gil

bert

and

Ste

ad(1

999)

,S

tud

y2

Eth

nic

min

orit

ies

GE

NL

abor

ator

yS

IG—

—H

eilm

an,

Bat

tle,

Kel

ler,

and

Lee

(199

8),

Stu

dy

2W

omen

WP

T,

SP

T,

GE

NL

abor

ator

yS

IGS

IGS

IGH

eilm

anan

dB

lad

er(2

001)

Wom

enG

EN

Lab

orat

ory

SIG

—S

IGH

eilm

anet

al.

(199

8),

Stu

dy

3W

omen

WP

T,

SP

T,

GE

NL

abor

ator

yM

IXM

IX—

Hei

lman

etal

.(1

992)

,S

tud

y1

Wom

enG

EN

Lab

orat

ory

SIG

NS

NS

Hei

lman

etal

.(1

992)

,S

tud

y2

Eth

nic

min

orit

ies,

Wom

enG

EN

Fie

ldS

IGS

IGS

IGH

eilm

an,

Blo

ck,

and

Sta

that

os(1

997)

,S

tud

y1

Wom

enG

EN

Lab

orat

ory

SIG

——

Hei

lman

etal

.(1

997)

,S

tud

y2

Wom

enG

EN

Lab

orat

ory

SIG

—S

IGH

eilm

anet

al.

(199

6)W

omen

SP

TL

abor

ator

yS

IGN

S—

Jaco

bson

and

Koc

h(1

977)

Wom

enS

PT

Lab

orat

ory

——

SIG

Mai

oan

dE

sses

(199

8)E

thn

icm

inor

itie

sG

EN

Lab

orat

ory

MIX

—M

cGla

shan

,W

righ

t,an

dM

cCor

mic

k(1

995)

Wom

enS

PT

Lab

orat

ory

——

NS

Res

end

ez(2

002)

,S

tud

y1

Eth

nic

min

orit

ies

GE

NL

abor

ator

yM

IX—

NS

Res

end

ez(2

002)

,S

tud

y2

Eth

nic

min

orit

ies

WP

T,

SP

TL

abor

ator

yS

IG—

SIG

Sei

jts

and

Jack

son

(200

1)E

thn

icm

inor

itie

sW

PT

Lab

orat

ory

NS

——

Su

mm

ers

(199

1)W

omen

GE

NL

abor

ator

yN

S—

—T

rave

ran

dA

llig

er(1

999)

Wom

enG

EN

Lab

orat

ory

NS

——

Un

zuet

a,L

ower

y,an

dK

now

les

(200

8),

Stu

dy

2E

thn

icm

inor

itie

s,W

omen

SP

TF

ield

SIG

——

%si

gnif

ican

t60

%33

%46

%%

mix

ed20

%17

%15

%%

non

-sig

nif

ican

t20

%50

%38

%

aG

EN

�ge

ner

ic;

WP

T�

wea

kp

refe

ren

tial

trea

tmen

t;S

PT

�st

ron

gp

refe

ren

tial

trea

tmen

t.b

SIG

�si

gnif

ican

tef

fect

inth

ep

red

icte

dd

irec

tion

;M

IX�

mix

edsu

pp

ort

(e.g

.,u

seof

mu

ltip

lem

easu

res,

som

eof

wh

ich

pro

du

ced

sign

ific

ant

resu

lts)

;N

S�

non

-sig

nif

ican

tef

fect

.D

ash

esin

dic

ate

that

the

dep

end

ent

vari

able

was

not

mea

sure

din

the

stu

dy.

Som

eef

fect

sre

por

ted

inth

eta

ble

cou

ldn

otbe

incl

ud

edin

the

met

a-an

alys

isbe

cau

seth

en

eces

sary

stat

isti

csw

ere

not

pro

vid

ed.

TA

BL

E2

Stu

die

sof

Sel

f-R

eact

ion

sam

ong

AA

PT

arge

ts

Stu

dy

AA

PT

arge

tG

rou

p(s

)A

AP

Typ

e(s)

aR

esea

rch

Set

tin

gP

erce

ived

Ste

reot

ypin

gbP

erce

ived

Com

pet

ence

bS

tate

Aff

ectb

Per

form

ance

Eva

luat

ion

sbO

bjec

tive

Per

form

ance

b

Bro

wn

etal

.(2

000)

,S

tud

y1

Wom

enW

PT

,S

PT

Lab

orat

ory

——

NS

—M

IXB

row

net

al.

(200

0),

Stu

dy

2E

thn

icm

inor

itie

s,W

hit

esS

PT

Fie

ld—

SIG

——

SIG

Bru

tus

and

Rya

n(1

998)

Wom

enS

PT

Lab

orat

ory

—N

SN

SN

SN

SC

hac

ko(1

982)

Wom

enS

PT

Fie

ld—

—M

IX—

—D

eitc

h,

Bri

ef,

Rob

erso

n,

and

Blo

ck(2

001)

Eth

nic

min

orit

ies

GE

NF

ield

SIG

——

——

Hat

tru

p(1

998)

Wom

enS

PT

Lab

orat

ory

——

—S

IGN

SH

eilm

anan

dA

lcot

t(2

001)

,S

tud

y1

Wom

enS

PT

Lab

orat

ory

SIG

SIG

SIG

——

Hei

lman

and

Alc

ott

(200

1),

Stu

dy

2W

omen

SP

TL

abor

ator

yS

IGS

IGS

IG—

—H

eilm

anet

al.

(199

8),

Stu

dy

1W

omen

GE

N,

WP

T,

SP

TL

abor

ator

y—

MIX

MIX

MIX

—H

eilm

anet

al.

(199

0)W

omen

,M

enS

PT

Lab

orat

ory

—S

IGS

IGS

IG—

Hei

lman

etal

.(1

991)

,S

tud

y1

Wom

en,

Men

SP

TL

abor

ator

y—

MIX

SIG

——

Hei

lman

etal

.(1

991)

,S

tud

y2

Wom

enS

PT

Lab

orat

ory

—S

IGN

S—

—H

eilm

anet

al.

(198

7)W

omen

,M

enS

PT

Lab

orat

ory

—S

IGN

SS

IG—

Mor

os(2

004)

Wom

en,

Men

,E

thn

icm

inor

itie

s,W

hit

esS

PT

Lab

orat

ory

——

—N

S—

Nac

oste

(198

5)W

omen

SP

TL

abor

ator

y—

—S

IG—

—N

iem

ann

and

Dov

idio

(200

5)E

thn

icm

inor

itie

sS

PT

Fie

ld—

SIG

——

—N

acos

tean

dL

ehm

an(1

987)

Wom

enS

PT

Lab

orat

ory

NS

——

——

Sta

nu

sh,

Art

hu

r,an

dD

over

spik

e(1

998)

Eth

nic

min

orit

ies

GE

NL

abor

ator

yS

IGN

SS

IG—

—T

aylo

r(1

994)

Wom

en,

Eth

nic

min

orit

ies

GE

NF

ield

—N

SN

S—

—

Tu

rner

and

Pra

tkan

is(1

993)

Wom

enS

PT

Lab

orat

ory

——

—M

IXN

SU

nzu

eta,

Gu

tiér

rez,

and

Gh

avam

i(2

010)

,S

tud

y1

Wom

enG

EN

,S

PT

Fie

ld—

SIG

——

—

Van

Laa

ret

al.

(200

8)E

thn

icm

inor

itie

sG

EN

Fie

ldN

S—

—S

IGS

IG%

sign

ific

ant

67%

62%

46%

50%

33%

%m

ixed

0%15

%15

%25

%17

%%

non

-sig

nif

ican

t33

%23

%38

%25

%50

%

aG

EN

�ge

ner

ic;

WP

T�

wea

kp

refe

ren

tial

trea

tmen

t;S

PT

�st

ron

gp

refe

ren

tial

trea

tmen

t.b

SIG

�si

gnif

ican

tef

fect

inth

ep

red

icte

dd

irec

tion

;M

IX�

mix

edsu

pp

ort

(e.g

.,u

seof

mu

ltip

lem

easu

res,

som

eof

wh

ich

pro

du

ced

sign

ific

ant

resu

lts)

;N

S�

non

-sig

nif

ican

tef

fect

.D

ash

esin

dic

ate

that

the

dep

end

ent

vari

able

was

not

mea

sure

din

the

stu

dy.

Som

eef

fect

sre

por

ted

inth

eta

ble

cou

ldn

otbe

incl

ud

edin

the

met

a-an

alys

isbe

cau

seth

en

eces

sary

stat

isti

csw

ere

not

pro

vid

ed.

statistic for each study. The SAMD statistic cap-tures the deviation of each effect size from theweighted average of the remaining effect sizes (Huf-fuctt & Arthur, 1995). Two studies had SAMD val-ues greater than 3 and were thus identified as po-tential outliers (Arthur et al., 2001). The authors ofone of these studies used an idiosyncratic method-ology, not used in any other study, by asking Whiteparticipants to imagine they were Black. Given theartificiality of this task, we excluded this study. Wereviewed the other study that was identified as apotential outlier and did not detect any method-ological anomalies. We therefore included thestudy in the analyses; however, excluding it had noimpact on our statistical conclusions.

We used the corrected effect sizes (�s) to con-duct meta-analytic path analyses (Shadish, 1996;Viswesvaran & Ones, 1995). Meta-analytic pathanalysis requires effect sizes for the relationshipsamong the dependent variables. We thereforegathered intercorrelations among the dependentvariables from the studies in our sample and usedthem to calculate corrected meta-analytic effectsizes (see Table 3). Sample sizes varied acrossstudies and we therefore used the harmonic meanacross studies as the sample size for the pathmodels (Viswesvaran & Ones, 1995). We con-cluded that a path model fit the data well if thecomparative fit index (CFI) was at least .95, theroot mean square error of approximation (RMSEA)was .06 or less, and the standardized root meansquare residual (SRMR) was .08 or less (Hu &Bentler, 1999).

To calculate a given hypothesized indirect (i.e.,mediated) effect, we multiplied the path coefficientsinvolved in that indirect effect. To test for statistical

significance, we used the delta method to calculatethe standard errors of the indirect effects (Muthén &Muthén, 2007). We were unable to use bootstrappedconfidence intervals, a more common approach fortesting indirect effects, because the dataset includedcorrelations rather than individual observations. Inaddition to the hypothesized models, we tested plau-sible alternative models. If an alternative model wasnested within the hypothesized model, we used achi-square difference test to compare the fit of the twomodels. If an alternative model was not nested withinthe hypothesized model, we could not assess relativemodel fit. Instead, we evaluated the alternative mo-del’s absolute fit.

RESULTS

We first calculated the meta-analytic effect sizefor each dependent variable to determine whetherAAPs have generalizable effects. We then used pathanalyses to test Hypotheses 1–6.

Meta-Analytic Effect Sizes

For studies of others’ reactions to AAP targets,AAPs had a significant effect on each outcome inthe expected direction (see Table 4, top half). AAPswere negatively related to others’ perceptions oftargets’ competence (� � �.27, CI95 � �.33 to �.17)and this finding held when we separated measuresof ability-based (� � �.30, CI95 � �.38 to �.19) andagency-based (� � �.25, CI95 � �.39 to �.05) com-petence. We used the program provided by Lipseyand Wilson (2001) to compare the magnitude of theeffects for the two competence measures and foundthat the effects did not differ (QB � 0.24, n.s.).4

AAPs were also negatively related to others’ per-ceptions of targets’ warmth (� � �.28, CI95 � �.45to �.05) and performance (� � �23, CI95 � �.30 to�.10). Moreover, the effect sizes did not differ forperceived (� � �.16, CI95 � �.22 to �.06) andexpected (� � �.26, CI95 � �.37 to �.09) perfor-mance (QB � 0.74, n.s.).

For studies of AAP targets’ own reactions, AAPsalso had significant effects on each dependent vari-able in the expected direction (see Table 4, bottom

4 This procedure is analogous to conducting an analy-sis of variance, in which the effect sizes are the depen-dent variable and a categorical variable that captures themeasure type (ability versus agency) is the independentvariable. This analysis produces a QB statistic, which hasa chi-square distribution and is analogous to an F-test.

TABLE 3Relationships among the Dependent Variables

Ratings of Others 1 2 3 4

1. Competence —2. Warmth .71 —3. Performance evaluations .65 .67 —

Ratings of the Self 1 2 3 4

1. Perceived stereotyping —2. Self-competence �.36 —3. State affect �.36 .24 —4. Performance evaluations �.25 .49 .40 —5. Objective performance �.11 .28 .09 .33

Note. Values are corrected correlation coefficients (�s).


half). AAPs were positively related to perceivedstereotyping (� � .26, CI95 � .06 to .41) and nega-tively related to self-competence (� � �.22, CI95 ��.27 to �.13), state affect (� � �.28, CI95 � �.37to �.15), performance evaluations (� � �.32, CI95 ��.38 to �.21), and objective performance (� ��.19, CI95 � �.26 to �.08). Due to insufficientsample sizes, we could not compare the effect sizesfor ability- and agency-based competence or ex-pected and perceived performance. In all, the re-sults indicate that AAPs have a significant, gener-alizable effect on each outcome of interest.

Path Models for Others’ Reactions to AAPTargets: Hypotheses 1–2

We tested the fit of the hypothesized model andsignificance of the hypothesized indirect effects(see Figure 1A). We used the overall effect sizes forcompetence and performance because the effect

sizes did not differ for the two competence (ability,agency) and two performance (perceived, expected)measures. We allowed competence and warmth tocovary in all models because both may be driven byothers’ general impressions of the target.

The hypothesized other-driven model fit thedata well (CFI � 1.00; RMSEA � .00; SRMR �.01; �2(1) � 0.40, n.s.; see Figure 2A). Moreover,the path coefficients revealed that the presence ofan AAP was negatively related to both perceivedcompetence (� � �.27, p � .01) and perceivedwarmth (� � �.28, p � .01), and that both com-petence (� � .35, p � .01) and warmth (� � .42,p � .01) were positively related to performanceevaluations.

We hypothesized that AAPs have indirect nega-tive effects on performance evaluations throughperceived competence (Hypothesis 1) and per-ceived warmth (Hypothesis 2). In support of Hy-potheses 1 and 2, the indirect effects of AAPs on

TABLE 4Meta-Analytic Estimates for the Effects of AAPs

Dependent Variable kb Nc rd Var. r CI95e �f Var. � CrI80

g Qh kFSi

Others’ reactions to targetsCompetencea 20 2,791 �.25 .04 �.33 to �.17 �.27 .03 �.50 to �.03 111.33** 1,419

Ability-based 19 2,621 �.29 .04 �.38 to �.19 �.30 .04 �.56 to �.05 126.63** 1,713Agency-based 5 742 �.22 .04 �.39 to �.05 �.25 .04 �.50 to .01 31.38** 73

Warmth 5 686 �.25 .05 �.45 to �.05 �.28 .05 �.57 to .02 38.71** 90Performance evaluations 13 1,995 �.20 .03 �.30 to �.10 �.23 .03 �.46 to .01 71.43** 434

Perceived 5 626 �.14 .01 �.22 to �.06 �.16 .00 �.23 to �.09 6.37 21Expected 8 1,369 �.23 .04 �.37 to �.09 �.26 .04 �.53 to .01 62.45** 248

Self-reactions among targetsPerceived stereotyping 6 1,060 .23 .05 .06 to .41 .26 .05 �.03 to .55 54.83** 197Self-competence 13 2,009 �.20 .02 �.27 to �.13 �.22 .01 �.38 to �.07 37.53** 433

Ability-based 12 1,690 �.24 .01 �.30 to �.17 �.26 .01 �.36 to �.14 24.73** 425Agency-based 1 — — — — — — — — —

State affect 9 1,204 �.26 .03 �.37 to �.15 �.28 .02 �.48 to �.08 37.39** 295Performance evaluations 7 1,017 �.29 .01 �.38 to �.21 �.32 .01 �.45 to �.19 17.13** 220

Perceived 6 744 �.31 .02 �.42 to �.20 �.34 .01 �.49 to �.18 16.88** 148Expected 1 — — — — — — — — —

Objective performance 6 1,122 �.17 .01 �.26 to �.08 �.19 .01 �.31 to �.07 14.67* 78

a Some studies of others’ reactions to AAP targets included multiple competence measures and thus contributed effects to the ability-and agency-based competence estimates. To maintain independence of samples, the effects for all competence measures in a study wereaveraged to calculate the overall competence effect. For this reason, the sum of studies for the more specific competence measures is greaterthan the total number of studies for the overall competence measure.

b k is the number of studies.c N is the number of participants across studies.d r is the mean sample-weighted effect size.e CI95 is the 95% confidence interval around r.f � is the mean corrected effect size.g CrI80 is the 80% credibility interval around �.h Q is the chi-square test statistic for homogeneity of effect sizes.i kFS is the fail-safe k statistic.* p � .05. ** p � .01.


performance through competence (� � �.10, p �.01) and warmth (� � �.12, p � .01) were bothsignificant, as was the total indirect effect of AAPson performance through both competence andwarmth (� � �.21, p � .01).

We compared the hypothesized model to threeplausible alternatives. First, scholars have sug-gested that others view AAPs as unfair and thusnegatively evaluate AAP targets on a number ofdimensions (e.g., Heilman, McCullough, & Gilbert,1996). A fairness-based perspective suggests thatAAPs have a direct effect on performance, in addi-tion to the direct effects on competence andwarmth, but does not suggest indirect effects onperformance through competence and warmth. Wetherefore tested a model that included direct effectsfrom AAPs to competence, warmth, and perfor-mance, but no indirect effects, and found it did notfit the data well (CFI � .56; RMSEA � .57; SRMR �.27; �2(2) � 388.75, p � .01). Second, unlike the

SCM, the discounting principle suggests that AAPshave a direct effect on competence, but does notsuggest a direct effect on warmth. We thereforetested a competence-only model by removing thepath from AAPs to warmth from the hypothesizedmodel to test the alternative possibility that anyeffect of AAPs on warmth is accounted for by thecorrelation between competence and warmth. Thefit of this model was poor (CFI � .94; RMSEA � .20;SRMR � .13; �2(2) � 48.49, p � .01) and signifi-cantly worse than the fit of the hypothesized model(��2 (1) � 48.09, p � .01).

Third, our theory suggests that competence andwarmth fully mediate the effect of AAPs on perfor-mance evaluations. We investigated the alternativepossibility of partial mediation by testing whetherAAPs have a direct effect on performance evalua-tions, after accounting for the indirect effectsthrough competence and warmth. We could not usepath analysis to test this model because the model

FIGURE 2Path Model Results

A) Others’ Reactions to AAP Targets: Hypothesized and Final Model; B) Self-Reactions among AAPTargets: Hypothesized Model; C) Self-Reactions among AAP Targets: Final Model

-.27

-.28

.35

.42

Presence of an AAP

Perceived Competence

Performance Evaluations

Perceived Warmth

.69

.10

-.33

-.31

.30

.26 .27

- .14

-.20

.42

Presence of an AAP

Self-competence

Objective Performance

StateAffect

Perceived Stereotyping

Performance Evaluations

.02

.24

.10 .22

.39

-.33

-.31.26

.26 .25

-.14

-.20

-.16

.01

Presence of an AAP

Self-competence

Objective Performance

StateAffect

Perceived Stereotyping

PerformanceEvaluations

-.14

B)

C)

A)

Note. Standardized path coefficients (�s) are reported. Solid paths are significant at p � .05; dashed paths are not significant. n � 589for Figure 2A; n � 301 for Figures 2B and 2C.


has 0 degrees of freedom. Instead, we regressedperformance evaluations on competence, warmth,and AAPs simultaneously. Competence (� � .35,p � .01) and warmth (� � .42, p � .01) werepositively related to performance evaluations, butAAPs were not (� � �.02, n.s.), indicating that theeffect of AAPs on performance evaluations is fullymediated by perceived competence and warmth.We therefore adopted the hypothesized model asthe final model for others’ reactions to AAP targets.

Path Models for Self-Reactions among AAPTargets: Hypotheses 3–6

We tested the fit of the hypothesized model andsignificance of the hypothesized indirect effects(see Figure 1B). To mirror the other-driven model,we used the overall effect sizes for competence andperformance evaluations; however, using ability-based competence and perceived performance pro-duced the same conclusions. We allowed perceivedself-competence and state affect to covary becauseboth variables may reflect general evaluations ofthe self. We also allowed self-evaluated and objec-tive performance to covary because they are likelyrelated.

The hypothesized model fit the data reasonablywell (CFI � .97; RMSEA � .09; SRMR � .04;�2(4) � 13.73, p � .01; see Figure 2B), although theRMSEA was above the .06 cutoff. The coefficientsrevealed that AAPs had negative effects on bothself-competence (� � �.14, p � .05) and state affect(� � �.20, p � .01), and a positive effect on per-ceived stereotyping (� � .26, p � .01). Perceivedstereotyping was negatively related to both self-competence (� � �.33, p � .01) and state affect(� � �.31, p � .01). Finally, self-competence waspositively related to both evaluated (� � .42, p �.01) and objective (� � .27, p � .01) performance,and state affect was positively related to evaluated(� � .30, p � .01), but not objective (� � .02, n.s.),performance.

We hypothesized that AAPs have an indirectnegative effect on (a) self-evaluated and (b) objec-tive performance through self-competence (Hy-pothesis 3) and state affect (Hypothesis 4). In sup-port of Hypotheses 3a and 3b, the indirect effects ofAAPs on performance, through perceived self-com-petence, were significant (self-evaluated: � � �.06;objective: � � �.04; both ps � .05). The indirecteffect of AAPs on performance, through state affect,was significant for self-evaluated (� � �.06, p �.01), but not objective (� � .00, n.s.), performance.

Thus, Hypothesis 4a was supported, but Hypothe-sis 4b was not.

We also hypothesized that AAPs have indirecteffects on (a) self-evaluated and (b) objective per-formance through perceived stereotyping and self-competence (Hypothesis 5) and through perceivedstereotyping and state affect (Hypothesis 6). In sup-port of Hypotheses 5a and 5b, the indirect effects ofAAPs on performance, through perceived stereo-typing and self-competence, were significant (self-evaluated: � � �.04, objective: � � �.02, bothps � .01). The indirect effect of AAPs on perfor-mance, through perceived stereotyping and stateaffect, was significant for self-evaluated (� � �.02,p � .01), but not objective (� � .00, n.s.), perfor-mance, which supports Hypothesis 6a, but not Hy-pothesis 6b. The total indirect effects of AAPs onself-evaluated (� � �.18, p � .01) and objective(� � �.07, p � .01) performance were alsosignificant.

We assessed the fit of three plausible alternativemodels. First, given that prior work has focused ondirect effects rather than mediated relationships,we tested a direct effects model in which the pres-ence of an AAP has a direct effect on perceivedstereotyping, self-competence, state affect, and self-evaluated and objective performance, but no indi-rect effects. This model did not fit the data well(CFI � .47; RMSEA � .25; SRMR � .16; �2(8) �159.14, p � .01). Second, theory grounded in self-doubt suggests that the presence of an AAP affectsperceived self-competence, but not state affect orperceived stereotyping. We therefore tested a com-petence-only model in which AAPs do not have adirect effect on perceived stereotyping and stateaffect, and any effect of AAPs on these variables isinstead accounted for by their correlations withself-competence. Specifically, this model includeda directional path from AAPs to competence, acorrelation between competence and perceived ste-reotyping, a correlation between competence andstate affect, and directional paths from competence,perceived stereotyping, and state affect to self-eval-uated and objective performance. This modeldid not fit the data well (CFI � .60; RMSEA � .24;SRMR � .15; �2(5) � 89.70, p � .01).

Third, we theorized that self-competence, stateaffect, and perceived stereotyping mediate the ef-fects of AAPs on performance. We investigated thepossibility that the mediation is only partial bytesting a model that differed from the hypothesizedmodel in that it included additional paths fromAAPs to both performance outcomes (see Fig-


ure 2C). The resulting model fit the data well(CFI � 1.00; RMSEA � .00; SRMR � .01; �2(2) �0.45, n.s.) and significantly better than the hypoth-esized model (��2(2) � 13.28, p � .01). The pathsfrom AAPs to self-evaluated (� � �.16, p � .01)and objective (� � �.14, p � .01) performance weresignificant; however, the direction and significanceof all other paths and indirect effects remainedunaltered. Given the superior fit of the partial me-diation model, we adopted this alternative modelas the final model and concluded that perceivedstereotyping by others, perceived self-competence,and state affect partially mediate the effects ofAAPs on targets’ self-evaluated and objectiveperformance.

In the final model, we used 95% confidence in-tervals to compare the magnitude of the total indi-rect effects and direct effects of AAPs on perfor-mance. The magnitude of the total indirect effectsand direct effects did not differ for self-evaluated(total indirect effect: � � �.16, CI95 � �.22 to �.10;direct effect: � � �.16, CI95 � �.26 to �.06) orobjective (total indirect effect: � � �.05, CI95 ��.10 to �.01; direct effect: � � �.14, CI95 � �.25 to�.02) performance. Thus, the portion of the effectof AAPs on performance that is explained by ourtheory (i.e., the indirect effects) is roughly equiva-lent to the portion that is not (i.e., the directeffects).

Robustness Checks

Our theory suggests that the negative effects ofAAPs hold across a number of potential boundaryconditions, including the group targeted by theAAP, the AAP type, and the research setting. Weused the program provided by Lipsey and Wilson(2001) to conduct post-hoc analyses to determinewhether our findings are robust across potentialmoderators. We focused the moderated analyses oncompetence because the sample was too small forother outcomes.

Stereotyping theories posit that any group can benegatively stereotyped in certain settings, even ifthe group is not chronically negatively stereotypedin the society (Aronson et al., 1999; Fiske et al.,2002). Notably, this assumption differs from priorclaims that AAPs only have negative consequencesif the target is chronically negatively stereotyped(e.g., Brown, Charnsangavej, Keough, Newman, &Renfrow, 2000; Gilbert & Stead, 1999; Heilman etal., 1987; Nacoste, 1985; Resendez, 2002). Moststudies in our meta-analysis investigated AAPs that

target chronically negatively stereotyped groups—specifically, women and ethnic minorities—but afew studies of targets’ self-reactions investigatedAAPs that target groups that are not chronicallynegatively stereotyped (e.g., Whites, men; Brown etal., 2000; Heilman, Rivero, & Brett, 1991). We com-pared the effect of AAPs on competence for groupsthat are (� � �.24, CI95 � �.31 to �.12, k � 13) andare not (� � �.16, CI95 � �.24 to �.05, k � 4)chronically negatively stereotyped and found thatthe magnitude of the two effect sizes did not differ(QB � 2.70, n.s.).

AAPs may be implemented in different ways thatvary in strength and scholars have suggested thatsome AAP types are more likely to produce stigmathan others (e.g., Evans, 2003). In spite of this vari-ability, all AAPs provide extra help to target groupmembers and should therefore stigmatize their tar-gets. We tested if the effect sizes differed for AAPsthat provide OE, WPT, SPT, or are generic (i.e.,strength is not specified; Harrison et al., 2006). Forstudies of others’ reactions to targets, the stigma ofincompetence did not differ for SPT (� � �.41,CI95 � �.56 to �.20, k � 7), WPT (� � �.32, CI95 ��.55 to �.05, k � 5), or generic AAPs (� � �.38,CI95 � �.49 to �.21, k � 10; QB � 1.20, n.s.). Forstudies of targets’ self-reactions, the stigma of in-competence did not differ for SPT (� � �.32,CI95 � �.35 to �.21, k � 11) and generic AAPs (� ��.20, CI95 � �.40 to .02, k � 3; QB � .01, n.s.).(There were insufficient studies to calculate effectsfor studies of OE and studies of targets’ self-reac-tions to WPT.)

Finally, some scholars have suggested that AAPsmay not have stigmatizing effects in the field, argu-ing that AAP targets who are aware they are quali-fied will not question their competence (Crosby,Iyer, & Sincharoen, 2006; Taylor, 1994). Althoughaware of absolute qualifications, AAP targets maybe unaware of their qualifications relative to thequalifications of others and may thus question theircompetence. We compared the effect of AAPsacross research settings. The stigma of incompe-tence occurred in the laboratory and the field andthe effect sizes did not differ across settings forstudies of both others’ reactions (laboratory: � ��.25, CI95 � �.31 to �.15, k � 17; field: � � �.39,CI95 � �.64 to �.08, k � 3; QB � 1.45, n.s.) andself-reactions (laboratory: � � �.25, CI95 � �.32 to�.14, k � 9; field: � � �.19, CI95 � �.29 to �.06,k � 4; QB � 0.75, n.s.).


DISCUSSION

Organizations around the globe have taken af-firmative action to counteract continued discrim-ination and facilitate employment success forunderrepresented groups. AAPs increase the rep-resentation of women and ethnic minorities inmanagerial positions, but also have the potential toproduce ironic effects by negatively affecting thevery employees they are intended to benefit. Priortheory suggests that perceptions that AAP targetslack competence, both in the eyes of others and inthe eyes of targets themselves, explain why AAPsnegatively affect targets’ performance. Yet thebroader social psychological literature suggests thatadditional mechanisms also play a role. Using the-ories of stereotyping, we developed a more compre-hensive theory of the unintended negative effects ofAAPs, which we tested using meta-analytic pathanalyses.

Drawing from the SCM, we theorized and foundthat a perceived lack of both competence andwarmth explain why others evaluate AAP targets’performance negatively. Drawing from STT, wetheorized and found that self-competence, state af-fect, and perceived stereotyping by others eachcontribute to explaining why AAP targets experi-ence low performance. The other- and self-drivenmodels derived from stereotyping theories fit thedata better than models derived from alternativetheoretical frameworks. The present research there-fore substantiates that stereotyping theories pro-vide a useful lens for understanding the negativeconsequences of AAPs. In the other-driven model,competence and warmth fully mediated the effectsof AAPs on performance; however, in the self-driven model, self-competence, state affect, andperceived stereotyping partially mediated the ef-fects of AAPs on performance. Thus, although thefindings for self-reactions among AAP targets sup-port our hypotheses, additional mechanisms notincluded in our theory likely exist.

Implications for Theory

We advance theory by demonstrating that stereo-typing theories provide a more comprehensive un-derstanding of the unintended negative effects ofAAPs on targets’ performance outcomes, as com-pared to prior theory grounded in discounting andself-doubt, which focuses on the stigma of incom-petence. Specifically, we propose that competenceis only one of several mechanisms that explain why

AAPs negatively affect targets’ performance out-comes. At the same time, our theory does not con-tradict prior work on the stigma of incompetence.Scholars have previously argued that AAPs and theassociated possibility that demographics influenceemployment decisions lead others to assume thatAAP targets lack competence and similarly leadtargets to doubt their self-competence. In the other-driven model, we propose that AAPs and the asso-ciated assumption that demographics play a role inselection signals that AAP targets lack status,which leads others to assume that AAP targets areincompetent. Thus, although our SCM-based expla-nation invokes notions of status, the two explana-tions are convergent, not contradictory. Similarly,in the self-driven model, we propose that AAPs andthe associated assumption that demographics playa role in selection creates a stereotype that AAPtargets lack competence, which leads targets todoubt their self-competence. Thus, although ourSTT-based explanation invokes the concept of ste-reotyping, it also converges with prior theory.

Our primary contribution is using stereotypingtheories to build a more comprehensive under-standing of the unintended negative effects ofAAPs. Although discounting and self-doubt havebeen the primary theoretical frameworks in thisliterature, a handful of prior studies have similarlydrawn from stereotyping theories. We extend pre-vious work by more fully developing the implica-tions of stereotyping theories for understanding theeffects AAPs on targets’ performance. For example,Aquino and colleagues (2005) used the SCM tohypothesize that AAPs affect warmth, but used dif-ferent rationales to justify why AAPs affect otheroutcomes. Thus, whereas past work suggests thatmultiple theories are needed to understand all ofthe dimensions along which others devalue AAPtargets, we propose that the SCM provides a single,unifying framework that explains how AAPs affectothers’ evaluations of targets’ competence, warmth,and performance, as well as the interrelationshipsamong these outcomes. Similarly, a few prior stud-ies of self-reactions among targets have drawn fromSTT (Brown et al., 2000; Van Laar, Levin, & Sin-clair, 2008), but have focused on the implicationsof STT for only a subset of the outcomes and path-ways through which AAPs affect performance thatare included in our theory. Thus, we build on priorwork by using STT to develop a fuller understand-ing of the mechanisms that contribute to the nega-tive effect of AAPs on targets’ performance.


Our theory and findings also suggest that thereare differences in the other- and self-driven mech-anisms that link AAPs to performance. Specifi-cally, we propose that the other-driven mecha-nisms include competence and warmth, while theself-driven mechanisms include self-competence,state affect, and perceived stereotyping. Notably,the rationale for why others perceive AAP targets aslacking warmth does not apply to AAP targets’ ownreactions and the rationale for why AAP targetsexperience negative affect and perceived stereotyp-ing by others does not apply to others’ reactions toAAP targets. It is therefore not surprising that thesepossibilities have not been investigated empiricallyand, as a result, our meta-analytic approach did notallow a test of whether the unique mechanisms inour other-driven model apply to the self-drivenmodel and vice versa. The present research there-fore offers suggestive, rather than definitive, evi-dence of differences in the other- and self-drivenmechanisms that link AAPs to performance.

We also contribute to theory by proposing andfinding that AAPs have negative effects on anygroup they target, even if the group is not chroni-cally negatively stereotyped in the broader society.This possibility is consistent with stereotyping the-ories, which indicate that stereotypes are contextdependent and that any group can be negativelystereotyped in certain settings. At the same time,our stereotyping-based perspective differs fromprior claims that AAPs only have negative conse-quences for target groups that are chronically neg-atively stereotyped (e.g., Brown et al., 2000; Gilbert& Stead, 1999; Heilman et al., 1991; Nacoste, 1985;Resendez, 2002). Theoretical and empirical evi-dence that groups that are not chronically nega-tively stereotyped are negatively affected by AAPshas implications that extend beyond the AAP liter-ature. For example, our stereotyping-based theorymay be relevant to other types of preferential treat-ment, such as nepotism or legacy admissions tocolleges and universities.

In addition, the present work has several impli-cations for the SCM. First, the SCM posits thatstereotypes are often ambivalent, in that they in-volve trade-offs between competence and warmth.As a result, most groups are stereotyped as lackingin either competence or warmth, but not both(Bergsieker, Leslie, Constantine, & Fiske, 2012;Fiske et al., 2007). For example, individuals gener-ally perceive housewives as warm, but not compe-tent, and perceive businesswomen as competent,but not warm (Fiske et al., 2002). Thus, business-

women’s gain in competence comes at the cost oflow warmth. Although competence–warmth trade-offs are common, our theory suggests that, becauseAAPs simultaneously indicate that AAP targetgroups have low status and increase their ability tocompete for resources, AAPs stigmatize their tar-gets as low in both competence and warmth. Thus,AAP targets provide a relatively rare example of agroup that is negatively stereotyped on both funda-mental dimensions of person perception.

Second, although SCM scholars have suggestedthat competence and warmth stereotypes are rele-vant for workplace evaluations (Cuddy et al., 2011),SCM research has focused largely on the conse-quences of competence and warmth outside of or-ganizations. For example, research has assessedwhether competence and warmth stereotypes areassociated with the tendency to help versus de-mean stereotype targets (e.g., Cuddy et al., 2007).Moreover, there is some question regardingwhether competence, warmth, or both affect perfor-mance evaluations. Theoretically, perceptions oflow competence are more likely than perceptionsof low warmth to result in passive harm, such asrating targets’ performance lower than deserved,although some evidence suggests that competenceand warmth both predict passive harm (Cuddy etal., 2007). We develop a rationale for why percep-tions of warmth, in addition to competence, arelikely to affect performance evaluations, and thushelp to advance theory regarding the relevance ofthe SCM—a social psychological theory—for work-place phenomena.

Our theory similarly has implications for STT.Scholars have argued that stereotype threat re-search is narrow in scope, given that it focusesprimarily on the consequences of negative stereo-types for students’ performance on academic tests(e.g., Shapiro & Neuberg, 2007) and less is knownregarding whether and how stereotype threat oper-ates in the workplace (e.g., Roberson & Kulik,2007). We provide insight into these issues by the-orizing that AAPs, which are common workplacepolicies, activate negative stereotypes about the so-cial groups they target and therefore trigger stereo-type threat. Moreover, we demonstrate that the in-terrelationships among the negative consequencesof AAPs are consistent with predictions derivedfrom STT.

We also make strides toward advancing under-standing of the mechanisms that drive stereotypethreat. Many mechanisms have been proposed yetambiguity persists regarding the processes that un-


derlie stereotype threat, in part because prior stud-ies have tended to focus on a single explanatorymechanism (e.g., Schmader et al., 2008). In usingSTT to build theory regarding the self-driven pro-cesses that link AAPs to targets’ performance out-comes, we propose that three distinct mechanisms—self-competence, state affect, and perceivedstereotyping by others— each play a role. More-over, we contribute to recent theory that has differ-entiated self- and other-as-source stereotype threat(Shapiro & Neuberg, 2007) by theorizing and find-ing that these two types of stereotype threat operatesimultaneously and independently and that AAPsnegatively affect targets’ performance through fourself-driven pathways (i.e., Hypotheses 3–6).

Finally, the present work contributes to theoryon the SCM and STT by demonstrating that boththeories apply to groups with permeable boundar-ies, not just groups with impermeable boundaries.Specifically, research on the SCM and STT hasoften focused on groups based on immutable char-acteristics, such as gender and race. We demon-strate that the principles of the SCM and STT ad-vance understanding of the experiences of a groupthat is defined by the situation (i.e., AAP targets),with the result that individuals may belong to thegroup in some settings but not others (e.g., organi-zations with and without AAPs).

Implications for Practice

Our theory and findings also offer practical in-sights. As shown in Tables 1 and 2, the percentagesof prior studies that find a significant effect ofAAPs on the outcomes included in our theoryrange from 33 to 67, which raises questions regard-ing the robustness of this phenomenon andwhether it should be of concern for organizations.Our meta-analyses demonstrate that AAPs have asignificant, generalizable effect on each outcome ofinterest, which indicates that prior null findingsstem from statistical artifacts (e.g., sampling error,unreliability), not the absence of true effects. Sim-ilarly, some scholars have questioned the robust-ness of the stigma of AAPs by suggesting that it isan artifact of the laboratory (Crosby et al., 2006;Taylor, 1994) or that it only applies to AAPs thatinvolve strong preferences (e.g., Evans, 2003). Incontrast to these claims, we find that the stigma ofAAPs occurs in the laboratory and in the field andthat the magnitude of the effect does not differ forAAPs that vary in strength. Our findings thereforesubstantiate that the stigma of AAPs is a real-world

phenomenon with the potential to derail organiza-tional efforts to create and maintain a diverseworkforce.

The present research also offers novel insights forpreventing AAPs from having negative effects ontheir targets. We found that perceived incompe-tence and low warmth both explain why othersnegatively evaluate AAP targets’ performance.Eliminating the negative effects of AAPs thereforerequires addressing stigmatization along both di-mensions. Publicizing the qualifications of an AAPtarget—a well-supported and commonly recom-mended strategy for preventing the stigmatizingconsequences of AAPs (e.g., Heilman, Lucas, & Ka-plow, 1990; Heilman et al., 2001)—eliminates thestigma of incompetence, but is unlikely to affectperceptions that AAP targets lack warmth. Lowwarmth stereotypes are driven by perceptions thatAAPs increase targets’ ability to compete for work-place resources and thus threaten non-targets’ out-comes. Our theory and findings suggest that organ-izations should address perceptions that AAPs are atodds with the interests of non-targets, perhaps bystressing that the increased diversity associated withAAPs has the potential to improve organizational per-formance and thus benefit everyone in the organiza-tion (e.g., Ely & Thomas, 2001).

Similarly, we theorized and found that AAPshave direct effects on the self-competence and stateaffect of AAP targets (self-as-source stereotypethreat), as well as indirect effects through perceivedstereotyping by others (other-as-source stereotypethreat). It follows that eliminating the self-drivenprocesses that link AAPs to performance requiresaddressing each of these pathways. For example,AAP targets need to know not only that they arequalified, to prevent low perceived self-compe-tence, but also that others are aware of their quali-fications, to prevent perceived stereotyping byothers.

Whereas prior work suggests that AAPs only stig-matize groups that are chronically negatively ste-reotyped, we theorized and found that the stigma ofAAPs also applies to groups that are not chroni-cally negatively stereotyped. Thus, our researchsuggests that, in settings where groups that are nottypically negatively stereotyped are underrepre-sented (e.g., men in nursing), and may thus betargeted by AAPs, organizations must be vigilant toprevent stigma.

The present research reveals that AAPs have neg-ative implications for a range of outcomes, includ-ing AAP targets’ performance. Thus, in some ways,


our findings present a discouraging picture of thefuture of AAPs, and one could argue that organiza-tions should not implement these policies. At thesame time, it is important to acknowledge that,although important, our focus on performancepresents a narrow view of the potential effects ofAAPs. For example, the adoption of AAPs facili-tates increased representation of women and ethnicminorities in management positions (Holzer & Neu-mark, 2000; Kalev et al., 2006; Leonard, 1984). Inaddition, if AAPs increase the number of womenand ethnic minorities in high-level positions, theserole models may decrease implicit assumptionsthat women and ethnic minorities lack the abilityneeded for such positions, thus mitigating discrim-ination and the need for AAPs in the future. Theimplementation of AAPs, along with efforts to max-imize their effectiveness, is therefore more likely tolead to favorable long-term outcomes than is theelimination of AAPs.

Many organizations implement AAPs as part ofbroader diversity initiatives (Kelly & Dobbin, 1998),which may include efforts to facilitate a multicul-tural climate, in which differences are acknowl-edged and valued, and the implementation of flex-ible and other family-friendly policies intended toattract groups of employees who tend to have sig-nificant personal-life responsibilities (e.g., women,parents). Like AAPs, these efforts are largely effec-tive; multicultural climates facilitate engagementand reduce perceived discrimination (Avery, Mc-Kay, Wilson, & Tonidandel, 2007; Leslie & Gelfand,2008; Plaut, Thomas, & Goren, 2009) and flexibleand family-friendly policies improve job attitudesand reduce turnover intentions (Butts, Casper, &Yang, 2013; Gajendran & Harrison, 2007). Yet, justas AAPs have unintended consequences, multicul-turalism can lead non-targets to feel excluded (Plaut,Garnett, Buffardi, & Sanchez-Burks, 2011) and use offlexible and family-friendly practices can have nega-tive career consequences (Leslie, Manchester, &Dahm, 2013; Leslie, Manchester, Park, & Mehng,2012; Manchester, Leslie, & Kramer, 2010, 2013). Theparallels between AAPs and other diversity policiessuggest that strategies similar to those needed to pre-vent AAPs from having negative effects may beneeded to maximize the effectiveness of other aspectsof diversity initiatives.

Limitations and Future Research

Meta-analysis is a powerful empirical tool, but isalso limited by its reliance on existing studies. For

example, we found support for the hypothesizedmechanisms that link AAPs to performance, butour ability to make causal claims was somewhatlimited. Many studies in our sample were experi-ments that manipulated the presence on an AAPand demonstrate that AAPs have a causal impacton a number of outcomes, but the causal orderamong the outcomes is less clear. We tested alter-native causal models and generally found that ourhypothesized models fit the data better than thealternatives. Nevertheless, we cannot provide defini-tive evidence for the causal order among the endog-enous variables in our models. Similarly, all of thestudies in our sample investigated the stigmatizationof AAP targets by either others or the self. We weretherefore able to test the other- and self-models inde-pendently, but not simultaneously.

Another potential limitation is that the numberof studies contributing to some meta-analytic effectsizes was relatively small. Even when based on fewstudies, however, the number of individuals con-tributing to the effect sizes was large, which lendsstronger confidence to our conclusions than tothose based on single studies. In addition, our ef-fect sizes may be inflated to the extent that studieswith significant results were more likely to be pub-lished and are thus overrepresented in our sample(i.e., the file drawer problem). The file drawer prob-lem is likely not a concern, however, given that thefail-safe k-values for our main effect estimates weregenerally large (M � 414; see Table 4). Moreover,empirical evidence indicates that the file drawerproblem has little impact on the validity of meta-analytic estimates (Dalton et al., 2012).

In the other-driven model, perceived competenceand warmth fully accounted for the effect of AAPson performance evaluations. Conversely, in theself-driven model, AAPs had a direct effect on per-formance after accounting for all of the hypothe-sized explanatory mechanisms, which indicatesthat additional factors play a role in linking AAPsto targets’ performance in the self-driven model.Stereotype threat research indicates that negativestereotypes affect not only self-reported affect butalso biological anxiety responses (Steele et al.,2002). Thus, future research should explorewhether physiological anxiety, as well as othermechanisms, help explain the effect of AAPs onself-evaluated and objective performance.

State affect was related to self-evaluated, but notobjective, performance. One possible explanationis that the negative state affect associated withAAPs leads targets to believe that they failed to


perform to the best of their ability, but this beliefdoes not translate into reality. Alternatively, therelationship between state affect and performancemay be more complex. Consistent with the Yerkes–Dodson law (Yerkes & Dodson, 1908), the effectmay be curvilinear such that performance is weakerat low and high levels of negative state affect than itis at moderate levels. Alternatively, the null findingmay stem from the presence of boundary condi-tions. For example, the effect of state affect onobjective performance may be small, with the resultthat it only emerges for tasks that are particularlychallenging. Future work on these possibilities willhelp clarify whether the effects of AAPs on objec-tive performance are driven by state affect, in ad-dition to self-competence and perceived stereotyp-ing by others.

A key contribution of this research is using ste-reotyping theories to build a more comprehensiveunderstanding of the negative effects of AAPs ontargets’ performance. Linking stereotyping theoriesto the AAP literature may also spur future research.For example, evidence demonstrates that the detri-mental effects of stereotype threat can be avoidedby emphasizing learning goals (Goff, Steele, & Da-vies, 2008). Future work should therefore explorewhether fostering an organizational climate forlearning, instead of performance, can prevent thedetrimental effects of AAPs on the self-evaluatedand objective performance of AAP targets.

Finally, future research should integrate the lit-erature on the negative effects of AAPs with otherliteratures, such as identity management (e.g., Paet-zold, Dipboye, & Elsbach, 2008). Identity is a dy-namic construct and individuals may change thegroups they identify with to avoid a stigmatizedidentity (e.g., Hogg & Abrams, 1988). As a result,the presence of an AAP in a given organization andthe associated stigmatizing consequences maycause AAP targets to disassociate from their organ-izational identity, which could have additionalnegative consequences for workplace outcomes. In-vestigations that explore this and related questionswill help further link the stigma of AAPs to othercore topics in the management literature.

CONCLUSION

AAPs improve employment outcomes for womenand ethnic minorities (e.g., Kalev et al., 2006), yetalso have unintended consequences in that theycan negatively affect the very groups of employeesthey are intended to benefit. We used stereotyping

theories to advance understanding of these ironiceffects of AAPs. Specifically, we extended priorwork by proposing and testing a theory that speci-fies multiple pathways through which AAPs nega-tively affect the performance outcomes of their in-tended beneficiaries, and also suggests that thereare differences in the other- and self-driven mech-anisms that link AAPs to targets’ performance out-comes. A more comprehensive understanding ofthe unintended consequences of AAPs is useful fordeveloping strategies to prevent the negative effectsof AAPs on their intended beneficiaries and thusfacilitate social justice in the workplace and insociety at large.

REFERENCES5

*Aquino, K., Stewart, M. S., and Reed II, A., . 2005. Howsocial dominance orientation and job status influ-ence perceptions of African-American affirmativeaction beneficiaries. Personnel Psychology, 58:703–744.

Aronson, J., Lustina, M. J., Good, C., Keough, K., Steele,C. M., & Brown, J. 1999. When White men can’t domath: Necessary and sufficient factors in stereotypethreat. Journal of Experimental Social Psychology,35: 29–46.

Arthur, W., Jr., Bennett, W., Jr., & Huffcutt, A. I. 2001.Conducting meta-analysis using SAS. Mahwah, NJ:Lawrence Erlbaum Associates.

Asch, S. E. 1946. Forming impressions of personality.Journal of Abnormal Psychology, 41: 258–290.

Avery, D. R., McKay, P. F., Wilson, D. C., & Tonidandel,S. 2007. Unequal attendance: The relationships be-tween race, organizational diversity cues, and absen-teeism. Personnel Psychology, 60: 875–902.

Bergsieker, H. B., Leslie, L. M., Constantine, V. S., &Fiske, S. T. 2012. Stereotyping by omission: Elimi-nate the negative, accentuate the positive. Journal ofPersonality and Social Psychology, 102: 1214–1238.

Bertrand, M., & Mullainathan, S. 2004. Are Emily andGreg more employable than Lakisha and Jamal? Afield experiment on labor market discrimination.American Economic Review, 94: 991–1013.

Bhaskar-Shrinivas, P., Harrison, D. A., Shaffer, M. A., &Luk, D. M. 2005. Input-based and time-based modelsof international adjustment: Meta-analytic evidenceand theoretical extensions. Academy of Manage-ment Journal, 48: 257–281.

5 An asterisk indicates that the study was included inthe meta-analysis.


Blau, F. D., & Kahn, L. M. 2007. The gender pay gap: Havewomen gone as far as they can? Academy of Man-agement Perspectives, 21: 7–23.

Bobo, L. D. 2000. Racial attitudes and relations at theclose of the twentieth century. In M. J. Smelser, W. J.Wilson & F. Mitchell (Eds.), America becoming:Racial trends and their consequences, vol. 1: 264–301. Washington, D.C.: National Academy Press.

Brief, A. P., & Weiss, H. M. 2002. Organizational behav-ior: Affect in the workplace. Annual Review of Psy-chology, 53: 279–307.

*Brown, R. P., Charnsangavej, T., Keough, K. A., New-man, M. L., & Renfrow, P. J. 2000. Putting the “af-firm” into affirmative action: Preferential selectionand academic performance. Journal of Personalityand Social Psychology, 79: 736–747.

*Brutus, S., & Ryan, A. M. 1998. A new perspective onpreferential treatment: The role of ambiguity andself-efficacy. Journal of Business and Psychology,13: 157–178.

Butts, M. M., Casper, W. J., & Yang, T. S. 2013. Howimportant are work-family support policies? A meta-analytic investigation of their effects on employeeoutcomes. Journal of Applied Psychology, 98: 1–25.

Cadinu, M., Maass, A., Rosabianca, A., & Kiesner, J. 2005.Why do women underperform under stereotypethreat? Evidence for the role of negative thinking.Psychological Science, 16: 572–578.

Caprariello, P. A., Cuddy, A. J. C., & Fiske, S. T. 2009.Social structure shapes cultural stereotypes andemotions: A causal test of the stereotype contentmodel. Group Processes and Intergroup Relations,12: 147–155.

Cardy, R. L., & Dobbins, G. H. 1986. Affect and appraisalaccuracy: Liking as an integral dimension in evalu-ating performance. Journal of Applied Psychology,71: 672–678.

*Chacko, T. I. 1982. Women and equal employment op-portunity: Some unintended effects. Journal of Ap-plied Psychology, 67: 119–123.

Crocker, J., Major, B., & Steele, C. 1998. Social stigma. InS. Fiske, D. Gilbert & G. Lindzey (Eds.), Handbook ofsocial psychology, vol. 2: 504–553. Boston, MA:McGraw-Hill.

Crosby, F. J., Iyer, A., & Sincharoen, S. 2006. Understand-ing affirmative action. Annual Review of Psychol-ogy, 57: 585–611.

Cuddy, A. J. C., Fiske, S. T., & Glick, P. 2007. The BIASmap: Behaviors from intergroup affect and stereo-types. Journal of Personality and Social Psychol-ogy, 92: 631–348.

Cuddy, A. J. C., Fiske, S. T., & Glick, P. 2008. Warmth andcompetence as universal dimensions of social per-

ception: The stereotype content model and the BIASmap. In M. P. Zanna (Ed.). Advances in Experimen-tal Social Psychology, 40: 61–149.

Cuddy, A. J. C., Glick, P., & Beninger, A. 2011. Thedynamics of warmth and competence judgments,and their outcomes in organizations. Research inOrganizational Behavior, 31: 73–98.

Dalton, D. R., Aguinis, H., Dalton, C. M., Bosco, F. A., &Pierce, C. A. 2012. Revisiting the file drawer problemin meta-analysis: An assessment of published andnon-published correlation matrices. Personnel Psy-chology, 65: 221–245.

Dalton, D. R., Certo, S. T., & Roengpitya, R. 2003. Meta-analyses of financial performance and equity: Fusionor confusion? Academy of Management Journal,46: 13–26.

*Deitch, E. A., Brief, A. P., Roberson, L., & Block, C. J.2001. Stereotype threat in the workplace and thewell-being of African American professionals. Paperpresented at the annual meeting of the Academy ofManagement, Washington, D.C.

*DeMatteo, J. S., Dobbins, G. H., Myers, S. D., & Facteau,C. L. 1996. Evaluations of leadership in preferentialand merit-based leader selection situations. Leader-ship Quarterly, 7: 41–62.

Dunning, D., & Sherman, D. A. 1997. Stereotypes andtacit inference. Journal of Personality and SocialPsychology, 73: 459–471.

Ely, R. J., & Thomas, D. A. 2001. Cultural diversity atwork: The effects of diversity perspectives on workgroup processes and outcomes. Administrative Sci-ence Quarterly, 46: 229–273.

*Evans, D. C. 2003. A comparison of the other-directedstigmatization produced by legal and illegal forms ofaffirmative action. Journal of Applied Psychology,88: 121–130.

Feldman, J. M. 1981. Beyond attribution theory: Cogni-tive processes in performance appraisal. Journal ofApplied Psychology, 66: 127–148.

Fiske, S. T., Cuddy, A. J., & Glick, P. 2007. Universaldimensions of social cognition: Warmth then com-petence. Trends in Cognitive Sciences, 11: 77–83.

Fiske, S. T., Cuddy, A. J., Glick, P., & Xu, J. 2002. A modelof (often mixed) stereotype content: Competence andwarmth respectively follow from perceived statusand competition. Journal of Personality and SocialPsychology, 82: 878–902.

Gajendran, R. S., & Harrison, D. A. 2007. The good, thebad, and the unknown about telecommuting: Meta-analysis of psychological mediators and individualconsequences. Journal of Applied Psychology, 92:1524–1541.

Garcia, L. T., Erskine, N., Hawn, K., & Casmay, S. R. 1981.


The effect of affirmative action on attributions aboutminority group members. Journal of Personality,49: 427–437.

*Gilbert, J. A., & Stead, B. A. 1999. Stigmatization revis-ited: Does diversity management make a differencein applicant success? Group and OrganizationManagement, 24: 239–256.

Goff, P. A., Steele, C. M., & Davies, P. G. 2008. The spacebetween us: Stereotype threat and distance in inter-racial contexts. Journal of Personality and SocialPsychology, 94: 91–107.

Gomez-Mejía, L. R., & Balkin, D. B. 1992. Determinants offaculty pay: An agency theory perspective. Academyof Management Journal, 35: 921–955.

Gross, J. J. 2002. Emotion regulation: Affective, cognitive,and social consequences. Psychophysiology, 39:281–291.

Gunderson, E. A., Ramirez, G., Levine, S. C., & Beilock,S. L. 2012. The role of parents and teachers in thedevelopment of gender-related math attitudes. SexRoles, 66: 153–166.

Harrison, D. A., Kravitz, D. A., Mayer, D. M., Leslie,L. M., & Lev-Arey, D. 2006. Understanding attitudestoward affirmative action programs in employment:Summary and meta-analysis of 35 years of research.Journal of Applied Psychology, 91: 1013–1036.

*Hattrup, K. 1998. The role of self-perception in reac-tions to preferential and merit-based hiring. Journalof Applied Social Psychology, 28: 225–334.

Heilman, M. E. 1994. Affirmative action: Some unin-tended consequences for working women. Researchin Organizational Behavior, 19: 125–169.

*Heilman, M. E., & Alcott, V. B. 2001. What I think youthink of me: Women’s reactions to being viewed asbeneficiaries of preferential treatment. Journal ofApplied Psychology, 86: 574–582.

*Heilman, M. E., & Blader, S. L. 2001. Assuming prefer-ential selection when the admissions policy is un-known: The effects of gender rarity. Journal of Ap-plied Psychology, 86: 188–193.

*Heilman, M. E., Battle, W. S., Keller, C. E., & Lee, R. A.1998. Type of affirmative action policy: A determi-nant of reactions to sex-based preferential selection?Journal of Applied Psychology, 83: 190–205.

*Heilman, M. E., Block, C. J., & Lucas, J. A. 1992. Pre-sumed incompetent? Stigmatization and affirmativeaction efforts. Journal of Applied Psychology, 77:536–544.

*Heilman, M. E., Block, C. J., & Stathatos, P. 1997. Theaffirmative action stigma of incompetence: Effects ofperformance information ambiguity. Academy ofManagement Journal, 40: 603–625.

*Heilman, M. E., Lucas, J. A., & Kaplow, S. R. 1990.Self-derogating consequences of sex-based preferen-tial selection: The moderating role of initial self-confidence. Organizational Behavior and DecisionMaking Processes, 46: 202–216.

*Heilman, M. E., McCullough, W. F., & Gilbert, D. 1996.The other side of affirmative action: Reactions ofnonbeneficiaries to sex-based preferential selection.Journal of Applied Psychology, 81: 346–357.

*Heilman, M. E., Rivero, J. C., & Brett, J. F. 1991. Skirtingthe competence issue: Effects of sex-based preferen-tial selection on task choices of women and men.Journal of Applied Psychology, 76: 99–105.

*Heilman, M. E., Simon, M. C., & Repper, D. P. 1987.Intentionally favored, unintentionally harmed? Im-pact of sex-based preferential selection on self-per-ceptions and self-evaluations. Journal of AppliedPsychology, 72: 62–68.

Hogg, M. A., & Abrams, D. 1988. Social identifications:A social psychology of intergroup relations andgroup processes. London, England: Routledge.

Holzer, H. J., & Neumark, D. 2000. What does affirmativeaction do? Industrial and Labor Relations Review,53: 240–271.

Hu, L., & Bentler, P. M. 1999. Cutoff criteria for fit in-dexes in covariance structure analysis: Conventionalcriteria versus new alternatives. Structural Equa-tion Modeling, 6: 1–55.

Huffuctt, A. I., & Arthur, W. 1995. Development of a newoutlier statistic for meta-analytic data. Journal ofApplied Psychology, 80: 327–334.

Hunter, J. E., & Schmidt, F. L. 2004. Methods of meta-analysis: Correcting error and bias in researchfindings (2nd ed.). Newbury Park, CA: Sage.

Jacobs, J. E. 1991. Influence of gender stereotypes onparent and child mathematics attitudes. Journal ofEducational Psychology, 83: 518–527.

*Jacobson, M. B., & Koch, W. 1977. Women as leaders:Performance evaluation as a function of method ofleader selection. Organizational Behavior and Hu-man Performance, 20: 149–157.

Jain, H. C., Sloane, P. J., & Horwitz, F. M. 2003. Employ-ment equity and affirmative action: An interna-tional comparison. Armonk, NY: M. E. Sharpe.

James, E. H., Brief, A. P., Dietz, J., & Cohen, R. R. 2001.Prejudice matters: Understanding the reactions ofWhites to affirmative action programs targeted tobenefit Blacks. Journal of Applied Psychology, 86:1120–1128.

Jones, E. E., & Nisbett, R. E. 1972. The actor and theobserver: Divergent perceptions of the causes of be-havior. In E. E. Jones, D. E. Kanouse, H. H. Kelley,R. E. Nisbett, S. Valins & B. Weiner (Eds.), Attribu-


tion: Perceiving the causes of behavior: 79–94.Morristown, NJ: General Learning Press.

Kalev, A., Dobbin, F., & Kelly, E. 2006. Best practices orbest guesses? Assessing the efficacy of corporate af-firmative action and diversity policies. AmericanSociological Review, 71: 589–617.

Kaplan, S., Bradley, J. C., Luchman, J. N., & Haynes, D.2009. On the role of positive and negative affectivityin job performance: A meta-analytic investigation.Journal of Applied Psychology, 94: 162–172.

Keller, J., & Dauenheimer, D. 2003. Stereotype threat inthe classroom: Dejection mediates the disruptingthreat effect on women’s math performance. Person-ality and Social Psychology Bulletin, 29: 371–381.

Kelley, H. H. 1972. Attribution in social interaction. InE. E. Jones, D. E. Kanouse, H. H. Kanouse, H. H.Kelley, R. E. Nisbett, S. Valins & B. Weiner (Eds.),Attribution: Perceiving the causes of behavior.Morristown, NJ: General Learning Press.

Kelly, E., & Dobbin, F. 1998. How affirmative actionbecame diversity management: Employer responseto antidiscrimination law, 1961–1996. AmericanBehavioral Scientist, 41: 960–984.

Konrad, A. M., & Linnehan, F. 1995. Formalized HRMstructures: Coordinating equal employment opportu-nity or concealing organizational practices. Acad-emy of Management Journal, 38: 787–820.

Kovach, K. A., Kravitz, D. A., & Hughes, A. A. 2004.Affirmative action: How can we be so lost when wedon’t even know where we are going? Labor LawJournal, 55: 53–61.

Landis, J. R., & Koch, G. G. 1977. The measurement ofobserver agreement for categorical data. Biometrics,33: 159–174.

Leonard, J. S. 1984. The impact of affirmative action onemployment. Journal of Labor Economics, 2: 439–463.

Leslie, L. M., & Gelfand, M. J. 2008. The who and when ofinternal gender discrimination claims: An interac-tional model. Organizational Behavior and HumanDecision Processes, 107: 123–140.

Leslie, L. M., King, E. B., Bradley, J. C., & Hebl, M. R.2008. Triangulation across methodologies: All signspoint to persistent stereotyping and discriminationin organizations. Industrial and OrganizationalPsychology, 1: 399–404.

Leslie, L. M., Manchester, C. F., & Dahm, P. C. 2013.Ironic effects of flexible work practices on parents’career success. Academy of Management Proceed-ings.

Leslie, L. M., Manchester, C. F., Park, T.-Y., & Mehng,S. A. 2012. Flexible work practices: A source of

career premiums or penalties? Academy of Manage-ment Journal, 55: 1407–1428.

Leslie, L. M., Snyder, M., & Glomb, T. M. 2013. Whogives? Multilevel effects of gender and ethnicity onworkplace charitable giving. Journal of AppliedPsychology, 98: 49–62.

Lipsey, M. W., & Wilson, D. B. 2001. Practical meta-analysis. Thousand Oaks, CA: Sage Publications.

Lynch, F. R. 1992. Invisible victims: White males andthe crisis of affirmative action. New York, NY:Praeger.

*Maio, G. R., & Esses, V. M. 1998. The social conse-quences of affirmative action: Deleterious effects onperceptions of groups. Personality and Social Psy-chology Bulletin, 24: 65–74.

Major, B., & O’Brien, L. T. 2005. The social psychology ofstigma. Annual Review of Psychology, 56: 393–421.

Malle, B. G., & Knobe, J. 1997. Which behaviors do peo-ple explain? A basic actor-observer asymmetry. Jour-nal of Personality and Social Psychology, 72: 288–304.

Manchester, C. F., Leslie, L. M., & Kramer, A. 2010. Stopthe clock policies and career success in academia.American Economic Review, 100: 219–223.

Manchester, C. F., Leslie, L. M., & Kramer, A. 2013. Is theclock still ticking? An evaluation of the conse-quences of stopping the tenure clock. Industrial andLabor Relations Review, 66: 2–32.

*McGlashan, K. E., Wright, P. M., & McCormick, B. 1995.Preferential selection and stereotypes: Effects onevaluation of female leader performance, subordi-nate goal commitment, and task performance. SexRoles, 33: 669–686.

Miller, D. T., & Ross, M. 1975. Self-serving biases in theattribution of causality: Fact or fiction? Psychologi-cal Bulletin, 82: 213–225.

*Moros, A. 2004. The effect of preferential selection onbeneficiaries’ self-evaluations. Paper presented atthe annual meeting of the Society for Industrial andOrganizational Psychology, Chicago, IL.

Muraven, M., & Baumeister, R. F. 2000. Self-regulationand depletion of limited resources: Does self-controlresemble a muscle? Psychological Bulletin, 126:247–259.

Muthén, L. K., & Muthén, B. O. 2007. Mplus user’s guide(5th ed.). Los Angeles, CA: Muthén & Muthén.

*Nacoste, R. W. 1985. Selection procedure and responsesto affirmative action: The case of favorable treat-ment. Law and Human Behavior, 9: 225–242.

*Nacoste, R. W., & Lehman, D. 1987. Procedural stigma.Representative Research in Social Psychology, 17:25–38.


*Niemann, Y. F., & Dovidio, J. F. 2005. Affirmative actionand job satisfaction: Understanding underlying pro-cesses. Journal of Social Issues, 61: 507–523.

Nisbett, R. E., & Wilson, T. D. 1977. The halo effect:Evidence for unconscious alteration of judgments.Journal of Personality and Social Psychology, 10:385–397.

Orwin, R. G. 1983. A fail-safe N for effect size in meta-analysis. Journal of Educational Statistics, 8: 157–159.

Osborne, J. W. 2001. Testing stereotype threat: Does anx-iety explain race and sex differences in achieve-ment? Contemporary Educational Psychology, 26:291–310.

Paetzold, R. L., Dipboye, R. L., & Elsbach, K. D. 2008. Anew look at stigmatization in and of organizations.Academy of Management Review, 33: 186–183.

Parsons, J. E., Adler, T. F., & Kaczala, C. M. 1982. Social-ization of achievement attitudes and beliefs: Parentalinfluences. Child Development, 53: 310–321.

Phinney, J. S. 1996. When we talk about American ethnicgroups, what do we mean? American Psychologist,51: 918–927.

Plaut, V. C., Garnett, F. G., Buffardi, L., & Sanchez-Burks,J. 2011. “What about me?” Perceptions of exclusionand Whites. Journal of Personality and Social Psy-chology, 101: 337–353.

Plaut, V. C., Thomas, K. M., & Goren, M. J. 2009. Ismulticulturalism or color blindness better for minor-ities? Psychological Science, 20: 444–446.

Pronin, E., Lin, D. Y., & Ross, L. D. 2002. The bias blindspot: Perceptions of bias in self versus others. Per-sonality and Social Psychology Bulletin, 28: 36–381.

*Resendez, M. G. 2002. The stigmatizing effects of affir-mative action: An examination of moderating vari-ables. Journal of Applied Social Psychology, 32:185–206.

Roberson, L., & Kulik, C. T. 2007. Stereotype threat atwork. Academy of Management Perspectives, 21:24–40.

Rosenthal, R. 1979. The file drawer problem and toler-ance for null results. Psychological Bulletin, 86:638–641.

Schmader, T., & Johns, M. 2003. Converging evidencethat stereotype threat reduces working memory ca-pacity. Journal of Personality and Social Psychol-ogy, 85: 440–452.

Schmader, T., Johns, M., & Forbes, C. 2008. An integratedprocess model of stereotype threat effects on perfor-mance. Psychological Review, 115: 336–356.

*Seijts, G. H., & Jackson, S. E. 2001. Reactions to employ-

ment equity programs and situational interviews: Alaboratory study. Human Performance, 14: 247–265.

Shadish, W. R. 1996. Meta-analysis and the explorationof causal mediating processes: A primer of examples,methods, and issues. Psychological Methods, 1: 47–65.

Shapiro, J. R., & Neuberg, S. L. 2007. From stereotypethreat to stereotype threats: Implications of a multi-threat framework for causes, moderators, mediators,consequences, and interventions. Personality andSocial Psychology Review, 11: 107–130.

Shteynberg, G., Leslie, L. M., Knight, A. P., & Mayer,D. M. 2011. But affirmative action hurts us! Race-related beliefs shape perceptions of White disadvan-tage and policy unfairness. Organizational Behav-ior and Human Decision Processes, 115: 1–12.

Spencer, S. J., Steele, C. M., & Quinn, D. M. 1999. Ste-reotype threat and women’s math performance. Jour-nal of Experimental Social Psychology, 35: 4–28.

*Stanush, P., Arthur, W., & Doverspike, D. 1998. His-panic and African American reactions to a simulatedrace-based affirmative action scenario. HispanicJournal of Behavioral Sciences, 20: 3–16.

Steele, C. M. 1990. The content of our character: A newvision of race in America. New York, NY: St Mar-tin’s.

Steele, C. M. 1997. A threat in the air: How stereotypesshape intellectual identity and performance. Amer-ican Psychologist, 52: 613–629.

Steele, C. M., & Aronson, J. 1995. Stereotype threat andthe intellectual test performance of African Ameri-cans. Journal of Personality and Social Psychol-ogy, 85: 797–811.

Steele, C. M., Spencer, S. J., & Aronson, J. 2002. Contend-ing with group image: The psychology of stereotypeand social identity threat. Advances in Experimen-tal Social Psychology, 34: 379–440.

*Summers, R. J. 1991. The influence of affirmative actionon perceptions of a beneficiary’s qualifications. Jour-nal of Applied Social Psychology, 21: 1265–1276.

*Taylor, M. C. 1994. Impact of affirmative action onbeneficiary groups: Evidence from the 1990 generalsocial survey. Basic and Applied Social Psychol-ogy, 15: 143–178.

*Traver, H. A., & Alliger, G. M. 1999. Evaluations andattributions of beneficiaries and nonbeneficiaries ofaffirmative action in mental and physical tasks.Poster presented at the Society for Industrial andOrganizational Psychology annual Conference, At-lanta, GA.

*Turner, M. E., & Pratkanis, A. R. 1993. Effects of prefer-ential and meritorious selection on performance: An


examination of intuitive and self-handicapping per-spectives. Personality and Social Psychology Bul-letin, 19: 47–58.

*Unzueta, M. M., Gutiérrez, A. S., & Ghavami, N. 2010.How believing in affirmative action quotas affectsWhite women’s self-image. Journal of ExperimentalSocial Psychology, 46: 120–126.

*Unzueta, M. M., Lowery, B. S., & Knowles, E. D. 2008.How believing in affirmative action quotas protectsWhite men’s self-esteem. Organizational Behaviorand Human Decision Processes, 105: 1–13.

*Van Laar, C., Levin, S., & Sinclair, S. 2008. Social iden-tity and personal identity stereotype threat: The caseof affirmative action. Basic and Applied Social Psy-chology, 30: 295–310.

Viswesvaran, C., & Ones, D. S. 1995. Theory testing:Combining psychometric meta-analysis and struc-tural equations modeling. Personnel Psychology,48: 865–886.

Wojciszke, B., Bazinska, R., & Jaworski, M. 1998. On thedominance of moral categories in impression forma-tion. Personality and Social Psychology Bulletin,24: 1245–1257.

Yang, C., D’Souza, G., Bapat, A., & Colarelli, S. M. 2006.A cross-national analysis of affirmative action: Anevolutionary psychological perspective. Managerialand Decision Economics, 27: 203–216.

Yerkes, R. M., & Dodson, J. D. 1908. The relation ofstrength of stimulus to rapidity of habit-formation.

Journal of Comparative Neurology and Psychol-ogy, 18: 459–482.

Lisa M. Leslie ([email protected]) is an associate pro-fessor in the Management and Organizations Departmentat the Stern School of Business, New York University.She received her Ph.D. in organizational psychologyfrom the University of Maryland. Her research focuses ondiversity in organizations, cross-cultural organizationalbehavior, and conflict management.

David M. Mayer ([email protected]) is an associateprofessor in the Management and Organizations Area atthe Ross School of Business, University of Michigan. Hereceived his PhD from the University of Maryland. Hisinterests include social and ethical issues in organiza-tions and his research focuses on behavioral ethics, or-ganizational justice, and workplace diversity.

David A. Kravitz ([email protected]) is a professor inthe Management Area at the School of Management,George Mason University. He received his PhD in socialpsychology from the University of Illinois at Urbana-Champaign. His research focuses on affirmative actionand diversity in organizations. He is particularly inter-ested in closing the gap between diversity researchersand practitioners.


Copyright of Academy of Management Journal is the property of Academy of Managementand its content may not be copied or emailed to multiple sites or posted to a listserv withoutthe copyright holder's express written permission. However, users may print, download, oremail articles for individual use.

the stigma of affirmative action: a stereotyping-...

Documents