V. PSYCHOLOGICAL AND BEHAVIORAL EFFECTS OF AFFIRMATIVE ACTION ON
TARGET GROUP MEMBERS
In this section, we review the empirical research on psychological and
behavioral effects of affirmative action on target group members. The review is based in
part on Turner and Pratkanis (1994a). This research has included three categories of
outcomes: (a) measures of motivation and task interest, (b) selfevaluations of ability
and performance, and (c) performance. After discussing empirical research, we briefly
review theoretical explanations for the reported effects.
A. Measures of Motivation, Interest, Commitment, and Choice
Research concerning the effects of selection procedures on task
motivation, interest, commitment, and choice has been conducted primarily with female
respondents. Results of these investigations are rather inconsistent.
Some studies indicate that affirmative action and sexbased selection
procedures decrease motivation and interest. In a survey of 70 women in managerial or
supervisory positions in a variety of organizations, Chacko (1982) found that women who
believed their sex was the reason for their hiring reported lower job commitment and
satisfaction and greater role ambiguity and conflict than women who did not believe sex
played an important role in their hiring. Heilman et al. (1987), however, did not observe
deleterious effects of sexbased selection on measures of task motivation or interest.
Similarly, Turner, Pratkanis, and Hardaway (1991) and Turner and Pratkanis (1993) found
that selection on the basis of sex did not directly impair selfreported motivation for a
task.
In one of the few national field studies on this issue, Taylor (1994)
examined the attitudes of 319 White women, 40 Black women, and 32 Black men who said they
worked for firms that either did or did not employ affirmative action procedures. Results
showed that White women working for firms with affirmative action programs did not differ
significantly from White women working for firms without such programs in their
selfreports of job satisfaction, working by choice as opposed to money, desire for an
enriched job, life satisfaction, happiness and other measures. Of these measures, only two
differed significantly for Black men and women. First, Black respondents reported a
greater desire for an enriched job (i.e., a job that would require a greater degree of
ambition, chances for advancement, feeling of accomplishment) when they said they worked
for a firm with affirmative action than when they reported they worked for a firm without
affirmative action. Black men and women differed in their evaluations of the degree of
excitement in their lives. Black men working for firms without affirmative action reported
more positive evaluations than did Black men working for firms with affirmative action
programs. Black women showed the opposite pattern.
In a mail survey reported by Graves and Powell (1994), 188 MBA graduates
(61 of whom were male; 96 of whom were White) were asked to indicate the influence of sex
in general and of sexbased affirmative action in past and future selection decisions,
and to indicate their job satisfaction, organizational commitment, and intent to remain
with their organizations. Results indicated that men felt their gender benefited them in
selection and promotion decisions more than did women. Women felt that they benefited from
sexbased affirmative action procedures more than did men. Job satisfaction was affected
by respondent sex, perceptions of the overall influence of sex in selection decisions, and
perceptions of sexbased affirmative action. For men, satisfaction and commitment
increased as discrimination against males decreased, remained level when practices were
sex neutral or favored men somewhat, and increased when practices offered a great
advantage to men. For women, satisfaction was not related to the effect of sex on
selection practices. Commitment, however, increased as perceptions of overall influence of
sex increased. Unlike the satisfaction measure, sex moderated the relationship between
perceptions of affirmative action and commitment. Females were more committed when they
perceived that affirmative action practices were advantageous to females. For males,
commitment increased as affirmative action practices were perceived as sex neutral or as
favoring males, and then leveled off.
Several studies have examined how selection affects task or job choice.
Heilman and Herlihy (1984) found that 90 male high school students expressed less interest
in an occupation in which sexbased as opposed to merit criteria was used. On the other
hand, 85 female students expressed less interest only when treatment on the basis of sex
was given and the proportion of women in the occupation was relatively high that is,
28% of the employees were female versus 8%. Nacoste (1987) found that women who read a
scenario about a competitively awarded university research grant were less likely to
report that they would apply for a job at that university when sex alone was used as the
basis of the award than when both sex and qualifications were employed. Graves and Powell
(1994) found that intent to remain with an organization was generally unaffected by
respondents' perceptions of the overall influence of sex in selection and promotion
decisions and of sexbased affirmative action procedures, although there was some
indication that females expressed a greater intent to remain in organizations with
affirmative action procedures perceived as favorable to women. In two separate studies,
Heilman et al. (1987; Heilman, Lucas, & Kaplow, 1990) found that females' desire to
remain in leadership roles was adversely affected when they were told they were selected
for their positions on the basis of sex. Finally, Heilman et al. (1991) reported that
women selected on the basis of sex subsequently chose a less demanding task than did women
selected on the basis of merit. Males did not differ in their task choice as a function of
selection procedure. In a second study, the type of selection procedure influenced
females' task choices. Only females selected on the basis of sex and who were given no
information about their qualifications chose the less demanding task. Females who were
selected on the basis of sex and given positive information about their qualifications,
like females selected on the basis of merit, were more likely to choose the more demanding
task.
Three effects appear to hold when considering the relationships among
selection procedure and measures of task motivation and interest. First, women's reports
of task motivation, interest, and job satisfaction do not appear to be reliably affected
by sexbased selection or affirmative action. Second, women's task and job choice seem to
be more consistently affected by sexbased selection: Women were more likely to choose
easier versus harder tasks and to show less interest in occupations in which sex alone was
used as the basis for selection. This second finding may reflect the fact that being
selected solely on the basis of sex raises doubts about one's ability, and that recipients
may attempt to avoid the stigma that is often attached to being selected on the basis of
sex. Third, the effect on women's task choice is determined by the nature of the
implementation strategy. When the implementation strategy provides unambiguous, explicit,
and focused evidence regarding qualifications, affirmative action and sexbased selection
do not adversely affect task choice. Because only one study focused on race-based
selection procedures, we cannot draw any general conclusions about effects of such
procedures.
B. Selfevaluations of Ability and Performance
The bulk of research on effects of selection procedures on recipient
selfevaluations of ability and performance has also dealt with gender-based selection
procedures. Few investigations examine how ethnic minorities respond to equivalent race-
or ethnicity-based selection processes (but see articles by Arthur et. al, 1992,
Doverspike & Arthur, 1995, Nacoste, 1994, Taylor, 1994, for exceptions; see Eberhardt
& Fiske, 1994 for a discussion of the pitfalls of generalizing to other demographic
groups). This research suggests that affirmative action, under certain conditions, may
affect women's selfevaluations of abilities and performance.
Most experimental studies examining this issue use a paradigm developed
by Jacobson and Koch (1977). They paired male participants with a female confederate who
was assigned to a leadership position on the basis of sex (strong preferential treatment),
chance, or superior performance on a test (merit). After performing a oneway
communication task with the confederate, the dyad was told they had either succeeded or
failed at the task.
Heilman et al. (1987) used a similar paradigm to examine leaders'
selfperceptions. Female and male participants were told they were selected for a
leadership role in a oneway communication task on the basis of either merit or sex. In
the meritbased selection condition, participants were told that their scores on a test
purportedly measuring leadership ability indicated they were qualified to assume the
leadership role. In the gender selection condition, participants were told that
insufficient numbers of males or females had participated in the study and that because of
their respective sex they would assume the leadership role. Again, participants were told
their dyad had either succeeded (that is, scored in the top quartile of all participants)
or failed (scored in the bottom quartile). Results showed that females who were selected
on the basis of their sex evaluated their leadership ability and their global task
performance more poorly than did females who were selected on the basis of merit. Males,
in contrast, were largely unaffected by the selection procedure. Moreover, these results
occurred despite the level of objective feedback concerning task performance.
It is important to note that the sex-based selection procedure employed
in this study was ambiguous with regard to applicant qualifications (i.e., the purported
selection test was not scored). These results suggest that procedures that employ sex as
the sole selection criterion and leave the issue of qualifications open can adversely
affect members of the targeted group. Certainly, one question raised by this study is
whether indications of applicant qualifications would overcome these consequences. Several
studies have addressed this issue.
As a whole, this line of research suggests that overt (but not subtle)
indications of qualifications may overcome the effects of sex-based selection. Along these
lines, Turner et al. (1991) reasoned that the negative consequences of this type of
selection should be especially apparent when individuals are assigned to a job for which
they have low expectations of success (such as when individuals face
"sexinappropriate" tasks) but should be less apparent when expectations of
success are greater (such as when individuals face "sexappropriate" tasks). To
test these hypotheses, Turner et al. (1991) told 96 male and 96 female participants that
they were selected on the basis of either merit or sex. Participants were assigned to a
task that was described using either masculine or feminine sexrole attributes, and were
told they either succeeded or failed. Results suggested that males and females responded
differently to meritorious and sexbased selection procedures, and that the sextyping
of the job did little to affect these responses. Females selected on the basis of sex
evaluated their decision making ability, their decision making performance, their
counseling ability, and their counseling performance more negatively than did
meritoriously selected females. In contrast, the evaluations of males selected on the
basis of sex were slightly but generally not significantly higher than those of their
meritoriously selected counterparts. Contrary to some previous research, males' and
females' selfevaluations of their general, overall performance effectiveness were
affected only by feedback and not by the type of selection process they experienced.
Nacoste (1989) told 97 male and female participants that they were
selected for inclusion in the experiment on the basis of either merit (their score on a
qualifying test) or sex (a lack of male or female participants for the study). Based on a
prior survey, Nacoste also selected equal numbers of male and female participants who
thought affirmative action policies were fair or unfair. Respondents predicted their
creativity scores, and these predictions can be treated as ratings of self-efficacy. There
was an effect of procedure on self-efficacy among those who considered affirmative action
unfair, but not among those who considered affirmative action fair. Thus, selection on the
basis of a procedure considered to be unfair led to lower self-efficacy.
Brutus and Ryan (1994) told 84 female undergraduates they had been
selected on the basis of merit or direct preferential treatment. In an ambiguous selection
condition, the experimenter stated that selection was merit-based but an experimental
confederate said it was gender-based. Participants' explanations for their selection
varied with the interaction of procedure and a pre-manipulation measure of self-efficacy,
so that self-efficacy had the strongest effect when the selection procedure was ambiguous.
Self-evaluation, however, was not affected by selection procedure or self-efficacy.
Several aspects of these results are worth noting. First, feedback
directed specifically at the performance of the individual does seem effective in
overcoming some negative effects of sexbased selection procedures. Turner et al. (1991),
for example, found that providing success feedback (i.e., scores in the top quartile) to
participants selected on the basis of sex overcame the negative effects of sexbased
selection on selfevaluations of overall performance effectiveness. However, this general
feedback did not generalize to other more specific attributes of performance, nor did it
generalize to evaluations of ability. Thus, sexbased selection impaired females'
selfevaluations of specific components of performance but not overall effectiveness.
This finding is consistent with models of selfesteem maintenance, suggesting that one
way to deal with threatening information is to localize the threat to a single domain and
then distance or "disidentify" with that area (Steele, 1992). Finally, subtle
indications of qualifications, such as assignment to a sexrole consistent positions, do
not seem to mitigate the negative consequences of sexbased selection procedures.
Research on more salient indicators of qualifications reveals stronger
moderating effects. Nacoste (1985) asked 96 female undergraduates to read a scenario
describing the selection process for a competitively awarded university research grant.
Women who read a scenario in which the grant was awarded on the basis of both sex and
applicant qualifications reported more positive affective evaluations than did women who
read a scenario in which the grant was awarded on the basis of sex alone. (This summary
variable included measures of competency, relaxation, pleasure, etc.)
The previously-discussed research by Arthur et al. (1992) partially
supported these results. All participants reported more negative affective evaluations (a
composite of measures of feelings of incompetence, dissatisfaction, irritation,
displeasure, etc.) when the recipients were less qualified than when they were equally
qualified. In their partial reanalysis of these data, Doverspike and Arthur (1995) found
that affect was more negative when the candidates' qualifications were unequal than when
they were equal. Analyses of affect also revealed a significant interaction of Sex X Race
X Qualifications. Evaluations of competence were affected by a main effect of selection
policy (with race-only conditions evaluated more positively than race-and-qualifications
conditions), a main effect of qualifications (equal qualifications more positively
evaluated than unequal qualifications), and several second order interactions. However,
these findings were qualified by a significant four way interaction of sex, selection
policy, qualifications, and race. Generally, all participants evaluated themselves more
positively when (a) their qualifications were greater and when (b) both race and
qualifications were part of the procedures. The one exception was Black males who
evaluated themselves more positively in the race only condition than in the race and
qualifications condition. This may have been an example of the selfprotective properties
that can be induced by these policies (e.g., Crocker & Major, 1989).
Heilman et al. (1990) told male and female participants they were
assigned to a leadership task on the basis of merit or sex. In addition, participants
assigned on the basis of sex either were given no information, were told they did well and
scored better than a confederate, or were told they did poorly and scored worse than a
confederate on a qualifying test. Results showed that females selected on the basis of sex
who received either no information or negative information about their scores evaluated
their leadership ability and performance more poorly than did both meritoriously selected
females and females selected on the basis of sex who were given positive information about
their scores. Males' selfevaluations were less favorable only in the sexbased
selection procedure, negative information condition.
Finally, Major, Feinstein, and Crocker (1994) found that
selfevaluations of leader skill were affected by the interaction of participant sex with
selection procedure. When selected on the basis of sex alone, males evaluated their skill
more positively than females. The selfevaluations of males and females did not differ
when they were selected on the basis of merit and on the basis of both sex and merit.
Interestingly, somewhat in contrast to Heilman et al. (1990), selfevaluations of leader
skill were not affected directly by selection procedure. Major et al. (1994) also found
that leaders selected on the basis of sex believed their qualifying score and ability were
less important in their selection than did participants selected on the basis of merit and
on the basis of sex and merit. Additionally, men were less likely to say their sex was an
important factor in their selection than were women.
Taken together, the results of these studies provide clear evidence that
selection procedures can have both deleterious and beneficial consequences for women's
selfevaluations of ability and performance. The nature of the implementation strategy
appears to determine the characteristics of these responses (see Barnes Nacoste, 1990 for
a procedural justice interpretation). To combat the selfdeprecation effect, the evidence
regarding qualifications must be unambiguous in its confirmation of the woman's
competency. Tactics that provide no information about qualifications appear to indicate to
targeted women that their qualifications are deficient, which in turn results in
selfdeprecation. The evidence of the woman's qualifications should be explicit.
Strategies that provide only subtle indications of competencies clearly produce poorer
self-evaluations than do meritorious selection strategies. Finally, the evidence should be
focused. Feedback was ineffective in mitigating the poorer selfevaluations associated
with sexbased selection when it was not concerned with the recipient's performance on
specific components of the task. Because virtually all the research was limited to women's
reactions to sex-based selection procedures, we cannot yet draw conclusions about effects
of race-based procedures.
This research has revealed that sex-based selection procedures can
affect self-evaluations of ability and performance. One obvious question is whether such
selection actually affects performance. Thus, we now turn to research on effects of
selection procedure on performance.
C. Performance
Only three studies have experimentally investigated the effects of
sexbased selection on task performance. Brutus and Ryan (1994), discussed above, found
that performance varied with the interaction of selection procedure and self-efficacy.
Self-efficacy had a stronger effect on performance when the selection procedure was
ambiguous than when it was clearly merit-based.
Nacoste (1989) told 48 male and 49 female participants that they were
selected for participation in the experiment on the basis of either merit or sex. There
were equal numbers of male and female participants who thought affirmative action policies
were fair or unfair. On a brainstorming task, Nacoste found no differences due to
selection procedure or fairness perceptions.
A study by Turner and Pratkanis (1993) demonstrated that performance is
complexly affected by selection processes and conceptualizations of task requirements.
Female participants (N = 60) were told that they were assigned to a position on the basis
of either merit or sex, and that successful performance on the experimental task required
either effort or inherent capabilities. Participants performed eight trials of a standard
brainstorming task. Women who were told they were selected on the basis of sex performed
better when the task required capability rather than effort. In contrast, meritoriously
selected women showed the opposite pattern; they performed better when the task required
effort rather than capability. Interestingly, selfevaluations of performance
effectiveness were affected only by the selection procedures. Females selected on the
basis of sex evaluated their performance creativity, their performance quality, and the
degree of analytic judgment they demonstrated more poorly than meritoriously selected
participants. This finding replicates many of the results discussed earlier. Turner and
Pratkanis interpreted their results as supporting a selfhandicapping model of sexbased
selection whereby individuals who feel uncertain about their ability allocate less effort
to the task in an attempt to attribute possible failure to a lack of effort rather than a
lack of ability. Thus, females selected on the basis of sex performed more poorly when the
task was described as requiring effort (a selfhandicapping tactic). No such readymade
excuse was available for participants selected on the basis of sex who were told the task
required inherent capability, and their performance was not deleteriously affected.
In summary, laboratory findings reveal a complicated relationship
between sexbased selection procedures and individual task performance. Effects of
selection procedure on performance appear to be moderated by such variables as
self-efficacy and task characteristics, which may affect task performance through their
influence on selfhandicapping strategies. Future studies on this topic are needed.
Researchers should examine other selection strategies, additional measures of task
performance, and various potential modifiers of the relationship between selection
procedure and performance. In addition, there is a need for research on race-based
selection.
D. Theoretical Perspectives on Recipient Reactions to Affirmative
Action
A number of theories have been employed to interpret research findings
on recipient reactions to affirmative action. These predictive and explanatory models of
recipient reactions to affirmative action draw on several disciplines, including social
psychology, organizational behavior, sociology, and social cognition. In this section, we
review five such perspectives.
Pettigrew and Martin (1987) invoked the notion of triple jeopardy to
explain recipient reactions to affirmative action. They suggested that target group
members (and specifically Black Americans) may be subjected to three pressures in
organizational settings. Negative stereotypes, solo status (being the sole target group
member) and token status (the perception that the employee is incompetent because he or
she obtained employment through affirmative action) together are predicted to contribute
to such outcomes as exaggerated expectations, assumed dissimilarities, extreme and/or
distorted evaluations, and possibly distorted performance.
Barnes Nacoste (1990; 1994) used procedural justice theories to predict
reactions to affirmative action (see also Clayton and Crosby, 1992). He suggested that
affirmative action policies and procedures that conform to the requirements of procedural
justice (e.g., those that are perceived as fair, give primary weight to merit, are used to
overcome discrimination, etc.) are more likely to produce positive recipient reactions.
Policies that violate procedural justice requirements induce the stigma of incompetence
and engender negative recipient reactions. Heilman and her colleagues (see Heilman, 1994
for a review) suggested that sexbased selection and affirmative action policies that do
not strongly incorporate candidate qualifications raise questions about applicant
competency and induce negative selfevaluations. According to Heilman, programs that
incorporate information about qualifications do not have such negative consequences.
Similarly, Crocker and Major (e.g., Major et al., 1994) suggested that certain forms of
affirmative action induce attributional ambiguity on the part of recipients. According to
this perspective, affirmative action recipients may experience uncertainty about the
causes of their outcomes (such as pay or performance) because these outcomes may be
attributed to affirmative action rather than to any personal characteristics or behavior
(such as education, ability, or effort).
Finally, Turner and Pratkanis (1993, 1994a, 1994b; Turner et al., 1991)
developed a model in which affirmative action can be perceived as a form of help. Like
other forms of help or assistance, it can induce positive consequences, negative
consequences, or a combination of positive and negative consequences. According to the
model, affirmative action engenders immediate negative outcomes (such as negative
selfevaluations of ability and performance, negative affect, negative evaluations of the
procedure and the organization, coupled with a high degree of motivation to alter the
situation and defensive behavior) when it is selfthreatening (i.e., it provides negative
selfrelevant messages, violates important societal norms, and fails to provide
instrumental benefits by removing discriminatory institutional barriers). These
shortterm consequences are predicted to translate into longterm negative consequences
of learned helplessness and dependence when the recipient perceives a low level of
personal control to change the situation. However, when the recipient feels a high level
of personal control, more positive consequences such as attempts to alter the
organizational situation or attempts to improve the recipient's situation are possible. In
contrast, Turner and Pratkanis suggested that affirmative action induces positive outcomes
when it is selfsupportive for the recipient. When affirmative action provides positive
selfrelevant messages (i.e., it provides explicit, unambiguous, and focused evidence of
qualifications), conforms to important societal norms (i.e., it is procedurally fair,
promotes independence and selfreliance, and rewards excellence), and provides
instrumental benefits (i.e., it provides indications of future success and removes
barriers to success and advancement), it is likely to produce positive immediate and long
term outcomes such as positive evaluations of ability and performance, positive affect,
low motivation to alter the situation and low degree of selfprotective defensive
behavior.
E. Summary
This review suggests that affirmative action programs may have positive,
negative, or both positive and negative consequences for recipients. Women's task choices
and selfevaluations of ability and performance can be negatively affected by sex-based
selection procedures that provide no evidence of recipient qualifications. Programs that
provide explicit, unambiguous, and focused evidence of recipient qualifications, in
contrast, do not appear to impair task choice or selfevaluations of performance and
ability. The selection procedure does not appear to strongly affect women's motivation. As
these other results might imply, research suggests a complex relationship between
selection procedure and individual task performance. However, few studies have been
conducted on this topic. In sum, it appears that the implementation strategy assumes
primary importance in determining the nature of women's reactions to affirmative action.
Thus, as with any other organizational change effort, the consequences of the intervention
depend heavily on the specific characteristics of its implementation. Finally, virtually
none of this research has dealt with reactions of racial minorities to race-based
selection procedures, and there is a clear need for such research.