VIII. CONCLUSIONS, LIMITATIONS OF CURRENT KNOWLEDGE, AND NEEDED
RESEARCH
A. Conclusions
The strongest conclusion that can be drawn from the reviewed research is
that the structure of an AAP will influence reactions to it. Attitudes are inversely
related to the weighting of demographic status, and evaluations of selection procedures
are directly related to the superiority of the chosen candidate. The effect of AAP
structure on attitudes is mediated, at least in part, by judgments of fairness and
self-interest, and fairness ratings are highly correlated with attitudes. There is
slightly more support for AAPs directed at women and people with disabilities than for
AAPs directed at racial minorities, although this effect is moderated by respondent
demographic status in a manner consistent with self-interest. Minorities and women are
more supportive of affirmative action than are White males, but other demographic
variables (e.g., age, income, education) are of little consequence. Attitudes toward
race-based affirmative action are inversely related to racism, and limited research
suggests that attitudes toward gender-based affirmative action are inversely related to
sexism. It has been suggested that attitudes will be associated with judgments of relative
deprivation of the target group and the respondent's own group, but valid research is
limited.
Some limited evidence suggests that there is great variability in what
the public thinks AAPs entail, and that public opinions are flexible. Opinions can be
changed by providing the respondent with information about details of the AAP and by some
justification of the use of affirmative action. Support for affirmative action is stronger
if the respondent has personally experienced discrimination. In addition, support for
affirmative action is higher if the respondent believes or is told that the target group
has suffered discrimination. There is evidence, however, that many Whites believe
discrimination is no longer a problem, and that Blacks themselves are to blame for the
Black-White income gap. This belief is consistent with a structuralist ideology, and
attitudes are inversely related to acceptance of a conservative, structuralist political
ideology.
Individuals who are identified as being selected under an AAP are
perceived as less competent, by themselves and by others, unless information is provided
that clearly and unambiguously demonstrates their competence. Conclusions regarding
self-stigmatization must be qualified because almost all the relevant research has been
based on reactions of White women to gender-based selection procedures; it is not known
whether the results will generalize to ethnic minorities selected in the context of
race-based procedures. There is no empirical research on effects of affirmative action on
relations among groups, but theoretical work predicts that effects will be negative unless
the AAP is positively evaluated by all involved parties.
Our limited review of the economic literature found that implementation
of affirmative action is associated with improved employment conditions of women and
racial minorities, although the improvements have been relatively small and inconsistent
across subgroups. Other work revealed no apparent effects of affirmative action on
organizational effectiveness. In addition, research has found that formal charges of
discrimination led to a decrease in stock prices, and public recognition of affirmative
action excellence led to a temporary increase in stock prices. In short, there is evidence
that affirmative action helps target group members, but no empirical evidence that it
harms organizations.
B. Limitations and Needed Research
The conclusions drawn above must be tempered by limitations in the
research. Much of the research on target and non-target group members' reactions has used
experiments, which necessarily simplify matters. This has the advantage of increasing
internal validity, but does not permit exploration of the complexity of affirmative action
attitudes. Research using surveys, in contrast, is more likely to permit such exploration.
Survey research, however, lacks internal validity. Indeed, all research on individual
difference variables (e.g., gender, ethnicity, racism, political perspective) lacks
internal validity because these variables cannot be manipulated. Ideally, every question
should be addressed with multiple research strategies, and the weaknesses of one strategy
will be balanced by the strengths of others (McGrath, 1981). Unfortunately, as we have
mentioned throughout this review, many questions have been addressed in limited ways. For
example, research on self-stigmatization has been limited to reactions of women to
gender-based selection in (mainly laboratory) experiments.
Another potential limitation is the contextual realism of the research
and stimuli for participants. In comparison to actual AAPs, the stimuli used in most of
this research have been simplistic. In addition, except for a few field studies with
employed adults, this research has not been done within the context of an actual AAP.
Furthermore, much of the research has been done with undergraduates who typically have had
little direct experience with affirmative action. This lack of direct experience has
important implications for how "crystallized" attitudes towards AAPs may be, and
therefore how easy they are to change or how strongly they manifest themselves in
behavior. Insofar as research on attitudes toward AAPs is concerned primarily with content
differences across persons and contexts rather than process differences (Eberhardt
& Fiske, 1994), the external validity of studies done with undergraduates responding
to novel scenarios is uncertain until demonstrated otherwise.
On a related point, there is a great need for research on what
organizations actually do to implement their AAPs (cf. Konrad & Linnehan, 1995a). This
information could be used to improve the mundane realism of research on reactions to AAPs,
and to evaluate the accuracy of public beliefs. There is also a need for research on
public beliefs about what actions are and should be included in AAPs. This information
would help administrators develop AAPs that help the target groups without stimulating
resentment and opposition.
Confusion about what affirmative action entails has been increased by
inconsistent statements made by decision makers and published by the media. A similar
confusion reigns in the professional literature. In brief, operational definitions of
affirmative action have varied considerably, but this variation has not been accompanied
by a parallel variation in terminology. Thus, some people operationalize affirmative
action as quotas, others as preferential treatment (weak, strong, or unspecified), others
as recruitment, others as the elimination of discrimination, etc. Given the strong effect
of AAP structure on reactions, this inconsistency in operational definitions has sometimes
led to a parallel inconsistency in results. We recommend that researchers use more precise
terms when describing their research; "affirmative action" is too vague.
Most of the research we reviewed was performed in the United States,
although Canada, New Zealand, and the Netherlands were also represented. There is a need
for additional research on affirmative action in other countries, and for an additional
attention to cultural influences on reactions to affirmative action. In addition, the
political climate in the United States has changed considerably since 1965, when EO11246
was issued. A review linking affirmative action attitudes to other changes over time would
be a valuable contribution.
Another important question is how individuals' attitudes change over
time, especially as a function of experience with affirmative action; we know of no
longitudinal research on this question. Still another important question is how job
satisfaction, organizational commitment, and similar variables are related to reactions to
the organization's affirmative action plan. Witt (1990) reports the results of three
discriminant analyses on the job satisfaction of White male university faculty, and
concludes that attitudes toward affirmative action have little effect compared to the
effects of time demand and other types of stress. We know of no other research on this
important issue.
Another methodological concern is the possibility that people do not
honestly express their affirmative action attitudes. This potential problem was addressed
in a recent article by Coughlin (1995), who describes Timur Kuran's views on preference
falsification. Kuran argues that people are unwilling to express their negative opinions
of affirmative action because they are afraid of being labeled as racists. While this
criticism applies to some research, much of the published work is based on anonymous
responses. In addition, it seems inconsistent with the very negative evaluations of
preferential treatment. Nonetheless, researchers who study affirmative action (or any
other sensitive topic) should take steps to ensure that respondents can express their
opinions openly and without fear.
In conclusion, previous research provides numerous questions and
hypotheses about affirmative action attitudes and about how affirmative action affects
target group members and non-members. Some conclusions can now be drawn with confidence,
but much more research is needed before we can claim to thoroughly understand the
psychological and behavioral implications of affirmative action.