518 Prejudice, Racism, and Discrimination
self-focus and self-attention, and they subsequently monitor
their behavior more carefully to ensure that it conforms more
closely to their personal beliefs.
Critique of the Dissociation Model
The dissociation model’s contention that prejudiced and un-
prejudiced perceivers would be equally responsive to priming
by an automatic processing task has, however, been recently
criticized and questioned by several investigators. Lepore and
Brown (1997), for example, criticized Devine’s (1989) auto-
matic processing study for including both categorical cues
referring to Blacks as a social group and stereotypic traits of
Black people among the subliminal primes. As an alternative
to the dissociation model, Lepore and Brown argued that the
link between the category and the stereotypic features relating
to Blacks differentiates White perceivers varying in preju-
dice, with the link being much stronger and more chronically
accessible for highly prejudiced White persons than for less
prejudiced ones. If only categorical cues referring to Blacks
as a group comprise the subliminal primes on an automatic
processing task, one should observe highly prejudiced White
persons subsequently forming more negative impressions
than less prejudiced ones—a result that Lepore and Brown
(1997, Study 2), in fact, have obtained.
By contrast, subliminal cues that include stereotypic at-
tributes along with the categorical label also prime the stereo-
typic knowledge of both high- and low-prejudice White
perceivers, which has been shown to be highly similar. Thus,
subliminal cues containing both category references and
stereotypic attributes on an automatic processing task would
notbe expected to reveal differences between White persons
varying in prejudice, a prediction that Lepore and Brown
(1997, Study 3) also supported in a conceptual replication of
Devine’s (1989) automatic processing study. Null hypothesis
predictions have been rife on the issue of automatic process-
ing effects on impression formation as a function of the
White participants’ prejudice toward Blacks. Predicting
the null hypothesis, however, is problematic because tests of
such hypotheses often lack sufficient statistical power (see
Cohen, 1992).
Kawakami, Dion, and Dovidio (1998) further reinforced
Lepore and Brown’s conceptual analysis in two ways. They
found that high-prejudice White persons were more respon-
sive to primes on a single task where automatic and con-
trolled processing could both be experimentally manipulated
by varying stimulus onset asynchrony (i.e., the difference in
time between presentation of the prime and a subsequent, to-
be-responded-to stimulus). Second, individual differences in
stereotype attribution as assessed by a separate measure
correlated with stereotypic activation on the experimental
task when it allowed automatic processing.
With regard to Devine’s automatic processing findings,
Fazio, Jackson, Dunton, and Williams (1995) have suggested
that the MRS has become a reactive and insensitive measure
of racial prejudice. Consistent with this point, they showed
that the levels of modern racism in White American partici-
pants failed to moderate priming effects on a procedure
(described later) that was designed to elicit automatic activa-
tion of racial attitudes.
Taken together, the preceding critiques of the dissociation
model have important implications for prejudice and its re-
duction. According to Lepore and Brown’s (1997) alternative
perspective, low-prejudice White persons have never estab-
lished the bad habit of prejudice toward Black people in the
first place or established it much less firmly than their highly
prejudiced White counterparts. For low-prejudice White per-
sons, the link between the social category, Blacks, and the
culturally stereotypic information about them is already weak
and tenuous. Rather than unlearning a bad habit, those inter-
ested in reducing prejudice in White people presumably need
to focus on the highly prejudiced Whites and on weakening
the associative strength of the links between the category of
Blacks as a social group and negative stereotypic information
and content about them.Automatic Activation as Prejudice MeasuresAutomatic activation techniques are a means of unobtrusively
measuring racial and other intergroup attitudes and an alterna-
tive to traditional attitude scales, which are often compromised
by social desirability and transparency regarding the goal of
assessing prejudice. Even the MRS has recently been shown to
be sensitive to social desirability, yielding lower scores from
White participants when administered by a Black experi-
menter than by a White one (Fazio et al., 1995, Study 3). From
their findings in several studies, Fazio et al. (1995) have styled
the MRS as a measure of White Americans’ “willingness to
express” negative feelings or opinions about Blacks, one that
also confounds racism with political conservatism. Other re-
searchers have noted that correlations between old-fashioned
and modern and symbolic racism are higher than would be
expected if these were truly two separate constructs rather than
different aspects of a single construct (see Dovidio et al., 1997;
Swim et al., 1995).
As an alternative, Fazio et al. (1995) proposed a priming
paradigm using automatic activation of attitudes from mem-
ory as an unobtrusive measure of racial attitudes that is
demonstrably superior to the MRS. The priming procedure
consists of multiple trials on a computer in which the prime