These data should, however, be interpreted with caution, as the sample of Blackparticipants was
rather small (n = 8).
What Fazioetal. call “over compensation” seems similar to what has elsewhere been termed“overcorrection” (Wegener & Petty, 1995). Although I will consistently use the former term, I see The two
as interchangeable. In either case, an actoris attempting to undoor avoid The effects of a presumed
bias, but because of aflawedimplicit theory about the magnitude of the bias, the corrective or
compensating process is excessive.
Given the demographic s of the student body from whichparticipants were drawn(over whelmingly
majority White), we suspect that most participants assumed that the alternative to the Blackpartner
was, in fact, a White partner.
In pretesting, the majority of participants indicated that the ywould assume the owner of such a
sweatshirt was Black. Ithank Joanne Millerfor suggesting the use of the sweatshirtto imply the
In fact, a follow-up study suggested that the IAT did not predict scores on either the MRS or the
Attitudes Toward Blacks Scale(Sargent & The Il, 2002), each of which solicits re actions to only
Blacks. Importantly, this was true even among individuals low on either subscale of the Motivation
to Control Prejudiced Re Actions Scale(Dunton & Fazio, 1997). Just as the participants in the high
attributional ambiguity condition of the experiment just reportedwere presumed to be relatively
unconcerned about appearing or acting prejudiced because of the presence of a plausible nonracial
excuse for the ir behavior, so were participants lowin motivation to controlprejudiced re actions
assumed to besimilarly unconcerned. However even among that group, the IAT did not predict
either self-report measure of prejudice. It may be that, even under the most favorable of conditions
(e. g., among individuals unmotivated to over ride their implicit attitudes), the IAT will not predict
eitherre actions to Blacksor Whites separately but will only predict the difference between The two
or an explicitchoice between a Black and a White.
Unlike White and Harkins (1994), Petty et al. (1999) left unspecified whether message recipients
would be personally by the policydescribedin the persuasive message, so relevance was
Incidentally, this proposalimplies that primacy effects on recall might beobtained with stigmatized
sources. If, for example, a Black source's message begins with weak arguments, the n attention to,
and consequent recall of, later arguments might be weaker than for early arguments.
Of course, one wonders if successful inhibition of negative implicit racial attitudes will be by
subsequent rebound effects, as have been observed with stereotypes (Macrae, Bodenhausen, Milne,
& Jetten, 1994).