We tested the idea that the correlation between implicit and explicit measures of prejudice depends on whether or not groups are normatively protected against discrimination. A pilot study (N 31) showed that 13 categories varied widely in the degree to which it is acceptable to express negative opinions about them. The main study involving 89 Catholic subjects found that explicit (reward allocation, liking ratings) and implicit measures (linguistic intergroup bias) of prejudice were correlated for the outgroup that is not normatively protected against discrimination (Islamic Fundamentalists) but uncorrelated for the outgroup that is protected (Jews).

This article addresses the role of intentional control in intergroup favouritism and out-group discrimination. We hypothesise that explicit strategies of discrimination (reward allocation, trait attributions) are under people's intentional control, whereas the linguistic intergroup bias (LIB) represents a strategy that allows relatively little conscious control. An experiment is reported in which basketball fans (N = 88) of two different teams were compared, one known for its uninhibited expression of intergroup hostility, the other known for considering aggressive behaviours unacceptable. Aggressive fans showed greater out-group discrimination than unaggressive fans on reward allocation and trait attribution, but comparable degrees of LIB. This suggests that on explicit tasks such as reward allocation, out-group discrimination can be inhibited relatively easily, whereas intentional control appears considerably more difficult for subtle differences in language abstraction. The LIB may therefore constitute a useful tool for assessing prejudice while minimising the impact of social desirability considerations.