Their statistical analyses reveal that this cannot be ascribed to self-selection by the participants, or to a publication bias toward positive results, or to the rigor of the research (methodologically stronger studies yielded larger effect sizes). Roughly half of the studies focused on nonracial and nonethnic groups (as described by sexual orientation or physical or mental disability, for example), and the effect sizes seen within this subset were the same as that for the racial/ethnic targets that stimulated the historical development of intergroup contact theory. Furthermore, it appears that the effects on individual attitudes can generalize to other members of the outgroup and even to other outgroups.
How is this mediated? They find that Allport's four features (common goals, intergroup cooperation, equal status, and official sanction) contribute significantly to the reduction of prejudice but are not essential, and that the last of the four conditions may be the most important one. Greater contact may reduce feelings of uncertainty or discomfort that might otherwise coalesce into anxiety or perceived threat, which might in turn harden into prejudice. Yet these ameliorative shifts may not survive in the absence of normative or authoritarian support, and studies of why contact fails to curb prejudice are needed. -- GJC
J. Pers. Soc. Psychol. 90, 751 (2006).