Four studies examined the effect of positive versus neutral affect on prefe
rence among potential discussion partners who were members of two in-groups
, two out-groups, or both an in-group and an out-group (crossed targets). T
he importance of targets' category memberships was manipulated by idiograph
ically based selection. Positive affect elevated evaluation of crossed targ
ets with a dominant (differentially important) in-group (Study 1). When cat
egories were made equally important, positive affect had no impact (Studies
2 and 3). Study 4 presented crossed targets with both equally and differen
tially important group memberships and showed that differential category im
portance (dominance) is necessary for positive affect to influence judgment
s about them. These results are explained by the broadened categorization i
nduced by positive affect.