Arranged an exchange of evaluations after 121 male undergraduates judged the relative success of 2 students. 1/2 the ss were led to believe that they would get an evaluation from a group member who had the same information about the students, and 1/2 that the group member had different information. Ss were also led to believe that the group member was either similar or dissimilar in terms of his interpersonal orientation and style of judging other people. Then ss received agreeing or disagreeing evaluations. A similarity * information within agree interaction supports the attribution theory predictions that (a) when the other had the same information, his agreement would increase confidence more if he were dissimilar than if he were similar, but (b) when the other had different information, his agreement would increase confidence more if he were similar. A statistical trend suggests that similar disagreers reduced confidence more than dissimilar disagreers. (PsycINFO Database Record (c) 2016 APA, all rights reserved)