Self-knowledge and social inference: II. The diagnosticity of cognitive/affective and behavioral data.

This study explores the social perceptions of observers exposed to tape-recorded interviews in which speakers described themselves, either emphasizing past thoughts and feelings, past behaviors, or whatever mix of these speakers perceived as appropriate. Observers' subsequent impressions of speakers were measured using Qsort ratings and various affective and behavioral predictions, which both speakers and speakers' close friends had previously completed. It was found that the cognitive/ affective interviews produced more accurate social impressions, or at least impressions that were more in accord with speakers' self-assessments prior to the interviews and with the assessments made by their close friends, than did the behavioral or the mixed interviews. This greater congruence was shown to result both from real and from stereotype'* accuracy. The relevance of these findings to theory and research on self-perception and social perception is discussed. One's impressions about one's self and about other social actors may be influenced by two fundamentally different and at times conflicting sources of information. They may be based primarily on words, deeds, or other objective manifestations of the self, or they may be based primarily on covert thoughts and feelings, that is, subjective experiences that are inherently private and that are available to others only indirectly and insofar as an actor is willing and able to communicate them. The bases of self-knowledge that have been most emphasized in theory and research are those that are public and external to the