A total of 192 kindergarten children were randomly chosen from an urban school district. The pupils were divided equally between 5- and 6-year-olds, males and females, and blacks and whites. The four experimenters--a black female, a black male, a white female and a white male--presented five photographs to the children. The number of pictures correctly identified on a "something like me" basis, were entered in a 2x2x2x2x2 factorial analysis of variance. The only significant difference (p=.003) was the interaction between the child's race and the experimenter's sex. (Author/CS)