| Abstract: | The Rorschach protocols of 43 males consecutively admitted to a university outpatient clinic were scored for frequency of the 20 Wheeler signs of homosexuality. Based on case history data, patients were assigned to homosexual, sex-role disturbed, or normal-control groups. In addition to the traditional group comparison the results were analyzed to yield P (Ci/Rj); that is, the probability of criterion group membership given test indicator. Both the homosexual and sex-role disturbed group displayed significantly more Wheeler signs than normals. Furthermore, given a Wheeler sign score of 15%, .75 of the predicted-homosexual group would be correctly classified compared to a .21 baserate prediction. It was suggested that expressing results as P (Ci/Rj) provides information more relevant to the clinician than is provided by the traditional practice of reporting significant differences between groups. |
