The Incremental and Predictive Validity of the Rorschach Test in Personality Assessments of Normal, Neurotic and Psychotic Subjects.

The purpose of this study was to determine the optimal amount of Rorschach protocol data required for maximal accuracy in person­ ality assessment when experience level of judges and degree of pathology of the subjects are taken into account. Confidence in judgment as it relates to the same three factors was also examined. One hundred forty-four judges, consisting of 36 untrained undergraduates, 36 senior undergraduate psychology majors, 36 graduate clinical psychology students, and 36 Ph.D.’s who were Fellows of the Society of Projective Techniques, or had indicated a special interest in projective techniques in the APA Directory, were used. Six Rorschach tests were administered to persons represent­ ing Normal, Neurotic and Psychotic states of mental health. They were selected on the basis of their life situations, treatment condi­ tion, and results on the MMPI. Case data was then divided into the following separate increments: 1 -Basic Identifying Data (BID): age, sex, level of educa­ tion, marital status, past and present treatment status, and occupation. 2 BID plus Free Association and the Location Sheet (FA and Loc). 3 BID, FA, Loc, and Scoring based upon FA (Sc). 4 BID, FA, Loc and Inquiry (Inq). 5 BID. FA, Inq, Loc and Scoring based upon FA and Inq (Scor). 6 BID, FA, Inq, Loc, Scor and Testing of the Limits. Each judge received a packet containing only one of the foregoing information conditions on a single case. After consider­ ing the data, a judge was required to delineate the protocol subject, utilizing the California Q-sort as a descriptive device. The . criterion for each case consisted of a consensual CQ-sort obtained from persons well acquainted with the subjects. Confidence in judgment was expressed quantitatively In the form of "percentage right," and subjectively in the form of selfevaluation of performance on a .5 point scale ranging from 5 for "Very Well" to 1 for "Very Poorly." It was found that; 1 There was an inverse relationship between level of train­ ing and judgmental accuracy, regardless of type of proto­ col or amount of information available. 2 Increased amounts of data resulted in decreased levels of accuracy. 3 BID used by itself was productive of more accuracy than any of the Rorschach data conditions. It exceeded Total Rorschach Data at the .01 level, and the FA/Inq combina­ tion at the .05 level of significance. 4 The best Rorschach data source for use with Normals and Neurotics was the FA/Sc combination. Verbalization was more crucial in the assessment of Psychotics and the FA condition by itself was most productive for that level of pathology. 5 Psychotics could be discriminated from the Normals at the .01 level, and they could be discriminated from the Neurotics at the .05 level of significance. In general, there was a positive relationship between accuracy and the degree of pathology of the subject. 6 Less trained judges were most accurate with Normal re­ cords, and more trained judges were most accurate with Psychotic records. 7 There was positive relationship between training and the tendency to over-estimate pathology of subjects. 8 Confidence did not increase as the amount of data in­ creased, and it did not alter with regard to type of protocol. 9 Most of the judges estimated their performances as being Fair or average, and thus no difference in subjective estimates of confidence was found. There was an inverse relationship between the quantitative estimate of confid­ ence and training, however. This difference appeared to be a reflection of degree of sophistication in statistical matters rather than differential confidence levels. 10 Rorschach accuracy and Social Desirability of the subject were related positively, 11 Genotypic as well as phenotypic characteristics were missed by the more trained judges.