Repeated Rater-Subject Contacts in Measurement of Change: Some Theoretical and Methodological Issues
90 newly admitted, acutely ill psychiatric patients referred for double-blind treatment with phenothiazine drugs were randomly assigned to 3 groups. Each S was evaluated before starting treatment and at the 5th, 15th, and 30th days of medication by 2 clinicians, using the MSRPP and IMPS behavioral rating scales. Group I was evaluated by a constant (or continuing) rater and simultaneously by a second clinician who was replaced by a new rater at the 5th day; Groups II and III by the constant rater and a second rater replaced at the 15th and 30th days, respectively. 3 clinicians were responsible for rating all patients; individual raters were equally represented in both regular ratings and alternate ratings for each group. The design permitted comparisons to be made at each evaluation point between the mean rating of clinicians X, Y, and Z as continuing raters and their mean rating (of the same patients) as fresh raters. The hypothesized generalized effect of repeated rater-ratee contacts was not found. Occasion-judge-contact interaction was, however, the largest single source of variation in scores of two primary factor components (symptom scales) of an illustrative morbidity measure. Experimental data are used to elucidate the thesis that effects of previous rater-subject contacts on scores derived from behavioral scales have not been adequately isolated or related to psychological theory.