Abstract
Seven rheumatologists made judgments about the improvement or deterioration of identical sets of 50 'paper' patients on two occasions one year apart. The stability of their judgments over one year (rs = 0.70) compared favourably with the reliability of duplicate judgments on each occasion (rs = 0.76). Multiple regression analysis of the patient data in relation to the disease assessments provided a model of each clinician's underlying judgment policy. The stability of judgments predicted by these policy models was even higher (rs = 0.83).