A more familiar and efficient method for estimating the parameters of Cason and Cason's model was examined. Using a two-step analysis based on linear regression, rather than the direct search interative procedure, gave about equally good results while providing a 33 to 1 computer processing time advantage, across 14 cohorts of junior medical students rated by residents and attending physicians in a medicine clerkship. First, regression analysis of z-transformed ratings produced regression weights (b's) associated with each rater and subject. Then, the unit vector and person b's were converted into inter-person distances, then person locations on the underlying stringency and ability scale. Essentially equally good fit with the data was achieved by the new method in 12 of the 14 data sets. In the other two, fit was still quite good. Correlation between parameter values estimated by the two methods was very high in groups where equal fit was achieved. In the other two, moderately high correlations were observed. The new method provided equally good improvement in reliability and convergent validity of adjusted scores. The improved economy and ease of application of the new method further expanded the advantages of using the Casons' model to statistically control rater bias. (Author/GDC)