Cason and Cason's model of performance rating was used to determine the extent to which variation in reviewer standards affected the reliability and validity of the program review process used to select papers for inclusion in the annual program. Data analyzed were the overall recommendation for acceptance and ratings on seven quality criteria from each reviewer on each paper proposal in 1983, 1985, and 1986. The Casons' model fit each year's data. Significant rater stringency variance was found for each of the three years. Rater stringency persisted up to three years providing strong construct validation for the model. Removing the rater stringency effect improved reliabilities from .768, .722, and .739 to .813, .790 and .790. Construct validities also improved. Had adjusted ratings been used in 1986, up to 6 of the 35 papers accepted would have been rejected. There were no significant differences in mean rater standards year to year; however, mean paper proposal quality was sharply lower in 1985. In all years, mean paper quality of accepted proposals was significantly better than that of rejected proposals. Access to adjusted ratings at the time of the selection decision would ease the committee's task and probably improve the quality of its decisions. (Author/PN)