Performance Appraisal Ratings as a Function of Source of Ratings and Purpose of the Appraisal

This study investigated the effects of purpose of appraisal ratings and source of appraisal ratings on four dependent measures: leniency, halo, variability, and construct validity. The purpose factor was comprised of four different levels: merit pay, performance improvement, research only, and no defined appraisal purpose. The rating source factor was comprised of two different levels: incumbent self-ratings, and supervisor ratings. One hundred and nineteen nursing assistants provided the self-ratings, and 39 nurses provided the supervisor ratings. Both sets of ratings were made using a 13-dimension graphic-type rating scale. Analysis of variance procedures were used to test the effects of appraisal purpose and rating source on the dependent measures. Significant Rater Source x Dimension, and Rater Source x Purpose x Dimension effects were obtained for the leniency measure. Findings provided partial support for the hypotheses that the self-ratings would be more lenient than the supervisor ratings and that rater source would interact with appraisal purpose to affect leniency. No support was obtained for the hypothesized main effect of appraisal purpose on leniency. A significant Rater Source effect was obtained for the halo measure. This finding provided complete support for the hypothesis that the self-ratings would display less halo than the supervisor ratings. No support was obtained for the hypothesized interaction of appraisal purpose and rater source on halo. A significant Rater Source x Dimension effect was obtained for the variability measure. This finding provided mixed support for the hypothesis that the self-ratings would be less variable than the supervisor ratings. The self-ratings were less variable than the supervisor ratings for less than half of the dimensions examined. No support was obtained for the hypothesized interaction of appraisal purpose and rater source on variability. Significant convergent validity and discriminant validity were obtained for each of the appraisal purpose conditions. No method bias was obtained for any of the appraisal purpose conditions. However, the hypothesis that appraisal purpose would differentially affect construct validity was not supported. The significance of these findings and recommendations for future research examining the role of appraisal purpose were discussed