Cognitive categorization processes in performance evaluation: Confirmatory tests of two models

We tested a data-driven 'traditional' model and a schema-driven 'categorization' model of rater cognitive processes. Raters viewed one of two videotaped lectures on economics. Next, they completed dimensional and overall performance ratings, and scales designed to measure performance-based and non-performance-based aspects of raters' general impressions. We tested model predictions using a combination of confirmatory factor analysis and single-equation regression estimates of structural parameters. Neither model received unequivocal support, though results favored the categorization model. We conclude that (a) performance-based and non-performance-based aspects of a rater's general impression can be distinguished empirically, (b) raters' overall impressions of ratees may substantially reflect the integration of performance-based information, and (c) performance evaluation may be better viewed in the context of `evaluation-based' rather than 'memory-based' judgments.