ACCURACY OF PERFORMANCE EVALUATIONS - TES TING WHERRY THEORY

The research describes an experiment conducted with 205 subjects in wh ich three hypotheses derived from the central theorems of Wherry's the ory (Wherry & Bartlett, 1982) were tested. The theory deals with the f actors which influence the accuracy of evaluator judgements in the per formance appraisal context. The factors tested were the influence of p ositive or negative irrelevant information about the evaluatee, the ab sence or presence of evaluator training, time pressure during the eval uation, and the type of rating scales used. Four indicators were used to assess judgement accuracy: leniency/severity, validity, halo, and i nterrater reliability. Except for the reliability data, external exper ts provided the reference scores used to calculate all dependent varia bles. The results of the experiment indicate that evaluators having ne gative irrelevant information are more severe than those receiving pos itive information. However both groups produce evaluations of equivale nt validities. Compared to equivalent control groups, trained evaluato rs produce evaluations which are more reliable and more valid. Time pr essures do not appear to affect rater accuracy. Behavioural rating sca les produce evaluations which are more valid and which are less satura ted by halo errors. These results are discussed in light of Wherry's t heory.