Tests of Three Causal Models of Halo Rater Error

Abstract

Three different conceptualizations of halo rater error have been identified in the performance appraisal literature. Fisicaro and Lance (1990) referred to these as "General Impression" (GI), "Salient Dimension" (SD), and "Inadequate Discrimination" models. The purpose of this study was to compare the explanatory power of these halo rater error models in terms of their ability to account for observed halo rating error. We found that both the GI and SD models accounted for substantial proportions of rating variance, but the GI model did so much more parsimoniously. We discuss the need to determine the generalizability of the present findings to additional rating contexts and to the individual level of analysis.