Statistical control of halo: Clarification from two cognitive models of the performance appraisal process

Abstract

Criticizes the partial correlation technique proposed by F. J. Landy et al (see record 1981-00274-001 ) used for the control of halo. Categorization and traditional structural models are derived to suggest alternative causal orderings between a rater's dimensional evaluations and overall impression of a ratee. Using logic underlying path-analytic methodology, the inappropriateness of the partial correlation technique of either model is demonstrated. It is suggested that the partialing procedure yields unspecifiable and uninterpretable residuals and that there seem to be no reasonable circumstances in existing causal models of cognitive performance appraisal processes that allow justification of its use. These arguments support R. J. Harvey's (see record 1982-20094-001 ) contention that the partialing approach should be abandoned. (20 ref) (PsycInfo Database Record (c) 2025 APA, all rights reserved)