The Use of Partial Correlation to Control Halo in Performance Ratings

Landy, Vance, Barnes-Farrell, and Steele (1980) have presented a technique for statistically controlling halo error in ratings by partial correlation procedures. Various writers (Harvey, 1982; Hulin, 1982; Murphy, 1982) have pointed out what they perceive to be deficiencies in Landy et al.'s technique. Landy, Vance, and Barnes-Farrell (1982) have responded to these criticisms, concluding that their technique produces dimensions which, while not appropriate for administrative purposes, do have "several legitimate uses" (for example, test validation). The focus of this paper is to suggest that it is unlikely that dimensions derived from partial correlation techniques will be useful for even the specialized purposes which Landy et al. describe.