Statistical approaches to the correction of correlational bias


Examines the effectiveness of a mean rating strategy, in which correlations are derived from ratings pooled across a number of observers, and also presents a prophecy formula for determining the number of raters needed. Analysis indicated that the success of mean ratings in correcting for correlational bias depends on the reliability of ratings, the number of raters, and the differences between true and assumed correlations between traits. A more efficient correlational strategy that not only allows statistical correction for correlational bias but also permits tests of various hypotheses concerning the rating process is also discussed. (18 ref) (PsycINFO Database Record (c) 2006 APA, all rights reserved). © 1980 American Psychological Association.

Publication Title

Psychological Bulletin