Paired comparison ratings. I. The effect of ratings of reductions in the number of pairs

Abstract

Two groups of 50 industrial employees were rated independently by their respective foremen using the method of paired comparison; all possible pairs of employees were rated. A performance rating index was obtained for each individual of each group using an index table that is provided with the Personnel Comparison System. The purpose was to determine the extent to which it would be possible, in paired comparison ratings of employees, to use reduced numbers of pairings and still achieve essentially the same rating results as would be obtained from a complete pairing of all individuals within the group. Results showed that ratings obtained from partial pairings result in fairly high correlations with ratings based on complete pairings; the correlations are reduced rather systematically with reductions in the numbers of pairs per individual on which the ratings are based. (PsycInfo Database Record (c) 2025 APA, all rights reserved)