Difference score measures of organizational behavior variables: A critique

Abstract

The use of difference scores as measures of organizational behavior variables is discussed. The basic difference score paradigm is presented, and literature is reviewed to illustrate several variations on this paradigm. These include the constructs purportedly measured by difference scores, the source of component scores, and the means by which difference scores are expressed. The basic problems of difference scores include potential unreliability, systematic correlation with their components, and spurious correlation with other variables. Threats to the meaningfulness of difference scores include questionable construct validity, expression as indexes of profile dissimilarity, and inadequate attempts to deal with the direction of differences. It is concluded that the use of difference scores whose components are provided by a single individual should be abandoned, and that the use of between-person measures must be carefully justified on both empirical and theoretical grounds. Reasons for the persistence of the incautious use of difference scores in organizational research are explored.