THE DISCRIMINANT VALIDITY OF SCALES DEVELOPED BY RETRANSLATION

The retranslation procedure is designed to yield unambiguous sets of behaviors exemplifying conceptually important and distinct performance dimensions. Previous research has shown that dimensions developed with the retranslation procedure possess only low to moderate discriminant validity. This study proposed a multiple-group factor analysis of the scaled behavioral examples for improving the discriminant validity of retranslation dimensions. Multitrait-multimethod analyses of actual ratings indicated that the factor-analytic procedure led to an improvement in discriminant validity. Nonetheless, the amount of discriminant validity possessed by the dimensions was still low. Several suggestions were made for further improving the discriminant validity of dimensions developed with the retranslation procedure.