ABSTRACT
Convergent and Discriminant Validation (CDV) set forth a ritual for casting out two demons: method variance and the jangle fallacy. The multitrait-multimethod (MTMM) matrix is an ingenious “tie in” of the convergent and divergent principles. There has been much progress in techniques for its analysis, but little attention to the limitations of the mold into which it forces questions. The sharp separation of traits from methods in CDV echoes the attempt to distinguish theory from observation that was the hallmark of positivism. The MTMM matrix was introduced to simplify thinking and probably succeeded too well in that. It assumes two or more methods that can be applied to a common spectrum of traits, thus limiting the methods that can be investigated. Interpretation often takes protocols having superficial similarities as “the same method.” Conversely, realizations whose differences are best regarded as sampling errors are often spoken of as “different methods.”.
