MTMMdesign

MTMMdesign refers to the multi-trait multi-method design, a framework used in psychometrics to evaluate the validity of measurement instruments by examining how multiple constructs are assessed with multiple measurement approaches. The concept, introduced by Campbell and Fiske in 1959, aims to distinguish trait variance from method variance and to provide evidence of construct validity for a set of measures.

In a fully crossed MTMM design, researchers select several traits (constructs) and several methods (measurement approaches), such as self-report questionnaires, observer ratings, and behavioral tasks, and collect data for all trait-method combinations. The resulting MTMM matrix displays correlations between measures of each trait as assessed by each method. Diagonal cells reflect reliability or the consistency of a given trait-method score. Monotrait-heteromethod correlations (same trait, different methods) are used to assess convergent validity, and should be relatively high. Heterotrait-monomethod correlations (different traits, same method) and heterotrait-heteromethod correlations (different traits, different methods) inform discriminant validity and should be lower, indicating that measurements capture distinct constructs and limit method-driven inflation of associations.

Analyses can be descriptive, via inspection of the MTMM matrix, or formal, using MTMM confirmatory factor analysis. MTMM CFA models specify latent trait factors and method factors to partition variance into trait-related and method-related components, allowing explicit testing of convergent and discriminant validity and the disentangling of method effects.

Limitations include design complexity, substantial data requirements, and interpretation challenges when methods or traits are not well matched. MTMM designs remain widely used in psychology and organizational research to validate scales and measurement instruments.