Multitrait-multimethod Analysis

Abstract

The multitrait-multimethod (MTMM) analysis procedure is a powerful measurement design in construct validation research. It offers an uncompromising perspective on the meaning of theoretical variables that require indirect measurement. The MTMM is a fully crossed measurement model whereby multiple procedures are used to measure multiple theoretical constructs in a large number of persons. The objectives of MTMM analysis are to determine whether the measurements of each trait derived by multiple methods are concordant, to show that measurements of different traits obtained using the same method are discordant, and to estimate the influence of different methods on the measurement of traits. Traits are universal, and their existence should not depend on the choice of a method to measure them. Therefore, convergent validity requires empirical evidence that different methods can be used to measure a trait. Furthermore, all MTMM analysis procedures provide evidence pertaining to three types of construct validity: convergent validity, discriminant validity, and method effect. However, there are no universally accepted formal tests for construct validity. The analytic definitions of different validity types vary across different MTMM analytic procedures. Therefore, construct validity evidence should be interpreted under the framework of a particular MTMM procedure used in the analysis.