Test validity

Construct (python library)

Statistics

Social psychology

Discriminant validity

Latent variable

Computer science

Psychometrics

Structural equation modeling

Validation test

Confirmatory factor analysis

Construct validity

Psychology

Mathematics

Convergent validity

Programming language

Clinical psychology

Internal consistency

Common-method variance

The application of CFA to Multitrait-Multimethod Matrices (MTMM) is an elaborated method for the evaluation of construct validity in terms of the discriminant and convergent validity as well as method effects. It is implemented to evaluate the acceptability of a questionnaire as a measure of constructs or latent variables, during the structural and external stages of an instrument validation process. CFA can be carried out in MTMM matrices, like in any other covariance matrix, to study the latent variables of traits and methods factors. CFA-MTMM models can distinguish systematic trait or method effects from unsystematic measurement error variance, thus offering the possibility of hypotheses testing on the measurement model, while controlling possible effects. A plethora of CFA-MTMM approaches have been proposed, but the Correlated Methods Models and the Correlated Uniqueness Models are two CFA-MTMM methods most widely used. This work discusses how these two approaches can be parametrized, and how inferences about construct validity and method effects can be drawn on the matrix level and on parameters level. Then, the two methods are briefly compared. Additional CFA MTMM parameterizations are also discussed and their advantages and disadvantages are summarized.

Applied Psychometrics: The Application of CFA to Multitrait-Multimethod Matrices (CFA-MTMM)