Date of Award
1988
Degree Type
Dissertation
Degree Name
Doctor of Philosophy
Abstract
This thesis evaluated the validity of the two most promising methods for Multitrait-Multimethod (MTMM) analysis. The two procedures evaluated were the Confirmatory Factor Analytic (CFA) method and the Composite Direct Product (CDP) method. These methods differ in that the former assumes that trait and method factors combine in an additive manner, while the latter assumes that a multiplicative relationship holds. The first stage in the analysis involved a comparison of the two procedures using ten empirical MTMM matrices. Monte Carlo simulations were utilized in the second stage of the analysis to provide further information that was pertinent to a comparison of the CFA and CDP methods.;It was found that 75% of the time the CDP method provided a superior explanation of the MTMM covariance matrices. In the few cases where the CFA model was superior, it appeared likely that this occurred because of the greater flexibility of the CFA computer program. Recommendations were made regarding modifications to the CDP computer program that would enhance its flexibility. Preliminary evaluations of six fit indices that are often used in structural models were made possible through the Monte Carlo analysis. Serious problems were discovered in five of the six indices. Finally, the implications of this research for the analysis of MTMM matrices in the social sciences were discussed.
Recommended Citation
Goffin, Richard Douglas, "The Analysis Of Multitrait-multimethod Matrices: An Empirical And Monte Carlo Comparison Of Two Procedures" (1988). Digitized Theses. 1697.
https://ir.lib.uwo.ca/digitizedtheses/1697