ABSTRACT
Recent research has shown that Bayesian estimation can mitigate the estimation difficulties (e.g., failed convergence, inadmissible parameter estimates) commonly encountered when estimating ‘multitrait-multimethod structural equation models’ (MTMM-SEMs; i.e., structural equation models specifically designed for MTMM data). Yet, researchers have not examined the conditions (i.e., sample sizes and effect sizes) that would cause a researcher to correctly select the true data-generating MTMM-SEM among a set of a plausible MTMM-SEMs. The purpose of this chapter is to describe and exemplify a methodology for estimating the probability of selecting the correct data-generating MTMM-SEM among a set of plausible MTMM-SEMs (conceptually similar to power analysis via Monte Carlo simulation). The chapter outlines the methodology via eight sequential steps, which are demonstrated with R and Mplus code. Researchers planning to conduct MTMM studies may use the methodology in this chapter to gain insight into which MTMM-SEMs may be detectable based on their anticipated study conditions (i.e., sample size and effect size).
