SMABS 2004 Jena University
SMABS 2004 Home Organization About Jena Sponsors Links Imprint / Contact SMABS Home

European Association of Methodology

Department of methodology and evaluation research

Jena University

Contributions: Abstract

The appropriateness of the WLSMV estimator for complex MTMM-models: results of a simulation study

Fridtjof Nussbeck Michael Eid Tanja Lischetzke
Université de Genève

Since Campbell and Fiske's (1959) seminal work on convergent and discriminant validity, the multitrait-multimethod (MTMM) analysis has become the method of choice for validating personality questionnaires. For the last forty years of research, several methods for analyzing MTMM-data have been developed.

However, most of these models have been developed for the analysis of metric response variables. Nussbeck, Eid and Lischetzke (submitted) developed an MTMM-model for response variables with ordered outcomes based on the Graded Response Model (Samejima, 1969). The parameters of this model can be estimated and some implications can be tested by structural equation modeling for ordinal variables.

Nevertheless, it remains an open question under which conditions (number of latent variables, number of manifest variables, and number of observations) the newly developed wlsmv-estimator implemented in the computer program MPLUS (Muthén & Muthén, 2002) leads to proper solutions.

To investigate this issue, an extensive Monte-Carlo simulation study was conducted. Findings of this simulation study are reported consisting of models with six, nine, and twelve latent factors, between 12 and 72 response variables, and sample sizes ranging from 250 to 1000 observations.