Electronic Thesis and Dissertation Repository


Master of Science




Paul F. Tremblay


Mixture models are used for identifying profiles or combinations of profiles and dimensions that explain observed variables. Given that these techniques can be misapplied (Lubke & Miller, 2014), much research is needed to understand their properties when applied to various data sets. The current study tests and compares the fit of mixture models to factor analytic models of personality trait facets based on the HEXACO Personality Inventory-Revised (Ashton & Lee, 2009a). This study also examines the relative amounts of variance in the facet variables that can be explained by underlying dimensions, latent profiles, and other sources. Ashton and Lee (2009b) concluded from a cluster analysis of the HEXACO traits that profiles did not explain much variance in the observed trait measures beyond the variance explained by the factors themselves. The present study builds on that research using a more sophisticated modeling approach, namely factor mixture modeling at the facet level.