Skewed Within-Class Mixture Distributions in Latent Growth Mixture Modeling: An Assessment of Specification Errors and Class Enumeration

• In Event: Poster Session 2
  In Poster Session: Division D Section 2 Poster Session 1

Abstract

Recent advances have made it possible to model mixture components within latent growth modeling using robust, skewed mixture distributions. This feature allows for flexibility in handling non-normality in longitudinal data, through manifest or latent variables, by modeling skewed or heavy-tailed latent classes rather than assuming normality. The aim of this study was to assess through simulation the potential under- or over-extraction of latent classes in a growth mixture model when the underlying data follow either normal, skewed-normal, or skewed-$t$ distributions. Findings suggest that model comparison and fit measures are accurate in recovering the correct class structure when specifying the new skewed distributional forms. However, this finding is directly related to the amount of class separation between latent classes.

Authors

• Sarah Depaoli, University of California - Merced

• James Patrick Clifton, University of California - Merced

• Keke Lai, Arizona State University - Tempe