Parameter recovery in non-normal mixture models

A methodological research line using Monte Carlo simulation to investigate parameter recovery in non-normal mixture models — statistical models used to uncover latent subgroups in skewed psychological data.

The project compares the SKEWT, SKEWNORMAL and TDIST options in Mplus under varying degrees of distributional non-normality, asking which specification most reliably recovers true mixture parameters and class assignments when the normality assumption of standard mixture modelling is violated — a common situation in clinical and psychological data. See Publications for forthcoming papers from this line.