Handling Missing Data in Growth Mixture Models

A Monte Carlo simulation was performed to compare methods for handling missing data in growth mixture models. The considered the current study were (a) a fully Bayesian approach using Gibbs sampler, (b) full information maximum likelihood expectation–maximization algorithm, (c) multiple imputation, (d) two-stage imputation method, and (e) listwise deletion. Of five methods, it found that generally produce less biased parameter estimates compared or single although key differences observed. Similarities disparities among are highlighted general recommendations articulated.