The Robustness of Parameter Estimates in Latent Growth Curve Modeling with Nonnormal Random Intercept and Slope

A skewed observed data often comes from nonnormally distributed residuals, constructs, or both. Under the conditions of imperfect distribution of outcomes, this study found that the semiparametric Bayesian estimation outperformed other traditional parametric estimation methods (Maximum Likelihood and Bayesian estimation), provided more robust parameter estimates.