Growth curve models are often used to investigate growth and change phenomena in social, behavioral, and educational sciences and are one of the fundamental tools for dealing with longitudinal data. Many studies have demonstrated that normally distributed data in practice are rather an exception, especially when data are collected longitudinally. Estimating a model without considering the nonnormality of data may lead to inefficient or even incorrect parameter estimates. Therefore, robust methods become very important in growth curve modeling. Among the existing robust methods, the two-stage robust approach (Yuan & Zhang, 2012) from the frequentist perspective and the semiparametric Bayesian approach (Tong, 2014) from the Bayesian perspective are promising. The purpose of this study is to compare the performance of the two approaches through a Monte Carlo simulation study on a linear growth curve model, by varying conditions of sample size, number of measurement occasions, population distribution, existence of outliers, covariance between the latent intercept and slope, and variance of measurement errors. Simulation results show that both approaches provide more accurate and precise parameter estimates than the traditional growth curve modeling when the normal assumption is violated. The semiparametric Bayesian approach performs better when data come from a mixture of normal distributions. If data are normal, the two approaches estimate the model as well as the traditional growth curve modeling. A real-data example based on the analysis of a dataset from the National Longitudinal Survey of Youth 1997 Cohort is also provided to illustrate the application of the two robust approaches.

附：讲座人简介