544 – Mixed Effects Models for Longitudinal Data
A Comparison of Bootstrap Methods for Mixed Model Analysis of Longitudinal Data
Xiao Wang
Statistics and Data Corporation
Mark Reiser
Arizona State University, Tempe, Arizona
Jeanne Wilcox
Arizona State University, Tempe, Arizona
Shelley Gray
Arizona State University, Tempe, Arizona
A longitudinal study may have clustering at more than one level. For example, in a longitudinal study of school children, observations on the same student over time produce clustering at the child level, and observations on children from the same classroom (teacher) produce clustering at the classroom level. Clustering at more than one level leads to complications in calculating bootstrap estimates of standard errors of parameter estimates. In this study, the parametric bootstrap is compared to case-resampling bootstrap for standard errors of parameter estimates obtained under a mixed model with clustering at two levels. For case resampling, a case is defined in two ways: (1) all observations from the same student, and (2) all observations from the same classroom. Using Monte Carlo simulations, the comparison will also examine the effect of number of clusters and cluster size on validity of the bootstrap standard errors. Results are given for the bootstrap-t, percentile, and percentile-bca methods. An application to a longitudinal educational study is also presented.