A Comparison of the Three Bayes Point Estimates in Cross-Classified Multiple Membership Growth Curve Modeling

• In Event: Advances and Accessibility in Statistics

Abstract

This simulation study evaluated parameter recovery in Bayesian cross-classified multiple membership growth curve modeling under a variety of manipulated conditions including: the number of measurement occasions (3, 5), the number of groups (30, 50, 100), average group sizes (20, 40), and multiple membership rate (20%, 40%). To test the accuracy of parameter estimation, three Bayes point estimates were compared. The posterior mean estimates of between-school variance in the intercept and slope were over-estimated while the posterior mode estimates were under-estimated with 50 or less groups. The posterior median estimates of between-school variance in the intercept and between-subsequent-school variance in the slope were over-estimated with 50 or less groups while between-school variance in the first-school variance were under-estimated with 30 groups.

Authors

• Hyewon Chung, Chungnam National University

• Ryoungsun Park, Wayne State University

• Jiseon Kim

• Hyeonjeong Jeon, Chungnam National University