The Impact of Sample Size on Three-Level Quadratic Cross-Classified Multiple Membership Growth Curve Modeling Estimation

서지정보

· 발행기관 : 한국교육학회
· 수록지 정보 : 교육학연구 / 60권 / 7호 / 1 ~ 22페이지
· 저자명 : 정혜원, 전현정, 박소영

초록

This study proposes a new model, termed the “three-level quadratic cross-classified multiple membership growth curve model,” which includes a quadratic growth curve to fit a non-linear growth curve to cross-classified multiple membership data in the presence of student mobility. Moreover, to provide guidelines for a minimum sample size requirement for the three-level quadratic cross-classified multiple membership growth curve modeling, this study employed a simulation under a variety of conditions including the number of measurement occasions (4, 5, and 6), the number of groups (30, 50, 100, and 200), average group size (10, 30, and 50), and multiple membership rates (20% and 40%). Across all of the simulated conditions, no substantial bias was found for the fixed effect and level-one variance component estimates. However, in some conditions, substantial biases were found for between-school variance component estimates for intercept, slope, and quadratic slope. In some of the smallest number of groups (n = 30) and group size (n = 10), coverage rates of 95% credible interval for the school variance component were not less than 5% (i.e., the nominal non-coverage rate). It is recommended that researchers apply three-level quadratic cross-classified multiple membership growth curve modeling to data with five repeated measures and at least 200 groups with 30 participants each.

참고자료

· 없음