Application of the Bootstrap Methods in Multilevel Models for Analysis of Longitudinal Data

• In Event: Roundtable Session 97
  In Roundtable Session: Multilevel Modeling Methods and Applications III

Abstract

Multilevel modeling has been a prevailing approach in analyzing longitudinal data, in which within-cluster dependence or correlation exits. Even though multilevel models are useful in dealing with longitudinal data, the accuracy of the results relies on sufficiently large sample size and underlying distribution assumptions. The purpose of this study is to investigate the use of different bootstrap methods to longitudinal data that is analyzed through multilevel models. This study involves a simulation study with varying conditions including model specification, sample size, and distributional assumptions of random effects. Empirical results using project STAR (Student-Teacher Achievement Ratio) data will be also presented and discussed.

Authors

• In-Hee Choi, University of California - Berkeley

• Hyo Jeong Shin, University of California - Berkeley