Weighting in Multilevel Models

• In Event: AERA Poster Session 4
  In Poster Session: Division D Section 2 Poster Session 1

Abstract

The goal of this study is to examine the performance of multilevel pseudo maximum likelihood (MPML) estimation method under the informative and non-informative condition in the context of a two-stage sampling design with unequal probabilities of selection. Simulation results indicate that including sampling weights in the model still produce biased estimates for level-2 variance. In general, the weighted methods outperform the unweighted method in estimating intercept and individual-level variance while the unweighted method outperforms the weighted for cluster-level variance estimation in the informative condition. Under the non-informative condition, the unweighted method works best in most cases. Besides, the ICC has obvious effects on level-2 variance estimate in both conditions.

Author

• Bing Tong, Michigan State University