Using Simulations to Investigate Sample Size Requirements for Two-Level Multilevel Linear Modeling: A Meta-Analysis

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

Abstract

An important problem in Multilevel linear modeling (MLM) is the calculation of sufficient sample sizes that generate accurate estimates with appropriate statistical power. The current meta-analysis integrates simulation studies of bias and power in parameter estimates to summarize the need of sample sizes for conducting MLM. The findings also verified the more importance of group-level sample size in attaining accurate estimates, in comparison with the individual-level sample size. It can be pointed out that with the same sample sizes and intraclass correlation, the power of individual-level tests is higher than those of group-level and/or cross-level tests.

Authors

• Jing Lv, East China Normal University

• Ke Wang, Nicholls State University

• Jinlai Lu, Taiyuan University of Technology