The Performance of Multivariate Methods for Two-Group Comparisons With Small Samples and Incomplete Data

• In Event: Marriott Roundtable Session 29
  In Roundtable Session: Missing Data Across Statistical Models

Abstract

A simulation study was conducted to compare the performance of mixed model and conventional multivariate methods to assess two-group mean differences with two outcomes under small sample and missingness conditions. When total sample size was 40 or larger, the likelihood ratio test provided accurate Type I error rates and the greatest power for the omnibus multivariate test. Under these same conditions, mixed model conventional t tests provided accurate Type I error rates and the greatest power for tests of specific outcomes. For extremely small samples, the adjusted likelihood ratio test (α=.025) and the mixed model t test using the Kenward-Roger correction had the best overall performance.

Authors

• Keenan A. Pituch, The University of Texas - Austin

• Megha Joshi, The University of Texas at Austin

• Ryoungsun Park, Wayne State University

• Wanchen Chang, NWEA