Robust Means Modeling: An Alternative for Repeated Measures Designs

• In Event: 63.085 - Poster Session 13
  In Poster Session: 63.085-4 - Modeling Nonindependent Data

Abstract

In recent decades, structural equation modeling (SEM) has gained increasing attraction across various fields. For the repeated measures design, two aspects of SEM are directly relevant for repeated measures designs. First, SEM makes no assumptions about variances, thereby circumventing the sphericity assumption altogether. Second, a branch of SEM called structured means modeling (SMM) can be used with robust rescaling corrections to estimate parameters in repeated measures designs precisely while freeing us from the normality assumption.
The current simulation study compares the performance of the adjusted ANOVA-based methods with those derived from the robust SMM framework. This simulation examines a wide variety of real world data scenarios, thereby determining the optimum strategies for repeated measures design estimation.

Authors

• Xiulin Mao, University of Maryland

• Gregory R. Hancock, University of Maryland