Detecting Multivariate Group Mean Differences: Comparing MANOVA and Structural Equation Modeling

• In Event: Sunday Roundtable Session 11:25 am
  In Roundtable Session: Measurement Issues in Structural Equation Modeling (Table 13)

Abstract

In this Monte Carlo simulation study, we compared MANOVA and Structural Equation Modeling (SEM) as omnibus test for multivariate group mean comparisons. The findings revealed that MANOVA demonstrated higher efficacy in controlling type-I error rate. Both MANOVA and SEM achieved the desired power level only when the population group mean difference is large. Additionally, we examined the effectiveness of Bonferroni correction and Benjamini-Hochberg (BH) corrections in the follow-up test within the SEM framework. The follow-up test using SEM path analysis method showed controlled type I error rate inflation with either correction method in large sample conditions. Adequate power was achieved with BH correction when there was a large population mean difference and a large sample size.

Authors

• Qiyue Zhang, Texas A&M University

• Oiman Kwok, Texas A&M University

• Paul R. Hernandez, Texas A&M University