The Homogeneity of Covariances Assumption in MANOVA: Differential Impact of Heterogenous Variances and Covariances

• In Event: Saturday Roundtable Session 3:05 pm
  In Roundtable Session: Regression Modeling for Educational Research (Table 21)

Abstract

Many applied researchers have learned that MANOVA is not robust to violations of heterogeneous covariance matrices, and that, when Box’s M is statistically significant, indicating the probable violation of homogeneity, they should use Pillai’s Trace (V) as their multivariate statistic (e.g., Meyers, Gamst, & Guarino, 2017). There have been few studies since Olson (1974, 1976) and the issue remains confusing. While this Monte Carlo experiment investigates the robustness of multivariate statistics, its focus in on the diagnosis of the assumption of homogeneity of covariance matrices using Box’s M and the potential to use Levene’s test to examine homogeneity of variances.

Authors

• Yuqing Liu, Ohio University

• Frank A Oppong, Ohio University

• Nina Adjanin, Northwest Missouri State University

• Gordon P. Brooks, Ohio University