Type I Error Rate and Statistical Power of ANCOVA (Analysis of Covariance), Potthoff’s Modified Johnson-Neyman Method, and Wilcox’s Trimmed-Mean Method

• In Event: Poster Session 3
  In Poster Session: Division D Section 2 Quantitative Methods and Statistical Theory

Abstract

The purpose of this study is to investigate the empirical performance of three statistical procedures (the conventional ANCOVA, the modified Johnson-Neyman method, and the Wilcox trimmed-mean method) via Monte Carlo study. The Type I errors are examined in three categories: 1) all assumptions are met, 2) two assumptions are violated individually (individual violation; non-normality, heterogeneous error variance), and 3) two assumptions are violated simultaneously (dual violation; non-normality& heterogeneous error variance). For the each category, the unequal covariate variance and the equal and unequal sample sizes were included. For estimating power, unequal within regression slopes conditions are added. The initial results indicate that the Type I error rate and statistical power of three procedures depended on the simulated conditions.

Authors

• Soyoung Kim, Korea National Sport University

• Stephen Olejnik, University of Georgia

• Ju Sung Jun, Sung-Sil University