300 – Models: Assumptions and Novel Applications
A New Transformed Test for Analysis of Variance for Skewed Distributions
Khairul Islam
Eastern Michigan University
Tanweer Shapla
Eastern Michigan University
Analysis of variance (ANOVA) is one of the most popular statistical techniques for comparing different groups or treatments with respect to their means. One of the important assumptions for the validity of ANOVA F test is the assumption of normality of the groups being compared. However, many real-life data do not follow normal distributions. In the violation of normality, the non-parametric Kruskal-Wallis test is often preferable. In this paper, we propose a new transformed test for one way ANOVA for skewed distributions. The performance of the new test is compared with the standard F and the non-parametric analogue of ANOVA by examples and simulations. Our results suggest that the new transformed test is appropriate for estimating the level of significance and is more powerful than standard F test and the non-parametric test for skewed distributions.