<![CDATA[Analysis of Covariance (ANCOVA) is a fundamental statistical method in experimental research, but it often faces violations of parametric assumptions that can compromise the validity of statistical inferences. This study aims to evaluate the relative performance of various alternative nonparametric ANCOVA methods through simulation. The simulation study generated a dataset of 75 observations (25 per group) using normal, gamma, and lognormal distributions to simulate violations of normality and homoskedasticity assumptions. Seven alternative methods were evaluated: Quade ANCOVA, ANCOVA Rank Transformation, Aligned Rank Transform, van Elteren Test, NANCOVA Resampling, ANCOVA Permutation, and ANCOVA Bootstrap. Assumption testing revealed significant violations of residual normality and homoskedasticity, while homogeneity of slopes was met. ANCOVA's Rank Transformation yields an F-statistic of 54.54 (p]]>
Copyrights © 2025
