作者: Mark L. Davison , Anu R. Sharma
DOI: 10.1037/0033-2909.107.3.394
关键词:
摘要: Let Y be a continuous, ordinal measure of latent variable θ. In general, for factorial designs, an analysis variance the observed cannot used to draw inferences about main effects and interactions on θ even when standard normality equality assumptions hold.