Nonparametric Evaluation of ANOVA Designs by Local, Regional, and Global Contingency Testing
Abstract
Nonparametric analysis of ANOVA designs is proposed to be done by simultaneous contingency testing of discretised response variables. Selected treatment-response patterns are evaluated by predictive configural frequency analysis (local contingency). Treatment response associations are defined by regional contingencies. Total treatment-response relation is defined by global contingency. All methods are illustrated numerically by biometrical examples.