Nonparametric Evaluation of Learning Curves by Hierarchical Contingeney Analysis

Hierarchical contingency analysis (HCA) is derived from the Perli-Hommel-Lehmacher (1986) closed test procedure for nonparametrical evaluation of learning curves of a 2 x 2-factorial experiment. By HCA, univariate main effects are detected without Bonferroni alpha adjustment, as is shown by a numerical example from gold fish shock avoidance conditioning. Alternative approaches to nonparametrical evaluation of MANOVA designs with and without repeated measurements are discussed.