Confidence Regions and Approximate<i>p</i>-values for Classical and Non Symmetric Correspondence Analysis

Abstract

Recently, a procedure was developed for constructing 100(1–α)% confidence ellipses for points in a low-dimensional plot obtained from performing classical correspondence analysis. This article reviews the construction of confidence regions for classical and non symmetric correspondence analysis and proposes a simple procedure for determining p-values of each of the points in this space. Such features enable the researcher to determine the statistical significance of a category to the association structure between the categorical variables being analyzed. They also reflect the information contained in dimensions higher than those that typically allow for a visual inspection of the association structure.