Multivariate Analysis: Correspondence Analysis

Overview of the Correspondence Analysis

The Correspondence analysis performs simple correspondence analysis, which you can use to analyze frequency data and associations between two or more nominal variables. The correspondence analysis finds a low-dimensional representation of the rows and columns of a contingency table consisting of the counts for the variables.

While principal component analysis constructs directions in the space of variables that explain variance, correspondence analysis constructs directions (sometimes called principal coordinates) that explain inertia. Inertia is the total chi-square statistic divided by the total number of observations. Correspondence analysis computes directions that best explain deviations from expected values (assuming no association). The analysis graphically represents each row and column by a point in a configuration plot.

You can run the Correspondence analysis by selecting AnalysisMultivariate AnalysisCorrespondence Analysis from the main menu. The analysis is implemented by calling the CORRESP procedure in SAS/STAT software. See the documentation for the CORRESP procedure in the SAS/STAT User's Guide for additional details. For a general introduction to correspondence analysis, see Friendly (2000).