On the Interpretation of Canonical Analysis

Abstract

Canonical correlation analysis has been increasingly applied to marketing problems. This article presents some suggestions for interpreting canonical correlations, particularly for avoiding overstatement of the shared variation between sets of independent variables and for explicating relationships among variables within each set.