descriptive discriminant analysis
a statistical procedure that distinguishes between two or more groups of a categorical outcome on the basis of several initial explanatory or predictor variables. The focus is on a set of standardized weights linking each initial variable to the grouping variable; high weights (in absolute value) indicate variables that tend to differ across the groups. For example, a researcher may want to study what differentiates those who survive cancer from those who do not by using a set of background variables, such as age, weight, frequency of saturated fat in the diet, and family history of cancer. The descriptive discriminant analysis may reveal negative standardized weights in the moderate to high range, suggesting that within the sample examined more family history of cancer, older age, higher weight, and more saturated fat in the diet are significantly linked with a lower chance of surviving cancer. Compare predictive discriminant analysis.