Research Tips

Logistic Regression and Discriminant Analysis

There are many situations in which marketers want to better understand what distinguishes one group of people or companies from another. Some examples include:

High- versus low-performing sales people.
Loyal customers versus those who “shop around”.
Customers who pay on time versus those who pay late or not at all.

Logistic regression and discriminant analysis are techniques that are useful in situations like these. Each predicts the category a particular observation will fall into based on other variables. Logistic regression can only be used when there are only two possible categories (e.g., prospects who purchased versus those who did not), but discriminant analysis can accommodate a great number of categories. For example, a catalogue or Internet retailer could use discriminant analysis to determine which variables (demographics, purchase history, etc.) distinguish consumers who will buy full-priced products, versus those who will only buy products offered at a discount, versus those who don’t buy at all.