Levels of Measurement

Binomial and Multinomial Measures

Binomial measures

In essence, a binomial measure is a nominal variable with two types or categories.

e.g., The customer is male or female

e.g., The purchase was made (a) with a discount, or (b) without a discount

e.g., The customer (1) has purchased our product in the past, or (2) has never purchased our product.

You've probably heard this old saying: "You're either pregnant or you're not. You can't be 'sort-of' pregnant." That cliché is getting at the idea of a binomial variable. A binomial is either yes or no, on or off. There are no in‑betweens.

Multinomial measures

In essence, a multinomial measure is a nominal variable with three or more types or categories.

e.g., Each actual or potential customer is assigned to one of three segments: (a) loyal, (b) non-loyal, or (c) never‑purchased

e.g., The customer's location is (a) the United States, (b) Canada, or (c) elsewhere

Please note:

This distinction between binomial and multinomial matters because many of the techniques that are available to analyze binomial measures are unavailable for analyzing multinomial measures. As a result, multinomial variables more difficult to analyze; they're a more complex animal.