ABSTRACT

In many areas of social science, ordinal variables are collected more often than any

other variables. These variables take the form of Likert items, age categories, and

so on. A small amount of thought about the topic reveals that ordinal variables are

neither exactly categorical nor exactly continuous. This issue is often overlooked in

the analysis of ordinal data, however, where researchers treat the variables as contin-

uous. In such situations, the mean and variance of each variable are reported, and the

variables are analyzed through the use of Gaussian GLMs.