Ordinal Logistic Regression

This chapter focuses on ordinal logistic regression, used when a dependent measure contains ordered categories. It reviews communication studies that have used ordinal logistic regression. The chapter concerns the most popular ordinal logistic regression, cumulative odds, because it works well with the kinds of questions communication scholars ask, and because SPSS fits this model in its Polytomous Universal Model (PLUM) procedure. It also offers instruction on how to conduct an ordinal logistic regression analysis in SPSS. In doing so, the chapter draws on three categorical explanatory variables and one ordinal response measure from the 2008 American National Election Study. To begin an ordinal regression, a researcher should first select analyze, followed by Regression, followed by Ordinal. While the ordinal regression procedure handles continuous measures effectively, the procedure does require a sufficiently large sample in the presence of metric covariates.