ABSTRACT
Logistic regression is similar to the previously discussed regression analyses such that it is used to predict an outcome. The one major difference between the typical ordinary least squares (OLS) regression model and logistic regression is that the dependent variable in the latter is binary or multinomial (i.e., categorical). Logistic regression can provide the probability or odds of an event occurring or not occurring. Individual predictors can be evaluated for their contribution and importance in hypothesized models. Essentially, application of logistic regression is similar to OLS regression, but with a categorical outcome. Various outcomes can be modeled using individual predictors, or a set of predictors. Hierarchical models that are researcher driven can be designed, as can exploratory statistical models The types of research questions answered by using logistic regression are those that have a dichotomous outcome (i.e., yes/no), or are multinomial (three or more categories).
