ABSTRACT
A simple way to understand binomial logistic regression is to regard it as a variant of linear multiple regression ( Chapter 19 ). Binomial logistic regression, however, uses a dependent variable which is nominal and consists of just two nominal categories. By employing a weighted pattern of predictor variables, binary logistic regression assesses a person’s most likely classifi cation on this binary dependent variable. This prediction is expressed as a probability or using some related concept. Other examples of possible binomial dependent variables include:
l Success or failure in your driving test l Suffering a heart attack or not l Going to prison or not. If the dependent variable has three or more nominal categories, then multinomial logistic regression should be used ( Chapter 20 ). In other words, if there are three or more groups or categories, multinomial logistic regression is the appropriate approach. Often, but not necessarily, the independent variables are also binary nominal category variables. So gender and age group could be used as the predictor variables to estimate whether a person will own a car or not, for example.
