Logistic Regression: An advancement of predicting consumer purchase propensity

The capability of Logistic Regression (LR) has been understated in most multivariate statistics textbooks; for example, predicting the dichotomous dependent variable from a set of binary explanatory variables where their relationship is not linear but interact with each other. In principal LR is free from the restrictive assumptions required by ordinary least square regression. The aim of this study is to demonstrate how logistic regression can be used to predict consumer behaviour where the explanatory variables are dichotomous and interact with each other. Using the example of a tea take-away purchase with 117 respondents, a step by step procedure was provided. The study illustrates how to predict the tea take -away purchase propensity (yes/no) together with the odds. The results confirm that logistic regression is an effective analytical method in marketing research and it can be extended to other studies with a similar nature.