Tutorial on Logistic Regression Model Building

Introduction

Model building is a logical process in which we select a set of explanatory variables that explain a substantial amount of the variability in the outcome variable. While we strive to search for a subset of variables that can explain maximum variability in the outcome, we also aim to develop an economical model with a minimum number of variables. We seek to fulfil these two contrasting objectives in the model building process, which does add a hint of art to the science of model building!

The objective of this document is to give a brief introduction to this logical process, especially for those who are planning to use the macros available on this site or for those new to model building. This is not intended to be a comprehensive guide to logistic regression model building; to read about the model building process in greater detail, please refer to some resources listed here.

We assume you have familiarity with basic statistical and epidemiological terminology. Explanations are given, where deemed necessary.