This chapter provides an introduction to logistic regression, which is a powerful modeling tool paralleling ordinary least squares (OLS) regression. The difference between the two is that logistic regression models categorical rather than numeric outcomes. First, the case of a binary or dichotomous outcome will be considered. Then the cases of unordered and ordered outcomes with more than two categories will be covered. In all three cases, the method of maximum likelihood replaces the method of least squares as the criterion by which the models are fitted to the data. Additional topics include the method of exact logistic regression for the case in which the maximum likelihood method does not converge, probit regression, and the use of logistic regression for analysis of case-control data. © 2008 Elsevier B.V. All rights reserved.
Spitznagel, E. L. (2007). 6 Logistic Regression. Handbook of Statistics. https://doi.org/10.1016/S0169-7161(07)27006-3