ABSTRACT
This chapter details logistic regression for a binary response and discusses parameter interpretation. It discusses estimation and inference for this relatively simple model case. The chapter introduces the probit, complementary log-log, and linear probability models for binary responses after a brief introduction to the generalized linear model. Binary responses arise quite frequently in research and clinical studies. Further, as the treatment of such simplest discrete outcomes will provide a basic understanding of the regression approach and elucidate the development of more models for polytomous outcomes. The principal objective of a logistic model is to investigate the relationship between a binary response and a set of independent variables. The parameters in the logistic regression model are interpreted using odds ratios. Prospective study designs are the most popular in many areas of biomedical and psychosocial research, including randomized, controlled clinical trials and cohort studies.
