ABSTRACT
This chapter provides a general discussion on response functions for probabilities. It describes the attractive aspects of the logit response function from the viewpoint of interpretation. The chapter illustrates the interpretation for a simple logistic model and a 2 by 2 contingency table. It also provides an introduction to the probit link function. The chapter shows how a probit regression model can be viewed as a normal regression problem with missing data. It presents an overview of other response functions including the popular angular transformation and the use of family of link functions. A general statistical problem is to model a binomial probability in terms of a vector of explanatory variables. To check the relative popularity of the probit and logit response curves, the Current Index to Statistics was to used to count the number of published statistics articles containing probit or logit in the keywords or title during different periods.
