Section 6 Binary Dependent Variables

In this section, we review models specifically designed for estimating the relationships between our independent variables and a dichotomous dependent variable. This is where “logit” and “probit” regression models become relevant.

Resources

These concepts are discussed in King, Gary. 1998. Unifying political methodology: The likelihood theory of statistical inference. University of Michigan Press. 5.1-5.3 (available through Rutgers online library)

The 2020 Gelman et al. book referenced in the syllabus also has a chapter devoted to logistic regression.

Key limitations of OLS in the binary case are discussed here, and will be referenced at the end of the section.