Interpret Size of Effect

Interpret the estimated coefficients

We can use null hypothesis testing and confidence interval construction to interpret the output of the logistic regression model. One important caveat is that a coefficient estimate in the logistic regression represents the marginal effect of an independent variable on the log odds of the dependent variable. It’s hard to understand what log odds mean substantively. There are two ways to interpret the estimated coefficients in a substantively more meaningful way. The first method is to exponentiate a coefficient estimate. This allows us to interpret the new quantity as the marginal effect of $x$ on the odds ratio of the dependent variable. Take model2.logit as an example. We can exponentiate its coefficient estimates as follows: