Extracting results from nonlinear models
Maarten L. Buis · RePEc: Research Papers in Economics · 2010
In this presentation, I discuss a variety of ways in which you can interpret the results obtained from nonlinear models (e.g., logit, probit, poisson, and streg) and how these different ways of interpreting results are related. I pay particular attention to 1) the difference between average predictions or marginal effects and predictions or marginal effects for an individual with average characteristics; 2) the different ways in which interaction effects can be interpreted; and 3) the difficulties in giving an causal interpretation to effects in nonlinear models.