ECON20003 Lecture Notes - Lecture 16: Probit, Logit, Logistic Regression
Document Summary
The equation for estimation is: yi equals 0 or 1, yi = pi + i, model is a good approximation for middle values of p but for some it can give values outside the. Allows us to make predictions where the dependent variable is a dummy variable: estimated probabilities must fall in the (0, 1) range, model cannot be estimated by least squares, technique known as maximum likelihood estimation is used. Likelihood function, where p depends on betas: values of betas that maximise l are called maximum likelihood estimates, they maximise the probability of obtaining the observed sample. Interpreting coefficients: unlike the linear probability model, the slope is not constant. I is called the index and is defined as: The marginal effect of a change in an independent variable given by: the average marginal effect for all observations is the mean of the series mfx (eviews) "on average" a __% (increase/decrease) in x (increases/decreases) y by ___