Probit and logit models typically require a normalization on the error variance for model identification. This paper shows that in the context of sample mean probability decompositions, error variance normalizations preclude estimation of the effects of group differences in the latent variable...

Random effects probit and logit specifications are common when analyzing economic experiments. Stata's fitted values from these estimations, however, appear to fit data poorly compared to their pooled counterparts. This is entirely due to Stata reporting the median predictive value, when...

Regression models of a dependent variable that take on qualitative values 0 and 1 cannot be interpreted as conventional regression models. The fact that the dependent variable takes on two different values causes some problems in the model. These problems include the fact that the model’s...

This paper proposes a dichotomous choice model that is based on a transformed beta (or "z") distribution. This model, called betit, nests both logit and probit and allows for various skewed and peaked disturbance densities. Because the shape of this density affects the estimated relation between...