EVALUATING PSEUDO-R(2)S FOR BINARY PROBIT MODELS

Many applied researchers of limited dependent variable models found it disadvantageous that a widely accepted Pseudo-R2 does not exist for t his type of estimation. The paper provides guidance for researchers in choosing a Pseudo-R2 in the binary probit case. The starting point is that R2 is best understood in the ordinary least squares (OLS) case w ith continuous data, which is chosen as the reference situation. It is considered which Pseudo-R2 is best able to mimic the OLS-R2. The resu lts are surprisingly clear: a measure suggested by McKelvey-Zavoina pe rforms the best under our criterion. However, in the more likely case of low Pseudo-R2's, a normalization of a measure proposed by Aldrich-N elson which we suggest is almost as good as the McKelvey-Zavoina, and is in general easier to calculate. We also show that if the underlying R2 is predicted using cubic regressions given the Pseudo-R2, all meas ures perform much better,