marketing data. A weighted least squares logit model is compared with a maximum
likelihood logit model different from that mentioned by Green et ai. Empirical applications are
used to compare the models. Suggestions are presented for interpreting and reporting the
results of logit-type models, with special attention to interaction effects.