I have been working with the latent class extension to the nlogit, and would like to know if it is possible to identify case-specific probabilities of belonging to a specific class. Any advice would be appreciated. Thanks, Kristy Wallmo