Typically, marginal effects from baseline models coincide pretty closely with differences in weighted proportions. I’m predicting an outcome along which the group of interest has 4% representation, and my marginal effects are double the gap as evident by weighted proportions. I’m wondering if marginal effects ‘blow up’ when one of the groups is too underrepresented?
Viewing 1 post (of 1 total)
The forum ‘Default Forum’ is closed to new topics and replies.