Hi there,
I'm working with a mixed-effects logit model and then obtaining marginal effects postestimation. Normally, when I work with regular logit models, the margins output says something like "Expression : Pr(y), predict()". However, with the melogit model, it's saying "Expression : Marginal predicted mean, predict()". This makes me concerned that the marginal effects that margins is outputting are not differences in predicted probabilities - could this be the case? Could it be giving me the difference in log-odds or odds instead?
Thank you!
I'm working with a mixed-effects logit model and then obtaining marginal effects postestimation. Normally, when I work with regular logit models, the margins output says something like "Expression : Pr(y), predict()". However, with the melogit model, it's saying "Expression : Marginal predicted mean, predict()". This makes me concerned that the marginal effects that margins is outputting are not differences in predicted probabilities - could this be the case? Could it be giving me the difference in log-odds or odds instead?
Thank you!
Comment