Does anyone know where there is information on fit tests or something that can be done to help choose the best fitting family and link for a glm model? I have panel data that I am trying to run a regression with using a binary dependent variable (0/1). I am trying to decide between a glm with a poisson distribution and log link or a binomial distribution with logit link. Or is it best to just use a logit model?

If anyone knows where there is more information on how to correctly make these choices, either online, through fit tests, or in a specific text book, I would greatly appreciate the help!



Browse other questions tagged or ask your own question.