[Clyde Schechter]

Let me clarify something from my preceding response. My statement that the Poisson estimator is robust has nothing to do with the largeness of your standard errors. I'm simply making the point that for the purposes of testing an effect, the mean = variance assumption of the Poisson estimator can be violated without causing much difficulty if you don't need predicted values. For a fuller discussion, see https://www.statalist.org/forums/for...postestimation where Joao Santos Silva discusses it and provides some references.

None of that discussion applies to -xtreg-.