GLM model with Gamma distribution and log link - residuals

[Rachel Brennan]

Dear all,
I am new to running the GLM model in Stata. I have healthcare data which is positively skewed (as it often is). I have tried running a glm model with gamma distribution and log link as I have seen recommended by others (particularly work by Manning and Mullahy). I have a basic question - would one expect the residuals to be normally distributed with a glm model? Or is this not expected for this model given the distribution is not expected to be normal?
Kind regards,
Rachel

[Maarten Buis]

If you look at help glm postestimation##predict, you see that there is a whole range of residuals possible after glm. Some of these were designed to be (approximately) normally distributed in a good fitting model, others are not.

[Rachel Brennan]

Thank you Maarten that is very helpful I will look further into the help for predict.