meglm estimated variance for residuals

[Leslie Laam]

Hello!
I ran a basic mixed effects model using meglm and am only seeing one variance estimate in the output, var(_cons). I have read through the manual to see if I can add an option to get the variance for residuals, but no options appear. Each example I see includes both variance estimates. When I use the mixed command, I get both estimates. However, I have a gamma distribution so need to use meglm.
Any suggestion is appreciated.
Thanks,
Leslie

[Clyde Schechter]

When you use -mixed-, you are fitting a linear model. In such a model, the residual variance is a separate parameter of the model. But in a gamma-family model, residual variance is not part of the model because there is no one variance that applies to all of the residuals in this model. The gamma model is inherently heteroscedastic; the residual variance is proportional to the square of the expected value. This is similar to, say, a Poisson regression where the variance is equal to the expected value.