In the xtreg manual entry ( http://www.stata.com/manuals14/xtxtreg.pdf )
it states the following regarding the FE model:
I don't quite understand how you can estimate a FE model to fit a random-effects model. from my understanding if I had some factor variable (say idcode) which I want to estimate as a random effect in my model, I use the mixed command. can anyone clarify this?
it states the following regarding the FE model:
xtreg, fe can estimate σν and σ, although how you interpret these estimates depends on whether you are using xtreg to fit a fixed-effects model or random-effects model. To clarify this fine point, in the fixed-effects model, νi are formally fixed—they have no distribution. If you subscribe to this view, think of the reported σbν as merely an arithmetic way to describe the range of the estimated but fixed νi . If, however, you are using the fixed-effects estimator of the random-effects model, 0.355622 is an estimate of σν or would be if there were no omitted variables.