postestimation, predict, and pweights

[Dominique Chaufree]

Hi,

I use the following regression:

reg y x i.time [pweight=weight], r
predict yhat, xb
bysort time: egen mean_yhat = mean(y_hat)

Are the predicted values based on the weighted regressions? Or do I need to weight the predicted values again (when calculating mean_yhat).

Thank you so much for your help!

[Clyde Schechter]

The predicted values in yhat are based on the weighted regression.

And the value of mean_yhat is an unweighted mean of those values.

[Dominique Chaufree]

Thanks a lot. This is extremely helpful!