Plot predicted quantiles from xtqreg

[Greg Petroski]

To keep it simple here is a toy example
xtqreg Y Time, id(FID) ls quantile(.5)
I would like to plot the predicted median with confidence interval for each value of Time, but how do I find or calculate the predicted medians?

[Joao Santos Silva]

Dear Greg Petroski,

Check the options predict and save, but note you will get a different prediction per FID and that those predictions are unreliable unless T is large; this is a problem with any fixed effects estimator.

Best wishes,

Joao

[Greg Petroski]

Dear Greg Petroski,

Check the options predict and save, but note you will get a different prediction per FID and that those predictions are unreliable unless T is large; this is a problem with any fixed effects estimator.

Best wishes,

Joao

T is fairly large, 30 time points per FID, but what I was hoping for is a overall estimate of the median at time t. I clearly have a lot to learn, but is this not doable in the context of longitudinal quartile regression? Naively, it seems like such a natural request.

[Joao Santos Silva]

Dear Greg Petroski
With fixed effects (in any model, not just QR) you get a different prediction per individual; that is the meaning of fixed effects.

Best wishes,

Joao

[Greg Petroski]

My fault. "Prediction" was not the right terminology. Can one calculate an "expected" median for each time t?.

[Joao Santos Silva]

Yes, I understood what you meant...