Issue Getting First Stage F-Stat from xtivreg

[justin winikoff]

I am running a model with xtivreg with fixed effects and when I include "first" to get the first stage F statistic, there is just a dot. Clicking on the help button says that "Stata has done that so as to not be misleading, not because there is something necessarily wrong with your model."

I tried running the model with xtivreg2 but am told that no interaction terms are allowed. Is there a way I can get the appropriate estimates with either xtivreg or xtivreg2?


Thanks

[Phil Bromiley]

You can do interactions in xtivreg2, you just have to calculate them instead of using factor notation. You can also simply retain the sample and estimate the first stage separately.