[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: xtreg vs reg w/ dummies
One of the fixed effects was being dropped, I just had not noticed. I
appreciate the 2nd part as well. I figured there must be something
like that going on behind the scenes.
On 3/13/06, Austin Nichols <email@example.com> wrote:
> Did you check that the set of new_attid effects was complete? I.e.
> was one of the fixed effects dropped instead of pitt being dropped?
> There should be 1092 coefficients listed: _Inew_attid_2-1093. Is there
> variation in pitt across time within each panel?
> It is possible that collinearity can produce a dropped variable in one
> command and not another, since Stata checks for collinearity by
> checking for full rank conditions within certain tolerances, and the
> matrices checked can be different in different commands, but I would
> expect the standard errors to look very suspicious in Model 2 in that
> case, and the SE on pitt (.1575) does not look near infinity at all in
> Model 2.
* For searches and help try: