[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: Pseudo R2 for XTprobit
thanks for your reply. Do you have any suggestions for better goodness
of fit measures in probit regressions on panel data? McFadden R2 or
Pseudo R2 were the ones suggested in various sources and empirical
studies but I'm open for whatever turns out to be useful and meaningful.
I'm just puzzled that *xtprobit* does not return any measure at all. Is
there any theoretical motivation why I could report probit results
without any goodness of fit measure?
Thank you very much
Stas Kolenikov wrote:
type -ereturn list- to find out. The likelihood is stored in e(ll),
and then you can do whatever you need to it.
To my opinion, pseudo-R^2 are lousy measures, and more so in the panel
data. Even in the linear model, you would have the within panel and
the between panel and the total sum of squares and hence R^2. That's
something not so well quantifiable even in the linear regression: it
would involve subtracting one sum of squares from another, and you are
not even guaranteed to have your sum of squares positive -- this was
mentioned on the list recently. For probit regressions, you would have
a hard time convincing me any of the pseudo-R^2 gives the same
messages as does R^2 in linear model.
On Thu, 07 Oct 2004 16:27:04 +0200, Uwe Berberich <firstname.lastname@example.org> wrote:
I apologize for bringing up my likely trivial question yet again.
However, I still couldn't find a way to report a goodness of fit measure
for my Probit-regressions. I used random effects *XTPROBIT* on an
unbalanced panel set. I already checked STATAs FAQ and previous list
entries. Unfortunately, most entries either pointed towards the
*"fitstat" *command that doesn't work with XTPROBIT or towards manual
calculations unsing the constant-only model and the full model. I'm used
to this procedure in theory, though I don't know where to find the
corresponding numbers in my STATA output.
* For searches and help try: