Re: st: compare different GLS models

[Estrella Gomez <estrellastata@gmail.com>]

Dear Mareike

You can check a simple loss error function for each specification (the
lowest the number, the better). See an example:

reg lxi lGDP_i lGDP_j contig comla smctry ldist i.ichlnum i.jchlnum i.t, robust
predict plxi

* Bias:
gen losspfiw = plxipfiw - lxi

* Absolute error loss:
gen abslosspfiw  = abs(losspfiw)

You can also check the mean square error and the bias (again, the
lowest the better):

* Mean Square Error (MSE)
gen losspfiw2 = losspfiw^2

I hope this helps

Regards
Estrella




2010/6/12 Mareike <MareikeHahr@aol.com>:
> Dear Statalist,
>
> I'm running a generalized least squares model on panel data. I was wondering
> how I compare the goodness of fit of different specifications of such GLS
> models?
> This might be a very basic question, but I would really appreciate if
> anybody could provide me with an answer.
>
> Thanks in advance
>
> Mareike Hahr
>
> University of Freiburg
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
>

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/