Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.

# Re: st: e(rmse) for xtreg, re ?

 From Scott Merryman To statalist@hsphsun2.harvard.edu Subject Re: st: e(rmse) for xtreg, re ? Date Fri, 7 Sep 2012 06:35:49 -0500

```On Thu, Sep 6, 2012 at 9:10 AM,  <matif1@yahoo.com> wrote:
> Dear Stata users,
>
>
> I'd like to learn how Stata calculates the root mean square error (e(rmse))
> after fitting a simple linear random-effects model (xtreg, re)...can anyone
> shed some light for me?
>
> Specifically,
> (1) does the RMSE reported by Stata include all random-components, i.e. u_i
> and e_it,  or simply e_it?
> (2) model degree of freedom:  how is the model degree of freedom defined in the calculation? Does Stata take into account the number
> of clusters, in addition to the number of parameters in the slope of the linear model? One previous post has touched this issue but it didn't talk about random-effects model (http://www.stata.com/statalist/archive/2010-03/msg00941.html ) ?
>
>
> I checked the manual (
> http://www.stata.com/features/panel-data/xtreg.pdf)..it only describes how
> rmse is defined for be and fe options, but not re. I also checked xtreg.ado
> defined there.

-xtreg, re- uses the within residuals to calculate the idiosyncratic
error component.

webuse grunfeld,clear
qui xtreg invest kst, fe
predict double e, e
gen double e2 = e^2
sum e2
disp sqrt(r(sum)/(e(N) - e(N_g)  - 2 + 1 ))
qui xtreg invest kst, fe,
disp e(rmse)

Scott
*
*   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/
```