Re: st: xtgls, xttest2, xttest3

[Clive Nicholas <clivelists@googlemail.com>]

Amin Mohseni wrote:

> I want to make sure my logic below is correct.
>
> First I am run xtgls with the option panels(correlated) which uses
> heteroskedastic and correlated error structure for computing the
> standard errors.
>
> Then I run xttest2 and xttest3 I reject the null for both tests
> (confirming that I have groupwise hetero and cross-panel correlation)
>
> Now I can trust the results from the xtgls,  panels(correlated)
> because xttest2 and xttest3 indeed point to groupwise hetero and
> cross-panel correlation) which I have included in my xtgls model.

So does running -xtreg, i(panelvar) fe robust-. Have you compared what
both these models tell you? Is N low? If so, -xtgls- estimates and
standard errors tend to be unreliable. Also, what do you find if you
run -xtpcse-? You should show your models using -estout-, then we can
all judge.

-- 
Clive Nicholas

[Please DO NOT mail me personally here, but at
<clivenicholas@hotmail.com>. Please respond to contributions I make in
a list thread here. Thanks!]

"My colleagues in the social sciences talk a great deal about
methodology. I prefer to call it style." -- Freeman J. Dyson.

*
*   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/