RE: st: RE: Panel data with small T(24) and N(14) - how to estimate?

["Jacobs, David" <jacobs.184@sociology.osu.edu>]

Check out Beck, Nathaniel, and Jonathan N. Katz. 1995. "What to Do (And Not to Do) With Time-Series Cross-Section Data."  American Political Science Review 89:634-647. These authors have a number of subsequent pieces on this issue in Journals like Political Methodology.  

The reason xtgls is out of favor at least in political science is because the T-values tend to be too optimistic.  I've found results consistent with this claim in my own work.  

I guess you mean longitudinal heteroskedasticity when you mention heteroskedasticity within groups.  In the statistics package called EViews one can correct for this disturbance in panel models.  But I wonder how you have come to a conclusion that you have this problem as I suspect that this kind of heteroskedasticity is quite unusual.

Dave Jacobs

-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Maria Plich
Sent: Friday, May 27, 2011 1:57 PM
To: statalist@hsphsun2.harvard.edu
Subject: Re: st: RE: Panel data with small T(24) and N(14) - how to estimate?

Dear David.

I appreciate your help.

In Stata help I have read that using xtpcse one can assume only
panel-level heteroscedasticity (I understand heteroscedasticity
between groups). In my panel data there is also heteroscedasticty
within groups. How to deal with it?

Could you also explain me why xtgls is less proper than xtpcse?

In advance thank you very much for help.

Maria

2011/5/27 Maria Plich <mariahelenaplich@gmail.com>:
> Dear David.
> I appreciate your help.
> In Stata help I have read that using xtpcse one can assume only
> panel-level heteroscedasticity (I understand heteroscedasticity between
> groups). In my panel data there is also heteroscedasticty within groups. How
> to deal with it?
> Could you also explain me why xtgls is less proper than xtpcse?
> In advance thank you very much for help.
> Maria
> 2011/5/27 Jacobs, David <jacobs.184@sociology.osu.edu>
>>
>> I suspect the Stata procedure named -xtpcse- would be more appropriate.
>>  Some political scientists have claimed that -xtgls- standard errors are too
>> small.  The -xtpcse- procedure does everything that -xtgls- does, so why not
>> use that procedure?
>>
>> Dave Jacobs
>>
>> -----Original Message-----
>> From: owner-statalist@hsphsun2.harvard.edu
>> [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Maria Plich
>> Sent: Friday, May 27, 2011 7:17 AM
>> To: statalist@hsphsun2.harvard.edu
>> Subject: st: Panel data with small T(24) and N(14) - how to estimate?
>>
>> Dear Statausers,
>>
>> I am creating model based on panel data which has got relatively small
>> N (14) and T (24). What is important N represents whole population. I
>> expect problems with heteroscedasticity within and between panels as
>> well as cross-sectional and serial correlation. I wonder if use of
>> generalized least squares (xtgls) to estimate relations between data
>> is a good idea. If not, could you suggest me something better?
>>
>> In advance thank you very much for help.
>>
>> Maria
>>
>> *
>> *   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/
>
>

*
*   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/