Such generic/theoretical (i.e. not related to Stata) questions are hardly answered on Statalist.
According to some statisticians, your questions are re-phrased as: given that I have panel data, which is the better way to estimate standard errors?
You can see some answers at:
http://www.kellogg.northwestern.edu/faculty/petersen/htm/papers/se/se_programming.htm
Briefly:
- using cross-sectional commands to manage panel data is not uncommon (at least in some disciplines, that one reviewed by Petersen, and mine also - innovation management)
- independently of which are your beliefs, Stata tells you if individual effects really exist (e.g. -xttest0- or the test at the bottom of the results of a fixed effect model). Obviously, not correcting the standard errors accordingly may give biased results (the magnitude of the bias depends on the type of the applied correction -- not all corrections lead to unbiased results)
- if you have panel data, the use of GLS lets you to take advantage of the maximum amount of information available (coefficients may be biased, too)
Hope this helps,
Nicola
At 02.33 13/02/2007 -0500, "Vladimir V. Dashkeyev" wrote:
>4. Does existence of individual effects mean that once one believes in
>their presence and estimates a panel model, it automatically prohibits
>him/her to estimate a cross section model, since the letter will
>result in biased estimates because of omitted variable (individual
>effect) bias? Or a researcher can suppose that in the long run
>individual effects are insignificant and estimate a cross section
>model, and at the same time suppose that in the short run the effects
>are significant and estimate a panel model?
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
