st: RE probit, econometrics-related

[David Hamilton <dthamilton@mac.com>]

> I have a sample of firms which I assemble into portfolios at  
> quarterly time intervals. I measure if some event has occurred  
> within a year of those cohort dates, so the intervals overlap. I  
> could keep the first quarter of every year for each firm to  
> eliminate the overlap, but that's throwing away a lot of data. I am  
> planning on running a RE probit model clustering on firm ID (the are  
> about 5,000 unique firms). Here is my question: can I use my  
> quarterly overlapping data, or must I use the non-overlapping  
> sample? Since I am primarily interested in whether one or two of my  
> regressors are significant, my guess is that it is ok -- the RE term  
> should absorb the correlation induced by the overlapping intervals.  
> I would greatly appreciate comments from Statalisters. I've tried  
> researching the answer with no luck. Thanks in advance.
>
> DH

*
*   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/