Re: Re: st: Debugging estimation changes

[Ed Gerrish <egerrish@gmail.com>]

Thanks for the suggestion.

I don't think this is the issue, however.  Here is why -- This is a
fixed effects model using -ivxtreg2- .  In order to run xt models, we
have to -sort- and -xtset-.  The result is that the panel structure
remains in place throughout the data manipulation process.

Ed

--
Ed Gerrish
Doctoral Student
School of Public and Environmental Affairs
Indiana University


On Sat, Jun 23, 2012 at 10:23 AM, Clyde B Schechter
<clyde.schechter@einstein.yu.edu> wrote:
> Ed,
>
> If there is no sampling going on in your code (even hidden within ado files that others wrote) another possible source of indeterminacy of results is -sort-.
>
> If you -sort- your data on a varlist that does not uniquely identify the observations, the resulting sort order is indeterminate and Stata does not guarantee to produce the same results each time.  If anything from that point on depends on the sort order of the data, you will likely get different results with re-runs.
>
> If this is the problem, you should change any -sort- statements so that either you include enough variables in the varlist to uniquely determine the sort order, or use the -,stable- option.  With that option, Stata will preserve the original order of the data within the -varlist- groups.
>
> Hope this helps.
>
> Clyde Schechter
> Dept. of Family & Social Medicine
> Albert Einstein College of Medicine
> Bronx, NY, USA
>
>
> *
> *   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/