Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: st: inconsistent random numbers even using -set seed-
From
Phil Schumm <[email protected]>
To
Statalist Statalist <[email protected]>
Subject
Re: st: inconsistent random numbers even using -set seed-
Date
Thu, 30 Jan 2014 13:24:57 -0600
On Jan 30, 2014, at 12:51 PM, Seed, Paul <[email protected]> wrote:
> Thank you to Daniel Klein and Maarten L. Buis for their very helpful comments on my problem.
> I'd like to think that -sort, stable- solves the problem; but I wonder how stable it really is.
> Will I get the same results if I
> a) exit Stata and then re-start it?
> b) change the sort order of the original data?
> c) change the contents of variables
> d) add or delete variables?
> I am fairly certain that it will not be stable if I add or delete records.
>
> I think for the future (if I ever need it), I will make sure there is a unique identifier
> and sort by that, before setting the seed & invoking random variables.
This has been discussed many times before, e.g.,
http://www.stata.com/statalist/archive/2005-09/msg00582.html
The best way to avoid being bitten by this issue is to use -isid, sort- instead of -sort- any time you need to ensure that the order of the dataset is uniquely determined (i.e., any time you are doing a calculation/manipulation that depends on the order of the dataset). Using -sort, stable- doesn't help here -- it just pushes the problem upstream.
-- Phil
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/faqs/resources/statalist-faq/
* http://www.ats.ucla.edu/stat/stata/