st: Re: error message in survwgt rake

[Steven Samuels <sjhsamuels@earthlink.net>]

Shawn, Please send future communications to Statalist.  Questions  
like yours may be of interest to others--and others may provide  
better answers than I can.

In the file "survwgt.ado" there is a "MAXRep(int XXXXX)" setting,  
where "XXXX" is an integer.  You can try increasing the default value.

However in my experience, raking fails to converge when when there  
are too many adjustment cells, when the sample or population numbers  
in some cells are small, or when different margins are inconsistent.   
As an example of inconsistency: an age-gender margin and an age-race  
margin yield different counts for the age-only dimension.  See the  
reference: http://www.abtassociates.com/presentations/ 
raking_survey_data_2_JOS.pdf.

-Steve



On Jul 26, 2008, at 12:13 AM, Shawn DuBravac wrote:


> Steve,
>
>  I read your solution here:  http://www.stata.com/statalist/archive/ 
> 2007-08/msg00387.html
>
> But I’m getting the error, “Warning: variable old_wt reached  
> maximum iterations before convergence.”
>
>
> Any thoughts?
>
> Thanks,
>
> Shawn

*
*   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/