Pramod Adhikari <[email protected]> appears to be getting an
incorrect PSU count (6,333 instead of 402) from -svy:mean- even though
-svydes- reports the correct count:
> I just installed Stata version 9 and I am re-running some analysis that I
> previously used Stata 8.
> The survey data (n=6,333) I am using comes from a multistage design, but I
> only have information on strata (8) and PSU (403) and I treat this data as
> a single stage clustered design. There are a few PSU with one observations
> only.
>
> However, subsequent 'svy' estimates show the number of PSUs as 6,333
> instead of 403. It did not happen in Stata 8.
>
> . svyset psu [pw=wt], strata(stateur);
>
> pweight: wt
> VCE: linearized
> Strata 1: stateur
> SU 1: psu
> FPC 1: <zero>
>
> Then I 'svvydes' to see whether the 'svyset' defined the psu and strata
> correctly .
(output omitted)
> However, when estimating means, the number of PSUs is shown as 6,333.
>
> . svy: mean violence
> (running mean on estimation sample)
>
> Survey: Mean estimation
>
> Number of strata = 8 Number of obs = 6333
> Number of PSUs = 6333 Population size = 6.9e+06
> Design df = 6325
>
> --------------------------------------------------------------
> | Linearized
> | Mean Std. Err. [95% Conf. Interval]
> -------------+------------------------------------------------
> violence | .0712806 .0039479 .0635415 .0790198
> --------------------------------------------------------------
I do not recall this ever being reported before, and can't reproduce the
problem. Pramod should make sure Stata 9 is fully up-to-date (including the
-update swap- for the executable). If that doesn't fix it, Pramod can email
me directly for further assistance.
--Jeff
[email protected]
*
* 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/
