Arnold H. Levinson <[email protected]> asked about a strange result from
-svymean-. Basically, -svymean- returned zero for the estimated standard
error of a sample mean for one of the subpopulations. Here is the result:
***** BEGIN
. svymean daymore2, by(cnty40) subpop(yellowdaymore2)
Survey mean estimation
pweight: allpstratwt5 Number of obs = 3368
Strata: newstrat Number of strata = 8
PSU: abtid Number of PSUs = 2947
Subpop.: yellowdaymore2==1 Population size = 641670.75
------------------------------------------------------------------------------
Mean Subpop. | Estimate Std. Err. [95% Conf. Interval] Deff
---------------+--------------------------------------------------------------
daymore2 |
cnty40==0 | .6443423 .0114679 .6218564 .6668281 1.76899
cnty40==1 | .5815394 0 .5815394 .5815394 .
------------------------------------------------------------------------------
***** END
In a private email, Arnold sent me a copy of the data.
Using -svydes- I was able to find the reason for the zero. Notice below that
the sample data present in the subpopulation of interest contains only 1 PSU
(a singleton PSU). Stata did not complain here, as it usually does with
singleton PSU's, because the singleton PSU was hidden within a subpop.
The zero resulted from plugging the data into the variance formula. Since
there is only 1 PSU total, the variance is essentially
Var = c*(X - X)
where c is some constant, and X is the only nonzero PSU total (from the subpop
in stata 8).
***** BEGIN
. svydes daymore2 if cnty40==1 & yellowdaymore2
pweight: allpstratwt5
Strata: newstrat
PSU: abtid
#Obs with #Obs with #Obs per included PSU
Strata #PSUs #PSUs complete missing ----------------------------
newstrat included omitted data data min mean max
-------- -------- -------- -------- -------- -------- -------- --------
8 1* 15 2 16 2 2.0 2
-------- -------- -------- -------- -------- -------- -------- --------
1 1 15 2 16 2 2.0 2
------------------
18
13305 = #Obs with missing allpstratwt5,
-------- newstrat, or abtid
13323
***** END
--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/
