Hi Nick,
I have pasted the output of a svy: tab and a svy: prop from my
dataset. The instances when I get negative CI's (see proportions table)
is when I have included two over options (sex age10). As mentioned
before the CI's for tabulate and proportions differ (but I understand
this now to be a function of tabulate using the logit/inverse logit
function) to calculate the boundaries.
As you can see from the proportions table the upper CI does not fall
below the lower CI but the lower CI is negative which doesn't make sense
to me.
. svy, subpop(if age10==0): proportion sexmajor , over(sex age10)
(running proportion on estimation sample)
Survey: Proportion estimation
Number of strata = 2 Number of obs = 7270
Number of PSUs = 7270 Population size = 7122.17
Subpop. no. obs = 258
Subpop. size = 404.543
Design df = 7268
_prop_1: sexmajor = 0. no
_prop_2: sexmajor = 1. yes
Over: sex age10
_subpop_1: 0. female 0. 1619
_subpop_2: 1. male 0. 1619

 Linearized Binomial Wald
Over  Proportion Std. Err. [95% Conf. Interval]
+
_prop_1 
_subpop_1  .9809524 .0116001 .9582128 1.003692
_subpop_2  .9782609 .0134002 .9519926 1.004529
+
_prop_2 
_subpop_1  .0190476 .0116001 .0036919 .0417872
_subpop_2  .0217391 .0134002 .0045291 .0480074

. svy, subpop(if age10==0): tabulate sexmajor sex, col percent ci se
(running tabulate on estimation sample)
Number of strata = 2 Number of obs =
7270
Number of PSUs = 7270 Population size =
7122.1699
Subpop. no. of obs =
258
Subpop. size =
404.5432
Design df =
7268

 sex
sexmajor  0, femal 1, male Total
+
0, no  98.1 97.83 97.94
 (1.16) (1.34) (.9081)
 [93.85,99.43] [92.9,99.36] [95.16,99.14]

1, yes  1.905 2.174 2.056
 (1.16) (1.34) (.9081)
 [.5717,6.154] [.642,7.1] [.8595,4.835]

Total  100 100 100



Key: column percentages
(linearized standard errors of column percentages)
[95% confidence intervals for column percentages]
Pearson:
Uncorrected chi2(1) = 0.6443
Designbased F(1, 7268) = 0.0233 P = 0.8787
Thanks for your ongoing interest :)
Jase
Original Message
From: ownerstatalist@hsphsun2.harvard.edu
[mailto:ownerstatalist@hsphsun2.harvard.edu] On Behalf Of Nick Cox
Sent: Monday, 30 October 2006 11:01 AM
To: statalist@hsphsun2.harvard.edu
Subject: RE: st: Different confidence intervals from proportions and
tabulates (also in survey)
As before, your terminology of negative CIs is puzzling.
An interval is a length; if the upper limit falls below
the lower limit, then there is really is a bug. However,
as before, I assume you are talking about negative lower limits
(bounds).
There is a no attachments rule on Statalist. (The FAQ explains.)
The export to MS Excel is naturally independent of these issues
and your own choice. If you have Excelspecific problems that
are also relevant to Statalist, others will advise.
Nick
n.j.cox@durham.ac.uk
Jason Ferris
> Dear Jeff and Nick,
> Thanks for your replies. I did not attach my dataset;
> although I could
> attach the output of the svy: prop and svy: tab if you want. I
> thought the easy access example would provide enough for you
> to tell my
> why there is a difference. And you have (thanks again).
>
> My concern was exporting the matrix to excel and the
> resulting 1.96*e(V)
> was giving me the CI's equivalent to the svy: proportion (which in my
> case was giving me negative CI's). After reading both your
> responses I
> will transform the data to the logit and keep everything
> nicely position
> between 0 and 1. Hopefully, in this manner the mat2txt command or
> parmest will work nicely for me  and save me a lot of manual entry.
>
*
* 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/
*
* 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/