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.
Regards,
Jason
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Jeff
Pitblado, StataCorp LP
Sent: Monday, 30 October 2006 9:52 AM
To: [email protected]
Subject: Re: st: Different confidence intervals from proportions and
tabulates (also in survey)
Jason Ferris <[email protected]> is concerned that the confidence
intervals from -svy: proportion- are different from -svy: tabulate- even
though the point estimates and standard errors are the same:
> Hi. I have been running survey proportions and observing results with
> negative confidence intervals (which doesn't make sense). When I use
> survey tab (with column percent, se and ci) I get the same point
> estimates and standard errors but different 95% confidence intervals.
I
> assume this is an issue with the proportion calculations using
"Binomial
> Wald" for confidence intervals.
>
> I checked the survey manual and have not been able to find why:
>
> Paste the following command to see my dilemma:
> webuse nhanes2b, clear
> svy: proportion race
> svy: tab race, ci se
>
> The results show the same point estimates and standard errors (with
> rounding) but different CI's. As mentioned, for my data, I get some
> negative CI's for svy: proportions commands but not for the svy:
> tabulate commands. My ultimate concern is being able to automatically
> extract the CORRECT estimates to excel (from using matrix e(b) and
e(V)
> - and calculating 95% CI from square-root of e(V) *1.96).
>
> I am using the latest version of Stata 9.2, on Windows XP.
While -svy: proportion- uses the normal approximation formulas for
confidence
limits, e.g.
phat +/- sehat*t_critical_value
the -svy: tabulate- command uses a logit transform. The benefit of
using the
logit transform is that the confidence limits are guaranteed to stay
within
the parameter space for a proportion, i.e. [0,1].
The formulas for the logit transformed confidence limits used by
-svy: tabulate- are documented in the 'Methods and formulas' section of
'[SVY] svy: tabulate twoway'.
--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/
*
* 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/
