Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.

# Re: st: vce(analytic) for proportions-- what is it exactly?

 From "Tamer Farag" To Subject Re: st: vce(analytic) for proportions-- what is it exactly? Date Thu, 17 Jan 2013 15:09:33 -0500

```Steve, nevertheless, thank you kindly for clearing this up, and for generously adding the references.  Makes total sense now.  This is my first time using statalist (though I have been reading the archives for years) and I am very grateful for your quick and very useful response.

Many thanks,
Tamer

>>> Steve Samuels <sjsamuels@gmail.com> 1/17/2013 2:08 PM >>>
I apologize: you were looking at the _right_ command, but the wrong -help-.

Steve

Tamer,

You are looking at the wrong commands and, therefore, the wrong -help-.

If you have survey data, you must first run -svyset- and then use
Stata's survey commands, such as -svy: proportion-. The default vce
option for -svyset- is vce(linearized), also known as Taylor
linearized.. Formulas are contained in Stata's Survey Manual or in
any sampling text. (My favorite is Sharon Lohr, 2009. Sampling: Design
and Analysis. Boston, MA: Cengage Brooks/Cole.)

What are standard errors based on vce(analytic) in (non-survey)
-proportion-? In the absence of weights or clusters, they are the
formulas found in introductory texts: se = sqrt of
P(1-P)/n. For a complex survey design (multiple stages, strata,
weights), this formula would be quite wrong.

Steve

Steven Samuels
Consulting Statistician
18 Cantine's Island
Saugerties NY 12477 USA

On Jan 16, 2013, at 3:52 PM, Tamer Farag wrote:

Hello All,

I am calculating survey weighted proportions using the svy: proportion command.  The Stata help file states that the default for calculating variance is vce(analytic).  Link below of the stata help file for proportion:

http://www.stata.com/help.cgi?proportion

Specifically, it states:

----+ SE/Cluster +-------------------------------------------------------
vce(vcetype) specifies the type of standard error reported, which
includes types that are derived from asymptotic theory, that allow
for intragroup correlation, and that use bootstrap or jackknife
methods; see [R]  ( http://www.stata.com/help.cgi?vce_option )vce_option ( http://www.stata.com/help.cgi?vce_option ).
vce(analytic), the default, uses the analytically derived variance
estimator associated with the sample proportion.

But just what is an "analytically derived variance estimator", exactly?  Is there any reason not to use it?

Many thanks,
Tamer

Confidentiality Statement:
This email message, including any attachments, is for the sole use of the intended recipient(s) and may contain confidential and privileged information.  Any unauthorized use, disclosure or distribution is prohibited.  If you are not the intended recipient, please contact the sender by reply email and destroy all copies of the original message.

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/

Confidentiality Statement:
This email message, including any attachments, is for the sole use of the intended recipient(s) and may contain confidential and privileged information.  Any unauthorized use, disclosure or distribution is prohibited.  If you are not the intended recipient, please contact the sender by reply email and destroy all copies of the original message.

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/
```