Statalist The Stata Listserver


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: Fractional logit for complex survey designs


From   "Ben M. Gramig" <gramigbe@msu.edu>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: Fractional logit for complex survey designs
Date   Mon, 7 Aug 2006 14:31:39 -0400

I am currently running Stata v8. I probably can get access to Stata v9,
which I understand has many updates and enhanced capabilities relative to
handling survey design specifications.

I am analyzing livestock health and farmer management practice data from the
USDA's National Animal Health Monitoring System (NAHMS).  I am familiar with
the svy commands in Stata which allow me to account for survey design so
that standard errors are calculated correctly for inference purposes.  I am
interested in using Papke and Wooldridge's fractional logit (flogit) model
to predict livestock herd level prevalence (a value in the unit interval
[0,1]) conditional on management practices and other farm characteristics.
This is discussed in the context of Stata here:
http://www.stata.com/support/faqs/stat/logit.html

I have been using glm fam(binomial) link (logit) with robust std errors to
implement the model, but thus far have not accounted for survey design
(stratified sampling w/o replacement, for instance) and therefore reliable
inference is not possible.  In Stata 8 I don't seem to be able to use the
svy commands with glm, which is the only way I know that the fractional
dependent variable can be modeled to predict the outcome of interest (herd
level disease prevalence).  I looked through the online Stata help files for
the updated and expanded svy commands for dealing with these issues and glm
is not listed under the regression models that can be used with the

svy [vcetype] [, svy_options] : command

syntax in Stata v9.

Has anyone else encountered this or might it be possible to program this in
Stata.  This didn't look like it would be terribly complicated, but I am
having trouble determining if glm meets the requirements for svy to support
an estimation command as discussed in Stata help subject "program
properties".

Any guidance or suggestions on this point are greatly appreciated.

Many thanks,
Ben


~*~*~*~*~*~*~*~*~*~*~*
Benjamin M. Gramig

PhD Student 
Dept. of Agricultural Economics
Michigan State University
105 Cook Hall
East Lansing, MI  48824-1024
(517)355-9654
http://www.msu.edu/~gramigbe
 

*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index