I can't figure out how to obtain an intraclass correlation coefficient to
quantify "household clustering" of dichotomous variables in a stratified
cluster survey. It was a household interview survey where all adults in each
household were interviewed, and where households (PSUs) were systematically
selected PPS from within 10 strata were included. We need to incorporate
sample weights (that are multiplied by a post-stratification factor as well).
I'm only working with prevalence estimates, not multivariable analysis at this
point. How do we do it with STATA? Svy commands don't seem to help with
intraclass correlation coefficients or comparable measures.
I was going to neglect strata and use loneway to obtain point estimates of the
intraclass correlation coefficient and then bootstrap this command to obtain
confidence intervals (since our variables are mostly dichotomous and the
literature suggests that the confidence intervals will not be reliable). BUT
loneway doesn't allow for pweights, only aweights.
Any suggestions for how to make loneway work with pweights? For how to obtain
intraclass correlation coefficients with weighted survey data using another
command? For using another measure of household effect with another set of
commands?
Thanks for your consideration, Mark
--
Mark A. Veazie, Dr.P.H., M.P.H.
Epidemiologist
Native American Cardiology Program
Indian Health Service
U.S. Department of Health and Human Services
1215 N. Beaver Street, Suite 201
Flagstaff, Arizona 86001
928.214.3920
Fax 928.214.3924
email [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/
