Hello Statalisters!
I need to run a regression with multiple dependent variables and possibly
multiple independent variables (although I could settle for a single
independent variable seeing as this is not extremely necessary for my data
analysis). The issue is that all of my variables, dependent and
independent, are binary, and therefore I don't believe that the traditional
MANOVA would be appropriate in this situation. My guess is that I need some
sort of multivariate -logit- model that will provide me with the ability to
enter multiple binary dependent variables. It would be great if I could
also obtain an output of odds ratios from this type of regression, which is
available when using the Stata command for traditional logistic regressions,
-logit Y X1 X2 ... Xn, or-. I have spent quite some time Googling and
reading Stata website, and I have not run across a way to do this in my
Stata 10.1 program for Windows, nor have I found any examples of this type
of regression in papers or other statistical programs. I did review the
listserv archives and it seems as if this question has been posted before
(March 22, 2005), but answers varied from the use of -mvprobit-, to a
log-linear model with the -poisson- command, to a multinomial logit model
using -mlogit-. Although I first started using Stata several years ago in
Econometrics and Public Policy courses, my training with models that have
binary outcomes is not extensive. Any help with my problem or even a more
general explanation regarding the suggestions posted in March 2005 would be
greatly appreciated! Thank you for your consideration.
All my best,
Geneviève
Geneviève Michelle Tindall
Research Assistant
Duke University ? Institute for Genome Sciences & Policy
450 Research Drive ? LSRC, Wing B ? Room 342C
Box 91009 ? Durham, NC 27708
919.681.3499 (tel) ? 919.613.6448 (fax)
[email protected]
www.genome.duke.edu
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
