Date:         Fri, 3 Dec 2010 04:47:31 -0800
Reply-To:     dfisher@csulb.edu
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "Dennis G. Fisher, Ph.D." <dfisher@CSULB.EDU>
Organization: California State University, Long Beach
Subject:      Re: reference and software for multivariate regression models
              with              complex survey data
Comments: To: haiyan chen <haiyan_c@YAHOO.COM>
In-Reply-To:  <802836.39344.qm@web110213.mail.gq1.yahoo.com>
Content-Type: text/plain; charset="iso-8859-1"

You might want to investigate some kind of generalized logit model although
your situation is more complex than anything I have tried.  You can do the
generalized logit in PROC LOGISTIC.  It is also going to depend on what
sample size you have in each condition separately and in combinations.  Good
luck.
HTH

Dennis G. Fisher, Ph.D.
Professor and Director
Center for Behavioral Research and Services
1090 Atlantic Avenue
Long Beach, CA 90813
tel: 562-495-2330 x121
Fax: 562-983-1421
http://www.csulb.edu/centers/cbrs/

-----Original Message-----
From: SAS(r) Discussion [mailto:SAS-L@LISTSERV.UGA.EDU] On Behalf Of haiyan
chen
Sent: Friday, December 03, 2010 4:29 AM
To: SAS-L@LISTSERV.UGA.EDU
Subject: Re: reference and software for multivariate regression models with
complex survey data

I am open to any reasonable solution to my data.

My three responses are three medical conditions that are positively
correlated with each other. We are interested in each of their associations
with the same group of independent variables as well as correlation among
themselves.

My initial intention is to investigate both each of their relations with the
same group of covariates and association among them in one model -
Therefore, I think I want some reference and software for multivariate
regression with survey data. And soon enough I found out that I only googled
limited reference and no software.

a solution with 3 models adequate? I long for simpler yet working solutions.


--- On Fri, 12/3/10, Peter Flom <peterflomconsulting@mindspring.com> wrote:


From: Peter Flom <peterflomconsulting@mindspring.com>
Subject: Re: reference and software for multivariate regression models with
complex survey data
To: "haiyan chen" <haiyan_c@YAHOO.COM>, SAS-L@LISTSERV.UGA.EDU
Date: Friday, December 3, 2010, 5:51 AM



haiyan chen <haiyan_c@YAHOO.COM> wrote
>
>I have an upcoming project which has three binary variables as outcomes.
The data are from a complex health survey.
>
>This is the first time that I analyze data as such. I am trying to get some
help here on reference and software for multivariate regression models with
complex survey data.
>

Murphy Choy suggested SURVEYLOGISTIC and Haiyan Chen responded

haiyan chen <haiyan_c@YAHOO.COM> wrote
>
>Hello,
>thanks a lot for your rsponse.
>I guess I need multivariate (not multiple) logistic regession model for my
three correlated binary response variables adjusting for survey design. I
don't know that PROC SURVEYLOGISTIC conduct multivariate logistic
regression?
>
>Please note that in my definition:
>
>multivariate = have more than one response variable; multiple = have
>more than one independent variable.
>

You've said you have 3 binary responses.  But do you need one model (which
would be multivariate) or is a solution with 3 models adequate, or even
preferable?

The fact that your 3 response variables does not answer this question.

If you tell us what these variables are and what the purpose of your
analysis is, that would help.

Peter

Peter L. Flom, PhD
Statistical Consultant
Website: http://www DOT statisticalanalysisconsulting DOT com/ Writing;
http://www.associatedcontent.com/user/582880/peter_flom.html
Twitter:   @peterflom