Date:         Thu, 17 Jul 2003 12:45:47 -0400
Reply-To:     emaguin@acsu.buffalo.edu
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         emaguin@acsu.buffalo.edu
Subject:      case-control analysis (matched subjects)
In-Reply-To:  <200307171441.h6HEf3o22832@pilot28.cl.msu.edu>
Content-Type: text/plain; charset=us-ascii

All,

I am trying to understand how to do an analysis where pairs or triplets of
persons are matched on relevant characteristics. I'd like to consider two
cases. One case where the pairs or triplets are pairs or triplets of
individuals. The other case is where one element of the pair or triplet is
a group of homogenous persons. With respect to pairs, the idea would be
that homogenous groups of persons are formed and one randomly selected
person in the group is designated as the 'case'--the person to receive the
treatment. The remaining persons form the controls.

Suppose the dependent variable is dichotomous or the time to an event, so
that a logistic regression or cox regression would ordinarily be used.

The main question is whether spss can do these types of analyses and, if
so, how. In particular, how is is the matching accounted for in the
analysis?
The secondary question is relevant citations, websites, or other references.

Thanks, Gene Maguin