Date:         Tue, 29 Jan 2008 09:16:41 -0600
Reply-To:     "Reutter, Alex" <areutter@spss.com>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         "Reutter, Alex" <areutter@spss.com>
Subject:      Re: Conditional logistic problems
In-Reply-To:  A<002701c861ea$4adf5150$2845cd80@ssw.buffalo.edu>
Content-Type: text/plain; charset="us-ascii"

Gene,

This doesn't directly help your problems with COXREG, but if the dataset is matched 1:1 and not 1:many, you can alternatively use Multinomial Logistic Regression (NOMREG).  There's an example in Marija Norusis's Advanced Statistical Procedures Companion and another in the online help (Help>Case Studies, then Regression Models>Multinomial Logistic).

Cheers,
Alex


-----Original Message-----
From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU] On Behalf Of Gene Maguin
Sent: Monday, January 28, 2008 2:14 PM
To: SPSSX-L@LISTSERV.UGA.EDU
Subject: Conditional logistic problems

All,

I have been working with another list subscriber who has a matched (1:1)
dataset and is trying to analyze relationships between variables using
conditional logistic regression (CLR). I have no experience with this type
of model. However, I did find a posting by Marta (shown below) as well as
several discussion sites. Since spss logistic won't do a conditinal
analysis, the trick is to use the coxreg procedure to do so. I believe I
have faithfully followed Marta's directions but am getting no results and
so, there is something I don't understand. Per Marta's posting, I have

COXREG  ftime  /STATUS=outcome(1)  /STRATA=pair
  /METHOD=ENTER x /PRINT=CI(95).

Where
Ftime is the survival time variable computed such that ftime=outcome+1.
outcome is the case-control variable with 0=control, 1=case.
Pair is the strata variable, 161 pairs.
X is the IV.

My case processing box shows

Case Processing Summary
                                           N    Percent
Cases available in analysis     Event(a)        161     50.0%
                              Censored  0       .0%
                              Total     161     50.0%
Cases dropped     Cases with missing values     0       .0%
                  Cases with negative time      0       .0%
                  Censored cases before the
                earliest event in a stratum     161     50.0%
                          Total         322     100.0%
a       Dependent Variable: ftime

Variables in the Equation(b)
        Wald    df      Sig.
X       .       0(a)    .
aDegree of freedom reduced because of constant or linearly dependent
covariates
bConstant or Linearly Dependent Covariates
S = Stratum effect. x = .5093 + S;


If somebody can educate me about what I'm missing or need to look at, I'd
greatly appreciate it.

Thanks, Gene Maguin

=====================
To manage your subscription to SPSSX-L, send a message to
LISTSERV@LISTSERV.UGA.EDU (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
LISTSERV@LISTSERV.UGA.EDU (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD