How to conduct a correlated logistic regression in SPSS?

[Jan Warren-Findlow]

Hi all.

I have a cross-sectional dataset with 190 participants (95 parent-child pairs). I have data that indicates who the pairs are, and their relationship within the pair. I am conducting logistic regression analyses and my co-author indicates that I need to perform a correlated logistic regression and/or specify an exchange correlation structure. This is all SAS-speak and I am having difficulty translating this to appropriate SPSS syntax.

Several colleagues have indicated that I need to use GEE but my understanding of GEE is that it requires repeated measures data, which is not what I have. Does anyone have any suggestions?

Be advised I am not a statistician or epidemiologist, so you'll need to use verbage akin to "SPSS for dummies".

Thanks in advance.

[Hector Maletta]

What I would suggest is constructing a new dataset, in which each case
represents all the parent variables plus (with different names) all the
child variables. Then, an ordinary logistic regression can be run with some
predictors referring to the parent and some to the child. I suppose the
dependent variable is some state or event occurring to the pair, or to the
child, and this event or state can be predicted by characteristics of the
parent and characteristics of the child. For instance, school success might
be predicted by child's age and parent's age, by parent's education and
child's nutritional status, and so on.
Apart from the above, you may also want to run the same logistic regression
(separately) on the parent dataset and on the children data set, perhaps to
see whether the same predictors are significant or whatever.
Other than that, I do not know of any procedure called "correlated logistic
regression".

Hector

-----Mensaje original-----
De: SPSSX(r) Discussion [mailto:[hidden email]] En nombre de Jan
Warren-Findlow
Enviado el: Thursday, October 21, 2010 1:34 PM
Para: [hidden email]
Asunto: How to conduct a correlated logistic regression in SPSS?

Hi all.

I have a cross-sectional dataset with 190 participants (95 parent-child
pairs). I have data that indicates who the pairs are, and their relationship
within the pair. I am conducting logistic regression analyses and my
co-author indicates that I need to perform a correlated logistic regression
and/or specify an exchange correlation structure. This is all SAS-speak and
I am having difficulty translating this to appropriate SPSS syntax.

Several colleagues have indicated that I need to use GEE but my
understanding of GEE is that it requires repeated measures data, which is
not what I have. Does anyone have any suggestions?

Be advised I am not a statistician or epidemiologist, so you'll need to use
verbage akin to "SPSS for dummies".

Thanks in advance.
--
View this message in context:
http://spssx-discussion.1045642.n5.nabble.com/How-to-conduct-a-correlated-lo
gistic-regression-in-SPSS-tp3230830p3230830.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (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
Se certificó que el correo entrante no contiene virus.
Comprobada por AVG - www.avg.es
Versión: 8.5.448 / Base de datos de virus: 271.1.1/3206 - Fecha de la
versión: 10/21/10 06:34:00

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (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

[Bruce Weaver]

Jan Warren-Findlow wrote

Hi all.

I have a cross-sectional dataset with 190 participants (95 parent-child pairs). I have data that indicates who the pairs are, and their relationship within the pair. I am conducting logistic regression analyses and my co-author indicates that I need to perform a correlated logistic regression and/or specify an exchange correlation structure. This is all SAS-speak and I am having difficulty translating this to appropriate SPSS syntax.

Several colleagues have indicated that I need to use GEE but my understanding of GEE is that it requires repeated measures data, which is not what I have. Does anyone have any suggestions?

Be advised I am not a statistician or epidemiologist, so you'll need to use verbage akin to "SPSS for dummies".

Thanks in advance.

I suspect your co-author means "conditional" logistic regression.  SPSS has no canned procedure for that, but it can be obtained via the COXREG procedure.  Marta García-Granero posted a nice example to this list a few months ago--you should be able to find it in the archives at:

   http://www.listserv.uga.edu/archives/spssx-l.html

[Art Kendall]

It sounds like you do have repeated measures.  Repeats are not just based on times.  members of a couple would be repeats.  Parent/child would be repeats.  The same variable measured under different conditions. Split plots are repeats, as are twins, litter mates, left/right eye, pairs of almost anything, etc.


Some dialectics of statistics (e.g., SUDAAN) use the term "correlated data" to refer not to the usual situation of variables being correlated, but in the sense of the inclusion of a case being correlated with the inclusion of another case.

Is your DV dichotomous?  Then GEE sound likely?  For data that is not too discrepant from continuous, some form of the GLM such as repeated measures ANOVA would be a likely candidate approach.

Art Kendall
Social Research Consultants



On 10/21/2010 12:33 PM, Jan Warren-Findlow wrote:

Hi all.

I have a cross-sectional dataset with 190 participants (95 parent-child
pairs). I have data that indicates who the pairs are, and their relationship
within the pair. I am conducting logistic regression analyses and my
co-author indicates that I need to perform a correlated logistic regression
and/or specify an exchange correlation structure. This is all SAS-speak and
I am having difficulty translating this to appropriate SPSS syntax.

Several colleagues have indicated that I need to use GEE but my
understanding of GEE is that it requires repeated measures data, which is
not what I have. Does anyone have any suggestions?

Be advised I am not a statistician or epidemiologist, so you'll need to use
verbage akin to "SPSS for dummies".

Thanks in advance.
--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/How-to-conduct-a-correlated-logistic-regression-in-SPSS-tp3230830p3230830.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (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 [hidden email] (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

[Ryan]

Jan,

 

I responded recently to a similar question. My response is located at the link below.  If after reading the post you still have questions, then feel free to write back.

http://www.listserv.uga.edu/cgi-bin/wa?A2=ind1009&L=spssx-l&P=R38952

 

Ryan

On Thu, Oct 21, 2010 at 12:33 PM, Jan Warren-Findlow <[hidden email]> wrote:

Hi all.

I have a cross-sectional dataset with 190 participants (95 parent-child
pairs). I have data that indicates who the pairs are, and their relationship
within the pair. I am conducting logistic regression analyses and my
co-author indicates that I need to perform a correlated logistic regression
and/or specify an exchange correlation structure. This is all SAS-speak and
I am having difficulty translating this to appropriate SPSS syntax.

Several colleagues have indicated that I need to use GEE but my
understanding of GEE is that it requires repeated measures data, which is
not what I have. Does anyone have any suggestions?

Be advised I am not a statistician or epidemiologist, so you'll need to use
verbage akin to "SPSS for dummies".

Thanks in advance.
--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/How-to-conduct-a-correlated-logistic-regression-in-SPSS-tp3230830p3230830.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (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