Generalised Estimating Equations_error message-help!

[zannaclay]

Hi

I am trying to run a Generalised Estimating Equations (GEE) analysis but keep reaching this error message:

"There are at least two records with the same values for the subject and within-subject variables. No output will be displayed."

Here are my analyses. I have checked and there are NO replicates it speaks of, so must be something about how data is set up

I am analysing the effect of numerous social variables on the likelihood of reconciliation between two chimpanzees following a fight

I have 20 individuals who thus can therfore be either the aggressor or the victim. the number of fights between each dyad, and per individual is very variable.

To control for the identity of the victim and the aggressor, I entered Victim ID as subject variable and Aggressor ID as repeated-subjects variable (i.e. multiple fights within the same chimp pair)

this is when I get the error message....is it becuase the two variables-victim and aggressor are using the same codes?

Please help!!!

[Alex Reutter]

Based on the below, it sounds like what you want is to use both Victim ID and Aggressor ID as subject variables; your "subject" is the dyad.  Your repeated-subjects variable would be an index of the fights within that dyad.

Alex

[zannaclay]

Hi Alex

thanks very much for this idea, I will try it

but...the issue I see is that by combining the victim ID and the aggressor ID into one subject= the dyad, I will not be able to account for the fact that the same individual is a victim/or aggressor in multiple dyads...if you see what I mean?

anyway around this?

[Alex Reutter]

Hm... if you specify Victim ID as the subject variable and Aggressor ID and Fight Index as the within-subject variables, that would allow the covariance matrix to account for the correlation between fights where the victim is the same, but not fights where the aggressor is the same.  If you're not required to use GEE for some other reason, you might have more flexibility by using a mixed model and specifying the victim and aggressor as random effects.

Alex

[zannaclay]

Hi Alex

thanks a lot for your thoughts, this is worth a shot

yes- I think ill try with GLMMs too, see which works best 

best wishes

Zanna

On Thu, Feb 23, 2012 at 1:11 PM, Alex Reutter [via SPSSX Discussion] <[hidden email]> wrote:

Hm... if you specify Victim ID as the subject variable and Aggressor ID and Fight Index as the within-subject variables, that would allow the covariance matrix to account for the correlation between fights where the victim is the same, but not fights where the aggressor is the same.  If you're not required to use GEE for some other reason, you might have more flexibility by using a mixed model and specifying the victim and aggressor as random effects.

Alex

---

If you reply to this email, your message will be added to the discussion below:

http://spssx-discussion.1045642.n5.nabble.com/Generalised-Estimating-Equations-error-message-help-tp5508536p5508872.html

To unsubscribe from Generalised Estimating Equations_error message-help!, click here.
NAML

--
Zanna Clay, PhD

Post-doctoral Research Fellow, Living Links Primate Center

Department of Psychology, Emory University

36 Eagle Row

Atlanta, GA 30322

[phone]-0240 (cell); 404-727-6778 (office)

[Ryan]

How is "reconciliation" measured?

Ryan

On Fri, Feb 24, 2012 at 3:05 PM, zannaclay <[hidden email]> wrote:

Hi Alex

thanks a lot for your thoughts, this is worth a shot

yes- I think ill try with GLMMs too, see which works best 

best wishes

Zanna

On Thu, Feb 23, 2012 at 1:11 PM, Alex Reutter [via SPSSX Discussion] <[hidden email]> wrote:

Hm... if you specify Victim ID as the subject variable and Aggressor ID and Fight Index as the within-subject variables, that would allow the covariance matrix to account for the correlation between fights where the victim is the same, but not fights where the aggressor is the same.  If you're not required to use GEE for some other reason, you might have more flexibility by using a mixed model and specifying the victim and aggressor as random effects.

Alex

---

If you reply to this email, your message will be added to the discussion below:

http://spssx-discussion.1045642.n5.nabble.com/Generalised-Estimating-Equations-error-message-help-tp5508536p5508872.html

To unsubscribe from Generalised Estimating Equations_error message-help!, click here.
NAML

--
Zanna Clay, PhD

Post-doctoral Research Fellow, Living Links Primate Center

Department of Psychology, Emory University

36 Eagle Row

Atlanta, GA 30322

[phone]-0240 (cell); 404-727-6778 (office)

---

View this message in context: Re: Generalised Estimating Equations_error message-help!

Sent from the SPSSX Discussion mailing list archive at Nabble.com.