Mixed-model-binary logistic model with dependence between individual repeated measures

Ben Bolker, thank you for your suggestions.

Yes, it is suprising that I in SAS and STATA have to assume independence between the measurements within an individual. I do not want to assume that. In addition I would like to be able to chose other distributions than the normal for my random effect, which is not possible in SAS (proc NLMIXED).
The generalized estimating equation packages are probably not an option as I do not whant marginal models.
I will look at the references you suggested.
Thank you.
/Anna


___________________________________
Anna Ekman  (Grimby-Ekman)
PhLic Statistics, PhD (Dr Med Sci)
Occupational and Environmental Medicine
Sahlgrenska Academy and University Hospital
Box 414, SE - 405 30 Goteborg, Sweden
Phone +46 (0)31 786 31 23
www.amm.se

Akademistatistik inom EpiStat
Sahlgrenska akademin vid G?teborgs universitet
www.sahlgrenska.gu.se/forskning/epistat/

________________________________________
Fr?n: Ben Bolker [bbolker at gmail.com]
Skickat: den 7 januari 2011 16:43
Till: Anna Ekman
Kopia: r-sig-mixed-models at r-project.org
?mne: Re: [R-sig-ME] Mixed-model-binary logistic model with dependence between individual repeated measures

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi, I am a novice R user and do not know how to properly mail to this
list. I apologies if I do it in the wrong way.

I want to analyze my data using a random intercept (later extended
also to random slope) logistic model for a binary outcome (later
extended to a ordinal outcome). This I have been able to do in SAS if
assuming the repeated measurements within an individual to be
independent, but I want to be able to choose different covariance
structures for the individual measurements. This I cannot do directly
in either SAS or STATA, and therefore now turn to R. How can I do
this in R?

Anna

I'm surprised that you can't do this in SAS (PROC MIXED, NLMIXED, or
GLIMMIX?) or Stata <http://www.gllamm.org/>, but: if you want to do it
in R, your choices are glmmPQL in the MASS package or possibly one of
the generalized estimating equation packages (geese, geepack?)  I would
recommend the following references for getting started:

  Zuur et al Mixed models (Springer)
  Pinheiro and Bates 2000 (Springer), especially the material on
temporal autocorrelation models

  Extending to a ordinal outcome with temporal autocorrelation could be
tricky ...




-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAk0nNJsACgkQc5UpGjwzenMymACfbqx+gtOjyhoX9hHpPyO/vbVg
NXUAniLvM+8voyzKA6axKLyJLclqeBhY
=wBwK
-----END PGP SIGNATURE-----