s-news
[Top] [All Lists]

Re: Binomial multilevel modelling or Conditional logistic

To: Spencer Graves <spencer.graves@PDF.COM>, Noori Akhtar-Danesh <nakhtardanesh@cogeco.ca>
Subject: Re: Binomial multilevel modelling or Conditional logistic
From: John Fox <jfox@mcmaster.ca>
Date: Fri, 28 Feb 2003 13:49:37 -0500
Cc: s-news@lists.biostat.wustl.edu
In-reply-to: <3E5FA53C.6060906@pdf.com>
References: <001301c2df50$f127a480$bfaa7182@csu.mcmaster.ca>
Dear Spencer and Noori,

The original posting referred to multilevel modeling, for which glm, which assumes independent observations, is likely not appropriate. Generalized linear mixed models are discussed, e.g., in Sec. 10.4 of the fourth edition of Venables and Ripley's Modern Applied Statistics with S. The latest version of the MASS library also includes a function, glmmPQL for fitting these models.

I hope that this helps,
 John

At 10:06 AM 2/28/2003 -0800, Spencer Graves wrote:
Have you considered "glm"?

Spencer Graves

Noori Akhtar-Danesh wrote:
Hi all,
Perhaps a simple question for you but very important for me! Can someone kindly tell me how to do binomial multilevel modeling in S-Plus2000. I am only aware of lme and nlme routines. However, I can not define family argument in these routines. My question may be asked in another way! How can I run conditional logistic regression in S-Plus? It is related to a case-control study with binary outcome when each case was matched with two controls.
Thank you for your help in advance,
Noori

-----------------------------------------------------
John Fox
Department of Sociology
McMaster University
Hamilton, Ontario, Canada L8S 4M4
email: jfox@mcmaster.ca
phone: 905-525-9140x23604
web: www.socsci.mcmaster.ca/jfox
-----------------------------------------------------


<Prev in Thread] Current Thread [Next in Thread>