[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

RE: st: group size needed for mixed models (binary response)

From   "Verkuilen, Jay" <>
To   <>
Subject   RE: st: group size needed for mixed models (binary response)
Date   Tue, 27 Nov 2007 11:35:55 -0500

Susan Lingle wrote:

>>>Your response helps tremendously - I have been struggling with this
**quite** some time. I wanted to make the change to a mixed model, 
because it was in response to reviewers' comments (and it can be wise to

heed such advice).<<<

You are in a position with mixed models that is not so good given the
large amount of cases where there is only one observation per cluster.
I'd consider use "robust" standard errors with clustering by parent.
That should fix up the standard errors for the overdispersion due to the
existence of twins. (If your sample size isn't large, use clustered
bootstrapping instead.) Stata does this for you very nicely in both
logit and xtlogit. 

>>>This evening, I was considering comparing AIC (or BIC) values to
a model with or without the mother's identity included as a random 


I wanted to test a specific hypothesis (WT are more likely to die during

summer; MD during winter) so did not plan to undertake a model selection

procedure. Anyway, I think your response gives me enough food for 
thought now - checking the intraclass correlation is a sensible place to

start before making the next decision.<<<

The problem with intraclass correlations separate from the other
regressors is that they can change quite a bit. Also, IIRC your outcome
variable is binary so I'm not sure what the one-way ANOVA procedure is
going to give you, especially if the base rate of death is skewed.
Better to run xtlogit with no regressors and look at its estimate of the
intraclass correlation. 

Suggestion: Look at Burnham, K. P., and D. R. Anderson. 2002. Model
selection and multimodel inference: a practical information-theoretic
approach. 2nd Edition. Springer-Verlag, New York, New York, USA. 488 pp.
This book is written by people working in wildlife biology (one is a
biostatistician, the other a biologist) and should provide a very useful
citation for you. They have a decent discussion of the overdispersion
problem, too, and some basic solutions for it. 

J. Verkuilen
Assistant Professor of Educational Psychology
City University of New York-Graduate Center
365 Fifth Ave.
New York, NY 10016
Office: (212) 817-8286 
FAX: (212) 817-1516
Cell: (217) 390-4609

*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index