R help - Jul 2013 - repeated measures logistic regression

I have searched the r-help archive and saw only one unanswered post related
to mine.

My design is as follows.

   - y is Bernoulli response
   - x1 is continuous variable
   - x2 is categorical (factor) variable with two levels

The experiment is completely within subjects. That is, each subject
receives each combination of x1 and x2.

This is a repeated measures logistic regression set-up. The experiment will
give two ogives for p(y==1) vs x1, one for level1 and one for level2 of x2.
The effect of x2 should be that for level2 compared to level1, the ogive
should have a shallower slope and increased intercept.

I am struggling with finding the model using lme4. Here is a guess at it:

glmer(y~x1*x2 +(1|subject),family=binomial)

So far as I understand it, the 1|subject part says that subject is a random
effect. But I do not really understand the notation or how to specify that x
1 and x2 are repeated measures variables. In the end I want a model that
includes a random effect for subjects, and gives estimated slopes and
intercepts for level1 and level2.

Thanks very much for any help.

Stan

	[[alternative HTML version deleted]]

Stanislav Aggerwal <stan.aggerwal <at> gmail.com> writes:
> 
> I have searched the r-help archive and saw only one 
> unanswered post related
> to mine.
  Take a look at the r-sig-mixed-models (@r-project.org)
mailing list and archive ...
> 
> My design is as follows.
> 
>    - y is Bernoulli response
>    - x1 is continuous variable
>    - x2 is categorical (factor) variable with two levels
> 
> The experiment is completely within subjects. That is, each subject
> receives each combination of x1 and x2.
> 
> This is a repeated measures logistic regression set-up.
> The experiment will
> give two ogives for p(y==1) vs x1, one for level1 and one 
> for level2 of x2.
> The effect of x2 should be that for level2 compared to level1, the ogive
> should have a shallower slope and increased intercept.
> I am struggling with finding the model using lme4. Here is a guess at it:
> 
> glmer(y~x1*x2 +(1|subject),family=binomial)
 
> So far as I understand it, the 1|subject part says 
> that subject is a random
> effect. But I do not really understand the notation or
>  how to specify that x1 and x2 are repeated measures variables. 
> In the end I want a model that
> includes a random effect for subjects, and gives estimated slopes and
> intercepts for level1 and level2.
  I believe you want

glmer(y~x1*x2 +(x1*x2|subject),family=binomial,data=...)

 (I strongly recommend including the data= argument in your call)

This will give a population-level estimate of

intercept (log-odds in group 1 at x1=0)
treatment effect on intercept (log-odds(level2,x1=0)-log-odds(level1,x=0))
log-odds slope in level 1
difference in slopes

as well as among-individual variances in all four of these parameters,
and covariances among all the parameters (i.e. a 4x4 variance-covariance
matrix for these parameters).

  For binary data and estimating 4 fixed + 10 RE parameters
(i.e., variances and covariances), you're going to need a lot of data --
very conservatively, 140 total observations.

  It may help to center your x1 variable.

  see http://glmm.wikidot.com/faq 
(especially http://glmm.wikidot.com/faq#modelspec), 
and the r-sig-mixed-models mailing list.