[R-sig-ME] Bernoulli glmm question.

[Rolf Turner]

Thanks to Marko Bachl, Tibor Kiss, Robert Espesser, and off-list Kevin 
Thorpe and Robert Kushler who all provided useful feedback and advice.

The overall import of the replies was that I should include ***word*** 
as a random effect, which I shall do.  In retrospect it's kind of 
obvious that this effect should be included.  Duhhhh!

Robert Espesser also suggested that I might wish to include some 
interaction terms in the model. He's almost surely correct, but I'm a 
bit reluctant to do so as that would take me (and definitely) the 
graduate student well out of our comfort zones.  I'll play around with 
this idea and try the various models that Prof. Espesser suggested, and 
see how much difference they make to the inferences to be drawn about 
the fixed effects.

Thanks again.

cheers,

Rolf Turner