Hi bbolker,
First thanks for your quick answer!
2) I use the last version of lme4, but an older version of the R-book, so that I didn't know that quasibinomial family was no more supported...
3) I used the script you kindly provided:
- why do you add the data$obs variable? When I run model1 I got the following message: Error in mer_finalize(ans) : q = 143 > n = 125.
- However, the model0 works fine, no more error message! Thanks for the indications on the use of continuous variable in random effects!
4) I actually did look at the data beforehand, using a probit visualization... That is why I was quite surprized by the results of initial models I did and thus considered using lmer...
(I would send them to you but don't see how to attach documents to my message)
I tried to use the script you provided for visualizing the data but again got an error message:
Error in `[<-.data.frame`(`*tmp*`, var, value = list(weight = c(32L, 83L, : replacement element 1 has 125 rows, need 720
Thanks again for the help,
Have a good day,
Pierrick