Dear All, I am using lmer to fit a glmm model using the PQL approximation. My data is not a Bernoulli (0/1) but a binomial one. The first thing I noticed is that the method gives me the approximated covariance matrix for the random effects but nothing about the residuals. In addition, I have the following estimation that I do not know what deos it mean: Estimated scale (compare to 1 ) 75043946. However, when I use the glmmPQL method to fit this model then the problem doesn't arise, i.e. I have the estimated covariance matrix for the random effects as well as the standard deviation of the residuals. I would be very grateful for any comments and/or explanation. Thank you in advance, Bernard --------------------------------- [[alternative HTML version deleted]]