[R-sig-ME] response residuals from a logistic mixed regression model using lmer

[vahl]

Dear,

Using the lmer function, I have fitted the following logistic mixed 
regression model on an experimental data set containing three fixed 
factors (A, B & C) and two random variables (day & cage):

model1 <- lmer(cbind(k1, k2) ~A*B*C + (1|day) + 
(1|cage),family=binomial(logit),data=data),

where k1 is the number of successes and k2 the number of failures in a 
trial.

Presenting all data in the same figure turns out to result in a graph 
that is hard to read. Therefore, I am now planning to present my data in 
two graphs, one containing information on A and B, the other containing 
information on B & C.

In presenting my data regarding factor A and B visually, I would like to 
take out the (experimentally imposed) variation explained by the factor 
C and the two random block factors. To do so, I need the raw, response 
residuals, that is the residuals on the logit scale. I have studied the 
help-file of the lmer function as well as that of the mer class, and I 
have read the question posted by Andy Fugard on this list as well as the 
reply of Douglas Bates (both posted at 14 march 2008), but am not able 
to find the response residuals. Any advice would be kindly appreciated.

Wouter Vahl