Hi! am fitting a model using glmer in the lme4 and nlme packages in R for count and continuous variables. For the count variables, i used Poisson family link and gamma for continuous response variables. the Poisson model is OK but I have a problem in handling the continous variables using glmer glmer(Y ~ X+(1|a), family = gamma,data=data).
Besides, to check the homogenize of
error in residual vs. fitted plot int the glmer, i write a command plot(model name) as the ordinary glm or lme. But, not working al all. I tried to find relavant examples in the glmer package in fitting the plots, but, I couldn't figure out the problems. So, can any one help me please on these two case.
With regards,
AB