Hi
I am conducting a meta-analysis on proportions and I'm interested in using the exact binomial likeliood for the within-study effects rather than the approximate approaches and I don't want to use the quasi-likelihood approach in R's glmmPQL.
Does anyone know how I can do this using nlme? I can do it in SAS nlmixed but I would rather use R -this is my SAS code:
proc nlmixed data=work.temp ;
parms mup=2 vp= 08; /*initial values*/
rawprop=1/(1+exp(-truep)); /*unkown true p logit proportion*/
model numberofisolates~binomial(sumisolatestudy, rawprop) ;
random truep~normal(mup,vp) subject=studyid;
run;
Thanks
B. Aletta Nonyane, PhD
Assistant Scientist
Department of International Health
Johns Hopkins Bloomberg School of Public Health
615 N. Wolfe Street
Baltimore, MD, 21205
[[alternative HTML version deleted]]