pbc2$serBilirD <- as.numeric(pbc2$serBilir > 1.8) glmmFit.pbc <- glmmPQL(serBilirD ~ year, random = ~ year | id, family = binomial, data = pbc2)