Displaying 1 result from an estimated 1 matches for "totalpat".
Did you mean:
totalpath
2008 Mar 08
1
analysing mixed effects/poisson/correlated data
...if rates vary between hospitals.
There is first order autoregression in the data. I have attempted to use the lmer function (and lmer2) with correlation structure and weights:
test1 <-lmer(opsn1~timepoint+as.factor(hospital)+ skillmixpc + nursehrsperpatday +(timepoint|hospital) +offset(log(totalpats)),family=poisson, data=opsn.totals)
test2 <-lmer(opsn1~timepoint+as.factor(hospital)+ skillmixpc + nursehrsperpatday +(timepoint|hospital)+offset(log(totalpats)),family=poisson, data=opsn.totals, correlation=corAR1(form=~1|hospital))
test3 <-lmer(opsn1~timepoint+as.factor(hospital)+...