I am attempting to fit a hierarchical negative binomial model with bayesm. Though my data is proprietary, I was able to recreate the same error with the margarine dataset. The error I get is as follows:
> look <- rhierNegbinRw(Data = list(regdata = dat1), Mcmc = list(R = 1000,
nprint = 100))
Z not specified - using a column of ones instead
Error in alpha <= 0 :
comparison (4) is possible only for atomic and list types
I set up the mock data as follows(the regression is completely nonsensical -- just trying to get the thing to work):
data(margarine)
chpr <- margarine$choicePrice
chpr$hhid <- as.factor(chpr$hhid)
N <- nlevels(chpr$hhid)
dat1 <- vector(mode = "list", length = N)
for (i in 1:N) {
dat1[[i]]$y <- chpr[chpr$hhid==levels(chpr$hhid)[i], "PPk_Stk"]
dat1[[i]]$X <- model.matrix( ~ choice + PBB_Stk,
data = chpr[chpr$hhid == levels(chpr$hhid)[i], ])
}
I would greatly appreciate any insight into this issue.
