Can we get the residuals from JAGS output?

Question

Say we fit a Bayesian linear mixed model using JAGS (or WinBUGS), does the output object include the model residuals? How can we find the residuals?

Thanks!

Answer 1

A JAGS (BUGS) model simply outputs the values of the nodes in the model you have told it to monitor. To get residuals you need to define those in the model and then monitor them. For example

model {
 bResponse ~ dnorm(0, 5^-2)
 sResponse ~ dunif(0, 5)
 for (i in 1:length(Response)) {
   eResponse[i] <- bResponse
   Response[i] ~ dlnorm(eResponse[i], sResponse^-2)
   Residual[i] <- (log(Response[i]) - log(eResponse[i])) / sResponse
 }
} 

were Residual[i] defines a residual for each of the i values of the Response. Note the above example does not deal with specifying which values to monitor.