Jeromy Anglim's Notes

I was trying to run a JAGS model of a vector of data Y1; part of the code is extracted as follows:

Y1[i] <- Mu[i] + E[i]
Mu[i] ~ dnorm(Delta, TauPrecision)
E[i] ~ dnorm(0, SigmaPrecision)

When I ran jags.model(...), I received the following error:

Error in jags.model("m2-complete-jags.txt", data = jagsdata.Y1, n.chains = 4,  : 
  RUNTIME ERROR:
Compilation error on line 5.
Y1[1] is a logical node and cannot be observed

• What caused this?

Observed variables like Y1 need to be assigned a distribution. Thus, the code needed to be rewritten to

Y1[i] ~ dnorm(Mu[i], SigmaPrecision)
Mu[i] ~ dnorm(Delta, TauPrecision)