jags_covariates.txt

model {

  ## joint likelihood

  for (i in 1:n) {

    c[i] ~ dnorm(phi.c[i], lambda)
    e[i] ~ dnorm(phi.e[i], tau)
    
    phi.c[i] = mu.c + beta*(e[i] - mu.e) + gamma.c[status[i]]
    phi.e[i] = mu.e + gamma.e[status[i]]
  }

  lambda = 1/psi
  psi = sigma2.c - sigma2.e*pow(beta, 2)


  ## priors

  mu.c ~ dnorm(0, 0.0001)
  logsigma.c ~ dunif(-5, 10)
  sigma.c = exp(logsigma.c)
  sigma2.c = pow(sigma.c, 2)
  
  mu.e ~ dnorm(0, 0.0001)
  logsigma.e ~ dunif(-5, 10)
  sigma.e = exp(logsigma.e)
  sigma2.e = pow(sigma.e, 2)
  tau = 1/sigma2.e
  
  beta ~ dunif(-5, 5)
  
  gamma.c[1] = 0
  gamma.e[1] = 0
  
  for (j in 2:3) {
    gamma.c[j] ~ dnorm(0, 0.0001)
    gamma.e[j] ~ dnorm(0, 0.0001)
  }
  
  ## predictions
  
  for (k in 1:3) {
  
    c_pred[k] ~ dnorm(phi.c_pred[k], lambda)
    e_pred[k] ~ dnorm(phi.e_pred[k], tau)
  
    phi.c_pred[k] = mu.c + beta*(e_pred[k] - mu.e) + gamma.c[k]
    phi.e_pred[k] = mu.e + gamma.e[k]
  }
  
}