Commentary: robust estimation of population parameters with sparse data.

WinBUGS code to fit the multivariate Bayesian relative risk model: model{ for( i in 1 : N ) { for( j in 1 : K ) { Y[i, j] ~ dpois(lambda[i, j]) # distribution of observations lambda[i, j] E[i, j] * theta[i, j] theta[i, j] exp(phi[ i, j]) # log parametrization } phi[i, 1:K ] ~ dmnorm(mu[ ], Omega[, ]) } for(j in 1:K){ mu[j] ~ dunif( 2,2) } Omega[1:K, 1:K ] ~ dwish(R[, ], 12) # Wishart on prec. matrix }