2. In Example 9.3 (farmer suicides) analyse the data without any predictors under a convolution prior, namely yi Po(i Ei ), log(i ) =

2. In Example 9.3 (farmer suicides) analyse the data without any predictors under a convolution prior, namely yi ∼ Po(μi Ei ), log(μi ) = β1 + si + ui .

Use the discrete prior (9.11) on the variances of spatial and unstructured effects. Obtain the probabilities p(si > 0|y) and p(ui > 0|y), and assess the relative importance of the two forms of variation.