7. In Example 6.7 (DP analysis of eye-tracking anomalies) try monitoring the i to obtain posterior means (i for each of the 101 subjects and

7. In Example 6.7 (DP analysis of eye-tracking anomalies) try monitoring the θi to obtain posterior means (θi for each of the 101 subjects and so obtain the DIC using the definition D − D(θ). In BUGS this will also require including code to obtain the deviance at each iteration. Assume the hyperparameters of the gamma-mixing density are free. Does adopting a DPP (with α a free parameter) improve over the standard Poisson-gamma mixture

(Chapter 5)? Is this conclusion affected by setting α at particular values, e.g. α = 1 and

α = 5 rather than letting it be a free parameter?