Output: Hierarchical Model assuming heterogeneous
mortalities using 0’s trick for the prior
Posterior means, posterior standard deviations and 95%
credible intervals (Inference->Samples->stats)
Sample path (history)
Posterior probability densities
(Inference->Samples->density)
(Further, right click on the figure->Margins->
Special…->Smooth -> change from 0.2 to 0.1-> apply all)
Sample autocorrelation function
(Inference->Samples->auto corr)
(If you cannot get all results at once, for example, you
could specify “alpha” in node and click on
Auto corr).
Running quantile plot
(Inference->Samples->quantiles)
Scatter plots (Inference->Correlations…)
Shrinkage Estimates of mortality (Inference->
Compare-> node:B, axis:loge and click on modelfit)
(Further, right-click on the figure,
Properties->Titles -> x-axis: log(e), y-axis: B)
Boxplot of mortality (Inference-> Compare->
node:lambda and click on boxplot)
(Further, right-click on the figure, Properties…->
Margins -> Special…-> check “rank” and uncheck “showbaseline”. Titles
-> x-axis: Hospital)
*Hospital 85 has the smallest mortality rate
Posterior predictive analysis. pleft:Pr(ypred[i] =<
y[i]). pright: Pr(y.pred[i]>=y[i])
Inference-> Compare-> node:pleft and click on
caterpillar)
(Further, right-click on the figure, Properties…->
Margins -> Special…-> check “rank” and showbaseline:0.05)
Inference-> Compare-> node:pright and click on
caterpillar)
(Further, right-click on the figure, Properties…->
Margins -> Special…-> check “rank” and showbaseline:0.05)