1. How are residual analysis and posterior predictive simulation similar to one another?

2. How can posterior predictive simulation be considered a method to compare models and not simply a method to evaluate a single model? 164 6 Evaluating Markov Chain Monte Carlo Algorithms and Model Fit

3. Reconduct the bivariate normal distribution example from the last chapter, using an MH algorithm to estimate the parameters, but select extremely poor starting values for them. Run the algorithm five times with different sets of such poor starting values and compute the scale reduction factor R. Also construct trace plots. Does the R calculation lead you to a different conclusion concerning where to consider the burn-in to have ended than the trace plots?