applications). However, the DA algorithm often suffers from slow convergence. The 'sandwich' DA
algorithm is an effective tool developed over the past decade to improve the speed of the DA
algorithm at negligible computational cost. The first part of the talk will focus on a
theoretical comparison of the sandwich algorithm with the DA algorithm, in the traditional
'two-block' DA case. The second part of the talk will focus on the extension of the sandwich algorithm
to the 'multi-block' DA case. The methods proposed will be illustrated on the Bayesian lasso
algorithm of Park and Casella (2008) and the Bayesian probit regression algorithm of
Albert and Chib (1993). This is joint work with Jim Hobert and Subhadip Pal.