1: \begin{abstract}

2: Bayesian hierarchical modeling is a popular approach to capturing

3: unobserved heterogeneity across individual units.  However, standard estimation methods such as Markov

4: chain Monte Carlo (MCMC) can be impracticable for modeling outcomes

5: from a large number of units.  We develop a new method to sample from posterior

6: distributions of Bayesian models, without using MCMC.  Samples are

7: independent, so they can be collected in parallel, and we do not need to be concerned with issues like chain

8: convergence and autocorrelation.  The algorithm is scalable under

9: the weak assumption that individual units are conditionally

10: independent, making it applicable for large datasets. It can also be

11: used to compute marginal likelihoods.

12: \end{abstract}