Modeling the distribution of site level effects is an important problem, but it is also an incredibly difficult one. Current methods rely on distributional assumptions in multilevel models for estimation. There it is hoped that the partial pooling of site level estimates with overall estimates, designed to take into account individual variation as compared to site level variation, does not distort the overall distribution of the Empirical Bayes site-level estimates too badly. The authors plan on investigating this claim, and comparing the results of multilevel modeling to multilevel modeling where the assumption of the normally distributed site level effects is relaxed by replacing it with an unspecified distribution assigned a Dirichlet process prior. The authors hope, via simulations and analysis of actual data sets, to identify where multilevel modeling does work in practice, where it is likely to lead to erroneous conclusions, and what requirements one needs to use these approaches with reasonable levels of confidence. It may be the case that instead of attempting to measure the distribution of site level effects, they should instead attempt to directly measure summary statistics of these distributions (such as skewness and variance). These are important next steps, once they establish how this approach generally works in practice. One figure is appended. [SREE documents are structured abstracts of SREE conference symposium, panel, and paper or poster submissions.]