Title: Alternative Bayesian Summaries for Evaluating Acquisitions Benchmarks in Department of Defense Operational Testing

Abstract: The Department of Defense acquisitions process for new systems evaluates those systems against pre-established performance benchmarks---but the method of evaluating said benchmarks is left to the discretion of operational testers. The traditional approach that has been used for summarization in tests with continuous response variables, a grand mean based on experimental data, does not directly account for uncertainty in the experimental response or bias in the construction of the summary. To address this issue, we propose three Bayesian alternatives to the traditional summary used for benchmark evaluation: a Bayesian grand mean, a predictive criterion, and a mixture criterion we refer to as the Bayesian mission mean.