Title: Estimating the distribution of the preposterior mean using moment matching

Abstract: Preposterior analysis covers a wide range of approaches in many different disciplines and relates to any analysis concerned with understanding the properties of a future posterior distribution before relevant data have been collected. Embedding preposterior analysis in a decision making context implies that we are interested in the hypothetical values of the posterior mean utility for different decision options, known as the distribution of the preposterior mean. The distribution of the preposterior mean utility can be summarised with the Expected Value of Sample Information (EVSI) and used to aid funding decisions for future data collection exercises, which is especially important in health economic decision making. We present a method for estimating the distribution of the preposterior mean utility with a small number of hypothetical data samples using moment matching and samples from the prior of the utility. We discuss why this methodology is likely to accurately approximate the distribution of interest. We also suggest some settings where the moment matching method will be less accurate, particularly for small sample sizes of the future data collection exercise. We then illustrate the success of this methodology by calculating the EVSI in four varied examples, including a practical example from the health economic literature.