p-values, posterior probabilities, and Bayes Factors. Who is who?
This talk pretends to define the different ingredients involved in statistical testing their relation and mainly clarify the misinterpretation of p-values. As pointed out by several agencies and journals, ASA statement (2016), Editorial of the Basic and Applied Social Psychology journal (2015) and many others, there is abuse in using the p-value in the statistical practice. We will comment on some calibrations of the p-value (Bayarri, Benjamin, Berger and Sellke, 2016; Cabras and Castellanos, 2019; Held and Ott, 2017; Selke, Bayarri and Berger, 2001) to obtain more interpretable measures as the posterior probability of a hypothesis.
Prof. Maria Eugenia Castellanos Universidad Rey Juan Carlos, Spain https://mecastellanos.wordpress.com/
Bayesian approaches for addressing missing data in Cost-Effectiveness Analysis (alongside Randomised Controlled Trials)
Health economics has become an increasingly important discipline in medical research. Bodies such as NICE provide guidance based on health economic evaluation. Much of recent research has focused on using advanced statistical decision theoretic approaches in health economic evaluation studies where the interest is in the analysis of a multivariate outcome comprising of clinical benefit and associated costs. Missing data on one or both outcomes is common in these studies. The complex structure of the relationships between the outcomes makes handling missing data more challenging in these studies. In this talk the issues related to missing data in health economic evaluation studies will be discussed and novel statistical models developed to handle missing data using a full Bayesian approach will be described, using as motivating example a pilot study conducted at UCL.
Prof. Gianluca Baio University College London, U.K. https://gianluca.statistica.it/