The Department of Mathematics and Statistics at UMBC will hold the 6th Annual Probability and Statistics Day at UMBC during April 20-21, 2012. The event is funded by National Security Agency and will consist of a half-day workshop on Friday afternoon and a full day conference on Saturday. P/S Day is open to statisticians from all local universities, government agencies and private industries.
Participants
Mathangi Gopalakrishnan
Bayesian Equivalence Testing for Normally Distributed Surrogate Endpoints
Surrogate endpoints are commonly used in clinical trials when the true endpoint is difficult to achieve. R.L. Prentice (1986) proposed the definition and operational criteria for the validity of a surrogate endpoint within the statistical framework of hypothesis testing. The main criterion requires acceptance of the null hypothesis of conditional independence of a true endpoint and the treatment given the surrogate endpoint. In this current research, we evaluate the validity of the surrogate endpoint using the equivalence hypothesis under a Bayesian framework. We consider the true and the surrogate endpoints to be normally distributed. We test the equivalence hypothesis by computing Bayes factors under various sets of priors. This method for evaluating the surrogate endpoint is also applied to a real dataset.