Linking Statistical Analysis and Decision Analysis in Health-Related Research

Dalene Stangl, PhD
Department of Statistical Science
Duke University
*Candidate for Division Head in the Division of Biostatistics

Wednesday, April 8th
3:30pm
MoosT 1-450
Minneapolis Campus

Abstract:

Historically, quantitative analysis in health-related research has served to present statistical summaries of research data. Primarily this means providing point and interval estimates for quantities of interest such as means, relative risks, odds ratios, survival probabilities, effect sizes, …. The focus is typically an estimate of effect size, an interval reflecting uncertainty in our estimate, and a hypothesis test of whether we could observe, by chance alone, effects at least as large as those observed. While this research process is itself a sequence of decisions, using the statistical results within a formal decision theoretic framework to guide subsequent decision making rarely occurs, or if it does, it is not transparent. However, nowhere is the need for a formal transparent link between statistical analysis and decision analysis clearer than in health-related research. This talk describes how Bayesian methods can provide this link, a brief historical review of some attempts to progress this line of research, and some suggestions for directions for future research.

A social tea will be held at 2:45 P.M. in A434 Mayo. All are Welcome.
For more details contact 612-624-4655 or see http://www.biostat.umn.edu/seminar_academic.html