|
|
| Name |
Berger, James O. |
| Location
|
Duke University |
| Primary Field
|
Applied Mathematical Sciences |
 Election Citation
|
|
Berger is a leader in the study of statistical decision theory. His work has revolutionized the modern study of multivariate statistical settings and helped to establish the superiority of Bayesian estimators. In his interdisciplinary work, he has focused on the applicability of complex computer models to physical processes in nature. |
|
Research Interests
|
|
My research has primarily been in the areas of Bayesian statistics, statistical decision theory, foundations of statistics, and various interdisciplinary areas of science and industry. The work in Bayesian statistics has focused on developing objective Bayesian methodology, most lately for selection among statistical models and for hypothesis testing. A significant motivator of the latter work is the need to have alternatives to `p-values,' which are arguably the most misused tools in science. My interest in the foundations of statistics is based on the perhaps surprising fact that the statistics profession is still in the midst of a major debate as to what foundational approach is best; my own view is that the foundations of statistics should primarily be Bayesian, but with a strong component of frequentist statistics - the currently dominant approach. I also focus on developing statistical procedures that are simultaneously interpretable from all foundational perspectives. My work in statistical decision theory has concentrated on discovering optimal statistical procedures in decision-theoretic settings, most lately when prediction is of primary interest. My interdisciplinary work has most recently focused on research in various astronomical problems and on the endemic problem of evaluating and appropriately utilizing complex computer models of physical processes in nature and engineering. |
|
|
|