James O. Berger

James Oscar Berger (born 1950) is one of the most prominent and productive Bayesian statisticians of the contemporary era. A professor at Duke University and founding director of the Statistical and Applied Mathematical Sciences Institute (SAMSI), he has made landmark contributions to statistical decision theory, objective Bayesian analysis, Bayesian model selection, and the philosophical foundations of statistical inference. His textbook Statistical Decision Theory and Bayesian Analysis is a standard reference, and his work on reference priors and Bayes factors has been adopted across the sciences.

Education and Career

Berger received his PhD from Cornell University in 1974 under the supervision of Lawrence Brown. After positions at Purdue University, he moved to Duke University in 1997, where he has remained. He served as director of SAMSI from 2002 to 2010, building it into a major center for interdisciplinary statistical research. He has held visiting positions at institutions around the world and has been a central figure in the international Bayesian community.

Statistical Decision Theory

Berger's 1985 textbook Statistical Decision Theory and Bayesian Analysis (second edition) synthesized several decades of development in the field, presenting a unified treatment of loss functions, risk, admissibility, minimax procedures, and Bayesian methods. The book became the standard graduate text and introduced many statisticians to the deep connections between frequentist and Bayesian approaches illuminated by decision theory.

Objective Bayesian Methods

With José-Miguel Bernardo and others, Berger has been a leading developer of objective Bayesian methods—approaches that seek to derive prior distributions from formal principles rather than subjective judgment. Their reference prior methodology extends Jeffreys' ideas to multiparameter problems and provides a systematic way to construct noninformative or minimally informative priors. Berger has also developed methods for default Bayes factors that allow objective Bayesian hypothesis testing.

“The Bayesian approach is not a luxury; it is a necessity. The only question is whether we are honest about our prior assumptions or hide them.”— James O. Berger

Model Selection and Multiplicity

Berger has made major contributions to Bayesian approaches to model selection, including the development of intrinsic Bayes factors and fractional Bayes factors for comparing models when improper priors are used. He has also worked extensively on the multiplicity problem in testing—how to properly account for multiple comparisons in a Bayesian framework—with applications to genomics and other areas where thousands of hypotheses are tested simultaneously.

Interdisciplinary Impact

Berger's applied work spans physics, astronomy, medicine, and environmental science. He has collaborated with physicists at CERN on the statistical analysis of particle physics experiments, with astronomers on the detection of exoplanets, and with climate scientists on uncertainty quantification. His insistence on rigorous statistical methodology in high-stakes applications has been influential across the sciences.