Herman Rubin
Herman Rubin was a distinguished professor of statistics and mathematics at Purdue University, known for his prolific research and fundamental contributions across numerous mathematical disciplines. He died in 2018 at the age of 91.
Education and career
Born in 1926 in Chicago, Illinois, Rubin was considered a child prodigy who earned his undergraduate and graduate degrees from the University of Chicago, completing his doctorate in mathematics at the age of 21 in 1948; his official advisor was Paul Halmos. He served in the U.S. Army during the Second World War. Prior to joining Purdue University in 1967, Rubin held faculty positions at several other prestigious institutions.Contributions to statistics
Rubin was considered a polymath with a unique ability to solve complex problems. He was a prolific researcher, publishing over 130 papers that have become standard texts in various fields. His key contributions include:- Multivariate Analysis: He co-authored seminal papers with T.W. Anderson on the distribution theory of maximum likelihood estimators in factor analysis and structural equation models.
- Monotone likelihood ratio: He is credited with the widely used and fundamental concept of monotone likelihood ratio families.
- Bayesian Inference: Rubin was a lifelong Bayesian statistician who took the theory and axioms of Leonard Jimmie Savage literally, and he contributed significantly to the theory of Bayes risks.
- Econometrics: He worked on problems related to the simultaneous equations model in economics, including the development of the Limited information maximum likelihood estimator.
- Set Theory and Probability: His work also included contributions to set theory, the theory of moderate deviations, and the characterization of probability distributions.
- Hypothesis testing: in collaboration with Samuel Karlin, the Karlin-Rubin theorem provides a method for finding uniformly most powerful tests for one-sided hypotheses, a significant contribution to the theory of hypothesis testing.