Faculty Member: 1960-1984
David Duncan received his undergraduate training in Australia at the University of Sydney and then earned his PhD in statistics from Iowa State University in 1947. After several years at Virginia Polytechnic Institute and the University of North Carolina, he joined the Johns Hopkins Department of Biostatistics in 1960. His early work focused on regression analysis, a topic to which he made two important contributions: First, he was an early advocate of what has become known as the Kalman filter, a method for dynamic estimation of the regression equation, which has special application time series problems. Second, he and co-author Strother Walker discovered logistic regression analysis as early as 1967. It is unclear who was the first advocate of using logistic rather than linear regression for binary responses but Dr. Duncan was certainly one of the first, if not the first.
Dr. Duncan's career-long love, however, was methodology for dealing with the "multiple comparisons" issue - that is, how to adjust statistical error rates to account for making a large number of inference from one data set. Early in his work on this problem, he created the Duncan Multiple Range test, which was the standard in the field for two decades. His original paper on the topic is still one of the most cited papers in the medical literature. Later in his career, Dr. Duncan adopted a Bayesian approach perspective, creating the k-ratio methodology that is also widely used today.
Dr. Duncan retired to Carmel, California in 1984, where he continued both his work on multiple comparison problems and his efforts to defeat his wife Mary Ann at tennis. He passed away, a few days short of his 90th birthday, on June 12, 2006.