Theodore W. Anderson, a renowned scholar in mathematical statistics and econometrics at Stanford, dies at 98

Anderson, who taught at Stanford for 21 years before retiring in 1988, remained a visible presence on campus and continued to engage with and inspire younger faculty and students.

Stanford Professor Emeritus Theodore W. “Ted” Anderson, a towering intellectual figure in the world of mathematical statistics and econometrics, and a “prophet” of the era of big data, died of heart failure on Sept. 17 at Stanford Hospital. He was 98.

Anderson, who retired from teaching in 1988, continued his work and close association with Stanford colleagues. Until recently, he regularly attended campus seminars. He remained a visible presence at the Economics Department’s social gatherings, and continued to engage with and inspire younger faculty and students.

1987 photo of Theodore W. Anderson

Theodore W. “Ted” Anderson, 1918-2016 (Image credit: Courtesy Stanford Department of Statistics)

“Amazingly, Ted submitted his last technical paper less than a month ago,” said Emmanuel Candès, chair of the Department of Statistics at Stanford, who said Anderson has been a towering intellectual figure in the department for nearly 50 years.

“His wide-ranging contributions to statistics and econometrics have had an enormous influence on the field and a lasting value. His 1958 textbook, An Introduction to Multivariate Statistical Analysis, educated a generation of theorists and applied statisticians, rapidly assuming the status of a classic with its extraordinary impact.”

Anderson’s published papers in mathematical statistics as well as econometrics are famous for their rigor, their attention to detail and a clarity of exposition that makes difficult things seem easy, Candès said.

In 1990, Wiley published The Collected Papers of T. W. Anderson: 1943-1985, a two-volume set comprising 109 papers and 16 commentaries, edited by George P. H. Styan.

David L. Donoho, a professor of statistics at Stanford, said Anderson pioneered and systematized the science of using high-dimensional data to detect, characterize, classify and predict phenomena subject to multiple measurements.

“Among other things, I would say that Ted was a prophet of the big data era, where multiple measurements on a person’s behavior or characteristics, such as weight, height, age, IQ, GPA, could be used to predict other characteristics or properties,” he said.

“Today this sort of multivariate analysis is routinely used in all sorts of scientific studies. In everyday life we now see it used in consumer credit scoring, in prediction of survival probabilities in breast and prostate cancer, in fraud detection and in identifying risky portfolios of investments.”

John B. Taylor, a professor of economics at Stanford who wrote his doctoral dissertation under Anderson’s tutelage, recalled his “crystal clear, rigorous and insightful lectures, his intellectual leadership in the department’s econometrics seminar and the genuine personal interest he took in students.”

“Ted instilled a unique degree of rigor to econometric research through his influential book The Statistical Analysis of Time Series and the attention to mathematical clarity and precision he shared in his interactions with his students and colleagues,” Taylor said.

In statistics, his name is associated with the Anderson-Darling test of fit, which is widely used in a range of applications, the Anderson-Bahadur algorithm and the Anderson-Stephens statistic for data on a sphere.

Cheng Hsiao, a professor of economics at the University of Southern California, characterized Anderson’s influence in econometrics “as no less than extraordinary.”

“Ted pioneered the limited information approach to estimate complex economic systems, laid the rigorous foundation for inference with nonstationary time series models and interactive effects for panel data models, etcetera,” Hsiao said. “His name is associated with the Anderson-Rubin test of overidentification in simulataneous equations models, the Anderson-Taylor multiperiod least squares control of linear input-output models and the Anderson-Hsiao estimate in dynamic panel data models.”

Colleagues also pointed to the so-called Anderson’s Lemma as a frequently used tool in the theory of high-dimensional probability.

In 2008, Stanford held a two-day conference to celebrate Anderson’s 90th birthday. The invited speakers included many of his former students, co-authors and colleagues, including Stanford Professor Emeritus Kenneth Arrow, 1972 Nobel laureate in economics. In presenting their research, the speakers discussed Anderson’s fundamental contributions and the overarching influence of his early work.

That same year, at the 17th International Workshop on Matrices and Statistics, held in Tomar, Portugal, colleagues honored Anderson’s long and distinguished career with an address based on a 12-page annotated and illustrated bibliography of his work.

Anderson was elected to the National Academy of Sciences, as well as the American Academy of Arts and Sciences. He was a Fellow of the American Statistical Association, the Royal Statistical Society, the Econometric Society, the Institute of Mathematical Statistics and the American Association for the Advancement of Science.

Anderson, who was born June 5, 1918, in Minneapolis, earned a PhD in mathematics at Princeton University in 1945. After graduating, he spent a year at the University of Chicago working as a research assistant on the Cowles Commission for Research in Economics, which was established to facilitate the economic transition from wartime to peacetime. In 1946 he joined the faculty of Columbia University, where he taught until 1967, when he began his Stanford career.

He enjoyed many activities outside his professional life, including international travel, classical music, opera and fine arts, tennis, reading and family gatherings. He also enjoyed dancing with his wife, Dorothy, his love of 70 years. They met in 1946 when Dorothy was a graduate student in social work at the University of Chicago, and married in 1950.

Anderson’s family is discussing plans for a gathering of remembrance. Colleagues also are organizing a memorial for Anderson at the next Joint Statistical Meetings, which will take place in Baltimore next summer.

In addition to wife, Dorothy, of Stanford, Anderson is survived by son Robert L. “Bob” Anderson of Stanford; daughter Janet Anderson Yang and her husband, Thomas Yang, and their children, Nathan and Clara, of Los Angeles; and daughter Jeanne Anderson and her partner, Randall Bachman, and her children, Kenton, Michael and Johnathon, of Anchorage.