George E. P. Box

Box was born in Gravesend, Kent, in 1919. He studied chemistry at University College London, but his education was interrupted by World War II. During the war, he conducted experiments for the British Army, testing the effects of poison gases on animals. This experience sparked his interest in statistics, as he realized the importance of proper experimental design. After the war, he completed a degree in statistics at University College London and later earned a Ph.D. from the University of London.¹

Box held professorships at the University of North Carolina, Princeton University, and the University of Wisconsin–Madison, where he founded the Department of Statistics in 1970. His research spanned experimental design, response surface methodology, and time series analysis. With Gwilym Jenkins, he developed the Box–Jenkins method for ARIMA models, which became a standard tool for forecasting. He also introduced the Box–Cox transformation to stabilize variance and make data more normal, and the Box–Behnken design for response surface optimization.

Box was a leading figure in the statistical quality movement, collaborating with engineers at Imperial Chemical Industries (ICI) and later with Japanese industry. He emphasized the importance of sequential experimentation and the use of simple, robust designs. His book Statistics for Experimenters (with William Hunter and Stuart Hunter) remains a classic. He also developed the concept of evolutionary operation (EVOP), a method for continuously improving industrial processes.²

Beyond his famous methods, Box made many subtle contributions. He introduced the concept of “model checking” through residual analysis, and his work on Bayesian inference in experimental design was ahead of its time. He also coined the term “robustness” in statistics. Box was an accomplished musician and painter, and he often used analogies from art to explain statistical concepts. His aphorism “all models are wrong, but some are useful” is widely quoted, but he also said, “The only way to find out what will happen when a complex system is disturbed is to disturb it, not merely to observe it passively.”³

ARIMA

Autoregressive Integrated Moving Average, a class of models for time series forecasting.

Box–Cox transformation

A power transformation used to make data more normally distributed.

Box–Behnken design

A class of response surface designs with fewer runs than full factorial designs.

Evolutionary operation (EVOP)

A method for continuously improving industrial processes by making small, planned changes.