George E P Box (1919-2013)
Last Thursday (28 March 2013), George Box passed away at the age of 93. He was one of the great statisticians of the last 100 years, and leaves an astonishingly diverse legacy.
When I teach forecasting to my second year commerce students, we cover Box-Cox transformations, Box-Pierce and Ljung-Box tests, and Box-Jenkins modelling, and my students wonder if it is the same Box in all cases. It is. And we don’t even go near his work on response surface modelling, design of experiments, quality control or random number generation. Occasionally, a student wonders if boxplots are also due to GEP Box, but they were the brainchild of his good friend John W Tukey.
I often quote Box’s famous words to my students “All models are wrong but some are useful” (Box, GEP, 1979, Robustness in the strategy of scientific model building, _Robustness in Statistics, _Academic Press, pp.201-236.) This summarises my view of statistical modelling perfectly — no-one should believe their models; instead, treat them as tools to be used to assist in understanding the data.
I only met Box once, at a conference about 10 or 12 years ago, when he was in his early 80s. But I’ve worked in his shadow all my career, and have admired his ability to move between academic research and industrial consulting with ease. Tukey had a similar twin career as an academic and a consultant, and they both have inspired me to attempt the same.
Box was son-in-law to RA Fisher – he married Joan Fisher, the second of RA Fisher’s five daughters. Thus, he was a link back to the foundations of modern statistics.
There is an interesting interview with George Box, conducted by Daniel Peña, that was published in the International Journal of Forecasting in 2001. I have attached it here for those who do not have access to IJF papers. In addition, several people have written reminiscences of Box in the last few days. Here are two that I enjoyed:comments powered by Disqus