Remembering George Box: We Are Very Fortunate that George Box Chose Statistics as His Life’s Work!
By Ronald D. Snee
Snee Associates, LLC
I first met George Box at a Princeton Conference in mid-1960s. I was introduced to George by my major professor, Horace Andrews. Horace was a graduate student at North Carolina State when George had been a visiting professor there in the 1950s. He and George had become good friends over the years. At the urging of Andrews I read everything George wrote and attended his presentations at every opportunity.
My professional relationship with George grew over the years. He invited me to the University of Wisconsin to give some seminars. He was my featured speaker at the Gordon Research Conference on Statistics in Chemistry and Chemical Engineering that I chaired in 1978. I produced his filming as part of the American Statistical Associations Filming of Distinguished Statisticians in 1982 using the DuPont filming facilities.
George was clever, witty, a deep thinker, and spoke and wrote with great clarity. He was a great storyteller and enjoyed performing. He was great at capturing the essence of ideas as evidenced by famous quotes: "All models are wrong, but some are useful," and "If you want to know what happens to a process when you interfere with it, you have to interfere with it, not passively observe it."
He was a practicing statistician as well as world-class researcher. I had the privilege of editing Volume Four of his collected works, which focused on his work on applications of statistics. It was an enlightening and stimulating experience. I saw George iterating between practice and theory as he solved problems. I marveled at his clear and interesting writing style. He made extensive use of graphics and schematics of all types. I saw George, the talented scientist, as well as George, the statistician, at work. In one article he commented that "statisticians must learn to be good scientists, a talent which, I think, has to be learned by example."
As I reflect on George’s work I see the influence it had had on my own statistical practice. Briefly stated, I design screening experiments using the methods of Box and Hunter; conduct response surface optimization studies using the designs of Box and Wilson, Box and Hunter, and Box and Behnken; do modeling using the methods of Box and Cox and Box and Tidwell; analyze time series using the models of Box and Jenkins; and develop growth curve and object shape analysis procedures using George’s 1950 paper on growth curve analysis. These techniques are also widely used by others working in industry and other areas of application. Statisticians, scientists, engineers and other professionals are indeed fortunate that George Box chose statistics as his life’s work!