As a graduate student and editor of the campus humor magazine Chaparral in 1961, Bradley Efron was suspended briefly for publishing a parody of Playboy magazine. These days, his writing is of a different sort: a monthly column for the professional journal of the 17,000-member American Statistical Association, of which he is president. Efron, MS ’62, PhD ’64, is professor of statistics and of health research and policy at Stanford.
Newton was no statistician. Consider the discovery of the law of gravity. Newton was sitting under a tree one day and an apple fell and hit him on the head. That’s not statistical evidence: “Newton just said, ‘A-ha.’” However, if he had seen 40 apples shoot up in the air and 60 apples fall on the ground, that would have been the start of a statistical argument. “The essence of a statistical argument is evidence that arrives a little bit at a time.”
Don’t believe that stuff about “lies, damned lies and statistics.” Statisticians are trained not to lie to themselves or others, Efron says. And it’s a good thing, too. Adrift in a sea of numbers, “it’s awfully easy to fool yourself, where no one number is decisive. It’s very easy to start pushing things in a direction that you want them to go.”
Little boys don’t dream of becoming statisticians. Efron’s father was a truck driver and a salesman, and the three-cushion billiard champion of St. Paul, Minn. “He also loved amateur math, and when I was a little kid he taught me how to add numbers in my head, stuff like that.”
They dream of becoming mathematicians. Efron wanted to be a mathematician, but as an undergrad at Caltech he found he wasn’t very good at modern mathematics. “It was much more axiomatic than traditional mathematics—almost an aesthetic field—and maybe my aesthetic sense is a little wanting.” Statistics, by comparison, turned out to be a good field for a person with a “short attention span,” as Efron describes himself. “One of the charms of statistics is that you can peek in on other fields and see how things are going. I work with people in astronomy, and I help doctors analyze their data and plan experiments.”
My, but that’s a big cathode ray tube you’ve got. Efron’s UNIX server can do in a second what he used to do in a year. That’s good, since there’s been a complete change in scale in the past 40 years. When Efron started out, he was working with 20 numbers per study. In his latest collaboration with a researcher at the Medical School, he’s looking at 20,000 genes for each of 88 mice in an atherosclerosis study. Because earlier procedures weren’t designed for massive data sets, with thousands of questions being asked at once, today’s statisticians are “trying to figure out new theories, when we’re not consulting for somebody else,” Efron says.
The chances of having a bike accident on campus are high. And that’s not just anecdotal. Efron doesn’t want to get hit, and he keeps statistics on close calls. “One thing statistics show quite clearly is that I’m in much more danger from bicyclists than from cars. It’s terrible at night because students have no fear at all and no lights at all. I can’t believe they live to be adults, and some of them wouldn’t if I had my choice.”
