On Monday evening, Stephen Stigler, the Ernest DeWitt Burton Distinguished Service Professor in the Department of Statistics, discussed his book The Seven Pillars of Statistical Wisdom at the Seminary Co-op Bookstore.
In the book, Stigler proposes seven foundational principles of statistics that he argues differentiate the discipline from computer science and mathematics. Stigler’s pillars are aggregation, information measurement, likelihood, intercomparison, regression, experimental design, and the residual. In his talk, Stigler summarized several selected principles, using stories to illustrate their practical applications.
“Ten years ago, after I’d seen David Letterman go through one of his top 10s, I thought, ‘What’s the top 10 in statistics? What are the top 10 ideas?’…It was an attempt to try to explain what is the core of statistics as a science. What do statisticians do that is different from computer scientists or from mathematicians or from economists?” Stigler said.
One of the pillars Stigler explained was experimental design, which underscores the importance of using many different treatments assigned randomly to many different subjects to best identify the effects of each treatment. To illustrate the importance of randomization, Stigler used an example from the comic book Master of Kung Fu, which his son loved as a child. In one scene, the Master chooses one of his girlfriends’ records to listen to at random, remarking, “I’ve selected one from many by chance. It is refreshing to be free of bias.”
“It turns out that randomization is magical in a way that most people don’t even realize. It validates an inference. It gives you a way of producing an ironclad measure on your conclusions that is not dependent upon some modeling or assumptions about the material you’re working with,” Stigler said.
Stigler also spoke about the importance of statistics as a field, which he emphasized with a short story by Jorge Luis Borges, “Funes the Memorious.” In the story, a man’s mind changes drastically after he falls off a horse, giving him the ability to remember everything about his life in vivid detail. Yet because of his injuries, the man loses his ability to think abstractly and cannot meaningfully reflect on these memories.
“Funes is big data without statistics. Gathering data is not enough. It can be very good, but it’s not enough. You need a disciplined way of analyzing and guarding against misleading impressions,” Stigler said.