Naked Statistics

Every college student should read Naked Statistics. Actually, everyone who does not regularly use statistics in their work should read it. The author, Charles Wheelan, does a marvelous job conveying the key elements of quantitative reasoning that everyone should be able to use when confronted with a news headline claiming some connection between two or more things. Even better, the book is breezy reading and highly entertaining; it’s subtitle is “stripping the dread from the data” hence Naked Statistics. Interesting vignettes and helpful examples are aplenty, and Wheelan’s sense of humor is delightful – there are running gags throughout the book.

After going through the bare basics, Wheelan does a masterful job with the Central Limit Theorem in chapter 8 of 13. His emphasis on the importance of drawing good representative samples litters the book highlighting the power of statistics when done well and the dangers when done poorly. Each chapter builds on the ones before, and I can see Wheelan employing the strategies of an effective teacher as he guides the reader through the material. The meat comes after the Central Limit Theorem and the book subsequently covers inference, polling, regression analysis, and program evaluation. The examples are always interesting and enlightening, and a strength of the book is how Wheelan connects theory to application.

While there are examples in sports, economics, medicine, governance, and more, there are also several related to education. The chapter on program evaluation is starts by asking the question “How would going to Harvard affect your life?” But evaluating it is tricky. “Well, to answer that question, we have to know what happens to you after you go to Harvard – and what happens to you after you don’t go to Harvard. Obviously we can’t have data on both.” Unless you can replay the tape or you’re able to access the multiverse where every choice creates an additional alternate reality bubble.

Instead of jumping straight into the question of whether going to an “elite” school impacts your life in some significant way, Wheelan pivots to another question to illustrate key concepts. “Does putting more police offers on the street deter crime?” You’ll have to read the book to see the connections! In any case, Wheelan systematically goes through counterfactual strategies one can use when randomized controlled experiments cannot be carried out. Harvard wouldn’t agree to participate on having random students, and students (and their parents) would likely not want to be randomly assigned not to attend Harvard. This is one of the best parts of the book – I was impressed by the ingenuity of researchers trying to get at a difficult question as carefully as possible amidst complicated and confounding factors.

In the conclusion, Wheelan poses five big questions (no, it’s not an exhaustive list) that statistics can and should help answer. One is education related: “How can we identify and reward good teachers and schools?” This turns out not to be an easy question to answer despite the punditry that abounds on this topic. I read a lot of education news so I’m exposed to it on a very regular basis. Wheelan brings up the very interesting study by Carrell and West at the Air Force Academy that attempts to answer the question of which professors are the most effective. It makes use of the “natural experiment” where students are randomly assigned to introductory calculus sections every year. Syllabi and exams are similar across sections. The results are interesting and oft-quoted. (I won’t tell you here because I’m encouraging you to read the book!)

I wonder if the magical education world of Harry Potter has similar issues. Does a Hogwarts education make a difference? Is there an advantage to completing your education? (After all the Weasley twins seem to have done well without “graduating”.) Is there a good reason to send your budding wizard-of-a-kid to Hogwarts rather than Durmstrang? Does Durmstrang really churn out more “bad” wizards? Statistics could likely answer such questions. Come to think of it, maybe I could use some of these as examples in my classes, now that “quantitative reasoning” is a new core requirement at my institution. It would be fun to make up some of the data!

P.S. I’ve also read Teaching Naked. I wonder if there’s a new trend. Maybe zombies are finally starting to go out of fashion?