Another delightful addition to the stuff-you-think-you-know-that’s-wrong genre, á la Freakonomics, Outliers, and The Black Swan.
Baseball players who have a good year usually do worse the following year, but bad players often improve. So the good players are slacking off, and the failures are trying harder, right? Wrong. It’s a mathematical law no different from the Pythagorean theorem. Everyone knows why brilliant men marry less-than-brilliant women, but why do brilliant women tend to choose less-intelligent men? Unlike Malcolm Gladwell, Steven Levitt, or Stephen Dubner, all of whom range over many topics, Smith (Economics/Pomona Coll.; Standard Deviations: Flawed Assumptions, Tortured Data, and Other Ways to Lie with Statistics, 2014, etc.) covers one big concept that almost everyone gets wrong: regression to the mean. Put simply, it means that for any action in which chance plays a role—e.g., taking a test, playing a game, running a business, governing, research—things even out. If you get an extreme result, the next result will probably drift toward the average. A distressing example: medical journals regularly publish studies of effective treatments that, once approved, don’t work as well (mammograms, antidepressants) and sometimes not at all (arthroscopic knee surgery). Eliminating chance from research is difficult. Journals prefer studies that find something. Scientists yearn for a breakthrough, so, if results aren’t impressive, they look again—and again. “If you torture the data long enough,” said British economist and author Ronald Coase, “it will confess.” Smith argues his case with more graphs, studies, examples, and math than necessary, and the end result may leave readers in despair—convinced that almost everyone, experts included, believes an important piece of nonsense and will continue to believe—unless they read this book.
A welcome, widely applicable follow-up to the author’s equally useful first book.