Book Reviews

Lost in a Gallup

A narrative history of US pollsters, journalists, and politics sheds light and context on a struggle more than a century old.

Analyst Recommendation: Neutral.

The “pollsters were wrong” post-election finger-pointing has grown tiresome. People act as if it were a revelation. But pollster mischief, wide inaccuracies, questionable methodologies and outlandish predictions of all kinds are at least 100 years old—polling was fraught from the beginning.

That perspective is the value of W. Joseph Campbell’s Lost in a Gallup (2020). The book begins with five consecutive quotations that sum up the whole affair nicely, but none better than, “The only crime in this business is to be on the wrong side” of a prediction, from George H. Gallup himself. The ersatz PT Barnum of the industry, Gallup was a pioneer, practitioner and vocal pundit of all things polling. What his quotation presupposes, though, is that pollsters are predictors. After all, what’s the point of a poll if not to forecast the outcome of, say, a vote? But in their more defensive postures, most pollsters will insist they aren’t explicitly predicting—and that (what’s that thing we always say in the investing world?) past performance isn’t a predictor of future results.

The topic of polling through the years can be quite dull, frankly. But Campbell is deft in not veering too far into the many theories—statistical, methodological and behavioral—behind polling. Instead, he offers a history of pollsters clashing through the ages with journalists and politicians alike. It’s right in Campbell’s wheelhouse: He’s made a long career of journalism about journalism. His indispensable work on yellow journalism (of the early 20th century) is vital for dissecting today’s media. It’s not about left or right bias—you have both. It’s about the structure of the business. Campbell gets it, and he takes that sensibility to his analysis of pollsters in this well-crafted book.

There’s a kind of guru-ism to all this—we’ve seen it most recently with folks like Nate Silver and Scott Rassmussen. It parallels the hunger for Oracles in the investing world—both polling and stock-picking being crystal ball professions. With investing, anyone claiming to be right all the time is selling you a world you’re wishing to see, not reality. And while we’re fairly ruthless when our chosen stock soothsayer is wrong, we seem just as, if not more, vicious toward pollsters with wrong predictions, no? In close elections, perhaps the foolish are those who put faith in the polls to start.

There’s a bigger point here about our collective gullibility—both in polling and investing. Just how easily we can convince ourselves to believe in all these obtuse and obfuscated “scientific” methods. As if any of this is really more than a stab in the dark toward sorting out what happens next in a wildly complex and endlessly adaptive world, catalyzed by emotionally charged human behavior. The “quant” aspect of polling models masquerades as rigor, but such quant models are a dime a dozen in the world of investing. And most don’t work.

Sentiment, psychology and proper context will always matter as much as method and data. That’s because we’re dealing with futurity. The future is uncertain, but people’s psychological reaction to the future, particularly in the context of market behavior, is something that rhymes in roughly cyclical fashion—with a ton of noise, of course. Polling is not the same as pricing, but the psychological gullibilities we fall for again and again do have their parallels.

Will Campbell’s book put to bed the cult of pollsters for good? Not a chance. Oracles are as old and archetypal as anything in civilization. Only the methodology has changed. Polls will persist because there is information and utility in them, but the onus is still on us all to realize that +/- margin of error cited at the bottom of every poll is at least as relevant as the conclusion itself. Just like any very official-looking data analysis your local stock analyst pushes on you, polling is no panacea, and never was.