Proofiness - Charles Seife [44]
What went wrong? It’s simple: somebody’s lying. Either the men are exaggerating their exploits, or the women are downplaying theirs—or both. There’s no doubt. Even in a clinical environment, even when the pollsters ensure complete confidentiality, even when there’s absolutely nothing to gain by lying, people always lie about the number of sexual partners they’ve had. It happens over and over again. In France, in England . . . they tell fibs about their sexual history. 42 No matter how the poll is run, you’re guaranteed to find out that the data are wrong, because people are habitual liars.
People lie because, on some level, they want to be liked and respected by everyone—including the person who’s asking the questions in a poll. Even if they’re never going to see the person asking the questions again, subjects tends to give answers that (they think) put them in a favorable light. It’s not just on matters sexual. People try to project the noblest image they possibly can to a pollster, whether or not that image is grounded in reality.
For example, less than a week after Hurricane Katrina devastated New Orleans, the Associated Press, teaming up with the polling company Ipsos, sought a variety of opinions from Americans about the disaster. One interesting tidbit: more than two-thirds of Americans claimed to have already donated money to Katrina relief efforts. Conservatively, this would mean that about sixty million households had donated money—all within a week of the disaster. This, naturally, would mean that an enormous amount of money would quickly flow to the disaster victims.
How much money? It’s hard to say for sure. The Salvation Army reported at the time that its average Katrina-relief donation was around two hundred dollars. That’s probably the typical size of a Katrina donation, but to be on the safe side, let’s assume that other relief organizations didn’t do as well, and that the average donation was much smaller than what the Salvation Army saw—say, fifty dollars. So if we have sixty million households, each donating fifty dollars, that would mean a total donation of a whopping three billion dollars.
Americans can definitely be generous in times of crisis, but they weren’t the selfless philanthropists that the poll would imply. At the time the poll was conducted, the total amount of money raised for Katrina victims was about six hundred million dollars—around one-fifth of what you’d be led to expect from the poll numbers. If you take into account the fact that much of that six hundred million came from corporations and not from individuals, the mismatch is even worse. The conclusion is obvious: there’s no way that two-thirds of Americans had truly donated money in the week after the disaster. It’s a Potemkin number. The truth is probably less than half of what the poll reported. Of course, this means that the majority of the people who said that they had donated were probably lying. People lying is a systematic error, not a statistical one, so the 3 percent margin of error (dutifully reported by the news media) was more or less irrelevant to the accuracy of the poll.
Sometimes the reasons for lying go beyond simple self-aggrandizement. Sometimes there’s outright malignant deceit. People occasionally attempt to game a poll for their own benefit. Early in the 1996 Republican presidential primaries, exit polls—polls performed upon voters just as they leave the site where they cast their vote—showed that long-shot far-right candidate Pat Buchanan had beaten the more mainstream Bob Dole in Arizona. Pundits on the networks immediately started chatting about how Buchanan was a credible challenge to Dole. But when the returns—the actual count of ballots cast—came in, Dole had beaten Buchanan by more than two percentage points. Why? According to CNN’s political director,