Travelers are often advised to avoid certain places when a big holiday comes -- New Orleans on Mardi Gras might be too bacchanalian for you, and midtown Manhattan on New Year's Eve can get awfully crowded. So if you're thinking of visiting Japan, you might want to avoid Oct. 18. It's Statistics Day, and it gets pretty crazy.
OK, maybe not -- to be honest, I have no idea whether the Japanese treat Statistics Day with the solemnity it deserves or as just another excuse to blow off work and barbeque burgers. But the fact that they have an actual holiday whose purpose is celebrating the collection and analysis of data speaks of a culture that values precision and holds numbers in high esteem. When Statistics Day had its 30th anniversary in 2002, the Japanese government sifted through 2,934 entries to choose the winning slogan, "Statistical Surveys Owe You and You Owe Statistical Data." Not exactly "It's the real thing" or "I ♥ New York," but kind of sweet all the same.
If we in America had to pick a slogan for our own view of statistics, it would probably be, "Figures don't lie, but liars can figure." To put it mildly, we are more than a little ambivalent about what numbers can tell us.
We ought to be in a golden age of data. We have more data than we have ever had before, more computing capacity to analyze it, and an information delivery system -- the Internet -- we couldn't have dreamed of 20 years ago. With a few clicks, you can have at your fingertips the mountains of U.S. Census data. You can access the 36 years worth of data gathered by the General Social Survey or the 60 years of data collected by the National Election Studies. You can get oodles of numbers from the Bureau of Labor Statistics or the Organization for Economic Cooperation and Development or the CIA's World Factbook. I could go on -- almost forever.
What's more, the system that delivers all these wonderful data also allows the instructive analyses of experts to reach all of us, lest we misunderstand what the numbers represent. There are a hundred statisticians, economists, and political scientists with their own blogs, and numbers-focused political sites like pollster.com and FiveThirtyEight.com have become quite popular -- among those who care about that sort of thing.
Furthermore, we recently cast off a presidency in which scientific inquiry and the data it produces were treated like postmodern objects, ideas which had no inherent truth but were righteous or blasphemous depending on which political outlook they supported.
So why is it that our collective numerical baloney detectors seem no more sophisticated than they were when people still used abacuses to make calculations? Why is it that misleading or manipulative uses of numbers are no less likely to carry the day for their dishonesty?
We can certainly put part of the blame on the press. How many times in recent years has it treated some bogus figure put out by one side of a political debate as though it might be true, depending on how you look at it? To take just one example, consider the gift that The New York Times offered up to anti-union forces last November, when a now notorious article by Andrew Ross Sorkin claimed that Big Three autoworkers were being paid $70 per hour. It was false -- the average worker was actually making $28 an hour. The $70 figure came from disingenuously combining four separate expenses incurred by the automakers, only one of which is actual wages. But that didn't stop conservative opponents of aid to the automakers from turning factory workers into the villains of the story, a bunch of greedy layabouts sucking the companies dry and driving them to ruin. The truth didn't much matter -- the idea ricocheted around the media for weeks.
The right thing for any reporter to do when confronted with the claim would have been to say, "I'm sorry, Mr. Conservative Think Tank guy, but you and I both know that autoworkers don't make $70 an hour. Is there anything else you'd like to add -- that's not a lie -- that I can use in my story?" But reporters don't necessarily say that sort of thing. And this is just one case. Journalists' lack of even the most rudimentary understanding of statistics is evident on the news pages and broadcasts nearly every day.
We can't put all the blame on the press, though. Our society is full of arenas in which people ignore the fact that there are important questions you can answer empirically, and doing so gives you better results than going with your gut. For instance, Bill James created the first version of his Baseball Abstract in 1977, and the analysis of the copious statistics enshrined in box scores quickly became an underground sensation. Yet it took two decades before a baseball team actually began to use statistical analysis to make key decisions on matters like which players to trade for. (The first team to do so was the Oakland Athletics, as Michael Lewis chronicled in his book Moneyball. James is now an adviser to the Boston Red Sox.) For all that time, a data analysis more sophisticated than counting home runs and RBIs was considered an interesting oddity but not something you could really use to make judgments.
Meanwhile, easily consumed pseudoscience seems to have no more trouble spreading through the land than it did in the days when traveling salesmen peddled tinctures and unguents with magical powers to cure all ills. As a recent Newsweek cover story detailed, Oprah Winfrey regularly brings guests on to her show to peddle all manner of ratings-friendly quackery, whether it's miracle anti-aging cures or absurd assertions about genuine public-health issues. Among them is former Playboy model Jenny McCarthy, who in response to her son's autism has become an advocate for the imaginary links between childhood vaccination and the condition. When confronted with a statement from the Centers for Disease Control saying that there is no scientific evidence of such a link, McCarthy responded, "My science is named Evan, and he's at home. That's my science." Well no -- that's not "science" at all.
At this point, I feel obligated to mention that American students do worse -- sometimes much worse -- on math and science tests than their counterparts from other developed countries (Finland, by the way, apparently is to high school math what Brazil is to soccer). But the mean scores on an international test tell us only so much. We're still turning out more than our share of Nobel Prize winners, after all.
Of course, people who know a lot about numbers don't always make the best decisions, particularly when it comes to predicting the future of complex systems. Witness the story of Long-Term Capital Management, a hedge fund whose personnel included two Nobel Prize-winning economists and a raft of Ph.D. mathematicians. Using intricate statistical models and a complex trading strategy, they generated huge profits for their investors, from the company's founding in 1994 until its disastrous implosion in 1998. And much of our current financial difficulties were caused by people who were clever enough to create inscrutably complicated financial instruments that generated large short-term profits, but too dumb to see that they were constructing a house of cards.
The lesson isn't that one shouldn't listen to people who know a lot about numbers. It's that one has to know which questions can be answered by data and which can't. Too often, too many of us can't tell the difference.