Dan Gardner is the New York Times best-selling author of Risk, Future Babble, Superforecasting (co-authored with Philip E. Tetlock), and How Big Things Get Done (co-authored with Bent Flyvbjerg). His books have been published in 26 countries and 20 languages. Prior to becoming an author, Gardner was an award-winning investigative journalist. More >

Statisticians gone wild

Statisticians gone wildTo turn statistical methodology into a political controversy, a government has to really screw up. But to make statisticians shriek and flap their arms like wounded albatrosses, to cause policy wonks to turn purple with rage, to compel retired civil servants to dispense with a lifetime of discretion and denounce the government's gobsmacking jackassery to reporters ... Well, that's something special. Now, personally, I adore discussions of statistical methodology. Can't get enough. And so, I must admit, I was a little tickled when Industry Minister Tony Clement ordered Statistics Canada to change the census without the slightest public consultation. Henceforth, while Canadians will still be required by law to complete the short census form, the long form, previously sent to one-fifth of households, will be voluntary. The response rate will decline, of course, and so the minister also directed that the long form be sent to one-third of households to compensate. But as I explained in a column last week -- if only I could discuss statistical methodology every week! -- the switch from a mandatory to a voluntary form will bias the data in many ways and increasing the number of households that get the long form won't correct the biases. It will just produce more numbers. That are biased. And not comparable with past census data. Oh, and because more forms will be sent out, the cost of the census will go up substantially. Back when Tony Clement was in the Mike Harris government, they liked to use the slogan "Doing more with less!" This is like that. Except it's "Doing less with more!" In the past, I've been sent the occasional e-mail by statisticians and they're usually what you would expect an e-mail from a statistician to be. Cautious, precise, understated. Unemotional. Spock on an especially dull day. But after my column last week, I got e-mails from statisticians who were angry at the government. Livid. Furious. Hissing, spitting, eyes-bulging, hands-clutching-for-an-invisible-throat. Very angry. This surprised me. I bet it surprised the government. Then the campaigns and petitions began. A long, long list of organizations wrote to formally protest the government's "misguided decision" -- that's the phrase polite people use instead of "jackassery" -- and demand its repeal. And these weren't the usual pothead Marxists. It was the Statistical Society of Canada. The Federation of Canadian Municipalities. The Canadian Marketing Association. The Canadian Association for Business Economics. I doubt the membership of the Canadian Association for Business Economics spends a lot of time reading Das Kapital and taking bong hits. Then the Earth shook. The change to the census will produce "seriously biased" data, the legendary statistician Ivan Fellegi told this newspaper. It is "indefensible." Coming from a man who spent half a century at Statistics Canada, including 23 years as Chief Statistician, this was rather like Moses returning from the mountain and explaining to the wayward Israelites that, no, you can't worship a golden calf, you idiots. To be fair, not everybody opposed the government's decision. Clement was stoutly supported by an assortment of anonymous blog posters. And by Dean Del Mastro, a Conservative MP. On CBC Radio the other day, following a lengthy interview with Ivan Fellegi -- who patiently explained why it was a bad idea to worship golden calves -- Del Mastro defended the government's decision. Ivan Fellegi is wrong, Del Mastro said. And he should know. Why, he has taken "a number" of courses on statistics. Del Mastro did not mention that his crowning educational achievement is a Bachelor of Commerce degree from the University of Windsor. One assumes he was being modest. At this point, I expected the government would quietly rescind its decision. No harm, no foul. And everybody stops talking about statistical methodology, unfortunately. I should have known better. Remember when Senator Nancy Ruth was blasted for telling women's groups they should "shut the f--- up" about the Harper government's stand on foreign aid and abortion? Ruth was painted as a Conservative bully. That was wrong. Ruth was actually sympathetic to the women's groups and she was warning them that if they kept protesting the Harper government would dig in and get really nasty. And she was right. That's how this government operates. Facing an army of angry PhDs, Clement actually fought a Twitter battle about the change with economist Stephen Gordon. Needless to say, Gordon won, a fact confirmed when Fellegi came down from the mountain and smacked Clement with his stone tablets. The government's response? On Tuesday, Industry Canada issued a press release. "The government does not think it is appropriate to force Canadians to divulge detailed personal information under threat of prosecution," Clement says in the release. "To meet the need for additional information, and to respect the privacy wishes of Canada," the government has made the long form voluntary. "To promote data accuracy, this voluntary survey will be sent to a larger cross-section of households than the old long-form census." As news reports had amply demonstrated, there is no evidence of widespread popular concern about coercion, privacy, and the census -- sensibly enough, as StatsCan is absolutely obsessive about protecting privacy. Further, the expanded reach of the voluntary survey will do nothing to correct the data for bias -- as statisticians had been explaining over and over to anyone who would listen. By repeating these empty claims without the slightest acknowledgement of what the critics had been saying, the minister was sticking his fingers in his ears while loudly humming Rule Britannia. It was a gesture of contempt. "I can't hear you!" Clement mocked. "I can't hear you!" The same day, in The Globe and Mail, Bill Robson, president of the C.D. Howe Institute, gently agreed that changing the census is a mistake but he worried that "the reaction from many opponents risks cementing the government's resolve." Bill's a gentleman who would never approve of potty mouth but that sounds an awful lot like Nancy Ruth warning women's groups to "shut the f--- up" because they're dealing with a pack of vindictive knuckleheads. Which is fine with me. Let the government's resolve to do the unspeakably stupid be cemented, I say. Sure it will waste money, hurt public policy, hamper business, and make us increasingly ignorant even as information becomes increasingly valuable. But I'll have lots more chances to write about statistical methodology. Thanks, Tony. Keep tweeting.