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 >

Fun With Crime Numbers

Be afraid, Canadians. Crime has gone up so rapidly that the rate of assault in this country is now "more than double that of the United States." So is the rate of sexual assault. And even the rate of overall violent crime. Fear, panic, chaos: Canada is experiencing nothing less than "the sharpest rise in violent crime in the nation's history" -- according to a press release issued in January by Alan Gottlieb, the head of an American lobby group called the Citizens Committee for the Right to Keep and Bear Arms. Not a word of this is true, mind you, although in American conservative circles similar claims about crime in Canada have become routine. Sadly, it's not concern for the safety of their Canadian cousins that motivates people like Mr. Gottlieb; it's American politics. The point of this propaganda is to demonstrate the folly of liberal policies and to bolster support for the traditional conservative approach of a cop on every street, a gun in every purse and a black man in every cell. Canada is just the straw man they use for target practice. If this stuff stayed in the United States, it would be amusing. But like guns and cocaine, it is slipping across the border. "Canada's overall crime rate is now 50 per cent higher than the crime rate in the United States," wrote David Frum in the National Post. "Read that again slowly -- it seems incredible but it's true." Actually, it's false no matter how slowly you read it. Chris Selley, one of the best bloggers in Canada, asked Mr. Frum how he came to his rather shocking conclusion. Mr. Frum responded that he had compared data from the U.S. Bureau of Justice Statistics and Statistics Canada. That may sound like a reasonable thing to do, but it's not. The two agencies gather their numbers using different definitions and different methods and that makes them apples and broccoli. They cannot be compared. David Frum's conclusion looks absurd because it is absurd. (Mr. Selley's detailed look at this can be found on his website, www.tartcider.com/blog .) I don't want to be too hard on Mr. Frum. People who aren't familiar with crime statistics make this mistake all the time. Mr. Gottlieb did the same to come up with his scary numbers. The unfortunate reality is it's difficult to accurately compare crime in different countries. There are really only two ways it can be done. First, you can carefully analyse the definitions and methods used to generate data to ensure you only compare apples to apples and broccoli to broccoli. And yes, this is as complicated, laborious and boring as it sounds. The other method is to conduct a survey in multiple countries -- using consistent definitions and methods -- asking people if they have been victims of crime in the last year. Both these methods have been used to compare Canada to the United States. StatsCan used the first method to compare police-reported crime in the two countries between 1980 and 2000. It found "over the past 20 years, Canada recorded much lower rates of violent crime than the United States did. However, rates for property offences have generally been higher in Canada." As for crime trends, there's remarkably little difference in the two countries: "After peaking in 1991, rates for both violent and property crime generally declined throughout the 1990s." Incidentally, the parallel track Canadian and American crime trends have taken for so long continues to this day. Since 2000, there has been no clear trend up or down in either country. Next up: the survey method of comparison. The United Nations International Crime Victim Surveys (ICVS) are the gold standard. The last ICVS was released in 2000 but another is due in March. The 2000 ICVS found that 23.8 per cent of Canadians said they had been victims of crime in the previous year. Most of those incidents involved minor property crime. In the United States, the victimization rate was 21.1 per cent. The average for the 17 countries surveyed was also 21 per cent. Don't get excited about the gap between the American and Canadian results. The sample sizes for the ICVS vary from country to country and they are generally small, which means the margin of error is substantial. Small differences don't mean much. What's more interesting about the ICVS results are the trends they reveal over time. Canada's victimization rate peaked in 1992 at 28.1 per cent. It dropped steadily until 2000. The American rate peaked in 1989 at 28.4 per cent and then it, too, dropped steadily until 2000. So once again, the crime trends in both countries are similar. The ICVS also asked people if they feel safe walking alone in their neighbourhoods after dark. Eighty-three per cent of Americans said they do. The Canadian result? Precisely the same. Of course, none of this will matter to people who play politics with crime numbers. When I asked Mr. Gottlieb for his sources, he directed me to a StatsCan report that notes on its very first page that in 2003 "the national homicide rate fell seven per cent to its lowest level since 1967" and that "the 2003 (overall) crime rate was 15 per cent lower than a decade ago." That's the way zealots work: Pick out the useful bits and pretend you didn't see the rest. But for the rest of us, it's nice to know there's no reason to hide under the covers.