Andrew Spicer's Weblog - Current - Index - Email
Index for my Toronto Budget 2004 project
National Post Gets Weak on Statistics, Too

Yesterday, when I wrote about the National Post's editorial smear of David Miller's position on crime I felt there was still more to the story. There was something about the statistics used in the Post's article that had me wondering.

They begin with "this past weekend at least four people were killed by criminals in Toronto". At the time, I felt that this one was misleading. The article was written about the Mayor of the City of Toronto, but the statistics were including a fourth murder that occured outside the city limits. However, I have since learned that a Toronto weekend stabbing victim died Monday at noon. Since I only read the Post's editorial online, I don't know if it was published before or after this 4th City murder.

The first major quibble I have is with this line: "overall, violent crime in Toronto is up 5% since 1999". This claim is technically true, but certainly misleading. When you review Toronto police statistical reports, you find that the trend reveals something different.

Looking at the number of violent crimes between 1993 and 2002 -- the years that have easily accessible data -- I found that the average was 34,968. 1999 was the second lowest year in that span. After 1999, crime rates climbed 2000-2002 before dropping in 2003. (I don't know the 2003 figure, but "up 5% since 1999" is 34,777.) Looking at standard deviations, 1999 is the unusual year. 2003 looks very close to (and perhaps below, depending on the exact number) the mean.

The Post's intention in using these statistics was to show an alarming spike in violent crime in Toronto. The real numbers show something different. The Post's technique is akin to mutual fund managers telling us how much their fund has risen over a paticular number of years -- neglecting to mention that they are comparing vs. the bottom of the trough.

Just for your information, the mean number of murders in Toronto between 1991 and 2002 was 61.9. In 2003 there were 65. 1999 again happened to be an unusual year, with a count dramatically lower than any other.

The last statistical claim that concerned me read "America now has the second-lowest per-capita rate of violent crime in the G7, ahead of only Japan". Does this strike anyone as suspicious?

I've had trouble finding data on this. But, sure enough, this document shows that Interpol claims that the USA has a low crime rate, less than half that in Canada. (2001 numbers: 8,573 crimes per 100,000 Canadians and 4,161 crimes per 100,000 Americans.)

However, if you go to the source on the Interpol website and view the crime statistics for Canada and the United States you will see that those numbers were calculated on a very different basis.

Look at those two documents and compare the "volume of crime per 100,000 inhabitants" column. What you may not notice right away is that the total at the bottom of the column is different from the sum of the individual numbers. It looks like there is separate reporting of "total number of offences contained in national crime statistics". The Canadian total is actually a few times higher than the total of the listed offences, while the American total is simply the sum of "murder", "rape", "serious assault", and "theft".

The closest thing I can get to comparing like numbers puts Canada under 3,000 crimes per 100,000 citizens, vs. the USA's 4,161.

Now, I don't know if the National Post relied on these Interpol numbers, but they do show that Canada had fewer crimes and fewer violent crimes per capita than the United States in 2001. A quick look at the other G7 countries leads me to believe that the United States actually has the highest rate of violent crime among them. Well, this wouldn't be news to anybody if the National Post wasn't reporting the opposite.

The bottom line? The National Post was full of crap in their editorial yesterday, and were just looking to force-fit "facts" into supporting their misguided criticism of David Miller.

POSTSCRIPT: See also Fact Checking the Post: Uh, That Would Be Last, Not Second



spicer index: