Statistics without care

James Taranto at Best of the Web highlights a USA Today article on police suicide, where it's obvious an uncritical journalist was the one doing the statistical work. Citing an 18 per 100,000 suicide rate for police and a 11.1 per 100,000 suicide rate for the general population, they claim police suffer a higher rate of suicide.

However, Taranto, who isn't a statistician but put some thought into the statistics, tried to do an actual apples-to-apples comparison.

If we assume that the police are a representative sample of the population as a whole (apart from the sex disparity), we would expect a suicide rate of approximately 16.8 per 100,000, not much below the 18 per 100,000 that USA Today reports.

It turns out, though, that the overall nationwide suicide rate is skewed downward because it includes children and teenagers, who have a much lower propensity for suicide than adults. This effect is so pronounced that every age group over 20 has a higher-than-average suicide rate. In particular, the overall suicide rate for 25- to 44-year-old men was 22.2 per 100,000 in 2002, and for 45- to 64-year-old men it was 23.5 per 100,000.

I suspect the author was writing the story to fit the following argument:

Police bear the same stress from work, family and illness that civilians do. What's different is the stress of the street and the access to a gun.

Another journalist with an agenda? The probability is good.