Numbers, please

I think one of my blog goals this year will be to write about stories that give, at best, a superficial view of reality or, at worst, a distorted one, because they use numbers incorrectly or incompletely. Journalists are mostly innumerate, with neither the inclination nor the ability to convey a sense of proportion. We just don't do math.

Here's an example involving Indiana traffic fatalities for 2006. They went down nearly 10 percent for the year, which is reported as a big deal.:

Gov. Mitch Daniels said the drop in wrecks and traffic deaths for the first 51 weeks of 2006 reflects "not just good luck but some hard work" on the part of local law enforcement and state police, who issued 10 percent more traffic tickets.

But that "nearly 10 percent" is just 81 deaths in a population of more than 6 million -- so statistically insignificant that it could be attributed to mere chance, no matter how many more traffic tickets were issued. (Hint: Anytime percentages are used a lot, there's a good possibility that the actual numbers are small. If I wanted to alarm you about the number of people killed by refrigerators falling out of second-floor apartments, it would be much more effective to say the incidence of such increased 50 percent in one year rather than to say two people were killed that way this year, up from one last year.)

A better number, which can be found without a lot of heavy-duty research, is this: "Since 1990, the rate of motor vehicle deaths declined from 2.5 to 1.3 deaths per 100,000,000 miles driven." That's a much more useful statistic because it provides valuable context (per miles driven) and shows a trend rather than a one-year snapshot. That decline would actually give us something to work with if we wanted to find out if there were something effective we'd been doing in the last 16 years that we might want to do more of.

A nearly 50 percent decline, by the way. Good job!