How to mislead with statistics

In a front-page story yesterday Older drivers bridle at blanket criticism the Boston Globe led off with a damning graph purporting to show how dangerous older drivers are. Take a look at the graph and you'll see that old driver are a menace, with double the death rate of teens! The source, the Insurance Institute for Highway Safety (IIHS), is a good one. In fact, my father was head of research there for many years...But the real story is not so simple. Also from IIHS, here's the distribution of crash deaths by age and gender per 100,000 people:This graph shows death rates rising for ages 65 plus, but much less dramatically than the previous graph. It also shows the teenage years and early 20s as the danger zone. One important difference is that old people don't drive many miles, so while death rates per mile driven are high, the trend in death rate per person is not nearly so dramatic.In addition, female death rates at every age are much lower than male death rates. But you won't see the Globe running an article on why men shouldn't be allowed to drive!But even if we accept the first graph as the key one, it doesn't mean old people are actually getting into more crashes per mile, or endangering others as the article implies. Here's how IIHS explains it:

Per mile traveled, fatal crash rates increase starting at age 75 and increase markedly after age 80. This is largely due to increased susceptibility to injury, particularly chest injuries, and medical complications among older drivers rather than an increased tendency to get into crashes. Fragility begins to increase at ages 60-64. At age 75, older drivers begin to be markedly overinvolved in crashes, but fragility is the predominant factor explaining the elevated deaths per mile traveled among older drivers.

The issues of older drivers is an important one but the Globe did a lousy job with this article.