Granny You Can Drive My Car

According to an L.A. Times article, drivers over 75 are responsible for 11% of traffic fatalities in New Hampshire, as compared to 13% nationwide. The difference is attributed to the NH law that requires drivers over the age of 70 to pass driving tests in order to renew their licenses. Eugene points out three major problems with this analysis; I wish to point out two more.

First, neither the 11% nor the 13% figure means anything without a comparison statistic -- specifically, the percentage of all drivers who are over the age of 70. If that figure were significantly smaller than the numbers above, we could conclude that older drivers are in fact more dangerous than others; but if that figure were significantly larger, we could conclude that they are actually less dangerous. So what’s the real figure? I don’t have numbers for drivers, but as a proxy, look at this page with age figures for the U.S. in 2001. After excluding the age groups in which no one can legally drive, people over 70 constitute 11.7% of the remaining population. Hmm… seems pretty close to the 11% and 13% figures. Now, my figures shouldn’t be taken too seriously, because it’s possible that lots of older people are choosing not to drive or are unable to drive, and thus their representation in the driving population could differ from their representation in the population as a whole. But the figures are certainly close enough to give us pause.

[UPDATE: Eugene summarizes a study with more specific figures on the accident frequencies of different demographic groups. The bottom line is that, on average, people over 70 aren't the safest age group of drivers out there, but they're not the most dangerous either. They are safer drivers than teenagers and often safer drivers than people in their 20s.]

Second, a tougher licensing standard applied to any demographic group would reduce the percentage of traffic fatalities attributable to that group, whether or not that group is more dangerous than the general public. What if, for example, we passed a law saying that all Jews had to pass driving tests to renew their licenses (while non-Jews could renew automatically)? Then some Jewish drivers would be weeded from the driving pool, and as a result, the percentage of traffic fatalities attributable to Jews would probably fall. This would be true even if the driving tests were totally uncorrelated to driving competence, simply because they would reduce the total number of Jews on the road. But that would in no way indicate that Jews are less safe drivers than the rest of us. Substitute any other group for Jews and you’d get similar results.