Disraeli is always lurking.

I did some poking around and found the National Highway Traffic Safety Administration query system, but the data only goes back to the late 1990s, well after the change in drinking ages to 21. There don’t seem to be other reliable sources of data available.

Just plotting the raw fatality numbers by age for each year, no matter what they show, are apt to be misleading. There are a number of other factors to first control for. Among them:

1. Total miles driven by age by year. Raw numbers of accidents are not interesting.

2. Normalization by population size. Total number of miles driven might have gone up, but this is not interesting unless the total miles per person in the relevant age group has gone up, too.

3. Safety of vehicles and of the roads. Surely, the increase in reliability of cars, more SUVs, better roads etc. contribute to fewer fatalities.

4. Percent of driving that is done on highway versus small road.

And on and on, and it’s boring enough already. Most importantly, you cannot allow the arbitrary picking of dates to prove a point. A common trick is to pick the high point of fatalities before the 21-age change and then show how fatalities have gone down. But, needless to say?, before the high point, when the drinking age was < 21, the percent of fatalities was lower. What was it in the 1960s? 1950s? Not nearly so high. So it’s more of a cultural problem then an age problem, after all.

Anyway, you can’t trust 95% of statisticians.

I can also recall, when I was first in the service in 1982, thanking God that I was allowed to have a beer while risking my skin.