A Wall Street Journal statistical oops

The Wall Street Journal’s recent editorial “Safe at Any Speed” is a nice piece, but the Editor fails to realize that the data he uses to draw his conclusion actually suggests the opposite conclusion.

Here are some excerpts from the editorial.

In 2005, according to new data from the National Highway Safety Administration, the rate of injuries per mile traveled was lower than at any time since the Interstate Highway System was built 50 years ago. The fatality rate was the second lowest ever, just a tick higher than in 2004.

As a public policy matter, this steady decline is a vindication of the repeal of the 55 miles per hour federal speed limit law in 1995.

This may seem non-controversial now, but at the time the debate was shrill and filled with predictions of doom. Ralph Nader claimed that "history will never forgive Congress for this assault on the sanctity of human life." Judith Stone, president of the Advocates for Highway and Auto Safety, predicted to Katie Couric on NBC's "Today Show" that there would be "6,400 added highway fatalities a year and millions of more injuries." Federico Pena, the Clinton Administration's Secretary of Transportation, declared: "Allowing speed limits to rise above 55 simply means that more Americans will die and be injured on our highways."

We now have 10 years of evidence proving that the only "assault" was on the sanctity of the truth. The nearby table shows that the death, injury and crash rates have fallen sharply since 1995. Per mile traveled, there were about 5,000 fewer deaths and almost one million fewer injuries in 2005 than in the mid-1990s. This is all the more remarkable given that a dozen years ago Americans lacked today's distraction of driving while also talking on their cell phones.

The table, presented in the editorial, shows a 16% decline in the fatality rate per 100 million miles vehicle miles traveled. However, it also shows a 37% decline in the injury rate and a 33% decline in the crash rate. Since the fatality rate was down less than the injury and crash rates, the probability of fatality given an injury or a crash is up, and substantially. This suggests that crashes are more dangerous now, exactly what would be expected if driving speed is higher.

Pedestrian deaths also are down less than crashes. This suggests a higher probability of a pedestrian death given a crash. This also is what would be expected if driving speed is higher.

Of course, there are many other things going on and there is no test of statistical significance, so all of my comments, including the WSJ’s comments are speculative. Nevertheless, it is nice to see, in some gross sense, that the fear mongers were wrong.