THEY SAY SPEED KILLS

We ask, 'How do you know?'

BY PATRICK BEDARD

Speed kills. Everybody knows that. We hear it so often. Like those slogans hammered into the popular culture on TV--"Reach out and touch someone," and "Don't leave home without it"
The driver reading the map does not see the red light. The blue sedan has the green and is 5 mph over the limit. See next page...
The driver reading the map does not see the red light. The blue sedan has the green and is 5 mph over the limit. See next page...
--everyone has heard "speed kills" eleventy dozen times.

It's always said in such reproachful tones, too, the ones reserved for lectures on racial equality and wife beating. "Speed kills." You're supposed to hang your head when someone says that to you.

Moreover, we hear "speed kills" in highly specific detail. From sources thought to be infallible. The Wall Street Journal, in an article by Christina Binkley on April 10, 1996, said, "There are plenty of reasons to be concerned about speeding, which causes nearly 20 percent of automobile deaths in the U.S., according to the National Safety Council [emphasis added]." Even more damning, the National Highway Traffic Safety Administration (NHTSA), in a document specifically approved by its chief, Dr. Ricardo Martinez, said, "In 1992, speed was a contributing factor in approximately 32 percent of all fatal crashes."

Thirty-two percent? One out of every three? That's a body count that demands attention.

Body count? Said in those terms, we are reminded of another time, the Vietnam War, when the federal government cooked the numbers so as to manage the public's response. One in three highway deaths. That's a staggering indictment of speed, so disturbing that we must ask, of NHTSA and of the Journal and of everyone else who says, "Speed kills," how do you know?

How do you know speed kills?

In the case of Binkley's writing in the Wall Street Journal, she's simply wrong. She misquotes her source. The National Safety Council doesn't say that speed causes even one fatality. Quite to the contrary, what it actually says carefully avoids simplifying crashes down to an identifiable cause. "In most motor-vehicle accidents, factors are present relating to the driver, the vehicle, and the road, and it is the interaction of these factors which often sets up the series of events that results in an accident. The table below relates only to the driver, and shows the principal kinds of improper driving in accidents in 1994. Correcting the improper practices listed below could reduce the number of accidents. This does not mean, however, that road and vehicle conditions can be disregarded."

The table is titled "Improper Driving Reported in Accidents, 1994" and associates 19.5 percent of all fatal accidents with "speed too fast or unsafe."

The same table appears each year in the council's annual Accident Facts. Speed's share of the improper driving in all fatal accidents in the 1990s has ranged from 16.5 percent in 1992 and 1993 to 24.9 percent in 1990. The source of the information is always attributed to "state traffic authorities," though the number of states reporting each year has varied between 11 and 17.

Now that we've taken a closer look, we can see that Binkley's source has delivered with numerical precision a fact of ambiguous value; it has only quantified one factor in a group of interacting factors. So Binkley writes out the ambiguity. She makes the numbers mean what she wants them to mean. She says speed causes these fatals, even though the council meticulously avoids saying that.

NHTSA didn't say speed causes crashes, either, though readers may interpret the message that way. The agency said "speed was a contributing factor in approximately 32 percent of all fatal crashes." What does this terminology mean, exactly, and how does the agency know that information? To find out, we must dig into the agency's data-accumulation system.

What we find is a database of secondhand information known as the Fatal Accident Reporting System (FARS). FARS started in 1975 and collects data from all motor-vehicle crashes on public roadways that result in the death of a person within 30 days of the incident. Its two-decade history, together with its nationwide coverage, makes it the country's most comprehensive database on fatal crashes.

It's important to remember that FARS is not a primary investigation. FARS analysts don't go to the crash scene, and they don't talk to witnesses. Instead, they rely on police reports, DMV records, and other public databases for information as they fill out standardized forms to produce a four-level analysis of each crash: accident level, vehicle level, driver level, and person level. A 467-page manual tells how to fill out the forms.

The speed-as-a-contributing-factor data are entered on the driver-level form. They're the last items, after 20 questions of pure fact, about the licensing of this driver;

Body Count No. 1

What NHTSA says: "The fatality rate for the German autobahn in 1993 was almost six times that of the U.S. interstate system, i.e., four fatalities per 100 million vehicle miles traveled (VMT) on the autobahn compared to 0.7 fatalities for U.S. interstates." Source: NHTSA State Legislative Fact Sheet The facts: The fatality rate in 1993 for the autobahn was 0.87, identical to that of the [U.S.] Interstate. Sources: Mercedes Benz; Insurance Institute for Highway Safety
about the charges filed against him (alcohol or drugs? speeding? reckless driving?); about the driver's previous record (accidents? suspensions? violations? convictions?). The related-factors section asks, in effect, what else--in the opinion of an analyst who has never been to the crash scene--might be involved in this crash.

The menu of acceptable answers is in the manual, 94 different items along with a two-digit numerical code for each: "Driving too fast for conditions or in excess of posted maximum" is code 44; "Drowsy, sleepy, asleep, fatigued" is code 01; "Broken or improperly cleaned windshield" is code 75. The form offers spaces for three different two-digit codes. The analyst's obligation is extremely flexible in this section. He may enter the code for one related factor. Or two. Or three. Or all three can be left blank.

We've all heard about "garbage in, garbage out." This flexibility in the inputting of related factors isn't the same as garbage, but it does allow a certain imprecision in the database.

In FARS, a crash or a fatality is "speed related" if code 44 is entered on the driver-level form for even one of the drivers involved in that event. Regardless of fault. Regardless of how many other factors may have been entered.

Consider this example: Car A is traveling at 50 mph in a 45-mph zone. As it passes through an intersection, it is T-boned by Car B, which ran a stoplight.

According to FARS, this is a speed-related crash because Car A was speeding, even though the accident was caused entirely by Car B.

 Interview with Dr. Ricardo Martinez

next

Copyright © 1995, Hachette Filipacchi Magazines Produced by @radical.media