STATISTICS DISTORT DRUNKEN-DRIVING STANDARDS DEBATE

Imagine you're sitting at a pleasant sidewalk cafe having dinner. After
taking a few sips of wine, you get up to go to the bathroom. Suddenly a car
driven by a man who has had a heart attack jumps the curb, mows you down and
dispatches you to a better world.

When you look down from heaven, you are astonished to find your death listed
as an alcohol-related fatality.

Strange but true. Under the rules established by the National Highway
Traffic Safety Administration, accidents are routinely ruled alcohol-related
even when alcohol had no discernible role in them. This practice has so
distorted the debate about drunken driving that it might be a good idea for
us to simply stop discussing the matter entirely until we start using valid
data.

Instead, we're in the midst of a national effort to use that distorted data
to lower the national blood alcohol content (BAC) standard from 0.10 percent
to 0.08 percent. (California's standard is already 0.08 percent.) Frank
Lautenberg, a U.S. senator from New Jersey, is leading the charge. With bad
information. I quote: ``In 1996, 41 percent of some 42,000 deaths due to
traffic crashes were alcohol-related.''

Sounds horrible. Sounds like 41 percent of the drivers involved in these
crashes were drunk.

They weren't. The actual figure for drunken drivers (those with BAC over
0.10 percent) involved in fatal crashes is less than half that figure, about
19 percent, according to NHTSA.

So where does the 41 percent figure come from? From the very heart of
statistical mendacity. For one thing, it includes any BAC above 0.01, just
one-tenth of the present legal limit in most states. Many of these people
were, for all intents and purposes, sober when they entered the pearly gates.

But that's not the worst part of the distortion. The figure also includes
accidents in which the driver was stone cold sober. The fatality was that of
either a pedestrian or a bicyclist who was drunk. Good and drunk, by the
way. NHTSA's own figures show that about a third of pedestrians killed in
accidents had a BAC over 0.10. The bicyclists tended to have had a few
drinks as well.

Which brings us to the central question that is not being asked in this
math-phobic nation: Why are we setting drunken-driving standards using a
database of drunken bicyclists and walkers? This is a question we should
answer before we start targeting moderate drinkers.

No one on either side of the current debate opposes targeting the
hard-drinking drivers, those above 0.10 percent. But the problem with the
proposed change is that it would, if strictly enforced, put a large number
of harmless drivers in jail in an attempt to scare these truly dangerous
drivers.

The bill's proponents claim that the drivers in the range between 0.08 and
0.10 percent are a significant hazard. But the statistics don't show it once
you factor out the walkers and bicyclists.

Lautenberg's office, for example, has claimed that 9 percent of highway
fatalities involved drivers whose BAC was less than 0.10 percent. This is
not true. The actual figure, when pedestrians and bicyclists are excluded,
is 6.2 percent.

That's still a significant number. And that figure could be used to make a
good argument for cracking down on even moderate drinking before driving.

Except for two things: One is that the fatalities are spread fairly evenly
through this range. Those at the top don't appear to be driving much
differently from those at the bottom.

The other flaw is that statistics are meaningless in a vacuum. You need to
put them in context. In other words, if redheaded drivers make up 5 percent
of the population but are involved in 50 percent of the accidents, you'll
want to steer clear of redheads. However, if redheads make up 5 percent of
the population but are involved in just 1 percent of the accidents, you'll
want to compliment them on their driving skills.

To evaluate the effect of moderate drinking, you need to know how many
people are driving around with alcohol in their systems. The 1996 National
Roadside Survey, for example, found that about 13 percent of drivers tested
were in the BAC range between zero and 0.099 percent.

Put those two studies together. You find that 13 percent of people are
driving around with moderate amounts of alcohol in their systems but they
are involved in just 6.2 percent of fatal accidents. Statistically, these
drivers are actually safer, not more dangerous, than the average driver.

They aren't, of course. I'm playing with statistics. But so are the
proponents of this bill, in more ways than I can list in this short space.
Until they stop using bad numbers, they should be ignored.