Dirty DWI Secrets

In response to the discussion of DWI (sorry, Windy, in Texas if you’re a grownup it’s DWI; DUI is the crime someone under 21 commits when he drives with any alcohol in his system) that started here, Chicago’s Windy Pundit gives us the numerical rundown on the real danger of DWI:

Drunk driving isn’t as dangerous as they’d like you to believe.

In 2006 according to numbers provided by MADD, 17,602 people died in alcohol-related crashes. You might think that means that a drunk drivers killed 17,602 people, but the NHTSA study from which they apparently pulled that number counts as alcohol-related any crash “involving at least one driver, pedestrian, or pedalcyclist with a BAC of .01 or above.” In other words, for it to count as alcohol-related, the drinker doesn’t have to be drunk or driving.

A little further down, the NHTSA report gives the number of fatalites in accidents where a driver was over the legal limit as 8615, or about half the number MADD uses. And even this number is probably high, because it includes accidents where the drunk driver was not at fault.

MADD also says that a 2002 survey indicated that Americans took 159 million alcohol-impaired driving trips. If only 8615 of them ended in DUI fatalities, that means that alcohol-impaired driving has a 1 in 18,000 chance of a fatality.

Make no mistake, that’s very high, but it doesn’t mean that every drunk who hits the road is the moral equivalent of a murder waiting happen. In order to have a 50/50 chance of killing someone, you’d have to make 12,000 impaired trips—driving home drunk from the bar every night for 35 years.

NHTSA statistics suggest that there are about 65 people injured in traffic crashes for every person killed in a traffic crash. Assuming that the same ratio holds true for alcohol-related crashes, if there’s a 1 in 18,000 chance of a fatality for every alcohol-impaired driving trip, then there’s a 1 in 277 chance of an injury. In order to have a 50/50 chance of hurting someone (including yourself), you’d have to make 195 impaired trips — driving home drunk from the bar twice a week for two years. (Windy, I’m using your assumptions — doublecheck my math, please.)

According to the CDC, 1.4 million people were arrested for DWI for those 159 million self-reported alcohol-impaired trips, which means that alcohol-impaired driving has a 1 in 114 chance of resulting in an arrest. In order to have a 50/50 chance of getting arrested, you’d have to make 75 impaired trips — driving home drunk from the bar three times a week for six months.

I’ve written that the difference between a successful drive home from the bar and one that leaves someone in jail, the hospital, or the morgue, is nothing more than dumb luck. Windy helps us measure that luck. I’m all in favor of a steely economic analysis of things that go bump in the night; I think that Americans are much too willing to blow risks out of proportion and overreact to them, as well as to react in ways that don’t reduce the risk in proportion to their cost. (See last year’s post about Managing the Risk of Child Sex Abuse.) But the danger of an individual trip resulting in disaster, and the number of trips that it takes to give you an even chance of disaster, doesn’t say anything about the magnitude of the disaster.

Our economists (or personal injury lawyers) might be able to put some prices on the average alcohol-caused crash fatality, and the average alcohol-caused crash injury. I suspect that the numbers would be large.

Even without those numbers, though, consider this: in Texas a DWI arrest will cost the accused at least $5,000 if he doesn’t hire a lawyer, and more if he wants to try to avoid conviction by hiring a lawyer. Every time you drive while impaired you’re betting that you won’t get caught. The rational person would spend at least $43.85 ($5,000 divided by 114) to avoid driving under the influence. That’s two ten-mile Yellow Cab rides in Houston.

None of this is to say that the MADDness of our DWI laws is the right approach to DWI, but intoxicated drivers are a real problem. Like most all societal problems, if it’s not solved by us rational people it’ll be solved by the kooks.

About Mark Bennett

Mark Bennett got his letter of marque from the Supreme Court of Texas in May 1995. He is famous for having no sense of humor when it comes to totalitarianism.
This entry was posted in DWI, statistics. Bookmark the permalink.

6 Responses to Dirty DWI Secrets

  1. Windypundit says:

    Great post, especially about accident rates.

    (And I plugged your math into my spreadsheet, and it’s fine.)

  2. Interesting data. But you’d need a baseline number for non-alcohol related injury accidents to compare it to. Most accidents happen involving no alcohol at all, so the assumption that all alcohol-involved incidents were caused by alcohol likely overstates the risk.

    According to the NHTSA report you linked to, there were around 38,000 accident fatalities in 2004. If 8K were DWI-related, that means around 30K were not. Even citing the higher number used by MADD, the report said “The percent of alcohol-related fatalities has declined from 60 percent in 1982 to 39 percent in 2004″

    NHTSA uses miles traveled as their denominator for calculating frequency instead of the number of self-reported trips used by Windy, so I couldn’t find apples to apples numbers to compare offhand. But whatever the per-trip injury-crash rate for non-alcohol related accidents is should be subtracted from the DWI rate you calculated to identify how much EXTRA risk using alcohol imposes.

    Judging by other data in the NHTSA report, another good question is whether sober teenage boys pose a greater risk than an adult drunk! The stats on age and gender breakdowns of accidents make one (momentarily) sympathetic with insurance actuaries!

    Good post and series!

  3. Windypundit says:

    Good points, Grits, especially about using miles travelled. I’m just too lazy to try to track that down. I imagine it must be hard to get accurate self-reported survey results for miles driven drunk.

    I know I should have been more careful to look at excess deaths, but I wanted to err in favor of greater danger to avoid being accused of twisting the stats in my favor.

  4. Mark Bennett says:

    Excellent points. I don’t think Windy assumed that all alcohol-related accidents were caused by alcohol; I think he took NHTSA’s number for accidents with an intoxicated driver (which is half the alcohol-related number) instead, noting that this was probably high.

    When you pay $43.85 to take a cab home and then back to your car the next day, you’re still risking causing (indirectly) someone else’s death or injury (are cabbies less safe or safer than sober non-cabbies?). You’re paying to eliminate only the risk of a DWI arrest.

    If we could somehow figure how much more likely an intoxicated person is than a sober one to injure or kill someone per mile driven, and the cost of doing so, we might rationally call a stretch limo instead of a cab.

    I can believe that a sober teenager might be a greater danger to himself and others than an intoxicated adult.

  5. I didn’t mean to be critical, btw, just was suggesting ways to follow up to complete the analysis. I think where you’ve gotten with it so far is valid and useful.

    Windy, any footnote giving average miles per trip in the data you got that stuff from? If so, you might be able to get to an apples to apples estimate through the back door.

    Once you get that figure, simple subtraction will get you to the ADDED risk of accidents from DWI.

  6. Windypundit says:

    No, I couldn’t find that data, but in a more recent post I linked to an NHTSA summary report and a study from that report that attempts to estimate relative risk as a function of BAC, which I think is what we really want to know.

    I’m not sure I really understood it, though. Check it out.

Leave a Reply

Your email address will not be published. Required fields are marked *

*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>