Wednesday, April 6, 2011

New Review of the 21 Drinking Age

Economist Darren Grant (Sam Houston State University) has recently published online a study on the history, merits, and effectiveness (or lack thereof) of the 21 drinking age.  This study is one of a four-part series on drunk driving and related legislation.  Though he is not a member of Twenty-One Debunked, he nonetheless appears to echo much of what we have been saying all along.  Specifically, it exposes several biases in both the policy-making process and the studies used to support the 1984 federal drinking age law.

If his name sounds familiar, it is probably because we already discussed an older study of his on a previous post.  That study critically analyzed the effects of zero-tolerance drink-driving laws.  He also has a working paper that looks at both MLDA and drunk-driving laws in general, in which he found that hard-to-measure "social forces" were generally far more important than any of these laws, and that fact tends to skew the estimated effects of these laws.

One notable aspect of his current study concerns what we like to call "The Incredible Shrinking Effect Size" of the 21 drinking age.  For the studies that were presented at the federal hearings around 1984, and got the most attention, effect sizes of 30% or even higher (in terms of 18-20 year old traffic fatality reductions) were often reported.  However, several other studies found much smaller ones, especially the more recent regression-based ones.  Is it 30%?  Or maybe it's 20%.  Or maybe it's 13%.  Or maybe 11%.  Or maybe 7%.  Or maybe 5%.  Or maybe it's really zero, or even a net increase in deaths.  And yes, there are studies that found the latter.  But politics prevailed over science nonetheless.

One should also realize that NHTSA's estimate on the lives currently saved by the 21 drinking age (25,000 total, or 900 per year) is based on an unpublished in-house study from 1985 using only 13 states.  That study found a reduction of 13%, and NHTSA simply extrapolated that figure to the number of 16-20 year old fatalities in later years.  This number is thus junk, especially since studies like Miron and Tetelbaum (2009) found that the apparent effect wore off after the first year or two, was only found in the early-adopting states, and was offset by increases in fatalities among drivers under 18.  Other studies, like Asch and Levy (1987 and 1990), Males (1986), and Dee and Evans (2001) find that any lives of 18-20 year olds that appear to be "saved" are merely delayed a few years to the 21-24 age group.  Thus there is no net benefit in the long run to keeping the drinking age at 21.

It's time to put science ahead of ideology for once.  What better time than now?

No comments:

Post a Comment