Several studies have shown that lowering the legal drinking age increases, and that raising the age decreases fatal traffic accidents involving those affected by the drinking-age change. Recently, it was found that the minimum legal drinking age did not have a perceptible effect on the fatality experience of drivers in general, or young drivers in particular. Instead, they found indications that the high drunk driving involvement of young drivers may be due to their inexperience in drinking, and that raising the legal drinking age merely shifts the inexperience in drinking to higher age groups. If the hypothesis is true, then it would represent a very important finding. However, we found that the technical approach selected is not suitable to test this hypothesis because it selected arbitrary (though plausible) factors for analysis, and omitted other, possibly more important factors. We therefore propose to evaluate the same question using another approach. This approach will use control groups that are not affected by changes in the drinking-age law but are affected by other factors--including travel and highway conditions, law enforcement, and socioeconomic characteristics of the population--that affect the treatment groups. We believe that the use of control groups will control more reliably for the effects of those factors than an attempt to incorporate them explicitly into a regression model.
