I reexamine the effects of MDALs on alcohol consumption using two quasi-experimental statistical methodologies that control for unmeasured factors more completely than methodologies used in previous studies.
Several studies of the effect of MDALs on youth alcohol consumption have used individual-level data for a single year.
nonexperimental) specific trend in alcohol consumption of 21-year-olds, and thus 21-year-olds are acting as a control group for the group of 20-year-olds who are affected by the MDALs.
Thus, it is important to find that MDALs affect 20-year-olds as much, and maybe even more, than they do youths of younger ages.
However, when the sample was separated by sex, OLS estimates of the effect of MDALs were inconsistent.
Part of the motivation for this study was methodological, as a common flaw of many past studies of the effect of MDALs was their failure to control adequately for time-invariant and time-varying state effects.
On the one hand, the estimates of the effect of MDALS on youth alcohol consumption presented in this article lack a degree of robustness necessary to conclude that MDALs effectively reduce alcohol consumption.
An alternative explanation of the paradox is that studies of the effect of MDALs on alcohol consumption have used self-reported measures of alcohol use.
In conclusion, I think the results of this analysis suggest that there needs to be additional study of the effect of MDALs on alcohol consumption and driving fatalities aimed at finding an explanation to the paradox raised in this article.