The study pools data from both a 1992 survey and a 2002 survey, and finds that those who were exposed to a drinking age of less than 21 when they were ages 18-20 were about 1.3 times more likely to have an alcohol use disorder, and about 1.7 times more likely to have a substance use disorder other than alcohol, in the past year. Even (scratch that, especially) when the respondents were in their 40s and 50s. The researchers attempted to control for confounders, but we all know there can always be some that were missed. And we know that if you torture the data enough, it will confess to anything.
We at 21 Debunked think that the relationship between MLDA and later alcohol problems is likely spurious for the following reasons:
- At least part of the relationship seems to be mediated by self-selective cross-state migration (i.e. budding alcohol abusers moving to states with more lenient laws). But there was no data on state of residence at age 18, only birth state and current state.
- The relationship was apparently NOT mediated by earlier age of drinking onset, as would be expected if the relationship was truly causal. Also, there was no significant effect on drinking before 18, so we can rule out the often claimed "spillover effect" on younger kids.
- Drug addiction (especially hard drugs) was affected more strongly than alcohol addiction for whatever reason, the opposite of what would be expected assuming causality.
- No distinction was made between a drinking age of 18, 19, or 20, which may have a misleading impact on the results. Though this could bias the results in either direction.
- The effect on past-year alcohol use disorders was strongest among respondents in their 40s and 50s, and weakest (and statistically insignificant) among those in their 20s, the opposite of what would be expected from a causal relationship. Both drinking age exposed groups start out fairly close, disorders decline at the same rate at first, and divergence does not occur until around age 35.
- Regressions did not control for state-specific trends, religion, or completely for ethnicity. Whites were treated as a monolithic group (despite wide variation in drinking cultures), and no distinction was made for Native Americans (who tend to have higher alcoholism rates).
- Unemployment and education were controlled for, but not poverty.
- The odds ratio for the relationship between MLDA <21 and later alcohol use disorders was an anemic 1.33. In epidemiology, odds ratios and (relative risks) below 2.0 are difficult to interpret and are often due to residual or unmeasured confounding, bias, or even chance. This caveat is especially true for retrospective and non-longitudinal survey studies such as this one. For example, the New England Journal of Medicine would likely consider this association "weak" and thus unfit for publication in most circumstances.
- And there's always the usual recall bias, as well as good old-fashioned denial. They don't call it a "pink elephant in the room" for nothing.
And why does Canada (MLDA 18 or 19 depending on province) not have a higher alcoholism rate than America? Demographically, they probably should! Ditto for most other countries with lower drinking ages, even the notoriously binge-drinking British and Australians? Some may say culture, but Canada's culture is not radically different than ours, and Britain's should, if anything, be more conducive to alcoholism than ours.
Another new study, this one done in Australia (with a legal drinking age of 18), explores the "age at first drink" effect on alcohol dependence from a new angle. Studying numerous twins, they find that there appears to be a gene-environment interaction among those who begin drinking before 15, especially before 13. Such very early drinking, or some environmental factor closely linked to it, appears to activate the bad genes. Though this could simply be a common vulnerability from other differences in environment rather than a truly causal effect of age. Or perhaps the context of drinking matters. The older one starts drinking, the more the environment matters, and the less genes do. But interestingly, and most importantly for the purposes of this discussion, the effect of age at first drink levels off after 18, with no significant difference between those who begin drinking at 18, 19, 20, or even 23. (Funny how this "leveling" effect occurs precisely at the age that corresponds to the MLDA. Coincidence? Hmmm.....)
This jibes well with a Canadian study done in 2000 that found that, while those who begin drinking between 11-14 have the highest risk for later alcohol dependence, even when taking time since first drink into account, there was no significant difference (in the long run) between those who began at 17-18 and those who began at 19 or later. For the latter group, the onset of dependence was merely delayed. Again, no conclusive proof of causation, but the authors concluded that prevention programs that succeed in delaying drinking to even 15-16 would likely produce substantial benefits. This is probably a better idea than trying to make the perfect the enemy of the good, like those who support the 21 drinking age never cease to do.
In other words, we need to see the forest for the trees. Something we as a society fail to do time and again.