However, there are still reasons to be skeptical of these findings:
- Correlation is not causation, and there may still be selection bias, reporting bias, and residual or unmeasured confounding.
- Only a few states and localities had an age limit of 21 for tobacco in 2016-2017, especially when New York and Massachusetts are excluded.
- In some of these few Tobacco 21 states/localities, the number of individuals surveyed was in the single digits.
- Even if these results are 100% due to the hike of the age limit to 21, the study may only be measuring short-term effects since the laws are so recent and only data from 2016-2017 were used. More longitudinal data are needed.
- Such "early-adopter" effects may not be generalizable or durable, as we saw with the 21 drinking age according to Miron and Tetelbaum (2009).
- Data were collected from November 2016 through May 2017, and yet New Jersey was listed a Tobacco 21 state even though their law didn't go into effect until six months later in November 2017. Thus, we noticed at least one potential coding error.
- California raised the cigarette tax significantly as of April 1, 2017, within the period of the study. And Illinois and Chicago have raised their cigarette taxes several times in the years before and after Chicago's Tobacco 21 law that was implemented in 2016.
- Smoking was already on the decline nationwide long before any Tobacco 21 laws were passed, and the data are not adjusted for pre-existing trends.
- Vaping was not examined in this study, and in any case all of the data was from before the JUUL craze came on the scene.
- And most importantly, the study did NOT look at people under 18 at all.
Also, it just so happens that yet another recent and preliminary study was done, this time longitudinally using BRFSS data of 18-20 year olds from 2011-2016 compared to 23-25 year olds, and comparing the local tobacco age limits by metropolitan/micropolitan statistical area (MMSA). This study, which was driven by even earlier adopters (mainly city and county-level Tobacco 21 ordinances), did find statistically significant reductions in current established smoking by 18-20 year olds that were not found for 23-25 year olds. But the devil is really in the details, since the effect size was rather small (1.2 percentage points, and at most 3.1 percentage points in some models) for practical purposes, and may still have been driven by reporting bias, selection bias, and/or differential sensitivity by age to tobacco tax hikes at the same time. And given how effect sizes for later adopters of any given policies tend to shrink over time compared to earlier adopters, these results do not look particularly encouraging. Especially since a cursory look at the trendlines in the study finds that the slight divergence in smoking rates that emerges in 2014-2015 re-converges and essentially disappears by 2016, suggesting that the findings are likely driven by short-term effects rather than longer-term effects.
Bottom line: it looks like the supposed benefits of raising the smoking/vaping age to 21 were, shall we say, all smoke and mirrors, at least for people under 18. The supposed success of Needham, MA, for example, was likely a statistical fluke and/or a result of endogeneity, much like the "early adopter" effects of the first few states to raise the drinking age to 21 creating that particular mirage in the 1980s. Or perhaps increased enforcement in general relative to neighboring towns did the trick regardless of the age limit, like it did in Woodridge, IL and several other communities the 1990s with an age limit of 18. Studies show that whenever vendor compliance exceeds 90-95%, there is indeed a dramatic drop in teen smoking regardless, by as much as 50% compared with previously weak enforcement and low compliance rates, especially for the youngest teens. More recent research bears this out as well, for teen smoking as well as vaping. And keep in mind that those who make it to 18 without smoking are far less likely to take up this deadly habit later on.
This all should be food for thought for policymakers debating not just the age limit for tobacco, but also for alcohol, cannabis, or anything else for that matter. And even if such benefits of the 21 age limit were real, we at Twenty-One Debunked would still not support an age limit any higher than 18, on principle alone. Old enough to fight and vote = old enough to drink and smoke. 'Nuff said.