Do Tobacco 21 Laws Really Work?

A recent study claims that Tobacco 21 laws work to reduce teen tobacco use.  But upon closer examination, there is less here than meets the eye.  While this new nationwide study controlled for a host of other variables and spanned from 2012-2019, it still leaves the reader with more questions than answers.  

For example, the supposed effect in survey data in grades 8, 10, and 12 was statistically significant for combustible cigarettes, particularly for grade 12, but NOT for vape products at any grade level, while for the Nielsen sales data the apparent effect was much larger for vape products than combustible cigarettes.  How exactly can this glaring contradiction be explained away?  Also, no significant effect was observed on heavy smoking, which intuitively should have had the largest effect size of all due to heavy smokers facing the greatest "hassle costs" when cigarettes become even somewhat more difficult to obtain.  And even for cigarette smoking in general, the absolute difference-in-differences was a few percentage points at most, for a substance whose popularity and prevalence was already declining among all ages long before the age limit was hiked to 21 (in contrast to vaping, which increased dramatically among young people during this period).

Furthermore, the study did not look beyond the first two years post-enactment of such laws, so any observed effects may very well be merely short-term effects (cf. Miron and Tetelbaum 2009 for the 21 drinking age and alcohol-related traffic fatalities).

Interestingly, the study authors do concede that at least some of the observed effect on people under 18 is a result of increased ID checking and increased perception of risk of of tobacco products.  Of course, we know that it is not really necessary to raise the age limit higher than 18 to achieve such effects.

Overall, this nationwide Tobacco 21 study (the second of its kind) was rather less impressive that an earlier nationwide study by different authors, for reasons that are not entirely clear.  But it is very likely a repeat of what we like to call "The Incredible Shrinking Effect Size", a phenomenon that we have seen before with studies of the 21 drinking age over time.  Which means that it was most likely a statistical mirage all along, at least in the long run.

Regardless of what sort of effects, or lack thereof, that Tobacco 21 laws supposedly had, we at Twenty-One Debunked nonetheless oppose such laws on principle 100%, just like we oppose the 21 drinking age and toking age as well.  Let America be America Again!

UPDATE:  A recent pro-21 article made a Freudian slip about what happened since Texas raised their smoking age to 21 in 2019:

"Since then, the use of tobacco by teens decreased from 21.8% in 2019 to 19.1% in 2021, according to the CDC and the American Lung Association."

That is NOT a particularly large difference in teen tobacco use (which includes both smoking and vaping), especially since teen vaping had already peaked by 2019 and teen smoking had already been plummeting for decades.  Technically, the latter data were from the 2020 Texas School Survey, not 2021, and the 2019 data were from the CDC YRBS, as each survey is done only every other year.  But either way, the point still stands.  In fact, when compared to the 2018 TSS data, the 2020 TSS data are almost identical.

2023 UPDATE:  The study in question has been recently revised, but not in a way that materially alters any of the above conclusions or criticisms.  In fact, reading between the lines, not only does the results look even less consistent and statistically significant than before, but now the effect size on survey results seems to be much larger for vaping than for smoking at all grade levels, implying that some vapers have likely switched to smoking.  That makes the Tobacco 21 policy a net public health loser.

2024 UPDATE:  A 2021 study of California, one of the first two states to implement a statewide Tobacco 21 law in 2016, showed weak and unimpressive numbers in a difference-in-differences analysis.  Womp womp.