Science Lesson: Conflating age with inevitable temporality (i.e., some things first occur in youth merely because youth comes first)

by Carl V Phillips

A random science lesson, because I have not written a good “the conventional wisdom — how everyone looks at this and thinks is self-evidently true — is not the only plausible explanation” lesson in a while (other than tweet storms), and just want to. I was triggered on the topic by some chatter I saw about a recent paper, though neither of those is particularly important (so no links).

Consider an example from another realm: A large portion of significant original contributions in theoretical mathematics are figured out, or at least the seeds are completed, when the author is under 25-years-old, or even under 20. The conventional wisdom is — or was (I have been out of that field for a long time) — that people’s sheer physical brainpower in this area declines with age, and that this is the only time someone has the ability to outperform all who have come before them. It is like being a professional athlete. You can be a perfectly solid athlete or science geek at 60 if you have the natural skills and keep at it, but to be among the absolute best — among the 0.001% who can be a performance-level jock or breakthrough mathematician — you have to have both the natural skills and be at your lifecycle physical peak.

But there is a plausible alternative theory that was pointedly ignored in that conventional wisdom: Generations of mathematicians have already worked out everything, within the bounds of what occurs to them to work on, that can be done by just plugging away at it. Therefore, new breakthroughs only come when someone is wired enough differently to see something beyond that, either in terms of recognizing something outside the existing bounds to pursue or some striking insight into a within-bounds problem. That is, they need to not just be solid in the skills of the field, but have one little cognitive quirk that no one else had. Either they have that when they are 16 or they don’t. If they do, they make their breakthrough early because they can. It is not about age — if one was somehow prevented from making the breakthrough for a couple of decades (but managed to keep up his skills in the field and was not scooped), he would have made it later.

Perhaps the relative contributions of those two factors has been largely resolved — as I said, I have been out of that area a long time. In contrast with the tobacco realm, most everyone who is aware of that debate is a smart clear thinker, so they may have long since worked out how much each of the stories explains the association of age and breakthroughs. But the point is that the naive explanation for something being associated with age — that it must have been entirely caused by age itself — was not so obviously correct as the conventional wisdom had it.

This is a metaphor, of course, for all the claims about tobacco use initiation, habituation, “addiction”, and such that are attributed to age because they are associated with age. This is a fail for exactly the reason found in the alternative theory of math prodigies: If something were able/likely to happen sometime in someone’s life, but not in most people’s, the fact that it happened early among the former (because it could) is not informative.

So we have the conventional wisdom that because smokers (etc.) mostly start fairly early in life, if you stop them from starting early, they never will. This is undoubtedly true to some extent. Everyone gets more set in their ways about what they do and do not do after adolescence. For smoking specifically, having adult-level judgment and a more forward-looking mindset makes it much less appealing (though this is not true for low-risk and potentially net beneficial smoke-free products). But it is obviously not nearly as true as is generally claimed. Someone who would have used a product at 16, but is somehow kept from doing so for two years does not magically revert to having the average lack of interest (which means being below the line for inclination to use the product) at 18. The same is true if you substitute age pairings 18…21 or even 16…40.

My goal here is to just immunize readers against the common naive error by planting the idea, so I am not going to delve deeply into the data. But just notice that transitioning to “smoker” status has gone down sharply among 14-year-olds in the US population, but not 18-year-olds. It is down overall, of course, but it is impossible to not notice that some of the “success” at earlier ages consists of delay rather than elimination. If the conventional wisdom were true, we should not have seen the sharp rise in the average age for that transition; the conventional wisdom says that the people who are pulling that average up do not exist.

The issue is clearer still for claims about early-initiating smokers (etc.) being more habituated (usually called “addicted” of course, but my readers will understand why that is bullshit rhetoric). If there is any variation within the population in terms of who is inclined to become strongly habituated — and obviously there is, due to both biological and social factors — then of course we see this. Those who are most inclined quickly become regular consumers upon first trialing at, say, 13. Those eventual-smokers (etc.) who ramp up more slowly were not so enamored, and so waited until it was easier to do. The former group are undoubtedly less likely to quit, have higher “dependence” scores, etc. The rhetoric attributes all of this obvious confounding to causation.

This does not means that there is no biological effect of early smoking (etc.) that causes greater inclination later in life, of course. But it does mean that the main body of evidence deployed in support of that claim is worthless. My readers presumably understand that the evidence deployed in support of “gateway” claims is bullshit because it merely observes the inevitable association across individuals choosing to use very similar products. Any association that is inevitable due to confounding cannot be said to be evidence of any causation without further serious analysis, analysis that tobacco control “researchers” never do. The present case is a bit more subtle than the gateway case, but it is exactly the same problem.

Similarly, these observations do not mean that somehow preventing an incidence of initiation at 16 is always just be a delay rather than permanent prevention. There is some probability of each. There is ample reason to believe that the probability of mere delay is fairly high. Yet the claims based on the observed association almost always bake-in the unstated and unexamined assumption that the probability of it being mere delay is approximately zero.

I did not become a regular drinker until my 30s, or a regular user of nicotine products and sometimes [redacted because we live in a fucked-up anti-liberty police state when it comes to stuff like this] until later still. But I trialed all of these before I was 20 and did a bit during my 20s. Those who want to say “it is all about ‘youth’ initiation!!!” will spin this into supporting their claims. Look closely at their claims and you will see that most of them would attribute my later behavior to those largely forgotten moments from adolescence. I can tell you there was no causal continuity between the trailing and later period of ongoing use, except via the confounding pathways. Granted I am a bit unusual — I have taken up quite a few things at time in my odd life that very few people ever do if they do not start at a much younger age: professional popular writing, various sports, farming, having babies. But the oddity there just illustrates the point that acting upon willingness or interest gets mistaken for causation, because willingness and interest are usually not kept latent for so long.

Consider one more metaphor that illustrates a different angle on this: adults who choose to visit Disney World (i.e., because they like to, not just because they are roped into taking their kids). There is undoubtedly a huge association between this and having visited as a child. Undoubtedly it is causal to some extent, but it would be obviously stupid to assume the association is all causal. Among those negative for both traits are those with a religious or semi-religious objection to visiting, those who disdained the idea as children (often due to their particular subculture think of it as belonging to Others), and those for whom making the trip is unaffordable. Those traits tend to be fairly persistent through the lifecycle, and this alone creates an association. Among those positive for both traits are those who just love stuff like that, and so pushed their parents to taken them and later choose to go again when they could. This increases the association with no causation in sight yet. Finally, among those positive for both are those who go back because they remember how much they enjoyed it as kids, the causal group. The “logic” of the tobacco control literature and rhetoric would be to claim that the association is caused entirely by the latter group.

I would assume that the marketing people at DisneyCorp — who are presumably much better at their jobs than most tobacco researchers and pundits are — have this all worked out and make extensive use of that knowledge. It would undoubtedly be possible to form honest estimates that separate the contributions of causation-by-age and mere temporality in the tobacco space also. But few in that space even recognize this is an issue, and most of them want to pretend it is not, and few of them have the skills to do the (actually pretty simple) analysis to try to sort it out.

It is one more persistent set of lies (partially intentional, partially due to Dunning-Kruger) to be aware of when analyzing tobacco control claims.