Human Nature: Science, Technology, and Life.

The Smoking-Drinking Problem

There's one more contrarian study I want to pick up on this morning: a paper in the Journal of Public Economics that links smoking bans to drunk-driving accidents.

The authors, Scott Adams and Chad Cotti, report: "Using geographic variation in local and state smoke-free bar laws in the US, we observe an increase in fatal accidents involving alcohol following bans on smoking in bars that is not observed in places without bans." They present evidence suggesting two explanations: 1) "smokers driving longer distances to a bordering jurisdiction that allows smoking in bars," and 2) "smokers driving longer distances within their jurisdiction to bars that still allow smoking, perhaps through non-compliance or outdoor seating."

Not too many folks read the Journal of Public Economics or have the time to wade through the whole paper. But if you follow Fox News or live in a country like Australia, Colombia, India, or Turkey, you've probably seen the AFP wire story about this study. It quotes the authors as summarizing their findings this way: "Banning smoking in bars increases the fatal accident risk posed by drunk drivers."

The AFP story leaves non-Americans with the impression that we have some kind of national smoking ban. "A ban on smoking in American bars has caused the number of accidents from drunken driving to surge," it begins. Later, it adds, "The ban is spreading across the United States, but in a piecemeal fashion."

Ideally, at this point, the reader starts to smell something wrong with the story. "The ban" can't be piecemeal. If some jurisdictions ban smoking in bars and others don't, it must be a patchwork of independent bans—as, in reality, it is. Furthermore, if you think about the causal mechanism the evidence apparently supports—"smokers driving longer distances" to get to places where they can light up—you begin to realize that the problem isn't "banning smoking in bars." It's the fact that these bans are piecemeal and inconsistently enforced. If "the ban" actually existed as such—if bar smoking were effectively prohibited nationwide—there'd be no incentive to get in your car and drive somewhere else. The drunk-driving problem is just as good an argument for nationalizing the bar-smoking bans as for scrapping them.

I'm not proposing total tobacco prohibition. That'd be just as foolish as the failed experiment in alcohol prohibition. But we don't have to go that far. The true implication of the drunk-driving study, if you think it through, is that the safest place to let people smoke is the place that doesn't require them to get in a car at all. It's called home.

Contrarianism in Context

Impressive cluster of contrarian research in today's batch: Coffee can help prevent Alzheimer's; trans fats can be good for you; fat kids have fewer cavities; and the alleged benefits of drinking lots of water are unfounded. I love reports like these. I've flagged and commented on lots of them in the previous Human Nature news roundups. Part of it is that I just enjoy contrarianism. Part of it is that discoveries like these expose our overconfidence about what we know. Biology is enormously complex. Sometimes extra weight is bad for you; sometimes, at death's door, it can save your life. We vilify and prohibit alcohol as a sin, then discover it can help your circulation.

But I don't want to let the mischievous fun of medical contrarianism obscure reality. The reason why studies like these are surprising and intriguing is that they generally run against the grain of biology. By and large, trans fats are horrible for you. Relying on coffee instead of sleep for daily energy is dangerous. And even if being fat somehow improves your kid's dental health, the damage done to the rest of his body isn't worth it.

When you see a report about the benefits of booze or chocolate, always remember that the reason it's worth a headline is that these things, in their usual form and consumed quantity, are generally unhealthy. Not a very entertaining takeaway, I admit. But true.