Using evidence that a policy would do no good as a basis for proposing the policy (this time it is about airplanes)

There is a political faction that for years has been pushing for regulators to require that babies on planes be secured in a car seat, thus requiring that they be in a seat of their own rather than on an adult's lap.  It is not clear why anyone actually wants this.  The usual explanation for weird regulation (it helps the industry profit) does not make sense -- after all, the airlines could change their own rules themselves.  The only thing I can come up with is that there is a cabal of frequent fliers who are really sick of 18-month-old lap babies kicking the backs of their seats and want to restrain them.  The reason this is such a mystery is that health concerns about the babies cannot possible explain their motivations, because -- as has been pointed out in numerous analyses -- this would kill more babies than it saved and injure many many more, as well as killing a lot of grownups.

Why is that?  Because the extra cost of another airplane seat for a young family would be just enough to tip some of them into driving rather than flying, and driving is far more dangerous than commercial airline travel.  And not only is air travel a lot safer than car travel, but most of the fatalities that do occur are in crashes in which a safety seat could not provide any protection (roughly half of all fatalities in crashes are on flights where there are no survivors).  Thus, unlike safety restraints in cars, which save the lives of roughly half of all the people who would have died in unrestrained crashes, the potential benefit is low.  Of course, occasionally someone is saved from injury by restraints on a plane, but it is rare enough that the benefit would be outweighed by the shift to driving.

There is pretty good data for calculating what would happen (data about price, response to price, accident statistics), so this makes a great teaching example that I have used many times.  It is a wonderful lesson in how policies do not necessarily have a particular effect just because who ever proposed them says that they have that effect.  One year, using one particularly good version of my class's analysis, a student and I presented the results at the annual epidemiology meetings and she won an award for the work.  That version is not currently online, so no link, but anyone who is interested can find someone else's version:  every few years, going back to when I was a student, someone repeats this analysis and concludes that the policy would kill far more people than it saved.

In spite of how there is no upside for anyone to this policy (other than that kicking thing), it is the official position of the American Academy of Pediatrics that not only should the regulation be put in place, but pediatricians should take desperately valuable time during office visits to advise new parents to spend the extra money to buy their baby her own airplane seat.  This was published in Pediatrics about a decade ago, in an article that basically said "yes, the economic analyses say more young families will drive and die as a result of this, but we do not believe in economic evidence."  Keep in mind that this is long before Pediatrics decided to strive for the title of Dumbest Health Science Journal with their articles about "third hand smoke" and such.

The latest is a push by the U.S. National Transportation Safety Board to implement the safety seat regulation.  They do this periodically, and the Federal Aviation Administration that makes the rules pays attention to the evidence about what would really happen and refuse.  But what is remarkable is the basis of NTSB's latest recommendation:  They cite the recent crash of a small commercial flight in which everyone was killed and no one could realistically have survived, but where the bodies of some of the children on board were flung from the wreckage, and thus were apparently not restrained.  Horrible without a doubt, but how does this possibly support the call for safety seats?  Is it because if the babies had been restrained then they only would have been killed once due to the crash and not killed again because they were then flung away from the wreckage?  Being flung from a vehicle is often the cause of death in car crashes, but not plane crashes.  Perhaps NTSB should have put airplane experts in charge of this one rather than their highway guys.