Why speed cameras don't work as well as we think they do

Speed cameras have been put in at various intersections across the globe. Depending on who you ask, they were put in to bring in money or to reduce accidents. But here's one reason why they don't reduce accidents as much as we think they do.

In the past we've talked about how extraordinary results cause people to believe that their "luck" is going one way or another. There is The Gambler's Fallacy, which causes people to think that, if a coin has landed on one side an unusual number of times it is "due" to land on the other side on the next flip. Conversely, there's the hot hand's fallacy, which causes people to think that a person who has shot dice successfully a few times is on a hot streak and will keep shooting well. In fact, both hot streaks and cold streaks are part of the normal course of events. A coin flipped many times will have long streaks of heads or tails. A die thrown enough times will come up lucky for long streaks in a row.

In the more superstitious days of gambling, casinos would literally have "coolers." They are incarnations of good luck, who will sit near a lucky gambler and cool their hot streak. Some people say that professional coolers are just an urban legend, but there are a few gamblers who believe in amateur coolers. A person sits next to them at a table, and their luck turns. What's really happening is regression to the mean. Every hot streak is a slight blip caused by chance, a temporary break away from regular casino results. Keep flipping that coin long enough and no matter how many heads you once got in a row, you'll get an even distribution between heads and tails. Keep playing roulette long enough and no matter how much you win in the first hour, you'll lose in the long run because the odds are in the casino's favor.

What does this have to do with speed cameras? For the most part, they're placed at intersections that are "hot" and have an unusually high number of accidents. Some of those intersections are legitimately more dangerous and need the cameras, but even if all the intersections in every city were equally dangerous, some would have a strangely high number of accidents in a year. The next year it would most likely regress to the average number of accidents per year, even if nothing were done, for the same reason that a streak of red in roulette will eventually turn black. Because both the dangerous intersection and the non-dangerous intersection will most likely have a reduction in the accident rate the year after the speed camera is put in, it would take a careful analysis to distinguish which reduction was due to extra caution on the part of thrifty drivers, and which was just the result of chance.

In fact, this problem is the same for all safety measures. An unusually high number of deaths or accidents will cause people to demand something be done. Safety measures — walls, speed limits, laws, overseeing agencies — are put in, and the problem is reduced. But was there a real problem, or was there simply the occasional chance increase in numbers we'd expect to see in everything from average rainfall to lotto winners? And did the safety measure really help, or did things just go back to what we think of as normal?

[Via The Improbability Principle]