Efforts by state legislators to restrict county and city red-light camera programs have focused a great deal of attention on how cities (and counties) have gone about implementing these safety reforms. Some legislators have alleged that local governments have focused much more on the revenue-generation potential of these programs than the safety factors. Legislative initiatives on the subject are likely to punish local governments and restrict the use of red-light cameras.
My real interest here is not red-light cameras. My real interest is in the variety of ways in which evidence can be presented so as to support, wittingly or unwittingly, a particular cause or point of view.
The fundamental point I’ll make today is that, unfortunately for advocates of all stripes and for journalists, really good data analysis makes for lousy headlines.
The fact is that analyzing data in a systematically and thoughtfully leads to nuanced interpretations in most cases. Even a seemingly simple question typically gets a less-than-simple answer.
For instance, the underlying premise of red-light camera programs is that red-light cameras improve traffic safety. Seems like a simple idea, so it ought to be simple to test the premise. But what does “improve traffic safety” mean?
If one focuses on total number of reported accidents of all kinds (so this is anything from a dented fender or bumper to a car being totaled), a first glance at OPPAGA’s analysis of red-light camera intersection data would lead one to conclude that red-light cameras are counterproductive. Total crashes in the study intersections increased by 12% in the study period after installation of the cameras compared to the study period before.
The overwhelming majority of these additional crashes were of one type: rear-end crashes, most likely resulting from late-braking cars being hit by a car following close behind. Those accidents increased by over 1,400 . . . which actually is more than the net increase in accidents.
Wait . . . what?
The net increase in accidents at red-light intersections is less than the increase in rear-end collisions?
Put differently, if one considers all other kinds of accidents (excluding rear-end collisions), the net effect of red-light cameras on accidents is a decrease of 236 accidents . . . a small but still meaningful 2% reduction overall.
The least dangerous kind of accident, in terms of human injury and risk of fatality, is the rear-end collision.
So the tradeoff, in one sense, is between an increase in the least dangerous kind of accident in return for a reduction in the more dangerous kinds of accidents. In particular, it is worth noting, OPPAGA found substantial decreases in sideswipe and head-on collisions . . .curious results, but statistically compelling.
If one is most concerned about fatalities and serious injuries, the OPPAGA report offers even more compelling evidence of the positive contribution of red-light cameras to traffic safety. There was a 49% reduction in accidents resulting in fatalities in the red-light camera intersections studied by OPPAGA. Injury-accidents also decreased by a total of 68 incidents, though that amounts to only a 1% decrease in the total number of such accidents.
So what should the headline be?
What will sell papers is the large increase in rear-end crashes (or the bigger, bolder and technically true statement that accidents increased, though that obscures much). That, or the 49% reduction in fatalities.
Of course, the headlines on this subject are being driven (pun intended) by the political agendas of others, not necessarily of journalists. There’s more to be said about that, but that’s for another day.