“Red-light cameras have helped cut wrecks by 83% in Richmond,” cheers a recent headline from The Richmond Times-Dispatch. Wow! That sounds almost too good to be true. Let’s read on.
A little further into the story we learn that the 83 percent reduction was for a single intersection, not citywide, as the headline implies. And while the 83 percent drop in accidents sounds impressive, can we attribute the reduction to the presence of the red-light camera? Likely not, since, in addition to the camera, police had been conducting stepped-up enforcement around that intersection as well.
Nonetheless, a police spokesperson believes “the camera played a helpful role in reducing those numbers.” But who knows? Neither the reporter nor the police provide any analysis to determine if the drop was actually statistically significant (more on this later), or if it was just random chance.
And speaking of crash data, where did the numbers come from in the first place? The police? The red-light camera vendor, or someone else with a financial stake in making the cameras appear as successful as possible? Again, who knows? The reporter didn’t bother to ask, apparently. Camera vendors often “help” police prepare their red-light camera accident reports, so it’s a valid question.
This story template endlessly repeats itself throughout the media, as reporters fail to think critically about what they’re reporting. This serves nobody, except the camera companies and public officials who benefit from the rigged red-light camera game. There are notable exceptions like the outstanding work of WTSP reporter Noah Pransky in Tampa, who exposed the many willful deficiencies in Florida’s red-light camera program.
If media outlets and public policy types would dig into the numbers and analyze the true impact of red-light cameras, the template would look much different.
A great example of this comes from our California colleague Jay Beeber, executive director of Safer Streets L.A. Beeber recently released this in-depth report on the safety impact of red-light cameras in Los Alamitos and Garden Grove, California. In both cities, Beeber found no meaningful reduction in accidents at intersections where cameras were installed. This comes as no surprise, but the rigor of the methodology and analysis makes the results much more compelling than those touted by any police department or local politician.
First off, Beeber used the most authoritative and complete source available for traffic accident data in California, the California Highway Patrol’s Statewide Integrated Traffic Records System database. Second, he rightly focused on changes in accidents caused by red-light running—the accidents red-light cameras are supposed to prevent. Sounds like common sense, but many pro-camera accident reports give the cameras credit for reducing accidents at locations well away from the intersection.
Beeber also analyzed the results for statistical significance, which indicated whether or not the changes were simply random or could be correlated with the use of red-light cameras. This is critical given the few data points available in many cases. An intersection may experience two accidents before cameras and one accident after, for a 50 percent reduction. That sounds impressive, but given the small numbers involved, it means nothing. (A quick look at the low accident totals for each intersection prior to camera installation makes you wonder why the cameras were installed in the first place.)
Few, if any, news stories about red-light cameras bother to mention this. Of the 10 intersections studied in both cities, only one intersection showed a statistically significant reduction in accidents, and even then it can’t be known whether or not the camera caused the drop.
Beeber’s analysis also factors in the fact that rear-end accidents tend to increase at intersections after the cameras go up. Camera supporters often ignore this well-documented phenomenon or downplay the severity of these accidents. For Garden Grove Beeber calculated a Collision Severity Index which determined “whether the total severity of injuries increased or decreased in the presence of the cameras where red light running collisions decreased and rear end collisions increased.”
In the final analysis Garden Grove experienced a non-statistically significant 37 percent decrease in red-light related collisions but also a 61 percent increase in rear-end collisions. Based on the Collision Severity Index, Beeber concluded that this tradeoff “likely represents an overall decrease in safety on the city’s roadways.”
Will a red-light camera company or a police department ever produce a report like this? Not by choice. But reporters and policymakers should demand such rigorous analyses before passing judgment on the safety merits of any red-light camera program.