The study is near meaningless since it only uses a sample size of 10 intersections with cameras, several of which had very low accident rates beforehand and thus no room for the camera to lower them further. Also, their dependent variable was simple raw number of accidents which is confounded with every growing number of vehicles going through the intersections in the ever more congested and populated Orange County. Had they used a proportion accidents per vehicle it would have shifted all the results toward fewer accidents (increasing the size of reductions, creating reductions were no change was observed, and reducing the size of accident increases).
Also, as with most rule changes, the effect of a inconsistent partial implementation reveals zero about the effect of full and consistent implementation. The effects can easily be the polar opposite for partial and full implementation. This is true drug laws, gun laws, and traffic laws. IF 99% of the intersections these drivers encounter have no camera, then that will determine how they drive toward and into the 1% with cameras. If 100% or even just a majority have cameras, then that completely alter how people would drive toward and into those with cameras. For example, 99% without cameras creates the expectation of no camera and thus drivers don't notice the ones with cameras until the last second and react quickly and with panic and surprise. IF most had cameras, then drivers would approach all with the expectation of a camera, but more cautious to begin with and not be panicked or surprised when they encounter one.
This kind of thing is why I increasingly believe that no one without college level training in stats, research methods, and behavioral science should be allowed to hold public office.
As to whether safety is relevant to having the cameras, the politicians most certainly sell them to voters based on safety and not revenue. People for these policies and for the candidates that support them based on such safety assumptions. That said, avoiding accidents is not the only reason for intersection cameras. "Box blocking" is a huge problem in some intersections (when people enter the intersection even though they cannot clear it before their light turns red). It prevents cross traffic from moving when they have a green light, which increases congestion, fuel wasting, pollution, etc.. It also causes many drivers to avoid problematic intersections via residential side-streets which is generally a bad thing since they are not designed for heavy through traffic. Blocking the Box is completely avoidable and done only by selfish assholes and dangerously oblivious drivers.
