The problem with these studies is what they don't tell you. The crashes are due to rear-end collisions and due to the high/excessive speed at which drivers enter intersections and tailgating. It's hard to hold the cameras responsible for the driver reaction to them. Most of these studies also fail to take into consideration that red light cameras actually lower cross-traffic collisions (t-bones, etc.) which tend to be the more fatal collisions.

It's sort of like holding a driver doing 75 in the fast lane responsible for a collision because someone behind him wanted to go 80... granted the driver going 75 should move over, but the driver is also not responsible for the actions of someone who wants to get around him. We're so used to coddling drivers that even the research traffic engineers are constantly looking for possible excuses for bad driver behaviour.

All that being said, I'd rather have speed cameras--that takes the intersection issue out of the equation.