I do wonder how randomized the shutting down of lights were in that study. Did those towns shut off every light? Or, did they only shut off lights in low risk areas? And, does the study differentiate between dimmed lights and lights shut off completely? The argument put forward by the Automobile Association is to not shut off the lights on roads where the speed limit is greater than 40 mph since the stopping distance for cars is much longer at high rates of speed and that lighting helps prevents deaths in those situations. This makes sense to me - shut off your lights based on risk analysis rather than a blanket shut them all off.