The complete statement was:
"[E]ven though injury crash rates were higher for vehicles displaying emergency lights, the overall collision rate for emergency and non-emergency vehicles was the same."
In other words, they're just as likely to hit the light show, but they hit it harder. There are several ways to interpret that, and we can't really know how from what has appeared here. It seems highly unlikely that the non-emergency vehicles were struck under the same circumstances as the emergency vehicles. If emergency vehicles at accident and arrest scenes are getting hit at the same rate as non-emergency vehicles stopped at stop lights, that's not an indication that the emergency lighting is ineffective. The emergency vehicles tend to be stopped at more hazardous locations, where there's high-speed traffic moving by. That's why the injury crash rates are higher. If their lights had no effect, one might expect then to get hit more often than non-emergency vehicles, not at the same rate.
Obviously, though, I'm just guessing, since I have no way of knowing what controls went into the data collection or analysis.