For any number of reasons
I find this study suspect. It's plain that many people use phones in the car without having accidents. Also plain that people have accidents without using phones. Likewise, attention to driving, skill, conditions of all sorts [including not only road, but inebriation, traffic, light and so forth], amount of enforcement, etc. are all highly variable. The smaller the sample, the less likely it is to be accurate. Add to this, the caveats mentioned herein, and you've got one useless study — which even scientifically cannot be counted on unless replicated.
I wonder who paid for this mess?