non comparable statistics
One of the studies I read 20 years ago found that mobile phone use was comparable to drunk-driving: but ignored the fact that drunk drivers are drunk for the entire journey, mobile phone users are effectively 'drunk' only while making a call.
Another study derived usage figures by observations in [a location that higher than normal mobile phone users] at a time of day that had [higher than than normal mobile phone usage].
I think that both those studies gave false figures for the total expected benefit of banning phone use while driving, But they also implied a false figure for how much pain the enforcement would cause.
If the number of accidents caused by mobile phone users is small only because sensible people don't use their phone much while driving, then this relatively harmless law is doing good while not causing much inconvenience to many people.