He has a valid point. Unless there is serious injury, many bicycle crashes don't get reported to any kind of authority, making a lot of studies biased. Whenever I've crashed (which is a few times in my 40+ years riding) unless there was serious injury (once) I don't report it to anyone. Bike crash studies/statistics cannot be accurate unless every crash was reported. Real world studies on the effectiveness of helmets in actual crashes are laughable, unless (maybe) they are based on surveys.