I think you have a problem with large numbers.
If the accident rates, given the tens of thousands of accidents yearly, aren't changing, then it would take a stupendously stupifyingly coincidental alignment of the stars to then make the accident rates exactly cancel out the *entire effect* of millions upon millions of cellphones being owned (and presumably used) by almost every person of driving age in the United States.
That your *entire argument* is based on refuting yearly accident rate figures based on a minor estimation detail, is unbelievable.
Do you realize how MANY cellphones there are owned by people in the USA?
If those cellphones were being used, while driving, and if they were causing accidents, no amount of fudging of the data would show what the data actually shows.
There is a paradox, to be sure, but the answer is never going to be found in the puny numbers associated with *estimation errors* that you want it to.
You're grasping at straws if you truly feel that the *estimation errors* exactly cancel out the absolutely stupendous effect we presume cellphone ownership to have on accident rates, in both timing and in number.
It's just not possible,and, it's a bit scary that you believe it is. Does anyone else believe that the answer to the paradox is simply that estimation errors have skyrocketed, and then plateaued at exactly the same rate as cellphone ownership has?