Seems like they would not need to make so many assumptions if they just did a survey of say, odometer readings of used EVs. Also 2014-2017 is ancient history in EV terms. The annual miles driven data is nearly meaningless as an indicator of current driving behavior.