More boringly , maybe its selection on the circumstances too. For example maybe ev's tend to drive more in urban environments, more urban may mean more collision opprtunities per time spent driving.
Of course ram is a farmers vehicle is desgned for rural use, so must rarely be seen in built up areas. /s
edit: having glanced at the cited article - theres no obvious mention of any risk adjustment, the measures seem to be simple ratio of crashes per driver. No obvious control for whether the sub-population spend more or less time driving.
Rate per - place-specific-risk adjusted person-hour would work better.
As often with things like risk, it really helps to be able to do a multidimensional analysis. See if vehicle type/brand is significant after controlling for as many circumstantial factors and exposure related factors as you can reliably observe.