True; a really interesting way to do these comparisons would be by narrow year, vehicle category, and purchaser (e.g. don't conflate fleet sales to private) a

True; a really interesting way to do these comparisons would be by narrow year, vehicle category, and purchaser (e.g. don't conflate fleet sales to private) as it normalizes a bunch of factors all at once. Sample size can be a problem, I expect.