The state farm estimates are next to useless for deer management without more information. But if they could break it down by counties, rural areas, urban areas etc., it could be pretty darn good. Statistically they are probably pretty damn close on actual deer collisions, but what good is it for deer management if they chances are 1 in 50 in the twin cities area but the odds change to 1 in 150 outside of that area.
I have also heard of some jack pine savages that have a single vehicle accident after having too many beers and make a claim saying they hit a deer. Might not be statistically significant but I like some good anecdotal evidence.