What would be very informative, I think is if we had an overall safety rating for the vehicle along with the driver deaths per car sold. Then we could compare highly rated groups of small with highly rated groups of big. Since we are addressing a group concerned about safety, no use to even include the 'known bad' models of either size.
It seems "good" means foreign, so if small means subcompact and large means err... large, we end up comparing a Honda Civic to a Toyota Avalon. The Civic at 84 driver deaths per million cars per year is nearest to the median for foreign subcompacts, the Avalon is the only foreign car in the large group and has 40 driver deaths per million cars per year.
In fact the Avalon is the safest car for total deaths in the whole table, though for own driver deaths two of the minivans beat it. But there seems to be a consensus developing that the minivans don't count anyway.
The Avalon salesman can boast that his car kills fewer than half as many people as a Civic each year.
The Civic salesman can point out that the difference in safety between the two is only worth $66 per year to a motorist who values his life at 1.5 million dollars, a difference that is likely to be negligible when set alongside any other criteria that are important to a potential purchaser choosing between the two.