There are many other factors that should be included in this study - and it more likely that these other factors will correlate much better to injury/death rates: seat belt usage, SRS equipment, ABS, dynamic stability control, driver’s BAC level, gender, driver’s years of experience, actual vs. posted speed limit, cellphone, weather condition, etc. etc. etc.

The problem with simple analysis is that the conclusions are, well, simply naive.

But ..

1.) different people tend to buy different cars. I argue that people with a higher acceptance of risk buy more sportive vehicles, and

2.) people tend to behave more riskier in safe (and sportive) cars.

All in all, the above results seem to be a perfect confirmation of the “behavioural adaptation hypothesis” which states that people often compensate safety gains by more riskier behaviour (if it is possible). In the above case it seems that the net effect is even negative (i.e. the costs of the riskier behaviour are larger than the safety gains of a big car). If you would behave the same in a SUV as in a mid-size car it should be clearly more safe to drive in a SUV.

In that respect I find the above conclusion misleading.