[jcl]

Gladwell seems to be picking and choosing the data he discusses. Look at the data presented in his article, just averaging fatality rates (assuming all models are roughly equally popular): the minivans are staggeringly safer than the other classes, followed by mid-size, SUV, and large vehicles -- all roughly equivalent -- trailed by compact, sub-compact, and pickup vehicles, also roughly equivalent. This is pretty similar to the report that you linked.

Gladwell seems to be looking for data to support his thesis that SUVs feel safer but are really less safe, which overall the data doesn't seem to support; for instance, he chooses to compare the subcompact Jetta to the SUVs, instead of one of the other subcompacts, which are among the least-safe cars. He is right that we do not expect the Jetta to be more safe than the larger cars, and it would be interesting to know the reason behind it. But the reason seems unlikely to be a false sense of safety in SUVs or increased maneuverability in subcompacts, since the other subcompacts should then be roughly as safe.

On the other hand, the data does support his assertion that minivans are safer, and maybe this is because, as he says, they are driven by more experienced drivers with children on board. This would also explain why SUVs are somewhat safer than trucks, even though they are practically the same vehicles.