This is but one illustration and you can decide how to pull apart the numbers. Without going into too many other variables, keep in mind they track other things in accidents - at least involving fatalities. Was alcohol involved? How about Speeding? I don't have the facts to back it up, but I'll bet that fatalities in Corvettes where speeding was involved is higher than that of a VW Rabbit. I'm sure there are skews to the alcohol related collisions as well.
Part of the problem is that you can't separate the driver from the vehicle in these studies. I guess if you have enough numbers, it becomes less relevant - but I think it's much more complicated than some of the numbers might indicate.
All important issues but I think you missed a big one. What is the accident
rate for the various vehicles? By rating drivers deaths 2 vehicles may appear just as safe; however, 1 vehicle may have a 50% greater accident rate with a 33% lower death rate in an accident. When only drivers deaths are being counted the 2 appear equally safe. When injury or death is counted its likely the vehicle with less accidents is safer.
Who drives what kind of vehicle is certainly a big issue. If they try to sort the issues out to finely then the small sample size reduces confidence in the results. Speeding and driving under the influence are also major factors in traffic accidents.
I don't necessarily drive the safest vehicles. The EV Buggy would be a bad choice to get into an accident. The 1965 Dodge, my 1966 Datsun, and an assortment of VW Beetles I've driven over the years are nowhere near as safe as modern vehicles either (gotta love lap belts and a metal dash.) The point is, the largest vehicles available are not necessarily the safest either. The IIHS report you cited makes that clear. Even today some sedans and mini-vans rate very high in safety, bettering many larger vehicles.
Paul