Are auto crash test results almost useless due to methods

I am probably buying a new car later on this year. I have been on the Institute for Highway Safety website where you can search just about every auto on the market for the detailed crash tests results for safety. One major flaw I see is that there is a note that states cars are only tested against other cars in the same weight category. What use is that? In reality, in the unfortunate event of a crash, you’re not at all guaranteed to be hit by someone of with a vehicle in your vehicles “weight class.” They seem to divide the classes up into areas such as sub-compact, compact, mid-side, large, SUV, etc. This seems to be one area when you would not want to compare apples to apples because you’re not going to be encountering apples to apples in society. Here it would seem you need to compare apples to “the average.” So instead of the Institute on highway safety using their current apples to apples method, why can’t they take the “average weight” of all registered vehicles on the road (maybe exempting outliers such as tractor trailers) and then test all vehicles against the “nationally averaged weight of all registered vehicles on the road.” Wouldn’t that provide much better data to consumers? If you’re buying a subcompact, they currently can boast a 5 star safety rating, but that may not be true at all if struck by a full size pickup or SUV. Agree? Or am I missing something?