Your "novice" point of view has come up with some very good points.
Crash test data is not necessarily relevant in real life crashes. One seat manufacturer gave the example of a rope around the child's neck. It will decrease head excursion numbers significantly, but will it increase the protection of the child?
Britax seats (and the Evenflo Triumph & Comet) have EPS foam which
should help protect the child's head. However, it may also increase head excursion numbers (the head will start out slightly farther forward than if the seat did not have the added layer of foam). US crash tests don't measure the impact of the child's head on the seat, but they do measure head excursion. We can't quantify the theoretical benefits of EPS foam vs the small increase in head excursion. As a parent, I've concluded that I prefer to have a seat with foam for
my children.
Some German/Austrian and British seats are now being manufactured with "Side Impact Protection." I can only assume this is due, at least in part, to the ADAC/ÖAMTC crash tests, which showed parents how inadequate many seats (especially boosters) were in a side impact collision. The crash test videos on the ÖAMTC site were one of the biggest factors in my decision to keep my 40 lb, almost-5-year-old son in a seat with a 5 point harness (the Britax Marathon), rather than moving him to a belt positioning booster. I don't have the numbers to justify my decision, but the potential for injury if his head were to strike the interior of the vehicle during a side impact collision seems great enough in my Saturn SW2 that I wanted the extra, theoretical protection that a harnessed seat provides.
I can only take the information I have and try to make the best decision for
my children based on that information, even when I don't have solid data and statistics to back me up. (And trust me, I really would prefer to know the numbers and just go with whatever stacks the odds in my favor.)