I introduced correlations because that's the proper statistical analysis to do! And regarding the "restricted range" argument, remember how almost every relation between NFL stats is linear over most of the range? If it's linear, then you actually have evidence that the correlation over a restricted range (a little over half of it) most likely applies to the rest. That is, the default hypothesis is that the relation is linear in this case, and anyone arguing otherwise would need to prove that. I'm just fine with saying it's just the evidence we have and be done with it, but the point is that IF you care about restricted range, then the default argument (based on league-wide correlations) is that the relation is linear and what I found applies to the entire range.