I agree, Fossil. And, as Steve's post also implies, you have to have begun suspecting the footwear made a difference, so you build your samples and controls around that. But it is such a strange thing to hypothesise about that is seems probable that a different approach was taken. A difference in outcomes between two or more groups was observed. Lets measure every difference in inputs we can find, and then look at what correlates with what. That is not an absurd thibg to do, but it is not a proper experiment in itself. Rather, it is a precursor to an experiment. It might reveal footwear correlates with results. Good - now do the sort of experiment described ny fossil to find out is the is a genuine factor.