A few people have said proving the higher level correlations is not meaningful and cannot be the basis of a scientific test. I don't understand, why not? If we create reliable empirical measures, and there is a high correlation between the measures at an appropriate alpha level (which normally <5% for published research), why is that not enough?