Your understanding of an abstract concept is only as good as that of your best example.

This doesn't seem to be true for anything that's normally analyzed statistically: the stock market, for example, or large-scale meteorology.