I prefer a different test: If you can't make a quantitative prediction with better than 50% accuracy then you don't understand it.
I know people will argue that predicting is difference from understanding, but I don't think so. If you can't make somewhat accurate predictions (even at low precision) then what good is your knowledge? If I model lighting as an angry thunder god, that's certainly a simple explanation, but the fact that I can't make predictions means that I don't really understand lightning.