More foolproof teaching using resampling

In "resampling" we broadly include (i) Sampling from populations (Monte Carlo calculations of probability) and (ii) resampling from samples (simple bootstrapping for inference). Both forms of resampling were primarily developed for researchers to answer questions too complex for analytical tools. For many of the modest problems we teach undergraduates, however, analytical solutions are usually available that provide great generality and insight. What need is there for resampling?