A data scientist uses bootstrap resampling to estimate the sampling distribution of the median. Why is bootstrapping useful here?