A method for estimating standard errors or confidence intervals in which a small sample of observations is randomly selected from the original sample, estimates are calculated, and the sample is returned to the original sample. This process continues many times to produce a distribution upon which to base the estimates.