In statistics, bootstrapping is any test that uses random sampling with replacement (for example, mimicking the sampling process). It is a resampling method that assigns measures of accuracy—such as bias, confidence intervals and prediction errors—to sample estimates. One of the most important uses of this technique is testing hypotheses on the basis of observed data. But current bootstrap theory doesn’t give a satisfactory account of how well the bootstrap works.
As Canada Research Chair in Economics, Dr. Russell Davidson is reformulating bootstrap theory based on bootstrap iteration. In other words, he and his research team are applying the bootstrap to the bootstrap. Because this is a costly undertaking, they are also developing new techniques that harness the power of parallel computation, which uses several processors simultaneously to carry out multiple smaller calculations.