What is bootstrapping?: The process of repeatedly re-sampling from a data set in an attempt to understand and estimate the population from which the sample was drawn. Bootstrapping provides some insight into the possible variability of a metric across a population, when only the value for a sample is known.
What bootstrapping is not: A means of increasing accuracy. Using bootstrap analysis improves the precision of calculations by reducing the effect of random sampling errors within the data set but cannot increase the amount or accuracy of the data set itself. As such, bootstrapping may hide assumptions being made.
Why bootstrapping matters: Bootstrapping is useful for statistical analysis with unconventional distributions, such as when the sample size is insufficient for straightforward statistical inference. It provides a relatively straightforward method for attaining estimates of key metrics and testing the stability of a data set.
Enjoy this post? Download all 200+ terms!