Definition Bootstrap/bootstrapping

The bootstrap method, in statistic, is a resampling technique used to calculate various types of parameters and their related standard errors/confidence intervals. The bootstrap method consists of taking a random subsample from a sample many times. Each "time" is usually referred to as an iteration.

For example, let's imagine that we have a population of 10 million individuals and take a random sample of 100,000. If we want to know the mean height of this population, we can take the mean height of our sample of 100,000 as estimator. But what is the margin of error of our sample mean? To find out, we can use the bootstrap technique and, for instance, randomly create a subsample of 1,000 individuals and get their mean height. If this subsampling is repeated, for instance, 500 times, we would have the mean height of 500 random subsamples and, importantly, the variance of this distribution of mean heights. The margin of error based on these elements approximates the margin of error of our sample of 100,000 individuals for a population of 10 million individuals.