# Difference between revisions of "Bootstrap method"

A computer-intensive re-sampling method, introduced in statistics by B. Efron in 1979 ([a3]) for estimating the variability of statistical quantities and for setting confidence regions (cf. also sample). The name ‘bootstrap’ refers to the analogy of pulling oneself up by one’s own bootstraps. Efron’s bootstrap is to re-sample the data. Given observations $X_{1},\ldots,X_{n}$, artificial bootstrap samples are drawn with replacement from $X_{1},\ldots,X_{n}$, putting an equal probability mass of $\dfrac{1}{n}$ on $X_{i}$ for each $i \in \{ 1,\ldots,n \}$. For example, with a sample size of $n = 5$ and distinct observations $X_{1},X_{2},X_{3},X_{4},X_{5}$, one might obtain $X_{3},X_{3},X_{1},X_{5},X_{4}$ as a bootstrap sample. In fact, there are $126$ distinct bootstrap samples in this case.

A more formal description of Efron’s non-parametric bootstrap in a simple setting is as follows. Suppose that $(X_{1},\ldots,X_{n})$ is a random sample of size $n$ from a population with an unknown distribution function $F_{\theta}$ on the real line, i.e., the $X_{i}$’s are assumed to be independent and identically-distributed random variables with a common distribution function $F_{\theta}$ that depends on a real-valued parameter $\theta$. Let $T_{n} = {T_{n}}(X_{1},\ldots,X_{n})$ denote a statistical estimator for $\theta$, based on the data $X_{1},\ldots,X_{n}$ (cf. also statistical estimation). The object of interest is then the probability distribution $G_{n}$ of $\sqrt{n} (T_{n} - \theta)$ defined by $$\forall x \in \mathbf{R}: \qquad {G_{n}}(x) \stackrel{\text{df}}{=} {\mathsf{P}_{\theta}}(\sqrt{n} (T_{n} - \theta) \leq x),$$ which is the exact distribution function of $T_{n}$ properly normalized. The scaling factor $\sqrt{n}$ is a classical one, while the centering of $T_{n}$ is by the parameter $\theta$. Here, $\mathsf{P}_{\theta}$ denotes the probability measure corresponding to $F_{\theta}$.

Efron’s non-parametric bootstrap estimator of $G_{n}$ is now defined by $$\forall x \in \mathbf{R}: \qquad {G_{n}^{\ast}}(x) \stackrel{\text{df}}{=} {\mathsf{P}_{n}^{\ast}}(\sqrt{n} (T_{n}^{\ast} - \theta_{n}) \leq x).$$ Here, $T_{n}^{\ast} = {T_{n}}(X_{1}^{\ast},\ldots,X_{n}^{\ast})$, where $(X_{1}^{\ast},\ldots,X_{n}^{\ast})$ denotes an artificial random sample (the bootstrap sample) from $\hat{F}_{n}$, the empirical distribution function of the original observations $(X_{1},\ldots,X_{n})$, and $\theta_{n} = \theta \! \left( \hat{F}_{n} \right)$. Note that $\hat{F}_{n}$ is the random distribution (a step function) that puts a probability mass of $\dfrac{1}{n}$ on $X_{i}$ for each $i \in \{ 1,\ldots,n \}$, sometimes referred to as the re-sampling distribution; $\mathsf{P}_{n}^{\ast}$ denotes the probability measure corresponding to $\hat{F}_{n}$, conditionally given $\hat{F}_{n}$, i.e., given the observations $X_{1},\ldots,X_{n}$. Obviously, given the observed values $X_{1},\ldots,X_{n}$ in the sample, $\hat{F}_{n}$ is completely known and (at least in principle) $G_{n}^{\ast}$ is also completely known. One may view $G_{n}^{\ast}$ as the empirical counterpart in the ‘bootstrap world’ to $G_{n}$ in the ‘real world’. In practice, an exact computation of $G_{n}^{\ast}$ is usually impossible (for a sample $X_{1},\ldots,X_{n}$ of $n$ distinct numbers, there are $\displaystyle \binom{2 n - 1}{2 n}$ distinct bootstrap samples), but $G_{n}^{\ast}$ can be approximated by means of Monte-Carlo simulation. Efficient bootstrap simulation is discussed, for example, in [a2] and [a10].

When does Efron’s bootstrap work? The consistency of the bootstrap approximation $G_{n}^{\ast}$, viewed as an estimate of $G_{n}$ — i.e., one requires $$\sup_{x \in \mathbf{R}} \left| {G_{n}}(x) - {G_{n}^{\ast}}(x) \right| \to 0, \quad \text{as} ~ n \to \infty,$$ to hold in $\mathsf{P}$-probability — is generally viewed as an absolute prerequisite for Efron’s bootstrap to work in the problem at hand. Of course, bootstrap consistency is only a first-order asymptotic result, and the error committed when $G_{n}$ is estimated by $G_{n}^{\ast}$ may still be quite large in finite samples. Second-order asymptotics (cf. Edgeworth series) enables one to investigate the speed at which $\displaystyle \sup_{x \in \mathbf{R}} \left| {G_{n}}(x) - {G_{n}^{\ast}}(x) \right|$ approaches $0$, and also to identify cases where the rate of convergence is faster than $\dfrac{1}{\sqrt{n}}$ — the classical Berry–Esseen-type rate for the normal approximation. An example in which the bootstrap possesses the beneficial property of being more accurate than the traditional normal approximation is the Student $t$-statistic and, more generally, Studentized statistics. For this reason, the use of bootstrapped Studentized statistics for setting confidence intervals is strongly advocated in a number of important problems. A general reference is [a7].

When does the bootstrap fail? It has been proved in [a1] that in the case of the mean, Efron’s bootstrap fails when $F$ is the domain of attraction of an $\alpha$-stable law with $0 < \alpha < 2$. However, by re-sampling from $\hat{F}_{n}$ with a (smaller) re-sample size $m$ that satisfies $m = m(n) \to \infty$ and $\dfrac{m(n)}{n} \to 0$, it can be shown that the (modified) bootstrap works. More generally, in recent years, the importance of a proper choice of the re-sampling distribution has become clear (see [a5], [a9] and [a10]).

The bootstrap can be an effective tool in many problems of statistical inference, for example, the construction of a confidence band in non-parametric regression, testing for the number of modes of a density, or the calibration of confidence bounds (see [a2], [a4] and [a8]). Re-sampling methods for dependent data, such as the block bootstrap, is another important topic of recent research (see [a2] and [a6]).

How to Cite This Entry:
Bootstrap method. Encyclopedia of Mathematics. URL: http://encyclopediaofmath.org/index.php?title=Bootstrap_method&oldid=11753
This article was adapted from an original article by Roelof Helmers (originator), which appeared in Encyclopedia of Mathematics - ISBN 1402006098. See original article