Sufficiency (statistics)

In statistics, one often considers a family of probability distributions for a random variable X (and X is often a vector whose components are scalar-valued random variables, frequently independent) parameterized by a scalar- or vector-valued parameter, which let us call θ. A quantity T(X) that depends on the (observable) random variable X but not on the (unobservable) parameter θ is called a statistic. Sir Ronald Fisher tried to make precise the intuitive idea that a statistic may capture all of the information in X that is relevant to the estimation of θ. A statistic that does that is called a sufficient statistic.

Mathematical definition

The precise definition is this:

A statistic T(X) is sufficient for θ precisely if the conditional probability distribution of the data X given the statistic T(X) does not depend on θ.

An equivalent test, known as the Fisher's factorization criterion, is often used instead. If the probability density function (in the discrete case, the probability mass function) of X is f(x;θ), then T satisfies the factorization criterion if and only if functions g and h can be found such that

<math>

f(x;\theta)=g\left(T(x),\theta\right)h(x). <math>

This is a product in which one factor, h, does not depend on θ and the other depends on x only through T(x). The way to think about this is to consider varying x in such a way as to maintain a constant value of T(X) and ask whether such a variation has any effect on inferences one might make about θ. If the factorization criterion above holds, the answer is "none" because the dependence of the likelihood function f on θ is unchanged.

Examples

• If X₁, ...., X_n are independent Bernoulli-distributed random variables with expected value p, then the sum T(X) = X₁ + ... + X_n is a sufficient statistic for p.

This is seen by considering the joint probability distribution:

<math>

P(X=x)=P(X_1=x_1,X_2=x_2,\ldots,X_n=x_n). <math>

Because the observations are independent, this can be written as

<math>

p^{x_1}(1-p)^{1-x_1} p^{x_2}(1-p)^{1-x_2}\cdots p^{x_n}(1-p)^{1-x_n} <math>

and, collecting powers of p and 1 − p gives

<math>

p^{\sum x_i}(1-p)^{n-\sum x_i}=p^{T(x)}(1-p)^{n-T(x)} <math>

which satisfies the factorization criterion, with h(x) being just the identity function. Note the crucial feature: the unknown parameter (here p) interacts with the observation x only via the statistic T(x) (here the sum Σ x_i).

• If X₁, ...., X_n are independent and uniformly distributed on the interval [0,θ], then max(X₁, ...., X_n ) is sufficient for θ.

To see this, consider the joint probability distribution:

<math>

P(X=x)=P(X_1=x_1,X_2=x_2,\ldots,X_n=x_n). <math>

Because the observations are independent, this can be written as

<math>

\frac{H(\theta-x_1)}{\theta}\times \frac{H(\theta-x_2)}{\theta}\times\ldots\times \frac{H(\theta-x_n)}{\theta} <math>

where H(x) is the Heaviside step function. This may be written as

<math>

\frac{H\left(\theta-\max(x_i)\right)}{\theta^n} <math>

which shows that the factorization criterion is satisfied, again with h(x) being the identity function.

The Rao-Blackwell theorem

Since the conditional distribution of X given T(X) does not depend on θ, neither does the conditional expected value of g(X) given T(X), where g is any (sufficiently well-behaved) function. Consequently that conditional expected value is actually a statistic, and so is available for use in estimation. If g(X) is any kind of estimator of θ, then typically the conditional expectation of g(X) given T(X) is a better estimator of θ ; one way of making that statement precise is called the Rao-Blackwell theorem. Sometimes one can very easily construct a very crude estimator g(X), and then evaluate that conditional expected value to get an estimator that is in various senses optimal.