Home page for accesible maths 2.2 Properties of estimators

Style control - access keys in brackets

Font (2 3) - + Letter spacing (4 5) - + Word spacing (6 7) - + Line spacing (8 9) - +

2.2.1 Bias

The main reason for our interest in the expectation of an estimator is to show that it is unbiased.

Definition.

An estimator is unbiased for θ if

𝔼[θ^]=θ

where θ is the unknown true value of the parameter.

In the following example, suppose that we have an IID sample X1,,Xn from a population with mean μ and variance σ2.

TheoremExample 2.2.1 Sample mean

We can show that the sample mean is an unbiased estimator of the population mean since,

𝔼[X¯] =𝔼[1ni=1nXi]
=1ni=1n𝔼[Xi]
=1ni=1nμ
=nμn
=μ
TheoremExample 2.2.2 Sample variance

The sample variance is an unbiased estimator of the population variance,

𝔼[S2] =𝔼[1n-1i=1n(Xi-X¯)2]
=1n-1𝔼[i=1n(Xi2-2XiX¯+X¯2)]
=1n-1𝔼[i=1nXi2-nX¯2]
=1n-1(i=1n𝔼[Xi2]-n𝔼[X¯2])
=1n-1(i=1n(σ2+μ2)-n(σ2/n+μ2))
=1n-1(nσ2+nμ2-σ2-nμ2)
=σ2
Remark.

This result uses the definition of the variance

Var(X)=𝔼[X2]-𝔼[X]2

and the linearity properties of the expectation that were seen in Math230. It also uses the following results for the sample mean (also seen in Math230):

  1. 𝔼[X¯]=μ,

  2. Var(X¯)=σ2n.