Scribd
Upload
5 views

Interval Estimation

Uploaded by

Aditha Buwaneka
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
5 views

Interval Estimation

Uploaded by

Aditha Buwaneka
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

I NTERVAL E STIMATION

Dr. Romesh Thanuja

(U O M) CM2111 1/6
O UTLINE

1 I NTRODUCTION

2 I NTERVAL ESTIMATOR VS INTERVAL ESTIMATES

3 F INDING I NTERVAL E STIMATORS FROM P IVOTAL F UNCTIONS

4 C ONSTRUCTING AN INTERVAL FROM A PIVOTAL FUNCTION

(U O M) CM2111 2/6
I NTRODUCTION

I NTRODUCTION

In the previous chapter, we discussed methods for obtaining point estimators of unknown
parameters.
In this chapter, we focus on learning about interval estimators, which provide a range of
possible values for an unknown parameter rather than just a single point estimate.

(U O M) CM2111 3/6
I NTERVAL ESTIMATOR VS INTERVAL ESTIMATES

I NTERVAL ESTIMATOR VS INTERVAL ESTIMATES

Let X = {X1 , X2 , . . . , Xn } be a random sample from FX parameterised by θ. Let T1 (X) and T2 (X)
be satisfies such that Ti (X) ≤ T2 (X) for any observed sample x.

The random interval:


[T1 (X), T2 (X)]
is an interval estimator of θ.

If the observed sample is x, then the interval:

[T1 (x), T2 (x)]

is an interval estimate (or confidence interval) of θ.

(U O M) CM2111 4/6
F INDING I NTERVAL E STIMATORS FROM P IVOTAL F UNCTIONS

F INDING I NTERVAL E STIMATORS FROM P IVOTAL F UNCTIONS

A way to construct 100(1 − α)% confidence set for θ is by using a pivotal function.

P IVOTAL F UNCTION
Consider a random sample X from FX parameterised by θ, where θ is a scalar parameter.
Suppose that we are interested in constructing an interval estimator of θ.

Let g(X, θ) be a function of X and θ which does not involve any unknown parameters other than θ.

g(X, θ) is a pivotal function if its distribution is known and does not depend on θ.

Note that a pivotal function defines a random variable, say Y = g(X, θ). By definition, the
distribution of Y does not depend on θ.

(U O M) CM2111 5/6
C ONSTRUCTING AN INTERVAL FROM A PIVOTAL FUNCTION

C ONSTRUCTING AN INTERVAL FROM A PIVOTAL FUNCTION

Suppose that we have a random sample X from FX parameterized by θ. To construct an interval


estimator of θ with a 100(1 − α)% confidence level using a pivotal function, we use the following
procedure.
1 Find a pivotal function g(X, θ) based on a reasonable point estimator.
2 Use the distribution of the pivotal function to find values k1 and k2 such that:

P(k1 ≤ g(X, θ) ≤ k2 ) = 1 − α

3 Manipulate g(X, θ) ≥ k1 and g(X, θ) ≤ k2 to make θ the reference point. This yields
inequalities of the form:

θ ≥ T1 (X, k1 , k2 ) and θ ≤ T2 (X, k1 , k2 )

for some function T1 (·) and T2 (·) independent of θ.


4 Construct the following interval

[T1 (X, k1 , k2 ), T2 (X, k1 , k2 )]

which is then an interval estimator of θ with a confidence coefficient of 1 − α.

Note that endpoints T1 and T2 are usually functions of one (and only one) of k1 and k2 .

(U O M) CM2111 6/6

You might also like