STATISTICAL INFERENCE

PART I
POINT ESTIMATION

1
 STATISTICAL INFERENCE
• Determining certain unknown properties of
a probability distribution on the basis of a
sample (usually, a r.s.) obtained from that
distribution
̂
Point Estimation: ̂  5 

Interval Estimation: 3    8 ( ) 

Hypothesis Testing: H 0 :   5
H1 :   5
2
 STATISTICAL INFERENCE
• Parameter Space ( or ): The set of all
possible values of an unknown parameter,
; .
• A pdf with unknown parameter: f(x; ), .
• Estimation: Where in ,  is likely to be?

{ f(x; ),  } The family of pdfs

3
 STATISTICAL INFERENCE
• Statistic: A function of rvs (usually a
sample rvs in an estimation) which does not
contain any unknown parameters.
2
X , S , etc
• Estimator of an unknown parameter :ˆ
A statistic used for estimating .
ˆ : estimator  U  X 1 , X 2 , , X n 
X : Estimator
An observed value

4
x : Estimate : A particular value of an estimator
 POINT ESTIMATION
• θ: a parameter of interest; unknown
• Goal: Find good estimator(s) for θ or its
function g(θ).

5
METHODS OF ESTIMATION

Method of Moments Estimation,

6
 METHOD OF MOMENTS
ESTIMATION (MME)
• Let X1, X2,…,Xn be a r.s. from a population
with pmf or pdf f(x;1, 2,…, k). The MMEs
are found by equating the first k population
moments to corresponding sample moments
and solving the resulting system of
equations.
Population Moments Sample Moments
1n k
k  E  X k  Mk   Xi
n i1
7
 METHOD OF MOMENTS
ESTIMATION (MME)
1  M 1 2  M 2 3  M 3
so on…
1 1
E  X    X i E  X    X i3
n n
1n
E  X    Xi
2 2 3

n i1 n i1 n i1

Continue this until there are enough equations

8
 EXAMPLES
• Let X~Exp().
• For a r.s of size n, find the MME of .
• For the following sample (assuming it is
from Exp()), find the estimate of :
11.37, 3, 0.15, 4.27, 2.56, 0.59.

9
 EXAMPLES
• Let X~N(μ,σ²). For a r.s of size n, find the
MMEs of μ and σ².
• For the following sample (assuming it is
from N(μ,σ²)), find the estimates of μ and
σ²:
4.93, 6.82, 3.12, 7.57, 3.04, 4.98, 4.62,
4.84, 2.95, 4.22

10
 DRAWBACKS OF MMES

• Although sometimes parameters are

• If moments does not exist, we cannot find

11
 MAXIMUM LIKELIHOOD
ESTIMATION (MLE)
• Let X1, X2,…,Xn be a r.s. from a population
with pmf or pdf f(x;1, 2,…, k), the
likelihood function is defined by
L 1 , 2 , , k x1 , x2 , , xn   f  x1 , x2 , , xn ;1 , 2 , , k 

  f  xi 1 , 2 , , k 
n

i 1

  f  xi ;1 , 2 , , k 
n

i 1

12
 MAXIMUM LIKELIHOOD
ESTIMATION (MLE)
• For each sample point (x1,…,xn), let
ˆ1  x1 ,..., xn  , ,ˆk  x1 ,..., xn 
be a parameter value at which
L(1,…, k| x1,…,xn) attains its maximum as a
function of (1,…, k), with (x1,…,xn) held fixed. A
maximum likelihood estimator (MLE) of
parameters (1,…, k) based on a sample (X1,…,Xn)
is
ˆ1  x1 ,..., xn  , ,ˆk  x1 ,..., xn 
• The MLE is the parameter point for which the
observed sample is most likely. 13
 EXAMPLES
• Let X~Bin(n,p), where both n and p are unknown. One
observation on X is available, and it is known that n is
either 2 or 3 and p=1/2 or 1/3. Our objective is to
estimate the pair (n,p).
x (2,1/2) (2,1/3) (3,1/2) (3,1/3) Max. Prob.
0 1/4 4/9 1/8 8/27 4/9
1 1/2 4/9 3/8 12/27 1/2

2 1/4 1/9 3/8 6/27 3/8

(2,1/ 3) if x  0
(2,1/ 2) if x  1

   
ˆ
n , ˆ
p x 
(3,1/ 2) if x  2
(3,1/ 2) if x  3 14
 MAXIMUM LIKELIHOOD
ESTIMATION (MLE)
• It is usually convenient to work with the logarithm
of the likelihood function.
• Suppose that f(x;1, 2,…, k) is a positive,
differentiable function of 1, 2,…, k. If a
supremum ˆ1 ,ˆ2 , ,ˆk exists, it must satisfy the
likelihood equations
 ln L 1 , 2 , , k ; x1 , x2 , , xk 
 0, j  1, 2,..., k
 j
• MLE occurring at boundary of  cannot be
obtained by differentiation. So, use inspection. 15
 MLE
• Moreover, you need to check that you are
in fact maximizing the log-likelihood (or
likelihood) by checking that the second
derivative is negative.

16
 EXAMPLES
1. X~Exp(), >0. For a r.s of size n, find the
MLE of .

17
 EXAMPLES
2. X~N(,2). For a r.s. of size n, find the
MLEs of  and 2.

18
 EXAMPLES
3. X~Uniform(0,), >0. For a r.s of size n,
find the MLE of .

19
INVARIANCE PROPERTY OF THE
MLE
• If ˆ is the MLE of , then for any function
(), the MLE of () is  ˆ .

Example: X~N(,2). For a r.s. of size n, the

20
 ADVANTAGES OF MLE
• Often yields good estimates, especially for
large sample size.
• Invariance property of MLEs
• Asymptotic distribution of MLE is Normal.
• Most widely used estimation technique.
• Usually they are consistent estimators. [will
define consistency later]

21
 DISADVANTAGES OF MLE
• Requires that the pdf or pmf is known
except the value of parameters.
• MLE may not exist or may not be unique.
• MLE may not be obtained explicitly
(numerical or search methods may be
required.). It is sensitive to the choice of
starting values when using numerical
estimation.
• MLEs can be heavily biased for small
samples.
22