UNIVERSITY OF NAIROBI

SECOND SEMESTER EXAMINATIONS 2019/2020

THIRDYEAR EXAMINATIONS FOR THE DEGREE OF
BACHELOR OFSCIENCE
STA/303/SMA 342: THEORY OF ESTIMATION
TIME: 2.00 P.M. -4.00 P.M.
DATE: DECEMBER 9, 2019

INSTRÜCTIONS:
Answer Question I and any other TWO Questions.

Question 1 30 marks)
terms as used in of estimation giving an illustration in each
(a) Discuss the following theory
case
Unbiasedness
i) Efficiency
iii) Completeness
iv) Minimal sufficiencyy
(v) BAN estimators 5 marks]

Distinguish between simple consistency and mean squared error consistency. Let X,
(b)
denote the mean of a random sample of size n from a population having p.d.f.
Sx,0) = 0e

X>0

x Show that the sequence {X,; n =1,2,...} is both simple and MSE consistent estimator of

[7 marks

Define sufficient statistic according to Jersey Neyman. Let X1,X3,-.,X, be a random

(
YA (N
-
Page 1 of4
 X

e
Sample from a Poisson distribution with
parameter 2. Show that = X, is sufficient
statistic for 2 and hence that
P| X, =t is a binomial distribution independent
of .
8 marks]
() Discuss maximum
likelihood estimation. Prove that maximum likelihood estimator
necessarily a function of sufficient statistics.
a
is
d 6 marks]
rsu
CL

() Discuss ta

(i) Point and interval estimation-

OPo

(ii) Classical and Bayesian estimation

14 marks]
Question 2 (20 marks)
(a Suppose Xi,X2...Xn is a random sample from a population where X
O with takes the value 1 or
respective probabilities 0 and 1-0. Show that the statistic
8 nT) is an unbiased estimator of
nn-1) Var X, where T =X, .Confirm that
2,-x
n-1 (8 marks)

(b) Suppose X is the number of successes in n trials with

probability p of success and if
Z=and Z, =*are
n+2
two estimators of p, for what values
of p will the second
estimator possess a smaller mean squared error than the first?
[12 marks]
If t(x) is a sufficient statistic for 6 and u =
p(1), show that

(1) pt) is also a sufficient statistic for .

(ii) ) is also sufficient for o(0).

S marks]

Page 2 of4
Question 3(20 marks)
(a) Describe briefly the method of moments estimation of point estimators, including for two or
more parameters.

Given that

( x , 0 ) .x -0)**

X-1,2,3. Show that f(x, 0) is a probability distribution function.

1-(1-0
Hence obtain the method of moment estimator of 0if the frequencies at x=1,2,3 are
respectively 18, 15 and 17 [10 marks]

(b) ) Discuss how Rao and Blackwell showed that sufficient statistics can be used to
determine UMVUEs.

(ii) Given that the necessary and sufficient condition for the existence of UMVUE is
-6
thata log L-
where t is a function of the sample observations only and A is a constant
independent of 6, show that var(t) = 2. [10 marks]

tD

Question 4: (20 marks)

(a) Let X1, X2 ..., Xn be a random sample ofsize n from a Poisson population with parameter 2.
Show that

) T=X, is sufficient and complete.

i

(ii) i s unique UMVUE of Var X. [10 marks]

n

(b) Suppose the developer of anew complex claims that 3 out of 5 buyers will prefer a two-
bedroom unit while his banker claims that 7 out of 10 buyers prefer a two-bedroom unit. In
previous predictions of this type the banker had been twice as reliable as the developer. If 12
ofthe next 15 rooms sold in the complex are two bedroomunits. o p
) Find Bayes estimate of the proportion of buyers who prefer a two bedroom unit

i) Assuming a normal population with the mean = 0.50 and variance 8 =0.09, find the
95% confidence interval for the population proportion. [10 marks]

t(T)-n = nA2
Page 3 of 4

ET E(T) O + na } -
=

ET) nE (T) n:A2 =

(r) nE - n°2 ()
 F's 9 t (1-0)o
-1 +)

Question 5: (20 marks)

(a) Let X1, X2, .,
X be a random sample from the density.
f(x,0) = O(1+x)) x>0

Find the MLE of 6 and use Cramer-Rao lower bound to find large sample variance of the
estimator.
[8 marks]
(b) ) Define MVB unbiased estimator for unknown
parameter 0. State the
characteristic equation for determining MVBUE explaining all the terms in the
equation.

ii) Suppose a random sample of size n is taken from a

population X that is N(4,o*)
where is known. Illustrate that MVB unbiased estimators are
necessarily MLEs
by obtaining MVBUE estinater of o*. [12 marks]

Page 4 of 4