SQB7024 STATISTICAL INFERENCE
Tutorial 5
1. Casella and Berger, Exer 8.11 For samples of size n=1,4,16,64,100, from a normal
distribution with mean and known variance 2 , plot the power function of the following
tests. Take 0.05 .
(a) H 0 : 0 versus H1 : 0
(b) H 0 : 0 versus H1 : 0
2. Hogg, McKean and Craig, Exer 8.1.1 Let X1 ,....X n be a random sample from the
distribution that has the pdf
1 x 2
f x | = exp , x .
2 2
Let the simple hypothesis read H 0 : 0 versus H1 : 1 . Show that the
test of H 0 versus H1 may be carried out by use of the statistics X , and that if n=25 and
=0.05, the power of the test is 0.999 when H 0 is true.
3. Hogg, McKean and Craig, Exer 8.2.1 Let X have a pmf f x | = x 1 1 x , x 0,1 ,
1 1
zero elsewhere. We test the simple hypothesis H 0 : versus H1 : by taking
4 4
1
a random sample of size 10 and rejecting H 0 : if and only if the observed values
4
x1 ,....x10 of the sample observations are such that 10 X i 1 . Find the power function
i 1
1
, 0 .
4
4. Casella and Berger, Exer 8.12 Let X1 , X 2 be iid uniform , 1 . For testing H 0 :
0 versus H1 : 0 , we have two competing tests:
1 X1 : Reject H 0 if X1 0.95
2 X1 , X 2 : Reject H 0 if X1 X 2 C
(a) Find the value of C so that 2
has the same size as 1 .
(b) Calculate the power function of each test. Draw a well-labelled graph of each power
function.
5. Casella and Berger, Exer 8.18 In each of the following situation, calculate the p-value of
the observed data.
1 1
(a) For testing H 0 : versus H1 : , 7 successes are observedout of 10
2 2
Bernoulli trials.
(b) For testing H 0 : 1 versus H1 : 1 , X 3 is observed, where X ~Poisson .
