Refer to QUESTION 1.
(a) What are the distributions of:
(i)
(x_(i))/(\sigma \sqrt(i));
(ii)
(x_(i)^(2))/(\sigma ^(2)i); and
(iii)
\sum_(i=1)^n (x_(i)^(2))/(\sigma ^(2)i).
(b) Prove or disprove that the expression of the asymptotic confidence interval (
nlongrightarrow\infty ) for
\sigma ^(2) is
hat(\sigma )^(2)(1 -1.96\sqrt((2)/(n)))
where
del^(2) is the maximum likelihood estimate of
\sigma ^(2) found in Question 1 (d)
(c) Given:
n=30 and Q=(30hat(\sigma )^(2))/(\sigma ^(2))?\chi _(30)^(2)
where
hat(\sigma )^(2) is the maximum likelihood estimate of
\sigma ^(2) found in Question 1 (d) when
n=30. Find
a and
b which satisfy
0.025=P(Q<=a)=P(Q>=b) and hence write down a
95% confidence interval for
\sigma ^(2).
(d) How can the
95% confidence interval in part (c) above be used to test the hypotheses
H_(0) :
\sigma ^(2)=1 versus
H_(1):\sigma ^(2)!=1 ?
(e) Evaluate your
95% confidence interval in part (b) at
n=30 and compare it with that in part (c). Which is better? Justify you answer
(4)