Name:
Stat 608 Final Examination May 8, 2012 (6:00 pm - 8:00 pm)
Instructions:
The total score is 100 points. There are ve questions. Show ALL your work!! Some questions have more than one parts. Check carefully to ensure that you dont miss any parts. Dont scratch on the line marked Score on the bottom of each page. You are allowed to use two 8.5x11(letter size) double-sided formula sheets and a calculator in this exam.
Stat 608
Final Exam
May 8, 2012
1. Let X1 , . . . , Xn be an iid sample from Bernoulli(1, ). (4points) (a) Show that
n i=1 Xi
is minimal sucient and so is
n 2 i=1 Xi
(Hint: Xi =1 or 0). .
(5points)
, and the Fisher information at . (b) Find the maximum likelihood estimator for ,
Page 2 of 9
Score:
Stat 608 (2points)
Final Exam
May 8, 2012
. (c) Find the asymptotic distribution of
(6points)
(d) Explain how to get an approximate 100% (1 ) condence interval for by using the likelihood ratio statistic.
(3points)
). (e) Find the asymptotic distribution of exp(
Page 3 of 9
Score:
Stat 608
Final Exam
May 8, 2012
2. Let X = (X1 , . . . , Xn ) be an iid sample from an exponential distribution, f (x | ) = ex where x > 0 and > 0. (5points) (a) Using Q(T, ) = 2T where T = n i=1 Xi as a pivot, give the explicit expressions necessary to construct a 100(1 )% condence interval for based on Q(T, ).
(5points)
(b) Suppose one takes an improper prior for , f () distribution for , f ( | X ).
where > 0. Find the posterior
Page 4 of 9
Score:
Stat 608 (4points)
Final Exam
May 8, 2012
(c) What is the Bayes estimator of under squared error loss?
(6points)
(d) Describe how to construct 100(1 )% Highest Posterior Density(HPD) interval for .
Page 5 of 9
Score:
Stat 608
Final Exam
May 8, 2012
3. Let X = (X1 , . . . , Xn ) be an iid sample from Uniform(0, ) where > 0. (3points) (a) Find the MLE for .
(6points)
(b) Find the form of LRT(likelihood ratio test) for H0 : = 1 vs. H1 : = 1.
Page 6 of 9
Score:
Stat 608 (6points)
Final Exam
May 8, 2012
(c) Find the form of likelihood ratio condence interval for .
(5points)
(d) Let Cn = X(n) , X(n) /1/n where X(n) = max{X1 , . . . , Xn }. Show that Cn is a 100(1 %) condence intervals for .
Page 7 of 9
Score:
Stat 608
Final Exam
May 8, 2012
4. Consider the problem from genetics of crossing two types of peas. The Mendelian theory states that the probabilities of the classications (a) round and yellow, (b) wrinkled and yellow, (c) round and green, and (d) wrinkled and green are 9/16, 3/16, 3/16, and 1/16, respectively. If, from 160 independent observations, the observed frequencies of these respective classications are 86, 35, 26, and 13, are these data consistent with the Mendelian theory? That is, test, with = 0.05, the hypothesis that the respective probabilities are 9/16, 3/16, 3/16, and 1/16. Note that P (X > 7.815) = 0.05 where X has a chi-squared distribution with the parameter 3. (6points) (a) Set up the hypotheses.
(8points)
(b) Test the hypotheses in (a) using Pearsons test.
(6points)
(c) Describe how to compute the P-value in (b).
Page 8 of 9
Score:
Stat 608
Final Exam
May 8, 2012
5. Let X = (X1 , . . . , Xn ) be random sample from F (x) where F (x) = P (X x) is a distribution function on the real line. We can estimate F (x) with the empirical distribution function, Fn (x): 1 Fn (x) = n where I (Xi x) = (10points) 1 0 if Xi x if Xi > x
n
I (Xi x)
i=1
(a) Show that at any xed value of x, Fn (x) is a consistent estimator for F (x).
(10points)
(b) Find the asymptotic distribution of Fn (x) at any xed value of x (justication required).
Page 9 of 9
Score:
