Date post: | 02-Jun-2018 |
Category: |
Documents |
Upload: | jose-pinto-de-abreu |
View: | 213 times |
Download: | 0 times |
of 149
8/10/2019 10739308_10203914552004027_1698851518_n
1/149
Mathematical Statistics witApplicationStudents SolutionsManu
KandethodyM.RamachandDepartmentofMathematicsandSta
UniversityofSouthFl
Tam
ChrisP.TsoDepartmentofMathematicsandSta
UniversityofSouthFl
Tam
AMSTERDAM BOSTON HEIDELBERG LONDON
NEW YORK OXFORD PARIS SAN DIEGO
SAN FRANCISCO SINGAPORE SYDNEYTOKYO
Academic Press is an imprint of Elsevier
8/10/2019 10739308_10203914552004027_1698851518_n
2/149
8/10/2019 10739308_10203914552004027_1698851518_n
3/149
Elsevier Academic Press30 Corporate Drive, Suite 400, Burlington, MA 01803, USA525 B Street, Suite 1900, San Diego, California 92101-4495, USA84 Theobalds Road, London WC1X 8RR, UK
Copyright 2009, Elsevier Inc. All rights reserved.No part of this publication may be reproduced or transmitted in any form or by any means, electronic ormechanical, including photocopy, recording, or any information storage and retrieval system, withoutpermission in writing from the publisher.
Permissions may be sought directly from Elseviers Science & Technology Rights Department in Oxford, UK:phone: (+44) 1865 843830, fax: (+44) 1865 853333, E-mail: [email protected]. You may alsocomplete your request on-line via the Elsevier homepage (http://elsevier.com), by selecting CustomerSupport and then Obtaining Permissions.
Library of Congress Cataloging-in-Publication Data
Applications submitted
For all information on all Elsevier Academic Press publications
visit our Web site at www.elsevierdirect.com
Typeset by: diacriTech, India
09 10 9 8 7 6 5 4 3 2 1
ISBN 13: 978-0-08-096443-0
8/10/2019 10739308_10203914552004027_1698851518_n
4/149
Contents
CHAPTER 1 Descriptive Statistics . . . . . . . .. . . . . . . .. . . . . . . . .. . . . . . . .. . . . . . . . .. . . . . . . 1
CHAPTER 2 Basic Concepts from Probability Theory . . . . . . . .. . . . . . . . .. . . . . . . .. . . . . 11
CHAPTER 3 Additional Topics in Probability . . . . . . . .. . . . . . . .. . . . . . . . .. . . . . . . .. . . . . 21
CHAPTER 4 Sampling Distributions . . . . . . . .. . . . . . . .. . . . . . . . .. . . . . . . .. . . . . . . . .. . . . 37
CHAPTER 5 Point Estimation . . . . . . . .. . . . . . . . .. . . . . . . . .. . . . . . . .. . . . . . . . .. . . . . . . .. . 49
CHAPTER 6 Interval Estimation. . . . . . . . .. . . . . . . .. . . . . . . . .. . . . . . . . .. . . . . . . .. . . . . . . 59
CHAPTER 7 Hypothesis Testing . . . . . . . .. . . . . . . . .. . . . . . . . .. . . . . . . . .. . . . . . . .. . . . . . . . 69
CHAPTER 8 Linear RegressionModels. . . . . . . . .. . . . . . . .. . . . . . . . .. . . . . . . .. . . . . . . . .. 81
CHAPTER 9 Design of Experiments. . . . . . . . .. . . . . . . .. . . . . . . . .. . . . . . . .. . . . . . . . .. . . . 87
CHAPTER 10Analysis of Variance . . . . . . . . .. . . . . . . .. . . . . . . . .. . . . . . . .. . . . . . . . .. . . . . . . 91
CHAPTER 11Bayesian Estimation and Inference. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
CHAPTER 12Nonparametric Tests. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
CHAPTER 13Empirical Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
CHAPTER 14Some Issues in Statistical Applications: An Overview . . . . . . . . . . . . . . . . . . 127
8/10/2019 10739308_10203914552004027_1698851518_n
5/149
8/10/2019 10739308_10203914552004027_1698851518_n
6/149
Chapter1Descriptive Statisti
EXERCISES 1.2
1.2.1. The suggested solutions:
For qualitative data we can have color, sex, race, Zip code and so on. For quantitative datawe can have age, temperature, time, height, weight and so on. For cross section data we can
have school funding for each department in 2000. For time series data we can have the crude
oil price from 1995 to 2008.
1.2.3. The suggested questions can be
1. What types of data the amount is?
2. Are these Federal Agency get same amount of money? If not, why?
3. Which Federal Agency should get more money? Why?
The suggested inferences we can make is
1. These Federal Agency get different amount of money.
2. The differences of money between the Agencies are kind of big.
EXERCISES 1.3
1.3.1. For stratified sample, we can say suppose we decide to sample 100 college students from the
population of 1000 (that is 10% of the population). We know these 1000 students come
from three different major, Math, Computer Science and Social Science. We have Math 200,
CS 400 and SS 400 students. Then we choose 10% of each of them Math 20, CS 40 and SS
40 by using random sampling within each major.
For cluster sample, we can say suppose we decide to sample some college students from the
population of 2000. We know these 2000 students come from 20 different countries and
we choose 3 out of the 20 countries by random sampling. Then we get all the individual
information from each of the 3 countries.
8/10/2019 10739308_10203914552004027_1698851518_n
7/149
2 CHAPTER 1 Descriptive Statistics
EXERCISES 1.4
1.4.1. By minitab
(a) Bar graphBar graph for the percent of road mileage
35.00%
C2
C1
30.00%
25.00%
20.00%
15.00%
10.00%
5.00%
0.00%Poor Mediocre Fair Good Very good
(b) Pie chart
PoorVery goodGoodFairMediocre
Category
Pie chart of the percent of road mileage
1.4.3. (a) Bar graph
40.00%
30.00%
20.00%C2
C1
10.00%
0.00%
Bar graph
Renewable
Energy
PetroliumNyclear
Electric Power
Natural
Gas
Coal
8/10/2019 10739308_10203914552004027_1698851518_n
8/149
Students Solutions Manu
(b) Pareto chart
Renewable
Energy
Petrolium Nyclear
Electric
Power
Natural
Gas
Coal
Percentage
Percent
Percentage
Percent
Cum %
0.40
40.0
40.0
0.23
23.0
63.0
0.22
22.0
85.0
0.08
8.0
93.0
0.07
7.0
100.0
C1
1.0
0.8
0.6
0.4
0.2
0.0
100
80
60
40
20
0
Pareto graph
(c) Pie chart
CoalNatural GasNyclear Electric PowerPetroliumRenewable Energy
Category
Pie chart of speciesspecies
1.4.5. (a) Bar graph
6
5
4
3
2
1
0
Count
C1
A B C D F
Bar graph
8/10/2019 10739308_10203914552004027_1698851518_n
9/149
4 CHAPTER 1 Descriptive Statistics
(b) Pie chart
A
BC
DF
Category
Pie chartspecies
1.4.7. (a) Pie chart
Mining
Construction
Manufacturing
Transportation
Wholesale
Retail
Finance
Services
Category
Pie chartspecies
(b) Bar graph
Servi
ces
Finan
ce
Retail
Who
lesale
Transp
ortation
Manu
factur
ing
Constru
ction
Minin
g
8000
7000
6000
5000
4000
3000
2000
1000
0
C1
C2
Bar graph
8/10/2019 10739308_10203914552004027_1698851518_n
10/149
Students Solutions Manu
1.4.9. Bar chart
20001990198019601900
80
70
60
50
40
30
20
10
0
C1
C2
Bar graph
1.4.11. (a) Bar graph
300
250
200
150
100
50
Accid
ents
0
C2
Bar graph
C1
C
hronic C
C
ancer
Diabetes
Heart
Kidney
Pneu
monia
Stroke
S
uicid
e
(b) Pareto graph
268.0
38.7
38.7
119.4
28.8
67.6
58.5
8.5
76.0
42.3
6.1
82.1
35.1
5.1
87.2
34.5
5.0
92.2
23.9
3.5
95.6
30.2
4.4
100.0
Percentage
Percent
Cum %
C1
Othe
r
Diabetes
Accid
ents
Pneu
monia
Heart
Cancer
Stroke C
700
600
500
400
300
200
1000
100
80
60
40
20
0
Percentage
Percent
Pareto graph
8/10/2019 10739308_10203914552004027_1698851518_n
11/149
6 CHAPTER 1 Descriptive Statistics
1.4.13.
90807560
9
8
7
6
5
4
3
2
1
0
Histogram
C1
Frequency
1.4.15. (a) Stem and leafStem-and-leaf of C1 N = 20
Leaf Unit = 10
1 4 7
3 4 99
8 5 00011
10 5 22
10 5 4455
6 5 6667
2 5 9
1 6 0
(b) Histogram
600580560540520500480
5
4
3
2
1
0
C1
Frequency
Histogram
8/10/2019 10739308_10203914552004027_1698851518_n
12/149
Students Solutions Manu
(c) Pie chart
526
542
546
553
558
565
568
572
595
605
475
493
499502
503
506
510
517
525
Category
Pie chartspecies
EXERCISES 1.5
1.5.1. Mean is 165.6667 and standard deviation is 63.15397
1.5.3. Data is 3,3,5,13 and standard deviation is 4.760952
1.5.5. (a) lower quantiles is 80, median is 95, upper quantiles is 115 and inter quantile range
is 35. The lower limit of outliers is 27.5 and upper limit of outliers is 167.5.
(b) The box plot is
(c) Therefore there are no outliers.
8/10/2019 10739308_10203914552004027_1698851518_n
13/149
8 CHAPTER 1 Descriptive Statistics
1.5.7.
li=1
fi(mi x) =l
i=1fi(mi)
li=1
x = nx nx = 0
1.5.9. (a) Mean is 33.105, variance is 177.0430 and range is 48.19.
(b) Lower quantile is 24.9225, median is 32 and upper quantiles is 42.985. The inter
quantile range is 18.0625. The lower limit of outliers is 2.17125 and upper limit ofoutliers is 70.07875. Therefore there are no outliers.
(c)
(d)
Histogram of y
Frequency
0
0
2
4
6
8
10 20 30 40 50 60y
1.5.11. (a) Mean is 110, standard deviation is 83.4847.
(b) 68%, 95%, 99.7%.
1.5.13. (a) Mean is 3.7433, variance is 3.501 and standard deviation is 1.871323.
8/10/2019 10739308_10203914552004027_1698851518_n
14/149
8/10/2019 10739308_10203914552004027_1698851518_n
15/149
10 CHAPTER 1 Descriptive Statistics
(b) Mean is 74.0625, median is 74, variance is 7.223892 and standard deviation is
2.68773.
(c)
The lower limit of outliers is 66 and upper limit of outliers is 82. Therefore we have nooutlier.
8/10/2019 10739308_10203914552004027_1698851518_n
16/149
Chapter2Basic Concepts from Probability Theo
EXERCISES 2.2
2.2.1. (a) S= {(R, R, R), (R, R, L), (R, L, R), (L, R, R), (R, L, L), (L, R, L), (L, L, R), (L, L, L)}(b) P= 78(c) P= 48(d) P= 38(e) P= 28
2.2.3. (a) P= 536(b) P= 536(c) P= 48(d) P= 38
2.2.5. PAB = {(H, H ), (H, T ), (T, H)} .2.2.7. (a) Probability space is
S= {(1, 1), (1, 2), (1, 3), (1, 4), (1, 5), (1, 6), (2, 1), (2, 2), (2, 3), (2, 4), (2, 5), (2, 6),(3, 1), (3, 2), (3, 3), (3, 4), (3, 5), (3, 6), (4, 1), (4, 2), (4, 3), (4, 4), (4, 5), (4, 6),
(5, 1), (5, 2), (5, 3), (5, 4), (5, 5), (5, 6), (6, 1), (6, 2), (6, 3), (6, 4), (6, 5), (6, 6)}(b) P= 636(c) P= 136
2.2.9. (a) Probability space isS= {N, N, N, S, S}Nstands for normal and Sstands for spoiled.
(b) P= 35 24= 620(c) No more than one means no or just one.
P= 35 24+ 2 25 34= .92.2.11. P= p + 2q
8/10/2019 10739308_10203914552004027_1698851518_n
17/149
12 CHAPTER 2 Basic Concepts from Probability Theory
2.2.13. (a) SinceA Bthen letA Ac = B, we knowP(A) + P(Ac) = P(B)by the axiom 3. SinceP(Ac) 0 by axiom 1 we knowP(A) + P(Ac) P(A)thenP(A) P(B).
(b) Let C= AB, A= Aa C and B= Ba C, by axiom 3 we know P(AB)=P(Aa) + P(Ba) + P(C).
Since P(A) = P(Aa
)+ P(C) and P(B) = P(Ba
)+ P(C) by axiom 3 again, we knowP(Aa) = P(A) P(C) and P(Ba) = P(B) P(C). Plug them back in previous equa-tion P(A B) = P(Aa) + P(Ba) + P(C) and we get the following equation P(A B)=P(A) + P(B) P(C)=P(A) + P(B) P(A B). IfA B=then by axiom 1 we knowP(A B) = 0 and just plug in we complete the proof.
2.2.15. (a) From 2.2.13 we knowP (A B) = P(A) + P(B) P(A B)and from axiom 2 we knowP(A B) 1, we can see thatP(A) + P(B) P(A B) 1 and that complete the proofP(A) + P(B) 1 P(A B).
(b) From 2.2.13 we knowP(A1 A2)= P(A1) + P(A2) P(A1 A2). From axiom 1we knowP(A1 A2) 0 it meansP(A1 A2) 0 and we can get the following
inequalityP (A1 A2) = P(A1) + P(A2) P(A1 A2) P(A1) + P(A2).2.2.17. (a) P= .24 + .67 .09 = .82
(b) P= 1 .82 = .18(c) P= 1 .09 = .91(d) P= 1 .09 = .91(e) P= 1 .82 = .18
2.2.19. (a) P= .55(b) P= .3(c) P= .7
2.2.21. (a) P= 3
5 2
4+ 2
5 1
4= .4(b) P= 35 24+ 35 24= .6(c) P= 2 35 24+ 25 14= .7(d) P= 35 24= .3
2.2.23. Without loss of generality let us assumeAn is increasing sequence then A1 A2 . . .An . . .. We know that ifA1 A2 . . . An . . . then A1 A2 . . .An . . . =
i=1
Ai=
limn An. From the condition we know limn An=
i=1
Ai and if we take probability on
both sides then limn
P(An) = P
i
=1
Ai = P limnAn
EXERCISES 2.3
2.3.1. (a) 45
(b) 1
8/10/2019 10739308_10203914552004027_1698851518_n
18/149
Students Solutions Manual
(c) 10
(d) 5400
(e) 2520
2.3.3. 1024
2.3.5. 53130
2.3.7. 155117520
2.3.9. 440
2.3.11. (a) p = .4313 .44425 .46798 .5263 1 = .04719(b) p = .0001189(c) p = .4313 .21419 .10344 1 1 = .009557
2.3.13. 180
2.3.15.
(a) p
= 1 365
364
...(365
20
+1)
36520 = .4114(b) p = 1 .2936 = .7063(c) Ifn = 23 thenp = .4927
2.3.17. p = 1 .27778 .16667 = .55562.3.19. (a) 7776
(b) 3.954 1021(c) 5.36447 1028(d) 3.022285 1012
2.3.21. The question is asking when the cell does the splitting to produce a child. There will be a
cell with half of the chromosomes. According to this understanding we have(a) 223
(b)
239
223
= .097416
EXERCISES 2.4
2.4.1. (a) .999
(b) 13
2.4.3. (a) P(A
|B)
+P(Ac
|B)
= P(AB)
P(B)
+ P (AcB)
P(B)
= P(B|A)P(A)+P(B|Ac)P(Ac)
P(B)
= P(B)P(B)
=1
(b) (i) ifP (A|B) + P(A|Bc)= 1 then we knowP (A|Bc)= 1 P(A|B)= P(Ac|B)thatmeansA andB are symmetric in probability. But it is not always true.
(ii) ifP (A|B) + P(Ac|Bc) = 1 then we knowP (Ac|Bc) = 1 P(A|B) = P(Ac|B)That means B and Bcs conditional probability are same, which is same as A and B are
independent. But that is not always true.
8/10/2019 10739308_10203914552004027_1698851518_n
19/149
14 CHAPTER 2 Basic Concepts from Probability Theory
2.4.5. IfA andB are independent thenP (A B) = P(A) P(B)
(i) P(AcB) = P(B)P(AB) = P(B)P(A)P(B) = (1P(A))(P(B)) = P(Ac)P(B)then we knowAc andB are independent.
(ii) According to(i)
just switchA
andB
and we can prove it.(iii) P(Ac Bc)= P(Bc) P(A Bc)= P(Bc) P(A) P(Bc)= (1 P(A))(P(Bc))=P(Ac) P(Bc).
2.4.7. P(E|F ) = 113= 452= P(E)thenE and Fare independent2.4.9. .1948
2.4.11. (a) P= .031125(b) P= .06
2.4.13. .8
2.4.15. (a) P(a dime is selected) =12
i=2P(a dime is selected|boxi is selected)P(boxi is selected)
=12
i=2i
12 P(the sum of dies= i)
= 2(1)+3(2)+4(3)+5(4)+6(5)+7(6)+8(5)+9(4)+10(3)+11(2)+12(1)12(36)= .583333
(b) P(box 4 is selected|a penny is selected)
= P(box 4 is selected & a penny is selected)P(a penny is selected)
= (3/36)(8/12)1 P(a dime is selected)
= .133332.4.17. P= .65752.4.19. P= .609762.4.21. (a) P(Accident rate) = .25 .086 + .257 .044 + .347 .056 + .146 .098 = .066548
(b) P(gourp4|Accident) = .146.098.0665 = .2152.4.23. P= .166672.4.25. P(Working) = P(B, C) + P(A, B,notC) + P(A,notB, C)
= [1 P(notB) P(notC) P(notB,notC)] + P(A)P(B|notC)P(notC)+ P(A) [P(notB) P(notB,notC)]
= [1 0.1 0.05 0.75 0.05] + 0.85 0.25 0.05 + 0.85 [0.1 0.75 0.05]= 0.8875 + 0.010625 + 0.053125 = 0.95
2.4.27. (a) P(same type of blood) = 1840 1739+ 1640 1539+ 440 339+ 240 139= .358974(b) P(type isO|type isB) = 1839= .4615
2.4.29. LetE denoteA ends up with all the money when he starts withi.
8/10/2019 10739308_10203914552004027_1698851518_n
20/149
Students Solutions Manual
LetF denote A ends up with all the money when he starts with N i. ForA starts withN imeansBstarts withibecauseNis the total moneyAandBhas so if we gotP(E)thenP(F) = 1 P(E).LetHdenote the event that the first flip lands heads andp denote the probability to have
Hon the first flip.P(E) = P(E|H)P(H) + P(E|HC)P(HC )
This probability represents thatA gets a head and combined with the probability ifB win
the first coin.
Now we letP(E) = P(E|H )p + P(E|HC)(1 p) = Piand define this as the first round.New, given that the first flip lands heads, the situation after the first bet is thatA has i + 1units andB has N (i + 1)units.Since the successive flips are assumed to be independent with a common probabilityp of
heads, it follows that, from that point on, As probability of winning all the money is exactly
the same as if the game were just starting with A having an initial fortune ofi + 1 and B
having an initial fortune ofN (i + 1)Therefore,P (E|H) = Pi+1and P (E|HC) = Pi+1. Letq to be 1 p,
Pi= pPi+1 + qPi1 i = 1,2, . . . , N 1
By applying the condition thatP0= 0 andPN= 1
Pi= 1Pi= (p + q)Pi= pPi+1 + qPi1
Pi+1 Pi= q
p(Pi Pi1) i = 1,2, . . . , N 1
After plug ini
We got
P2 P1= q
p(P1 P0) =
q
pP1
P3 P2= q
p(P2 P1) =
q
p
2P1
PN PN1= q
p(PN1 PN2) =
q
p
N1P1
If qp= 1
ThenP2 P1= P1and P2= 2P1P3= 3P1
PN= NP1PN= 1
8/10/2019 10739308_10203914552004027_1698851518_n
21/149
16 CHAPTER 2 Basic Concepts from Probability Theory
Which meansP1= 1NThereforePi= iN
If qp 1, thenPN P1= 1 P1= P1
q
p
1 qp
N11
qp
= P1
qp q
p
N1
qp
1 = P1
qp
qp
N1 qp
+ 1 = P1
1
qp
N1 qp
P1= 1 qp
1 qp N
Add all equations, we got
PN
P1=
P1 q
p + q
p
2
+ + q
pN
1
Add firsti 1 of all equations, we got
Pi P1= P1
q
p+
q
p
2+ +
q
p
i1= P1
q
p
1
qp
i11 qp
Pi=
1 qp1
qp
N
1
qp
i1 qp
=
1
qp
i1
qp
N
Then if we start withN i, just replacep byq and i byN i.
Qi=
1
pq
Ni1
pq
N if p
q 1
Qi= N i
Nif
p
q= 1
Pi + Qi= 1
EXERCISES 2.5
2.5.1. (a) c = e(b) P= e(c) P= 1 e e 2e2
8/10/2019 10739308_10203914552004027_1698851518_n
22/149
Students Solutions Manual
2.5.3. F(x) =
0, where x 5.2, where 5 x 0.3, where 0 x 3.7, where 3
x
6
1, where 6 6
2.5.5. p(x) =
0, where x 1.2, where 1 x 3.6, where 3 x 9.2, where x 9
2.5.7. (a) c = 19(b) P= .7037
(c) F(x) = 0, where x 0
1
27 x
3
, where 0 x 31, where x 3
2.5.9. f(x) =
0, where x 02x
(1+x)2 , where x 02.5.11. p = .7013 .55809 = .1432
p = .7364
2.5.13. f(t) =
0, where t 0(t)1
e
(t) , where 0 t
EXERCISES 2.6
2.6.1. m(t) = 16
et + e2t + e3t + e4t + e5t + e6tVar(x) = 2.9167
2.6.3. (a) E(Y) = 3.6E(Y2) = 17.2E(Y3) = 95.3
VAR(Y ) = 4.24(b) My(t) = et .1 + .05 + e2t .25 + e5t .4 + e6t .2
2.6.5. E(X)
= x xP(X = x) =
n=12nP(X
=2n)
=
n=12n 1
2n
=
n=11
= 2.6.7. a = 12
b = 12.6.9. (a) E(c) = cf (x) = c
(b) E(cg(x)) = cg(x)f (x) = c g(x)f(x) = cE(g(x))
8/10/2019 10739308_10203914552004027_1698851518_n
23/149
18 CHAPTER 2 Basic Concepts from Probability Theory
(c) E
i
gi(x)
=
i
gi(x)fi(x) =
i
g(x)f(x) =
i
E(gi(x))
(d) V(ax + b)= E((ax + b) E(ax + b))2 = E(ax + b aE(x) b)2 = E(a(x E(x))2= a2V(x)
Plug inb = 0 get another one.2.6.11. E(X) = c 1 + 0 = c
V(X) = Ex2 (E(x))2 = c2 c2 = 0CDF isF (X) = (x c)where is indicator function
2.6.13. Mx(t) = e(et1)E(X) = Mx(0) = ete(e
t1)|t=0= E(X2) = Mx (0) = ete(e
t1) + (et)2e(et1)|t=0= + 2V(X) = E(X2) (E(x))2 = + 2 2 =
2.6.15. (a) p = 1x+1 letq = 1 p = xx+1 wherexstart from 0 to infinity means number of failures
before the 1st success. Therefore the total number of trails is x+
1.
E(x) =
x=0xp(1 p)x = p
x=0
xp(q)xq 1q= qp
x=0
x + 1
x
(q)x = qpp2 = q
p by nega-
tive binomial
Another way to prove is
E(x) ==
x=0xp(1 p)x =p
x=0
x(q)x1q = pq
x=0
d(q)x
dq
= pq
d
x=0(q)x
dq = pqd
11qdq = pq
1
(1 q)2= pq
p2= q
p
E(x2) =
x=0x2p(1 p)x =pq
x=0
x2(q)x1 = pq
x=0
d(xqx)
dq
= pqd
x=0
(xqx)
dq= pq
d
1p
x=0
(xpqx)
dq= pq
d
11q E(x)
dq
=pq
d 1
1qqp
dq =pq
d q
(1q)2dq =
pq1 2q + q2 + 2q 2q2
(1 q)4 =pq(1 q2)
p4 =pq pq3
p4
V(x) = E(x2) (E(x))2 = pq pq3
p4 q
2
p2= pq pq
3 p2q2p4
= pq pq2(q + p)
p4
= pq(1 q)p4
= p2q
p4 = q
p2
8/10/2019 10739308_10203914552004027_1698851518_n
24/149
Students Solutions Manual
(b) Mx(t) =
x=0extp(1 p)x = p
x=0
(etq)x = p1etq= p
1(1p)et
Whenetq 1 and that isetq 1 take ln both sides give us ln(et) ln 11p
That is whent ln(1 p).
2.6.17. E(x) = 10 x(x2)dx + 21 x6x2x232 dx + 32 x (x3)22 dx =18+ 1 + 38= 1.52.6.19. Mx(t) =
0 e
xt12 e
xdx + 0 ext12 exdx = 1(t+1)(t1)2.6.21. My(t)=
0 e
tyeydy= 0 etyydy= 1(t) 0 e(t)yd(t )y= 1(t) e(t)y|0 = 1
(t)= (t) t and 0 SinceMy(t)= Mx(t)= (t) and MGF uniquely define the PDF therefore we know thatxhas the same distribution as y and
g(x) =
ey, 0 and 0 x
0, otherwise
8/10/2019 10739308_10203914552004027_1698851518_n
25/149
This page intentionally left blank
8/10/2019 10739308_10203914552004027_1698851518_n
26/149
Chapter3Additional Topics in Probabili
EXERCISES 3.2
3.2.1. (a) P(X = 7) = 107 (0.5)
7(0.5)107
= 120(0.5)7(0.5)3= 0.117
(b) P(X 7) = 1 P(8) P(9) P(10)= 1 0.044 0.010 0.001= 0.945
(c) P(X >0) = 1 P(0)= 1 0.001= 0.999
(d) E(X) = 10(0.5) = 5Var(X) = 10(0.5)(1 0.5) = 2.5
3.2.3. (a) P(Z >1.645) = 0.05, soz0= 1.645.(b) P(Z 1.645) = 0.95, soz0= 1.645.
3.2.5. (a) P(X 20) = P
Z 20 105
= P(Z 2) = 0.9772
(b) P(X >5) = PZ >5 10
5 = P(Z > 1) = 0.8413(c) P(12 X 15) = P
12 10
5 Z 15 10
5
= P(0.4 Z 1) = P(Z 1) P(Z
8/10/2019 10739308_10203914552004027_1698851518_n
27/149
22 CHAPTER 3 AdditionalTopics in Probability
(d) P(|X 12| 15) = P(15 X 12 15) = P(3 X 27)
= P3 10
5 Z 27 10
5
=P(
2.6
Z
3.4)
=P(Z
3.4)
P(Z 0 andx 0.
Since eachp(x) 0, then
p(x) xp(x) = x exx! = ex x
x!= e
1 + 1
1!+2
2!+
= e(e) = 1 here we apply Taylors expansion one.
This shows thatp(x) 0 and xp(x) = 1.3.2.13. The probability density function is given by
f(x) =
1
10, 0 x 10
0, otherwise
8/10/2019 10739308_10203914552004027_1698851518_n
28/149
Students Solutions Manual
Hence,
P(5 X 9) =9
5
1
10dx = 0.4.
Hence, there is a 40% chance that a piece chosen at random will be suitable for kitchen use.
3.2.15. The probability density function is given by
f(x) =
1
100, 0 x 100
0, otherwise
(a) P(60 X 80) = 8060 1100 dx = 0.2.(b) P(X >90) =
100
90
1
100dx = 0.1.
(c) There is a 20% chance that the efficiency is between 60 and 80 units; there is 10% chancethat the efficiency is greater than 90 units.
3.2.17. LetX= the failure time of the component. And X follows exponential distribution with rate0.05. Then the p.d.f. ofX is given by
f(x) = 0.05e0.05x, x >0.Hence,
R(10) = 1 F (10) = 1 10
0
0.05e0.05xdx = 1 1 e0.5 = e0.5 = 0.607.3.2.19. The uniform probability density function is given by
f(x) = 1, 0 x 1.
Hence,
P(0.5 X 0.65|X 0.75) = P(0.5 X 0.65 and X 0.75)P(X 0.75)
= P(0.5 X 0.65)P(X
0.75)
=
0.650.5
1dx
0.750 1dx
= 0.150.75
= 0.2
3.2.21. First, findz0such thatP(Z > z0) = 0.15.P(Z >1.036) = 0.15, soz0= 1.036.x0= 72 + 1.036 6 = 78.22
The minimum score that a student has to get to an A grade is 78.22.
8/10/2019 10739308_10203914552004027_1698851518_n
29/149
24 CHAPTER 3 AdditionalTopics in Probability
3.2.23. P(1.9 X 2.02) = P
1.9 1.960.04
Z 2.02 1.960.04
= P(1.5 Z 1.5) = 0.866
P(X 2.02) = 1 P(1.9 X 2.02) = 0.13413.4% of the balls manufactured by the company are defective.
3.2.25. (a) P(X >125) = P
Z > 125 11510
= P(Z >1) = 0.16
(b) P(X z2) = 0.5 andP(Z > z3) = 0.8
Using standard normal table, we can find thatz1= 0.842,z2= 0 andz3= 0.842.Then
y1= 0 + (0.842) 0.65= 0.5473 x1= exp(y1)= 0.58, similarly we can obtainx2= 1 andx3= 1.73.For the probability of surviving 0.2, 0.5 and 0.8 the experimenter should choose doses 0.58,
1 and 1.73, respectively.
3.2.29. (a) MX(t) = E(etX) =
0
etx 1
()x1ex/dx
= 1()
0
x1 exp1 t
x
dx
= 1()
0
1 tu1
eu
1 tduby lettingu = 1 t
xwith 1 t >0
= 1
()
1 t
0
u1 exp(
u)du
note that the integrand is the kernel density of(, 1)
= 1()
1 t
() 1
= (1 t)when t < 1
.
8/10/2019 10739308_10203914552004027_1698851518_n
30/149
Students Solutions Manual
(b) E(X) = MX(0) = (1 t)1()
t=0= , and
E(X2) = M(2)X (0) = ddt[(1 t)1()]
t=0
= (
+1)(1
t)2(
)2
t=0=(
+1)2
Then Var(X) = E(X2) E(X)2 = ( + 1)2 ()2 = 2.3.2.31. (a) First consider the following product
()() =
0
u1eudu
0
v1evdv
=
0
x2(1)ex2
2xdx
0
y2(1)ey2
2ydv by lettingu = x2 and v = y2
= 2
0
|x|21 e
x2
dx
2
0
|y|21 e
y2
dy noting that the integrands are
even functions
=
|x|21 ex2 dx
|y|21 ey2 dy
=
|x|21 |y|21 e(x2+y2)dxdy
Transforming to polar coordinates withx = r cos and y = r sin
()() =2
0
0
|r cos |21 |r sin |21 er2 rdrd
=
0
r2+22er2
rdr
20
(cos )21(sin )21 d
=1
2
0
s+1esds
4
/20
(cos )21(sin )21d
by lettings = r2
= ( + )2/20
(cos )21(sin )21d
= ( + )21
0
t1/2(1 t)1/2 12
t(1 t) dtby lettingt= cos2
8/10/2019 10739308_10203914552004027_1698851518_n
31/149
26 CHAPTER 3 AdditionalTopics in Probability
= ( + )1
0
t1(1 t)1dt
= ( + )B(, )
Hence, we have shown that
B(, ) = ()()( + ) .
(b) E(X) = 1B(, )
10
x x1(1 x)1dx = B( + 1, )B(, )
10
x(+1)1(1 x)1B( + 1, ) dx
= B( + 1, )B(, )
1 = ( + 1)()( + + 1)
( + )()()
= ()()
( + )( + )( + )()() =
+ , and
E(X2) = 1B(, )
10
x2 x1(1 x)1dx = B( + 2, )B(, )
10
x(+2)1(1 x)1B( + 2, ) dx
= B( + 2, )B(, )
1 = ( + 2)()( + + 2)
( + )()()
= ( + 1)()()( + )( + + 1)( + )
( + )()()
= ( + 1)( + )( + + 1) .
Then Var(X)
=E(X2)
E(X)2
=
( + 1)
( + )( + + 1)
+
2
=
( + )2
( + + 1).
3.2.33. In this case, the number of breakdowns per month can be assumed to have Poisson
distribution with mean 3.
(a) P(X= 1)= e3311! = 0.1494. There is a 14.94% chance that there will be just onenetwork breakdown during December.
(b) P(X 4) = 1 P(0) P(1) P(2) P(3) = 0.3528. There is a 35.28% chance thatthere will be at least 4 network breakdowns during December.
(c) P(X7)=7
x=0e33x
x! =0.9881. There is a 98.81% chance that there will be at most 7network breakdowns during December.
3.2.35. (a) P(1< X
8/10/2019 10739308_10203914552004027_1698851518_n
32/149
Students Solutions Manual
The probability that an acid solution made by this procedure will satisfactorily etch a
tray is 0.6442.
(b) P(1< X
8/10/2019 10739308_10203914552004027_1698851518_n
33/149
28 CHAPTER 3 AdditionalTopics in Probability
Thus, ifc= 1/4, then 11 11 f (x, y)dxdy= 1. And we also see thatf (x, y) 0 for all xandy. Hence,f (x, y)is a joint probability density function.
3.3.5. By definition, the marginal pdf ofXis given by the row sums, and the marginal pdf ofY is
obtained by the column sums. Hence,
xi 1 3 5 otherwisefX(xi) 0.6 0.3 0.1 0
yi 2 0 1 4 otherwisefY(yi) 0.4 0.3 0.1 0.2 0
3.3.7. FromExercise 3.3.5we can calculate the following.
P(X = 1|Y= 0) = P(X = 1, Y= 0)fY(0)
= 0.10.3
= 0.33.
3.3.9. (a) The marginal ofXis
fX(x) =2
x
f (x, y)dy =2
x
8
9xydy = 4
9(4x x3), 1 x 2.
(b) P(1.5< X 1) =1.75
1.5
2
x
8
9xydy
dx =
1.751.5
4
9
4x x3 dx
= 49 2x2
x4
4 1.75
1.5
= 0.2426.
3.3.11. Using the joint density inExercise 3.3.9we can obtain the joint mgf of(X, Y )as
M(X,Y )(t1, t2) = E(et1X+t2Y) =2
1
2x
et1x+t2y8
9xydydx
=2
1
8
9xet1x
2
x
et2yydy
dx =
21
8
9xet1x
K x
t2et2x + 1
t22et2x
dx
whereK = e2t2
t22(2t2 1)
= 89
K
21
xet1xdx 89t2
21
x2e(t1+t2)xdx + 89t22
21
xe(t1+t2)xdx
= 89
K
x
t1et1x 1
t21et1x
21
8/10/2019 10739308_10203914552004027_1698851518_n
34/149
Students Solutions Manual
89t2
x2
t1 + t2e(t1+t2)x 2x
(t1 + t2)2e(t1+t2)x + 2
(t1 + t2)3e(t1+t2)x
2
1
+ 89t2
2
x
t1+
t2e(t1+t2)x 1
(t1+
t2)2e(t1+t2)x
2
1
After simplification we then have
M(X,Y) (t1, t2) =t1 + 3t2 t21 3t22 4t1t2 + t21 t2 + 2t1t22+ t32
t22 (t1 + t2)3 et1+t2 + (2t2 1)(1 t1)
t21 t22
et1+2t2
+
t1 3t2 + 2t21+ 6t22+ 8t1t2 4t21 t2 + 8t1t22 4t32t22 (t1 + t2)3
+ (2t2 1)(2t1 1)t21 t
22
e2t1+2t2
3.3.13. (a) fX(x) = n
y=0f (x, y) = n
y=0
6xy
n(n + 1)(2n + 1)
2 = 36x2[n(n + 1)(2n + 1)]2
ny=0
y2
= 6x2
n(n + 1)(2n + 1) , x = 1,2, . . . , n.
fY(y) =n
x=0f (x, y) =
nx=0
6xy
n(n + 1)(2n + 1)2
= 36y2
[n(n + 1)(2n + 1)]2n
x=0x2
= 6y2
n(n + 1)(2n + 1) , y = 1,2, . . . , n.
Giveny = 1,2, . . . , n, we have
f (x|y) = f (x, y)fY(y)
=
6xy
n(n + 1)(2n + 1)
2
6y2
n(n + 1)(2n + 1)
= 6x2
n(n + 1)(2n + 1) , x = 1,2, . . . , n.
(b) Givenx = 1,2, . . . , n, we have
f (y|x) = f (x, y)fX(x)
=
6xy
n(n + 1)(2n + 1)2
6x2
n(n + 1)(2n + 1)
= 6y2
n(n + 1)(2n + 1) , y = 1,2, . . . , n.
3.3.15. (a) E(XY) = x,y
xy f (x, y) =3
x=1
3y=1 xy f (x, y) =
35
12 .
(b) E(X) = x,y
x f (x, y) =3
x=1
3y=1
x f (x, y) = 53
, and
E(Y) = x,y
y f (x, y) =3
x=1
3y=1
y f (x, y) = 116
.
8/10/2019 10739308_10203914552004027_1698851518_n
35/149
30 CHAPTER 3 AdditionalTopics in Probability
Then, Cov(X, Y ) = E(XY) E(X)E(Y) = 3512
53 11
6= 5
36.
(c) Var(X) =
x,y[x E(X)]2 f (x, y) =
3
x=13
y=1 x 53
2 f (x, y) = 5
9, and
Var(Y ) = x,y
[y E(Y)]2 f (x, y) = 3x=1
3y=1
y 116
2 f (x, y) = 2336
.
Then,XY= Cov(X, Y )
Var(X)Var(Y )= 5/36
(5/9)(23/36)= 0.233.
3.3.17. Assume thata and c are nonzero.
Cov(U, V ) = Cov(aX + b, cY+ d) = acCov(X, Y ),
Var(U) = Var(aX + b) = a2Var(X), and Var(V ) = Var(cY+ d) = c2Var(Y ).
Then, UV
=
Cov(U, V )
Var(U)Var(V ) =
acCov(X, Y )
a
2Var(X)c
2Var(Y ) =
ac
(ac)
2XY
= ac|ac| XY=
XY, if ac >0
XY, otherwise .
3.3.19. We fist state the famous CauchySchwarz inequality:
|E(XY)|
E(X2)E(Y2)and the equality holds if and only if there exists some constant
and, not both zero, such thatP ( |X|2 = |Y|2) = 1.Now, consider
|XY| = 1
Cov(X, Y )Var(X)Var(Y )
= 1 |Cov(X, Y )| =
Var(X)Var(Y )
|E(X X)(Y Y)| =
E(X X)2E(Y Y)2
By the CauchySchwarz inequality we have
P
|X X|2 = |Y Y|2 = 1
P(X X= K(Y Y)) = 1 for some constantK P(X = aY+ b) = 1 for some constantsa andb.
3.3.21. (a) First, we compute the marginal densities.
fX(x) = x
f (x, y)dy = x
eydy = ex, x 0, and
fY(y) =y
0
f (x, y)dx =y
0
eydx = yey, y 0.
8/10/2019 10739308_10203914552004027_1698851518_n
36/149
Students Solutions Manual
For giveny 0, we have the conditional density as
f (x|Y= y) = f (x, y)fY(y)
= ey
1
yey
= 1y
, 0 x y.
Then,(X|Y= y)followsUniform(0, y). Thus,E(X|Y= y) = y2
.
(b) E(XY) = y xxy f (x, y)dxdy = 0y
0 xyeydx
dy = 0 12 y3eydy = 3,
E(X) = x
x fX(x)dx =
0 xexdx = 1, and
E(Y) = y
y fY(y)dy =
0 y2eydy = 2.
Then, Cov(X, Y ) = E(XY) E(X)E(Y) = 3 1 2 = 1.(c) To check for independence ofX andY
fX(1)fY(1) = e2 = e1 = f (1, 1).
Hence,X and Yare not independent.
3.3.23. Let2 = Var(X) = Var(Y ). SinceX and Yare independent, we have Cov(X, Y )= E(XY) E(X)E(Y) = E(X)E(Y) E(X)(Y) = 0. Then, Cov(X, aX+Y ) = aCov(X, X)+Cov(X, Y ) =aVar(X)= a2, and Var(aX + Y )= a2Var(X) + Var(Y )= (a2 + 1)2. Thus, X, aX+Y=
Cov(X, aX + Y )Var(X)Var(aX + Y ) =
a22(a2 + 1)2
aa2 + 1
.
EXERCISES 3.4
3.4.1. The pdf ofX is fX(x) = 1a if 0 < x < aand zero otherwise.
FY(y) = P(Y < y) = P(cX + d < y) = P
X 0.
3.4.7. LetU= X + YandV= Y.ThenX = U V andY= V, and
J=
x
u
x
v
y
u
y
v
=
1 10 1 = 1.
Then the joint pdf ofUand Vis given by
fU,V(u, v) = fX,Y(u v, v) |J| = fX,Y(u v, v).
Thus, the pdf of U is given byfU(u) = fX,Y(u v, v)dv.
8/10/2019 10739308_10203914552004027_1698851518_n
38/149
Students Solutions Manual
3.4.9. (a) Here let g(x) = x
, and hence, g1(z) = z+ . Thus, ddz
g1(z) = . Also,
fX(x) = 12
e 12
x
2, < x < . Therefore, the pdf ofZ isfZ(z) = fX(g1(z))
ddz g1(z) = 12 e 12 z2 , < z < , which is the pdf ofN (0, 1).(b) The cdf ofUis given by
FU(u) = P(U u) = P
(X )22
u
= Pu X
u
= Pu Z u
=
u
u
12
e12 z
2dz = 2
u
0
12
e12 z
2dz, since the integrand is an even function.
Hence, the pdf ofUis fU(u)=
d
duFU(u)
= 2
2e
u2
1
2u = 1
2u
12 e
u2 , u >0 and
zero otherwise, which is the pdf of2(1).
3.4.11. Since the support of the pdf ofV isv > 0, then g(v)= 12 mv2 is a one-to-one function onthe support. Hence, g1(y)=
2ym
. Thus, dde
g1(y)= 12
m2y
2m
. Therefore, the pdf ofE is
given by
f(y) = fV(g1(y)) ddy g1(y)
= c 2ym e2ym
12my
= c
2ym3
e2y
m , y >0.
3.4.13. LetU=
X2 + Y2 andV= tan1
YX
. HereUis considered to be the radius and Vis the
angle. Hence, this is a polar transformation and hence is one-to-one.ThenX = Ucos V andY= Usin V, and
J=
x
u
x
v
y
u
y
v
=cos v u sin vsin v u cos v
= u cos2 v + u sin2 v = u.
Then the joint pdf ofUand Vis given by
fU,V(u, v) = fX,Y(uv, v) |J| = 1
22exp
1
22
u2 cos2 v + u2 sin2 v
u.
= u22
e u2
22 , u >0, 0 v 0.
8/10/2019 10739308_10203914552004027_1698851518_n
39/149
34 CHAPTER 3 AdditionalTopics in Probability
Apply the result in Exercise 3.4.14with = 2. We have the joint pdf ofU= XY2 and V= YasfU,V(u, v) = 12 e(u+v), v > 2u, v >0. Thus, the pdf ofUis given by
fU(u) = v
fU,V(u, v)dv
=
0
1
2e(u+v)dv = 1
2eu, u 0
2u
1
2e(u+v)dv = 1
2eu, u
8/10/2019 10739308_10203914552004027_1698851518_n
40/149
Students Solutions Manual
3.5.5. Apply Chebyshevs theorem we have
P
Xn
n p
2) = P
X100 22/
100
> 2 22/
100
P(Z >0) = 0.5, whereZ N(0, 1).3.5.13. First note thatE(Xi) = 1/2 and Var(Xi) = 1/12. Then, by CLT we know thatZn= Snn/2n/12 =
X1/21/12/
n
approximately followsN (0, 1)for largen.
8/10/2019 10739308_10203914552004027_1698851518_n
41/149
8/10/2019 10739308_10203914552004027_1698851518_n
42/149
Chapter
4Sampling Distributio
EXERCISES 4.1
4.1.1. (a) There are 53 = 10 equally likely possible samples of size 3, so the probability for each
is1 10without replacement:
X M S
(2, 1, 0) 1 1 1(2, 1, 1) 2 3 1 21 3(2, 1, 2) 1 3 1 41 3(2,0,1) 1 3 0 21 3(2,0,2) 0 0 4(2,1,2) 1 3 1 41 3(1,0,1) 0 0 1(1,0,2) 1 3 0 21 3(1,1,2) 2 3 1 21 3(0,1,2) 1 1 1
(i)
X 1 2/3 1/3 0 1/3 2/3 1p(X) 1/10 1/10 2/10 2/10 2/10 1/10 1/10
(ii)
M 1 0 1p(M) 3/10 4/10 3/10
(iii)S 1
7 3 2
13 3
p(S) 3/10 4/10 1/10 2/10
8/10/2019 10739308_10203914552004027_1698851518_n
43/149
38 CHAPTER 4 Sampling Distributions
(iv)E
X = (1)1 10+ 2 31 10+ 1 31 10+ 0 1 10+ 1 32 10
+
23
1
10
+ (1)
1
10
= 0
E
X2 = (1)2 1 10+ 2 32 1 10+ 1 32 1 10+ 02 2 10
+
13
2 210
+
23
2 110
+ 12
1
10
= 1 3
Var
X = EX2 EX2 = 1
3 02 = 1
3
(b) We can get 53 = 125 samples of size 3 with replacement4.1.3. Population: {1,2,3}.p(x) = 1 3, forx in {1,2,3}
(a) = 1N
Ni=1
ci= 2, 2 = 1NN
i=1(ci )2 = 2 3
(b)
Sample X Sample X Sample X
(1,1,1) 1 (2,1,1) 11 3 (3,1,1) 12 3
(1,1,2) 11 3 (2,1,2) 12 3 (3,1,2) 2
(1,1,3) 12 3 (2,1,3) 2 (3,1,3) 21 3
(1,2,1) 11 3 (2,2,1) 12 3 (3,2,1) 2
(1,2,2) 12 3 (2,2,2) 2 (3,2,2) 21 3
(1,2,3) 2 (2,2,3) 21 3 (3,2,3) 22 3
(1,3,1) 12 3 (2,3,1) 2 (3,3,1) 21 3
(1,3,2) 2 (2,3,2) 21 3 (3,3,2) 22 3
(1,3,3) 21 3 (2,3,3) 22 3 (3,3,3) 3
X 1 11 3 12 3 2 21 3 22 3 3
p(X) 1/27 1/9 2/9 7/27 2/9 1/9 1/27
(c) E
X
=X
x p(x) = 2, E
X2
=X
x2 p(x) = 42 9, then VarX
= 2 9
4.1.5. Since
i
(xi x)2 =
i
x2i nx2, we have E(S)2 = 1n
i
EX2i nEX2nAssuming the sampling from a population with mean and variance2, we have
E
S2 = 1
nn
2 + 2
2
n+ 2
8/10/2019 10739308_10203914552004027_1698851518_n
44/149
Students Solutions Manual
= 2 2
n
=
n 1n
2 < 2 = E
S2
4.1.7. LetXbe the weight of sugarX N( = 5lb, = 0.2lb)
ThenX = 1n
ni=1
Xiis the mean weight, wheren = 15.
By Corollary 4.2.2,E(X) = , and Var(X) = 2n
. ThenX N(5, 0.22 15), and X50.22 15
= ZN(0, 12). Therefore, the probability requested is P(0.2< X 5< 0.2) = P(
15< Z
170) = P(Z >226) = 04.1.11. Let X be the time. X
N(
=95, 2
=102). Then X9510
=Z
N(0, 12). Therefore,
P( X< 85) = P( Z< 1) = 0.8413, or 84.13% of measurement times will fall below85 seconds.
4.1.13. According the information, = 215 and= 35.(a) Ifn = 55, we can assumeX N(, ), thenPX >230 = PZ > 2302153555
= 0.0007
(b) Ifn = 200, we can assumeX N(, ), thenPX >230 = PZ > 23021535200
= 0
(c) Ifn = 35, we can assumeX N(, ), thenPX >230 = PZ > 2302153535
= 0.0056(d) Increasing the sample size, decrement the probability
4.1.15. LetTbe the temperature.
Sincen=60, we assumeT N (98.6, 0.952). ThenT N (98.6, 0.952 60). Therefore, P (T99.1) = 0
EXERCISES 4.2
4.2.1. We have thatY 2(15)(a) We can see, for example in a table, thatP (Y 6.26) = 0.025. Theny0= 6.26(b) Choosing upper and lower tail area to 0.025, and since P(Y 27.5)= 0.975, and
P(Y 6.26)= 0.025, then P (a < Y < b )= 0.95, then b= 20.975,15 = 27.5, a=20.025,15= 6.26
(c) P(Y
22.307)=
1
P(Y 2) = 0.4232
8/10/2019 10739308_10203914552004027_1698851518_n
45/149
40 CHAPTER 4 Sampling Distributions
4.2.5. SinceX1, X2, . . . , X5are i.i.d.N (55, 223), thenY=5
i=1(Xi55)2
223 2(5)
(a) Since Z = Y n
X552223 , and
n
X552223 2(1), Zis Chi-square distributed with 4 degrees
of freedom andYis Chi-square distributed with 5 degrees of freedom
(b) Yes
(c) (i)P (0.62 Y 0.76) = 0.0075 (ii)P (0.77 Z 0.95) = 0.0251
4.2.7. Since the random sample comes from a normal distribution, (n1)S2
2 2(n1).
Setting the upper and lower tail area equal to 0.05, even this is not the only choices,and using
a Chi-square table withn 1 = 14 degrees of freedom, we have (n1)b2
= 20.95,14= 23.68,and (n1)a
2 = 20.05,14= 6.57. Then, with= 1.41,b = 3.36, anda = 0.93
4.2.9. SinceT t8(a) P(T 2.896) = 0.99(b) P(T
1.860)
=0.05
(c) Sincet-distribution is symmetric, we finda such thatP(T > a) = 0.012 . Thena = 3.35
4.2.11. According with the information,=11.4, n=20, y= 11.5, ands= 2, thent= ys
n=
0.224. The degrees of freedom are n 1 = 19, so the critic value is 1.328 at = 0.05-level.Then, the data tend to agree with the psychologist claims.
4.2.13. If X 2(v), then X
= v2 , = 2
, then E(X)= v2 (2)= v and Var(X)= v2 (2)2 = 2v
4.2.15. IfX1, X2, . . . , Xnis fromN (, 2)
then, by Theorem 4.2.8,
(n
1)S2
2 is from2
(n1)then, by Exercise 4.2.13, Var
(n1)S2
2
= 2(n 1)
Since Var(aX) = a2Var(X), (n1)24
Var(S2) = 2(n 1)Simplifying after multiplying by
4
(n1)2 , we obtain Var(S2) = 24
n1
4.2.17. IfX and Yare independent random variables from an exponential distribution with com-
mon parameter = 1, then using 4.2.16 with n= 1, 2X 2(2) and 2Y 2(2) thenXY= 2X2Y F (2, 2)
4.2.19. IfX
F (9,12)
(a) P(X 3.87) = 0.9838(b) P(X 0.196) = 0.01006(c) F0.975(9,12) = 0.025 thenF0.975= 3.4358.
0.025 = P(X < F0.975) = P
1X
> 1F0.975
, where 1
X F (12,9)
Then 1F0.975
= 3.8682 andF0.975= 0.258518. Thus,a = 0.2585, b = 3.4358
8/10/2019 10739308_10203914552004027_1698851518_n
46/149
Students Solutions Manual
4.2.21. IfX F (n1, n2)the PDF is given by
f(x) =
[(n1 + n2)/2](n1/2)(n2/2)
n1
n2
n1
n2x
n121
1 + n1n2
x
(n1+n2)/2, 0< x <
0, otherwise
Then
EX =
0
x[(n1 + n2)/2]
(n1/2)(n2/2)
n1
n2
n1
n2x
n121
1 + n1n2
x
(n1+n2)/2dx
= [(n1 + n2)/2](n1/2)(n2/2)
n1
n2
n1
n2
n121
0
xn12
1 + n1
n2x
(n1+n2)/2dx
Let y
=1
1
+ n1n2
x1
then x
= n1n2
1y(1
y)1 and dx
= n1n2
1(1
y)2dy and
limx
1
1 + n1
n2x1= 1, then
EX = [(n1+n2)/2](n1/2)(n2/2)
n1n2
n22
n2n1
n22+1 1
0
yn22 (1 y)n12 dy, which converges forn1 >2.
For > 0, > 01
0
y1(1 y)1dy = ()()(+) , where () =
0 x
1exdx with the
property() = ( 1)( 1).
ThenEX = [(n1 + n2)/2](n1/2)(n2/2)
n1
n2
n22
n2
n1
n22+1 (n1/2 + 1)(n2/2 1)
[(n1 + n2)/2]
= n2n1
[(n1 + n2)/2](n1/2)(n2/2 1)(n2/2 1)
n12
(n1/2)(n2/2 1)[(n1 + n2)/2]
= n2n2 2
, n2 >2
Similarly,
EX2 = [(n1 + n2)/2](n1/2)(n2/2)
n1
n2
n12
n2
n1
n22+2
10
yn12 (1 y)
n223dy, which converges forn2 >4
= [(n1 + n2)/2](n1/2)(n2/2
1)(n2/2
2)(n2/2
2)
n2
n12 (n1/2 + 1)(n1/2)(n1/2)(n2/2 2)
[(n1
+n2)/2
]=
n2
n1
2 n1(n1 + 2)(n2 2)(n2 4)
, n2 >4.
Now, Var(X) = EX2 (EX)2. Therefore,
EX = n2n2 2
, n2 >2 and Var X =n22(2n1 + 2n2 4)
n1(n2 2)2(n2 4)
8/10/2019 10739308_10203914552004027_1698851518_n
47/149
42 CHAPTER 4 Sampling Distributions
4.2.23. IfX1, X2, . . . , Xn1 is a random sample from a normal population with mean 1 and vari-
ance2 and ifY1, Y2, . . . , Yn2 is a random sample from an independent normal population
with mean2 and variance 2, then X N
1,
2
n1
, Y N
2,
2
n2
,
(n1 1)S212
2(n11), and (n
2 1)S2
22
2(n21).
ThenX Y N
1 2, 2n1 + 2
n2
and
(n11)S212
+ (n21)S222
2(n1+n22)then XY(12)
2n1
+ 2n2
N(0, 12)and (n11)S21
2 + (n21)S22
2 2(n1+n22)
Then, since the samples are independent, we have by definition that
X Y (1 2)2
n1+
2
n2
(n1 1)S212
+ (n2 1)S22
2
[n1 + n2 2]
T(n1+n22)
This after simplification becomes:
X Y (1 2)(n1 1)S21+ (n2 1)S22
n1 + n2 2
1
n1+ 1
n2
T(n1+n22) Q.E.D.
4.2.25. IfX 2(v)withv >0, then the pdf ofX is given by
f(x) =
1
(v/2)2v/2ex/2xv/21, 0 < x <
0, x 0
Then, by definition of MGF,
MX(t) = 1
(v/2)2v/2
0
ex(t1/2)xv/21dx =
1
1 2tv/2
0
ewwv/21(v/2)
dw
= (1 2t)v/2, t 0, then the cumulative distribution of1is
F1 (t) =t
0
1
10ex/10dx = ex/10
t
0= 1 et/10.
Let Y represent the life length of the system, then Y=min(1, 2) and FY(y) = 1[1 Fi (y)]2, then the pdf ofYis of the formfY(y) = 2fi (y)[1 Fi (y)] and is given by
fy(y) =
1
5ey/5, 0 < y <
0, otherwise
4.3.3. X1, X2take values 0, 1;X3take values 1,2, 3, andY1= min {X1, X2, X3}Since the values ofX1, X2are less or equal to the values forX3, Y1take values 0, 1
Since the values ofX3are greater than the values forX1, X2, thenY3= max{X1, X2, X3}take values 1,2, 3
SinceY1 Y2 Y3, Y2take values 0, 14.3.5. LetX1, X2, . . . , Xn be a random sample from exponential distribution with mean , then
the common pdf is given byf(x) = 1
ex , ifx >0
Using Theorem 4.3.2, the pdf of thek-th order statistic is given by
fk(y) = fYk (y) = n!
(k 1)!(n k)! f(y)(F(y))k1(1 F(y))nk, whereF(y) = 1 e y
thenfk(y) = n!(k1)!(nk)! f(y)
1 e yk1
ey
nk
Then, the pdf ofY
1isf
1(y)
=nf (y) e
yn1 = ne
ny
, which is the pdf of an exponentialdistribution with mean n
, and the pdf ofYn is
fn(y) = nf (y)[F(y)]n1
= n
ey
1 e ny
n1
8/10/2019 10739308_10203914552004027_1698851518_n
49/149
44 CHAPTER 4 Sampling Distributions
4.3.7. X1, . . . , Xna random sample are i.i.d with pdff (x) = 12 , 0 x 2thenF(x) = x0 12 dx = x2 , if 0 x 2
thenF(x) = 1, x >2
x
2 , 0 x 20, x 10) = 1 P(Yn 10)
The CDF Fn(y)ofYnis [F(y)]n, whereF(y)is the cdf ofXevaluated inyThen,P (Yn
y)
=Fn(y)
= [F(y)
]n
= [P(X
y)
]n
andP (X y) = P
Z y102
ThenP (Yn y) =
P
Z y102n
ThenP(Yn > y) = 1 P(Yn < y)= 1
P
Z y102n
8/10/2019 10739308_10203914552004027_1698851518_n
50/149
Students Solutions Manual
ThereforeP (Yn >10) = 1 [P(Z 0)]n= 1 (0.5)n
4.3.11. X1, . . . , Xnis a random sample from Beta(x = 2, = 3)The joint pdf ofY1and Yn, according Theorem 4.3.3, is given by
fY1,Yn (x, y) = n!
(1 1)!(n 1 1)!(n n)! [F(x)]11[F(y) F(x)]n11[1 F(y)]nnf(x)f(y)
= n(n 1)[F(y) F(x)]n2f(x)f(y), if x < y
SinceXi Beta(X = 21 = 3)fori = 1,2, . . . , n, the pdf is
f(x) = ( + )()()
1(1 x)1, x [0, 1]
and, the DF is
F(x) =
0, x 0( + )()()
x0 t
1(1 t)1dt, 0 x 1
1, x 1In our case,
f(x) = (5)(2)(3)
x21(1 x)31 = 4!1!2! x(1 x)
2 = 12x(1 x)2, if 0 x 1
and
F(x) =
x
0
12t(1 t)2
dt= 12x2
2 2x3
3 +x4
4
, if 0 x 1
Then, the joint pdffY1,Yn (x, y), using the 4.3.3, is given by
fY1,Yn (x, y) = n(n 1)
12
y2
2 2y
3
3 + y
4
4
10
x2
2 2x
3
3 + x
4
4
n212x(1 x)212y(1 y)2
= 12nn(n 1)
1
2
y2 x2 2
3
y3 x3+ 1
4
y4 x4n2 xy(1 x)2(1 y)2,
if 0 x y 1, andfY1,Yn (x, y) = 0, otherwise
EXERCISES 4.4
4.4.1. X1, X2, . . . , Xn, wheren = 150, = 8, 2 = 4, thenx= 8 and2x= 4
By Theorem 4.4.1: limn P
Z X x
x
= 1
2k
z e
2 2du
thenP (7.5< X
8/10/2019 10739308_10203914552004027_1698851518_n
51/149
46 CHAPTER 4 Sampling Distributions
4.4.3. LetTbe the time spent by a customer coming to certain gas station to fill up gas
SupposeT1, T2, . . . , Tn are independent random variables, with t= 3 minutes, 2t = 1.5minutes, andn = 75
ThenY
=
n
i=1
Tiis the total time spent by then customers
Since,Y= 3 hours = 180 minutes,P(Y
8/10/2019 10739308_10203914552004027_1698851518_n
52/149
Students Solutions Manual
4.4.13. SIDS occurs between the ages 28 days and one year
Rate of death due to SIDS is 0.0013 per year.
Randon sample of 5000 infants between the ages 28 day and are your
LetX be the number of SIDS related deaths
p = 0.00103,n = 5000ThenX Bin(n = 5000, p = 0.00103)The probability requested is
P(X >10) P
Z >10 np 0.5
np(1 p)
= 0.0274
8/10/2019 10739308_10203914552004027_1698851518_n
53/149
This page intentionally left blank
8/10/2019 10739308_10203914552004027_1698851518_n
54/149
8/10/2019 10739308_10203914552004027_1698851518_n
55/149
50 CHAPTER 5 Point Estimation
5.2.11. We have
E(X) = = xVariance(2) = 1
n
Xi X
2
2
= 1
n x2i x2
=
1n
Xi
2
Xin
2The method of moments estimator for i s given by T (X1, . . . . . . .Xn) =
1n
Xi
2
Xin
2
5.2.13. The method of moments estimator for = T(X1, . . . . . . .Xn) = E(X) = XT(X1, . . . . . . .Xn) = XSince and2 both are unknown.2 = 1
n
Xi X
2This implies2
= (n1).1
n.(n
1) Xi
X2,2 = n1n s2Lets2 = n1
n s2
method of moments estimator for2 is given by:
s2 = 1n
ni=1
Xi X
2
EXERCISES 5.3
5.3.1. f(x) = nx
px(1 p)nx
L(p, x1, x2, . . . .xn) = log
n
i=1n
xi
+
Xi logp + n
Xi
log(1 p)
plogL(p, X1, X2, . . . .Xn) =
Xi
p+ n
Xi
1 p (1)
plogL(p, X1, X2, . . . .Xn) =
Xi
p n
Xi
1 p
For maximum likelihood estimator ofp
plogL(p, X1, X2, . . . .Xn) = 0
Xi
p
n
Xi
1 p
=0
(1 p)Xi pn Xi = 0. This implies Xi= pXi + n Xip =
Xi
n ,
p = X.Hence, MLE ofp = p = X.By invariance propertyq = 1 p is MLE ofq.
8/10/2019 10739308_10203914552004027_1698851518_n
56/149
Students Solutions Manual
5.3.3. f(x) = 1
ex impliesL() = 1
ne
Xi
ln L() = n ln
Xi
Now taking the derivatives with respect to and and setting both equal to zero, we have
ln L = n + Xi2 = 0n+Xi= 0.= XFrom the given data:
MLE ofis given by= X = 1+2++7+214 . = 6.07.5.3.5. Here pdf ofX is given by
f(x) =
2x2
ex2
3 if, x >0
0, otherwise
(5.1)
L(, X1, . . . . . . . .Xn) = 2(2)2n
i=1 Xie x2
2
ln L = ln 2 2 ln +ni=1 ln Xi ni=1 Xi22
ln L = 2n
+ 2
3
ni=1 X
2i
= 0 implies, 2n
+ 2
3
ni=1 X
2i= 0
n2 +
X2 = 0
2
= X2i
n
=
X2in
.
5.3.7.
f(x) =
x1e x
if, x 00, otherwise.
L(, , X) = nn i=1 nX
1i e
Xi
L(, , x) = n ln ln + ( 1) ln Xi Xi 2
L(, , x) = n
+ni=1 ln Xi n ln Xi ln Xi
L(, , x) = n Xi (a)1
L(, , x) = n+ (1) Xi
8/10/2019 10739308_10203914552004027_1698851518_n
57/149
52 CHAPTER 5 Point Estimation
For maximum likelihood estimator of:
ln L = 0. This implies;n+ni=1 ln Xi n ln Xi ln Xi = 0
similarly, n+ (1) Xi
= 0 n+ (1)
Xi
n
Xi
= 0 solving for we get =
Xin
+1
Hence,
n
2+
Xi
Xi
1 Xin
+1
ln
Xi
ln
Xin
+1
There is no closed form solution for and . In this case, one can use numerical methodssuch as Newton-Raphson method to solve for, and then with this value to solve for.
5.3.9. f(x) = (2)()2
[x(1 x)](1), 0ln L(, x) = n ln (2) ln ()2+ ( 1)ni=1 ln(Xi 1)
ln L(, x) = n
2(2)(2) 2
()2(0)2
+ni=1 ln Xi(Xi 1).
5.3.13.
f(x) =
13+2 if, 0 x 3+ 20, otherwise.
This impliesL(, x) = 1(3+2)2 for 0 x 3+ 2
When 3+2 max(Xi), the likelihood is 1(3+2)2 . Which is positive and decreasing functionof(forfixedn). However, for < max(Xi)23 , thelikelihood drops to 0, creating discontinuityat point max(Xi)23 .Hence we will not be able to find the derivative. The MLE is the largest order statistic=max(Xi)2
3 = Xn.5.3.15. HereX N(, 2)
ln L(, , x) = n2
ln 2 n2
ln 2 n
i=1
(xi )2
22
L
=
ni=1
(Xi )
Similarly L2
= n22
+ 122
ni=1(Xi )2
For maximum likelihood estimates of and ;
(Xi ) = 0 implies = X
8/10/2019 10739308_10203914552004027_1698851518_n
58/149
Students Solutions Manual
Similarly, for2,
n22
+ 122
ni=1
(Xi )2 = 0
=
(Xi X)n
.
5.3.17. f(x) = 1
e x
It is given that the reliabilityR(x)= 1 F(x). This implies F(x)= f(x). Hence F(x)= 1
2e
x .
Thus,R(x) = 1 F(x)and
L(x, ) =n
i=1
1 + 1
2e
Xi
.
EXERCISES 5.4
5.4.1. E(X) = E
Xin
i.e E(X) = 1
n
E(Xi). Where, E(X) =
xe(x)dx. By integration by
partsE(X) = (1 + ). ThusEX = 1n
(1 + ). This impliesE(X) = 1 + .
5.4.3. Sample standard deviations =
1n1
(xi X)2
E(s) = E
1
n 1
(Xi + X)2
E(s) = E
1
n
(xi )2 (X )2
E(s) =
1
n
E(xi )2 E(X )2
E(s) =
1
n
2
2
n
5.4.5. LetY
=C1X1
+C2X2
+ +CnXn.
For an unbiased estimate, we need to have E(C1E(X1) + + CnE(Xn)) = That is, C1 + Cn = Which is possible if and only ifCis= 1n for all i= 1, 2 . . . . . . . . . n . This implies 1n +
1n
+ + 1n
=
nn= i.e = . Verified.
8/10/2019 10739308_10203914552004027_1698851518_n
59/149
54 CHAPTER 5 Point Estimation
5.4.7. Xi U(0.).Yn= maxX1, . . . . . Xn= Ynis the MLE of.(a) By method of moment estimatorE(X)
=X
= +0
2 . This implies
=2X.
Hence the method of moment estimator= 2X(b)E() = E(Yn)E() = E{max(X1, . . . Xn)}E() = n
n+1E() = E(2X).E() = 2.E(/2). That is, E() = . Hence, 2 ia an unbiased estimate of .(c)E(3) = n+1n E()This impliesE(3) = .is an unbiased estimate of.
5.4.9. Here,Xi N(, 2)
f(x) = 122
exp
(x )
2
22
We have E() = E(X) = is an unbiased estimate for. By definition, the unbiased estimatethat minimizes themean square error is called the minimum variance unbiased estimate (MVUE) of.
MSE() = E( )2
That is MSE()= var(). Minimizing the MSE implies that the minimization of Var().
is the MVUE for.
5.4.11. E(M) = E(X) = . Thus, sample median is an unbiased estimate of population mean .Now, Var(X) = 1
n
(X )2
Var(M) = 1n
(M )2
Where Var(X) VarM5.4.13.
f(X) = 1
2exp
|X|
for < x <
The likelihood function is given by:
f (X1, X2, . . . . . . . . Xn, ) = 12n
1n
exp |Xi|
.
Takeg |Xi|, = 1nexp |Xi| andh(X1, X2 . . . . . . . Xn) = 12n
|Xi| is sufficient for.
8/10/2019 10739308_10203914552004027_1698851518_n
60/149
8/10/2019 10739308_10203914552004027_1698851518_n
61/149
56 CHAPTER 5 Point Estimation
5.4.21. The likelihood function is given by
f(x1, . . . . . . xn) =
nn
i=1(xi)1 for 0< x 00 otherwise
LetU= (X1, X2, . . . . . . . . Xn)theng(x1, . . . . . . . Xn, ) = n
ni=1 x
1i andh(x1, . . . . . . xn) = 1. Therefore U is sufficient for.
5.4.23. The likelihood function is given by
f (x1, . . . . . . . . xn) = 2n
n
n
i=1Xi
e
x2
i
Letg
x2i, = 2n
ne
x2
i andh(x1, . . . . . . . . xn) =
ni=1 xi
Hence, x2i is sufficient for the parameter.
EXERCISES 5.5
5.5.1. ln L(p, x) = lnn
i=1
nxi
+ xi ln pni=1(n xi) ln(1 p) For MLE (1 p) xi p
(n xi) = 0. This implies p =
xin2
. Suppose Yn =
xin
. Thus p = Ynn
. Where
E
Ynn
= 1
nE(Yn)= 1n2 nnp=p.pis an unbiased estimate ofp. Similarly Varp=Var
Ynn=
Var
xin2
This implies Varp = nn4
np(1 p). Thus Varp 0 as n . Ynn
is consistent.
5.5.3. E(Xi) = i, E(X2i) = 2and E(Xi) = 4.E(s2) =
1n
E
Xi + X2This impliesE(s2) = n22
n . That isE(s2) = (n1)2
n .
S2 is an biased estimator of2
HereS2 = n1n
S2
Var(S2) = n1n2
2 0 as n .Where (n1)s
22
2(n1).Thus, Var
(n1)s2
2
= 2(n 1). This implies (n1)2
4 var(S2) = 2(n 1).
Finally, Var(S2) = 2n1
2.
Moreover, Bias(s2) = E(s2) 2. This implies2n
0 as n .Thuss2 is an unbiased estimate of2.
5.5.5. Here E(x) = and var(x)= 2 (For exponential distribution). Now E(X = E) andVar() = 2
n 0 as n . X is an unbiased estimate of.
5.5.7. Here, ln(, X) = n ln + 1
ni=1 ln(Xi).
Differentiating above equation and equating to zero we get =n
i=1ln Xin
8/10/2019 10739308_10203914552004027_1698851518_n
62/149
Students Solutions Manual
5.5.9. Here, var(1) = 112n and Var(2) = 112 .Thus the efficiency of2relative to 1is (2,1) = 1n 1
Similarly,e
2,3 = n+23 >1 ifn >1.
2is efficient than 3ifn >1.
8/10/2019 10739308_10203914552004027_1698851518_n
63/149
58 CHAPTER 5 Point Estimation
5.5.17. It can be easily verified thatE
1 = , E2 = and E3 =
Similarly, Var
1 = 3181 2, Var2 = 6n1725(n3) 2.
Var
3
= 2
n.
Now the corresponding efficiencies are given bye 2,1=31775(n3)81
(6
n17
), e 3,1 =
31n
81
.
e
3,2 = (6n17)n25(n3) .
5.5.21. The ratio of the joint density function of two sample points is given by;
L(x1, . . . . xn)
L(y1, . . . yn)= exp
122
ni=1
X2in
i=1Y2i
2
ni=1
Xi n
i=1Yi
.
For this ratio to be free of and 2, we must haven
i=1 Xi=n
i=1 Yi andn
i=1 X2i =n
i=1 Y2i .
Thusn
i=1 Xiandn
i=1 X2i are jointly minimal sufficient statistics for and
2. SinceXis
unbiased estimate forands2 is an unbiased estimate for2. The estimators are functions
of the minimal sufficient statistics. This implies theXand s2 are MVUES for and2.
5.5.23. The ratio of joint density function at two sample points, we have
L(x1, . . . . xn)
L(y1, . . . yn)=n
i=1(Yi)ni=1(Xi)!
Xi
Yi .
For the ratio to be free of we must have
Xi
Yi= 0. Thus
Xi, form the minimal
sufficient statistics for.
5.5.25.
L(x1, . . . . xn)
L(y1, . . . yn) =eXiYi
For the ratio to be independent of, we need to have
Xi=
Yi. Thus
Xiis minimal
sufficient for. NowE
Xi = nXis UMVUE, by RaoBlackwells theorem.
5.5.27.
L(x1, . . . . xn)
L(y1, . . . yn)=n
i=1(Yi)ni=1(Yi)!
e
X2i
Y2i .
The ratio to be free of, we must have
X2i
Y2i =0. Therefore
X2i is MVUE for.
Moreovers2 is an unbiased estimator for2.
8/10/2019 10739308_10203914552004027_1698851518_n
64/149
8/10/2019 10739308_10203914552004027_1698851518_n
65/149
60 CHAPTER 6 Interval Estimation
p
1
21/2<
2
(n1)s2 < 12/2
= 1
p
(n1)s2
2/2< 2 5So the given data can be approximated as a normal distribution.
Here 1 = 0.98 = 0.02/2 = 0.01z/2= z0.01= 2.325
Thus the 98% confidence interval is given byp z/2
p(1p)
n
=
925 2.325
925 1625
50
= (0.202, 0.518)
6.2.7. n = 50
x = 11.4= 4.51 = 0.95 = 0.05z/2= 1.9695% confidence interval is
x z/2 n
=
11.4 1.96
4.550
= (10.153, 12.647)
6.2.9. n = 400p = 0.3np = 120> 5n(1 p) = 280> 595% confidence interval is given by
p z/2
p(1p)n
=
0.3 1.96
0.30.7400
= (0.255,0.345)
6.2.11. Proportion of defectionp = 40500= 2251 = 0.9/2 = 0.05z/2= 1.64590% confidence interval is given by
225 1.645
225 2325500
= (0.06, 0.1)
8/10/2019 10739308_10203914552004027_1698851518_n
68/149
Students Solutions Manual
6.2.13. x N(, 16)p(x 2< < x + 2) = 0.95z/2
n= 2
n = (z/22 )2 = 1.9642 2 = 15.37 16
6.2.15. n = 425p = 0.45np >425 0.45> 5n(1 p) = 425 0.55> 595% confidence interval is given by
p z/2
p(1p)n
=
0.45 1.96
0.450.55425
= (0.403, 0.497)
For 98% confidence interval
1 = 0.98 = 0.02z/2= 2.335
0.45 2.335
0.450.55425
= (0.394, 0.506)
6.2.19. p = 52601 = 0.95/2
=0.025
z/2= 1.96The 95% confidence interval is given by
p z/2
p(1p)n
=
5260 1.96
5260 860
60
= (0.781, 0.953)
6.2.21. = 35E = 15E = z/2 nn
= z/2
E 2 = 1.9635
15 2 = 20.92 216.2.23. x = 12.07
= 12.911 = 0.98/2 = 0.01z/2= 2.335
8/10/2019 10739308_10203914552004027_1698851518_n
69/149
64 CHAPTER 6 Interval Estimation
98% confidence interval for mean is given byx z/2 n
=
12.07 2.335 1.9135
= (11.32, 12.82)
EXERCISES 6.3
6.3.1. (a) When standard deviation is not given and there is not enough sample size, we use
t-distribution.
(b) As differences decreases the sample sizen increases which means we are closing in on
the true parameter value of.
(c) The data are normally distributed, and the values ofx and thesamplestandard deviation
are known.
6.3.3. x = 20s = 4
1 = 0.95(a)
x t/2,4 sn
=
20 t0.025,4 45
(b)
20 t0.025,9 410
(c)
20 t0.025,19 420
6.3.5. x = 2.22s = 0.005n
=26
98% confidence interval for isx t/2,25 sn
=2.22 2.485 1.67
26
= (3.03,1.41)
6.3.7. x = 0.905s = 1.671 = 0.98n = 1098% confidence interval for is
0.905 t0.025,90.00510
6.3.9. Similar to 6.3.8
6.3.11. x = 410.93s = 312.87
8/10/2019 10739308_10203914552004027_1698851518_n
70/149
Students Solutions Manual
95% confidence interval for isx t/2,14 sn
=
410.93 2.145 312.8715
= (237.65, 584.21)
6.3.13. x
=3.12
s = 1.04n = 1799% confidence interval foris
x t/2,4 sn
=
3.12 2.921
1.0417
= (2.40, 3.84)
6.3.15. x = 3.85s = 4.55n = 2098% confidence interval foris
x t/2,19 sn
=
3.85 2.5394.5520
= (2.64, 5.06)
6.3.17. x = 148.18s = 1.91n = 1095% confidence interval foris
x t/2,9 sn
=
148.18 2.262
1.9110
= (147.19, 149.17)
EXERCISES 6.4
6.4.1. x = 2.2s = 1.421 = 0.90 = 0.10n = 2090% confidence interval for2 is given by
(n1)s22/2,19
, (n1)s2
21
/2,19 = 19(1.42)2
32.85 ,19(1.42)2
8.90655 = (1.1663, 4.3015)6.4.3. x = 60.908
s2 = 12.661 = 0.99 = 0.01n = 10
8/10/2019 10739308_10203914552004027_1698851518_n
71/149
66 CHAPTER 6 Interval Estimation
99% confidence interval for2 is given by(n1)s2
2/2,9, (n1)s
2
21/2,9
= 912.6623.58 , 912.661.73 = (4.8321, 65.8613)
6.4.5. x = 2.27
s2 = 1.021 = 0.99 = 0.01n = 1899% confidence interval for2 is given by
(n1)s22/2,9
, (n1)s2
21/2,9
= 171.0227.58 , 171.025.69 = (0.6287, 3.0475)
6.4.9. From excel or by calculation, sample variance s2 = 148.44, sample mean x = 97.24, n = 2599% confidence interval for population variance is given by
(n1)s22/2,24
, (n1)s2
21/2,24
= 24148.4436.41 , 24148.449.886 = (97.8456, 360.3642)
6.4.11. x = 13.95s2 = 495.0851 = 0.98 = 0.02n = 25
98% confidence interval for2
is given by(n1)s22/2,24
, (n1)s2
21/2,24
= 24495.08542.97 , 24495.08510.85 = (276.5194, 1095.1189)
EXERCISES 6.5
6.5.1. For procedure I,x1= 98.4,s21= 235.6,n1= 10For procedure II, x2= 95.4,s22= 87.15,n2= 10 = 0.02/2 = 0.01z/2= 2.98598% confidence interval for difference of mean is
x1 x2 z/2
s21n1
+ s22n2
=
98.4 95.4 2.985
235.610 + 87.1510
= (13.9580, 19.9580)
8/10/2019 10739308_10203914552004027_1698851518_n
72/149
Students Solutions Manual
6.5.3. x1= 16.0, s1= 5.6, n1= 42x2= 10.6, s2= 7.9, n2= 45 = 0.01/
2 = 0.005z/2= 2.575
99% confidence interval for difference of mean is
x1 x2 z/2
s21n1
+ s22n2
=
16.0 10.6 2.575
(5.6)2
42 + (7.9)2
45
= (1.6388, 9.1612)
6.5.5. x1= 58, 550, s1= 4, 000, n1= 25x2= 53, 700, s2= 3, 200, n2= 23
Since21= 22 but unknown, we can use pooled estimatorS2p= (n11)s
21+(n21)s22
(n1+n22)
S2p= 24(4,000)2+22(3,200)246
Sp= 3639.398The 90% confidence interval is
x1 x2 t/2,(n1+n22) Sp
1n1
+ 1n2
= 58, 550 53, 700 2.326 3639.398 125+ 123 = (2404, 7296)6.5.7. x1= 28.4, s1= 4.1, n1= 40
x2= 25.6, s2= 4.5, n2= 32(a) MLE of1 2is given by(x1 x2)(b) 99% confidence interval for1 2is
x1 x2 z/2
s21n1
+ s22n2
=
28.4 25.6 2.565
(4.1)2
40 + (4.5)2
32
=(0.1678, 5.4322)
6.5.9. x1= 148, 822, s1= 21, 000, n1= 100x2= 155, 908, s2= 23, 000, n2= 1501 = 0.98 = 0.02
8/10/2019 10739308_10203914552004027_1698851518_n
73/149
68 CHAPTER 6 Interval Estimation
/2 = 0.01z/2= 2.57598% confidence interval for difference of mean is given by
x1 x2 z/2
s21n1
+ s22n2
=
148, 822 155, 908 2.2
(21,000)2
100 + (23,000)2
150
= (508, 13, 664)6.5.11. x1= 35.18, s21= 19.76, n1= 11
x2= 38.76, s22= 12.69, n2= 131 = 0.9 = 0.1
90% confidence interval for
21
22 is given bys21s22
1Fn11,n21,1/2
, s21s22
1Fn11,n21,/2
=
19.7612.69
1F10,12,0.95
, 19.7612.691
F10,12,0.05
= 19.7612.69 12.75 , 19.7612.69 2.91 = (0.5662, 4.5313)6.5.13. x1= 68.91, s21= 287.17, n1= 12
x2= 80.66, s22= 117.87, n2= 121 = 0.95 = 0.05/2 = 0.025(a) 95% confidence interval of difference of mean is
x1 x2 z/2
s21n1
+ s22
n2
=
68.91 80.66 1.96
281.1712 + 117.8712
= (23.0525, 0.4475)
(b) 95% confidence interval for 2122
is given by
s21s22
1
Fn11,n21,1/2 ,
s21
s22
1
Fn11,n21,/2 = 281.17117.87 1F11,11,0.975 , 281.17117.87 1F11,12,0.025 = 281.17118.87 13.48 , 281.17118.87 3.48 = (0.6855, 8.3055)
8/10/2019 10739308_10203914552004027_1698851518_n
74/149
Chapter7Hypothesis Testin
EXERCISES 7.1
7.1.1. (a) H0: = 0H1: > 0
(b) H0: = 0H1: >1.20
7.1.3. H0 : p = 0.5H1 : p >0.5
n = 15(a) = probability of type I error= p(rejectH0|H0is true)
= p(y 10|p = 0.5) = 1 p(y 10|p = 0.5)
=1
15
y=10
c(15, y)(0.5)y(0.5)15y
= 1 0.941= 0.059
(b) = p(acceptH0|H0is false)= p(y 9|p = 0.7)
=9
y=0c(15, y)(0.7)y(0.3)15y
= 0.278(c)
=p(y
9
|p
=0.6)
=9
y=0c(15, y)(0.6)y(0.4)15y
= 0.597
8/10/2019 10739308_10203914552004027_1698851518_n
75/149
70 CHAPTER 7 Hypothesis Testing
(d) For = 0.010.01 = p(y k|p = 0.5)From binomial table = 0.01 falls between k = 2 and k = 3. However, fork= 3, =0.018 which exceeds 0.01. If we want to limit to be no more than 0.01, we will take
k= 2. That is we rejectH0ifk 2.For = 0.010.03 = p(y k|p = 0.5)From binomial table = 0.03 falls between k = 3 and k = 4. However, fork= 4, =0.059 which exceeds 0.05. If we want to limit to be no more than 0.05, we will take
k= 3. That is we rejectH0ifk 3.(e) When = 0.01. From part d, rejection region is of the form(y 2).
Forp = 0.7 = p(y 2|p = 0.7)
= 1 p(y 10
(a) = probability of type I error= p(rejectH0|H0is true)= p(x >11.2| = 10)
= P x/n > 11.2104/25| = 10= P(z >1)= 0.1587
(b) = p(acceptH0|H0is false)= p(x 11.2| = 11)= P
x
/
n 11.211
4/
25| = 11
= P(z 0.25)= 0.5787
(c) 0= 10a
=11
z= z0.01= 2.33z= z0.8= 0.525
n = (z+z)22(a0)2
= (2.330.525)242(1110)2
= 52.13 rounded up to 53
8/10/2019 10739308_10203914552004027_1698851518_n
76/149
Students Solutions Manual
7.1.9. 2 = 16H0 : = 25H1 : = 25n
= (z+z)22
(a
0)2
= (1.645+1.645)216(1)2
= 173.19 rounded up to 174
EXERCISES 7.2
7.2.1. H0 : = 0H1 : = 1L() = 1
2n
nexp
(xi)222
L(0)
= 12nn exp
(xi0)222
L(1) = 12n
nexp
(xi1)222
L(0)L(1)
= exp
(xi0)222
+
(xi1)222
= exp
(xi1)2
(xi0)2
22
ln
L(0)L(1)
=
(xi1)2
(xi0)2
22
=
x2i 2nx1+21
x2i 2nx02022
= 2nx(01)(01)(0+1)
22
= (01)
xi2
(2021)22
Therefore, the most powerful test is to rejectH0if
(0 1)
xi
2
20 21
22
ln k
(0 1)
xi
2 ln k +
20 21
22
xi
2 ln k +
2021
2
(0 1)
= C
Assume0 < 1, rejection region for = 1is given byxi C
WhereC =
2 ln k+
2021
2(01)
8/10/2019 10739308_10203914552004027_1698851518_n
77/149
72 CHAPTER 7 Hypothesis Testing
Rejection region for = 0is given by xi C
7.2.5. H0: = 0
H1: < 0f(y) = 2y
2ey2/2 x >0
L() =n
i=1
2yi2
ne
y2i/2
L(0) =n
i=1
2yi20
ne
y2i/20
L(1) =n
i=1
2yi21
ne
y2i/21
L(0)
L(1) =
n
i=12yi20
ni=1
2yi21
n e y2i/21 y2i/20=
10
2ne
y2i/21
y2i/20
ln
L(0)L(1)
= 2n ln
10
+ y2i/21 y2i/20 ln k
y2i CWhereC=
ln k 2n ln
10
20
21
20217.2.7. H0: p = p0
H1: p = p1wherep1 > p0f(p) = px(1 p)1xL(p) = p
xi (1 p)n
xi
L(p0) = p
xi0 (1 p0)n
xi
L(p1) = p
xi1 (1 p1)n
xi
L(p0)L(p1)
=
p0p1
xi 1p01p1
n xi kTaking natural logarithm, we have
xi ln
p0
p1
+
n
xi
ln
1 p01 p1
ln k
lnp
0p1
ln1 p01 p1 xi + n ln1 p
01 p1
ln k
xi
ln k n ln
1 p01 p1
ln
p0
p1
ln
1 p01 p1
To find the rejection region for a fixed value of, write the region asxi C, whereC=
ln k n ln
1p01p1
ln
p0p1
ln
1p01p1
8/10/2019 10739308_10203914552004027_1698851518_n
78/149
Students Solutions Manual
EXERCISES 7.3
7.3.1. f(x) = 12
exp (x)2
22
L(1)
= 12nn
1
exp(xi)2
221 Here0=
20
anda= R
20
Hence
L
20
= max 120
nexp
(xi )2220
Since the only unknown parameter in the parameter space is 2, < 2 < ; maximumlikelihood function is achieved when2 equals to its maximum likelihood estimator
2mle= 1
n
n
i=1
(xi x)2
=
21
20
n/2exp
(xi )2
221
(xi )2220
=
(xi x)2n20
n/2exp
n
(xi )22
(xi x)2
(xi )2
220
The likelihood ratio test has the rejection region:
Reject if k, which is equivalent ton
2
ln(xi x)2
n
2
lnn20+
n(xi)2
2
(xix)2
(xi)2
220
ln k
7.3.3. f(x) = 12
exp (x)2
22
L(1) = 1
2n
n1
exp
(xi)2221
L(2) = 12n
n2
exp
(yi)2222
Let = L(22 )
L
21
= 21
nexp
(yi)2
222
(xi)2221
Thus the likelihood ratio test has the rejection region
RejectH0if kn ln
2
1
+
(yi )2222
(xi )2221
ln k
(yi )2
22
(xi )221
2 ln k 2n ln
2
1
8/10/2019 10739308_10203914552004027_1698851518_n
79/149
74 CHAPTER 7 HypothesisTesting
21= 1
n
ni=1
(xi x)2
22= 1
n
n
i=1
(xi x)2
n
(yi 2)2(yi y)2
n
(xi 1)2(xi x)2
2 ln k 2n ln
2
1
The rejection region is n
(yi 2)2(yi y)2
n
(xi 1)2(xi x)2
C
7.3.5. f(x) = 1
ex/ forx >0
L() = 1n
exp
xi
L(0)
= 1n
0
expxi
0
L(1) = 1n1 exp
xi1
L(0)L(1)
= n1n0
exp
xi1
xi0
=
10
nexp
xi
1
xi
0
We reject the null hypothesis if
n ln
1
0
+
1
1 1
0
xi ln k
xi ln k n ln1
0
100 1
Where we reject null hypothesis if
xi m1or
xi m2
Provided m1=
ln k n ln
10
1001
andm2= 1
ln kn ln
10
0110
EXERCISES 7.4
7.4.1. n = 50, = 0.02, x = 62, s = 8H0 : 64H1 : 1.769) = 0.0384
8/10/2019 10739308_10203914552004027_1698851518_n
80/149
Students Solutions Manual
(c) Smallest = 0.0384p-value > 0.02
We fail to reject the null hypothesis.
7.4.3. H0:
=0.45
H1: 0.85
} =0.1977
(b) Herez = 0.85z/2= z0.005= 2.58Rejection region is {z < z0.005, z > z0.005}, i.e. {z < 2.58, z >2.58}p-value = min{z < 0.85, z >0.85} = 0.3954
(c) Assumptions: even though population standard deviation is unknown, because of the
large sample size, normal distribution is assumed.
7.4.5. x = $58, 800, n = 15, s = $8, 300(a) P(rejectH0|H0is true) = 0
Since the probability of rejecting null hypothesis equals to zero. Therefore, the null
hypothesis is accepted.(b) = 0.01H0: = 55, 648H1: >55.648
t= xs/
n= 58,80055,648
8,300/
15 = 31522143.05= 1.47
Rejection region is {t > t0.01,14}t0.01,14= 2.624Sincet= 2.642 is greater than 1.47, we fail to reject the null hypothesis
7.4.7. H0: p0= 0.3H
1:p
0>
0.3p = 5501500= 0.366z = pp0
p0q0n
= 0.3660.30.30.71500
= 0.0660.0118= 5.593
Rejection region is {z > z0.01}z0.01= 2.33
8/10/2019 10739308_10203914552004027_1698851518_n
81/149
76 CHAPTER 7 Hypothesis Testing
i.e. {z >2.33}Yes, customer has preference over ivory color
7.4.9. (a) x = 42.9, s = 6.3674, = 0.1H0 :
=44
H1 : = 44The data is normally distributed.
z =x s/
n
= 42.9 446.3674/
10
=1.12.013
= 0.5644
Rejection region forz is|z| < z0.05}wherez0.05= 1.645
z0.05= 1.645< 0.56444We fail to reject the null hypothesis
(b) 90% confidence interval for is
x z/2 sn , x z/2 s
n
= (42.9 1.645
2.013, 42.9
+1.645
2.013)
=(39.588, 46.21138)
(c) From a, we can see that it is reasonable to take = 44. The argument is supported bythe confidence interval in b.
7.4.11. x = 13.7, s = 1.655, n = 20H0 : = 14.6H1 : z0.01}z0.01= 2.33
We reject the null hypothesis. Thus there is a statistical evidence to support this claim.
7.4.13. x = 32,277, s = 1,200, n = 100, = 0.05H0 : = 30,692H1 : >30,692
z = 32,27730,6921,200/
100
= 13.20Rejection region is
z > z0.05
z0.05= 1.645Hence we reject the null hypothesis that the expenditure per consumer is increased from
1994 to 1995
7.4.15. H0 :
=1.129
H1 : = 1.129t= 1.241.129
0.01/
24 = 0.1110.00204= 54.41
t0.05,23= 2.807Wheret= 54.41> t0.05,23Thus we reject the null hypothesis. That is price of gas is changed recently.
8/10/2019 10739308_10203914552004027_1698851518_n
82/149
Students Solutions Manual
EXERCISES 7.5
7.5.1. H0: 1 2= 0H1: 1 2= 0
z = (
y1
y2)
(
1
2)
s21n1
+ s22
n2
= 74
71
8150 + 10050 =
31.9026= 1.5767
Rejection region forzis|z| > z0.025
z0.025= 1.96Sincez0.025 >1.57, we fail to reject the null hypothesis. To see the significant difference we
need to have = 0.0582 level of significance.7.5.3. x1= 58, 550, x2= 53, 700, s1= 4000, s2= 3200, n1= 25, n2= 23
H0: 1 2= 0H1: 1 2 >0S2p
=
(n11)s21+(n21)s22(n1
+n2
2)
=
24(4000)2+22(3200)225
+23
2
=13245217
Sp= 3639.3979t= (x1x2)0
Sp
1n1
+ 1n2
= (58,55053,700)03639
125 + 123
= 4.61288
Rejection region is {t > t0.05,46} i.e. {t >1.679}Since1.679 < 4.61288, werejectthe null hypothesis. Thus implies that there existssignificant
evidence to show that the males salary is higher than that of female.
7.5.5. x1= 105.9, x2= 100.5, s21= 0.21, s22= 0.19, n1= 80, n2= 100H0: 1 2= 0H1: 1 2= 0
Uset=
(
x1
x2)
0
s21n1
+ s22n2
and use two sidedt
-test.
7.5.7. x1= 7.65, x2= 9.75, s1= 0.9312, s2= 0.852, n1= 10, n2= 10(a) H0: 1 2= 0
H1: 1 2 >0S2p= (n1
1)s21+(n21)s22(n1+n22) =
9(0.9312)2+9(0.852)218
Sp= 0.892479t= (x1x2)0
Sp
1n1
+ 1n2
= (7.659.75)00.89
110+ 110
= 5.276
Rejection region is {t > t0.05,18} i.e. {t >1.734}Thus fail to reject the null hypothesis.
(b) H0: 21= 22
H1: 21= 22
Test statisticF= s21
s22= (0.9312)2
(0.852)2= 1.1945
From theF-tableF0.025(9,9)= 4.03
8/10/2019 10739308_10203914552004027_1698851518_n
83/149
78 CHAPTER 7 HypothesisTesting
F0.95(9,9)= 14.03= 0.248Rejection region isF > 4.03 andF < 0.248
Since the observed value of the statistic 1.1945 < 4.03, we fail to reject the null
hypothesis
(c) H0 : d= 0H1 : d >0
d= 2.1Sd= 1.1670t= dd0
Sd/
n= 2.1
1.670/
10= 3.16
Fromt-tablet0.05,9= 1.833Wereject the null hypothesis, this implies hat the down stream is less than the upstream.
7.5.9. (a) x1= 2.04, x2= 3.55, s1= 0.551, s2= 0.6958, n1= 14, n2= 14H0 : 1 2= 0H1 : 1 2 1.77, we fail to reject the null hypothesis.
(b) Test statisticF= s21
s22= 0.6281
From theF-tableF0.025(13,13)
=3.11
F0.95(13,13)= 13.11= 0.321Rejection region isF >3.11 andF
8/10/2019 10739308_10203914552004027_1698851518_n
84/149
Students Solutions Manual
Rejection region isF >2.585 andF 9.48
By using contingency table
2 = 43.862 falls in the rejection region at= 0.05, we reject the null hypothesis. That is collectivebargaining is dependent on employee classification.
7.6.3. O1= 12, O2= 14, O3= 78, O4= 40, O5= 6We now computei(i = 1,2,3,4,5)using continuity correction
1= p(x 55)2= p
z 65.5704
3= p
z 75.5704
4= p
z 85.5704
5= p
z 95.5704
Taking above probability we need to findEiand follow 6.6.5.
7.6.5. E1= 950 0.35, E2= 950 0.15, E3= 950 0.20, E4= 950 0.30O1= 950 0.45, O2= 950 0.25, O3= 950 0.02, O4= 950 0.282 = 834.7183From Chi-Square table
20.05,3= 7.81Thus2 = 834.71> 7.81, we reject the null hypothesis, at least one probabilities is differentfrom hypothesized value.
8/10/2019 10739308_10203914552004027_1698851518_n
85/149
This page intentionally left blank
8/10/2019 10739308_10203914552004027_1698851518_n
86/149
Chapter8Linear Regression Mode
EXERCISES 8.2
8.2.1. (a) Proof see example 8.2.1
(b) SSE
2 follows central Kai square with degree of freedomn 2E
SSE2
= n 2 and2 is a constant
ThereforeE(SSE) = (n 2)28.2.3. (a) Least-squares regression line isy = 84.1674 + 5.0384x
(b)
30 40 50 60 70
50
100
150
200
250
x
y
8.2.5. (a) Check the proof in Derivation of
0and
1.
(b) We know the line of best fit is y = 0+1x and plug in the point (x, y) we get0= y 1x. Complete the proof.
8.2.7. y = 1x + SSE = n
i=1e2i=
ni=1
[yi (1xi)]2
8/10/2019 10739308_10203914552004027_1698851518_n
87/149
82 CHAPTER 8 Linear Regression Models
(SSE)1
=
ni=1
[yi(1xi)]2
1= 2 n
i=1[yi (1xi)]xi
= 2 ni=1
[xiyi (1x2i)] = 0
1=n
i=1 (xiyi)n
i=1(x2i)
y = 40.175 + .9984x8.2.9. Least-squares regression line isy = .62875 + .83994x
20 30 40 50 60
20
25
30
35
40
45
50
x
y
8.2.11. Least-squares regression line isy = 2.2752 + .00578x
50 100 150
1
2
3
4
x
y
8/10/2019 10739308_10203914552004027_1698851518_n
88/149
Students Solutions Manual
EXERCISES 8.3
8.3.1. (a) Least-squares regression line isy = 57.2383 .4367x(b)
40 50 60 70 80 90
15
20
2
5
30
35
40
45
50
x
y
(c) The 95% confidence intervals for0is (40.5929, 73.8837)
The 95% confidence intervals for1is (.6806, .1928)
8.3.3. 1andyare both normally distributed. In order to show these two are independent we justneed to show the covariance of them is 0. (Property of normal distribution)
Cov
1, y
= E
1 y
E
1
E (y)
= E
Sxy
Sxx y
1 E
1
n
ni=1
yi
=E
ni=1(yi y) (xi x)n
i=1(xi x)2
y 1
1
n
n
i=1
(0
+1
x)
= 1 (0 + 1x) 1 0 21x