CHAPTER 4

Section 4.1

1.
a. P(X  1) =

b. P(.5  X  1.5) =

c. P(X > 1.5) =

2. f(x) = for –5  x  5, and = 0 otherwise

a. P(X < 0) =

b. P(–2.5 < X < 2.5) =

c. P(–2  X  3) =

d. P(k < X < k + 4) =

3.
a.
0.4

0.3

0.2
f(x)

0.1

0.0

-3 -2 -1 0 1 2 3
x

b. P(X > 0) =

c. P(–1 < X < 1) =

136
 Chapter 4: Continuous Random Variables and Probability Distributions

d. P(x < –.5 OR x > .5) = 1 – P(–.5  X  .5) = 1 –

= 1 – .3672 = .6328

4.

a.

b. P(X  200) =

P(X < 200) = P(X  200)  .8647, since x is continuous.

P(X  200) = 1 – P(X < 200)  .1353

c. P(100  X  200) =

d. For x > 0, P(X  x) =

5.

a. 1=

1.6

1.4

1.2

1.0

0.8
f(x)

0.6

0.4

0.2

0.0

0.0 0.5 1.0 1.5 2.0

x

b. P(0  X  1) =

c. P(1  X  1.5) =

d. P(X  1.5) = 1 –

137
 Chapter 4: Continuous Random Variables and Probability Distributions

6.
a.
0.8

0.7

0.6

0.5

0.4

f(x)
0.3

0.2

0.1

0.0

2.0 2.5 3.0 3.5 4.0 4.5

x

b. 1 =

c. P(X > 3) = by symmetry of the p.d.f

d.

e. P( |X–3| > .5) = 1 – P( |X–3|  .5) = 1 – P( 2.5  X  3.5)

=1–

7.

a. P(X  6) = (after u = x – .5)

b. 1 – .562 = .438; .438

c. P( 5  Y  6) = P(Y  6) – P(Y  5)  .562 – .491 = .071

138
 Chapter 4: Continuous Random Variables and Probability Distributions

8.
a.
0.20

0.15

f(y)
0.10

0.05

0.00

0 2 4 6 8 10
y

b. =

c. P(Y  3) =

d. P(Y  8) =

e. P( 3  Y  8) = P(Y  8) – P(Y < 3) =

f. P(Y < 2 or Y > 6) =

9.
a. f(x) = for 25  x  35 and = 0 otherwise

b. P(X > 33) =

c. E(X) =

30  2 is from 28 to 32 minutes:
P(28 < X < 32) =

d. P( a  x  a+2) = , since the interval has length 2.

10.
a.

139
 Chapter 4: Continuous Random Variables and Probability Distributions

b.

c. P(X  b) =

d. P(a  X  b) =

Section 4.2
11.
a. P(X  1) = F(1) =

b. P(.5  X  1) = F(1) – F(.5) =

c. P(X > .5) = 1 – P(X  .5) = 1 – F(.5) =

d. .5 =

e. f(x) = F(x) = for 0  x < 2, and = 0 otherwise

f. E(X) =

g. E(X2) =

So Var(X) = E(X2) – [E(X)]2 = , x  .471

h. From g , E(X2) = 2

12.
a. P(X < 0) = F(0) = .5

b. P(–1  X  1) = F(1) – F(–1) =

140
 Chapter 4: Continuous Random Variables and Probability Distributions

c. P(X > .5) = 1 – P(X  .5) = 1 – F(.5) = 1 – .6836 = .3164

d. F(x) = F(x) = =

e. by definition. F(0) = .5 from a above, which is as desired.

13.

a.

b. cdf: F(x)= . So

c. P(x > 2) = 1 – F(2) = 1 – or .125;

d.

or .75

e.

141
 Chapter 4: Continuous Random Variables and Probability Distributions

14. for –1 ≤ x ≤ 1

a. P(Y = .5) = P(X ≥ .5) = = .25

b. P(Y = –.5) = .25 as well, due to symmetry. For –.5 < y < .5, F(y) = .25 +
= .25 + .5(y + .5) = .5 + .5y. Since Y ≤ .5, F(.5) = 1 and F(y) = 1 for y > .5 as well. That
is,

1.0

0.8

0.6
F(y)

0.4

0.2

0.0

-1.0 -0.5 0.0 0.5 1.0

y

15.
a. F(x) = 0 for x  0, = 1 for x  1, and for 0 < x < 1,

4
1.0

0.8
3

0.6
2
f(x)

F(x)

0.4

1
0.2

0 0.0

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
x x

b. F(.5) = 10(.5)9 – 9(.5)10  .0107

c. P(.25  X  .5) = F(.5) – F(.25)  .0107 – [10(.25)9 – 9(.25)10]

 .0107 – .0000  .0107

142
 Chapter 4: Continuous Random Variables and Probability Distributions

d. The 75th percentile is the value of x for which F(x) = .75

 .75 = 10(x)9 – 9(x)10  x  .9036

e. E(X) =

E(X2) =

V(X)  .6818 – (.8182)2 = .0124, x = .11134.

f.    = (.7068, .9295). Thus, P( –   X   + ) = F(.9295) – F(.7068) = .8465

– .1602 = .6863, so the probability X is more than 1 sd from its mean equals 1–.6863 =
3137.

16.
a. F(x) = 0 for x < 0 and F(x) = 1 for x > 2. For 0  x  2,
F(x) =

1.0

0.8

0.6
F(x)

0.4

0.2

0.0

0.0 0.5 1.0 1.5 2.0

x

b. P(X  .5) = F(.5) =

c. P(.25  X  .5) = F(.5) – F(.25) =

d. .75 = F(x) =  x3 = 6  x  1.8171

e. E(X) =

E(X2) =

V(X) = x = .3873

f.    = (1.1127, 1.8873). Thus, P( –   X   + ) = F(1.8873) – F(1.1127) = .8403

– .1722 = .6681
143
 Chapter 4: Continuous Random Variables and Probability Distributions

17. With X = temperature in C, temperature in F = so

so  = 3.6

18. If X is uniformly distributed on the interval from A to B, then

V(X) = E(X2) – [E(X)]2 = .

With A = 7.5 and B = 20, E(X) = 13.75, V(X) = 13.02

a. F(x) =

b. P(X  10) = F(10) = .200; P(10  X  15) = F(15) – F(10) = .4

c.  = 3.61, so    = (10.14, 17.36)

Thus, P( –   X   + ) = F(17.36) – F(10.14) = .5776
Similarly, P( – 2  X   + 2) = P(6.53  X  20.97) = 1

144
 Chapter 4: Continuous Random Variables and Probability Distributions

19.
a. P(X  1) = F(1) = .25[1 + ln(4)]  .597

b. P(1  X  3) = F(3) – F(1)  .966 – .597  .369

c. f(x) = F(x) = .25 ln(4) – .25 ln(x) for 0 < x < 4

20.

a. For 0  y  5, F(y) =

For 5  y  10, F(y) =

1.0

0.8

0.6
F(y)

0.4

0.2

0.0

0 2 4 6 8 10
y

b. For 0 < p  .5, p = F(yp) =

For .5 < p  1, p =

c. E(Y) = 5 by straightforward integration (or by symmetry of f(y)), and similarly V(Y)=

. For the waiting time X for a single bus,

E(X) = 2.5 and V(X) =

21. E(area) = E(R2) =

145
 Chapter 4: Continuous Random Variables and Probability Distributions

22.

a. For 1  x  2, F(x) = so

F(x) =

b.  2xp2 – (4 – p)xp + 2 = 0  xp = To find

, set p = .5  = 1.64

c. E(X) =

E(X2) = Var(X) = .0626

d. Amount left = max(1.5 – X, 0), so

E(amount left) =

23.
a. F(X) = =p  x = (100p)th percentile = A + (B – A)p

b.

c.

146
 Chapter 4: Continuous Random Variables and Probability Distributions

24.

a. E(X) =

b. E(X) = 

c. E(X2) = , so

Var(X) =

d. Var(X) = , since E(X2) = .

e. E(Xn) = , which will be finite if n – (k+1) < –1, i.e. if n < k.

25. Since X is uniform on [0,360], E(X) = = 180 and σX = = . Using

the linear representation of Y, E(Y) = (2π/360)E(X) – π = (2π/360)(180) – π = 0, and σY =

(2π/360)σX = (2π/360)( )= ≈ 1.814. (In fact, Y is uniform on [–π, π].)

147
 Chapter 4: Continuous Random Variables and Probability Distributions

26.
a. 1= = =  k = 2.4

b.
2.5

2.0

1.5
f(x)

1.0

0.5

0.0

0 1 2 3 4 5 6 7 8
x

c. E(X) = = = 0.5, or
$500. Similarly, E(X2) = =
= 0.625, so V(X) = 0.625 – (0.5)2 = 0.375, and σX
= = 0.612, or $612.

d. The maximum out-of-pocket expense, $2500, occurs when $500 + 20%(X – $500) equals
$2500; this accounts for the $500 deductible and the 20% of costs above $500 not paid by
the insurance plan. Solve: $2,500 = $500 + 20%(X – $500)  X = $10,500. At that
point, the insurance plan has already paid $8,000, and the plan will pay all expenses
thereafter.

Recall that the units on X are thousands of dollars. If Y denotes the expenses paid by the
company (also in $1000s), Y = 0 for X ≤ 0.5; Y = .8(X – 0.5) for 0.5 ≤ X ≤ 10.5; and Y =
(X – 10.5) + 8 for X > 10.5. From this,

E(Y) = = +

= 0 + 0.16024 + .00013, or $160.37.

27. P(Y  1.8 + 32) = P(1.8X + 32  1.8 + 32) = P( X  ) = .5

a. 90th for Y = 1.8(.9) + 32 where (.9) is the 90th percentile for X, since
P(Y  1.8(.9) + 32) = P(1.8X + 32  1.8(.9) + 32)
= (X  (.9) ) = .9 as desired.

b. The (100p)th percentile for Y is 1.8(p) + 32, verified by substituting p for .9 in the
argument of b. When Y = aX + b, (i.e. a linear transformation of X), and the (100p)th

148
 Chapter 4: Continuous Random Variables and Probability Distributions

percentile of the X distribution is (p), then the corresponding (100p)th percentile of the
Y distribution is a(p) + b. (same linear transformation applied to X’s percentile)

Section 4.3

28.
a. P(0  Z  2.17) = (2.17) – (0) = .4850

b. (1) – (0) = .3413

c. (0) – (–2.50) = .4938

d. (2.50) – (–2.50) = .9876

e. (1.37) = .9147

f. P( –1.75 < Z) + [1 – P(Z < –1.75)] = 1 – (–1.75) = .9599

g. (2) – (–1.50) = .9104

h. (2.50) – (1.37) = .0791

i. 1 – (1.50) = .0668

j. P( |Z|  2.50 ) = P( –2.50  Z  2.50) = (2.50) – (–2.50) = .9876

29.
a. .9838 is found in the 2.1 row and the .04 column of the standard normal table so c = 2.14.

b. P(0  Z  c) = .291  (c) = .7910  c = .81

c. P(c  Z) = .121  1 – P(c  Z) = P(Z < c) = (c) = 1 – .121 = .8790  c = 1.17

d. P(–c  Z  c) = (c) – (–c) = (c) – (1 – (c)) = 2(c) – 1

 (c) = .9920  c = .97

e. P( c  | Z | ) = .016  1 – .016 = .9840 = 1 – P(c  | Z | ) = P( | Z | < c )

= P(–c < Z < c) = (c) – (–c) = 2(c) – 1
 (c) = .9920  c = 2.41

149
 Chapter 4: Continuous Random Variables and Probability Distributions

30.
a. (c) = .9100  c  1.34 (.9099 is the entry in the 1.3 row, .04 column)

b. 9th percentile = –91st percentile = –1.34

c. (c) = .7500  c  .675 since .7486 and .7517 are in the .67 and .68 entries,
respectively.

d. 25th = –75th = –.675

e. (c) = .06  c  .–1.555 (.0594 and .0606 appear as the –1.56 and –1.55 entries,
respectively).

31.
a. Area under Z curve above z.0055 is .0055, which implies that
( z.0055) = 1 – .0055 = .9945, so z.0055 = 2.54

b. ( z.09) = .9100  z = 1.34 (since .9099 appears as the 1.34 entry).

c. ( z.663) = area below z.663 = .3370  z.633  –.42

32. From Table A.3, P(–1.96  Z  1.96) = .95. Then P( – .1  X   + .1) =

implies that = 1.96, and thus that

150
 Chapter 4: Continuous Random Variables and Probability Distributions

33.

a. P(X  18) = = P(Z  2.4) = (2.4) = .9452

b. P(10  X  12) = P(–4.00  Z  –2.40)  P(Z  –2.40) = (–2.40) = .0082

c. P( |X – 15|  1.5(1.25) ) = P( |Z| ≤ 1.5) = P(–1.5  Z  1.5) = (1.5) – (–1.5) = .8664

34. p = .10; n = 200; np = 20, npq = 18

a. P(X  30) = = (2.47) = .9932

b. P(X < 30) =P(X  29) = = (2.24) = .9875

c. P(15  X  25) = P(X  25) – P(X  14) =

(1.30) – (–1.30) = .9032 – .0968 = .8064

35.
a. P(X  10) = P(Z  .43) = 1 – (.43) = 1 – .6664 = .3336.
P(X > 10) = P(X  10) = .3336, since for any continuous distribution, P(x = a) = 0.

b. P(X > 20) = P(Z > 4)  0

c. P(5  X  10) = P(–1.36  Z  .43) = (.43) – (–1.36) = .6664 – .0869 = .5795

d. P(8.8 – c  X  8.8 + c) = .98, so 8.8 – c and 8.8 + c are at the 1st and the 99th percentile
of the given distribution, respectively. The 1st percentile of the standard normal
distribution has the value –2.33, so
8.8 – c =  + (–2.33) = 8.8 – 2.33(2.8)  c = 2.33(2.8) = 6.524.

e. From a, P(x > 10) = .3336. Define event A as {diameter > 10}, then P(at least one A i) =
1 – P(no Ai) =

36.
a. P(X < 1500) = P(Z < 3) = (3) = .9987; P(X ≥ 1000) = P(Z ≥ –.33) = 1 – (–.33) = 1
–.3707 = .6293

b. P(1000 < X < 1500) = P(–.33 < Z < 3) = (3) – (–.33) = .9987 – .2707 = .7280

c. From the table, (z) = .02  z ≈ –2.05  x = 1050 – 2.05(150) = 742.5 μm. The
smallest 2% of droplets are those smaller than 742.5 μm in size

151
 Chapter 4: Continuous Random Variables and Probability Distributions

d. P(at least one droplet in 5 that exceeds 1500 μm) = 1 – P(all 5 are less than 1500 μm) = 1
– (.9987)5 = 1 – .9935 = .0065

37.
a. P(X = 105) = 0, since the normal distribution is continuous;
P(X < 105) = P(Z < 0.2) = P(Z ≤ 0.2) = Φ(0.2) = .5793;
P(X ≤ 105) = .5793 as well, since X is continuous

b. No, the answer does not depend on μ or σ. For any normal rv, P(|X–μ| > σ) = P(|Z| > 1) =
P(Z < –1 or Z > 1) = 2Φ(–1) = 2(.1587) = .3174

c. From the table, (z) = .1% = .001  z ≈ –3.09  x = 104 – 3.09(5) = 88.55 mmol/L.
The smallest .1% of chloride concentration values are those less than 88.55 mmol/L

38. Let X denote the diameter of a randomly selected cork made by the first machine, and let Y be
defined analogously for the second machine.
P(2.9  X  3.1) = P(–1.00  Z  1.00) = .6826
P(2.9  Y  3.1) = P(–7.00  Z  3.00) = .9987
So the second machine wins handily.

39.
a.  + (91st percentile from std normal) = 30 + 5(1.34) = 36.7

b. 30 + 5( –1.555) = 22.225

c.  = 3.000 m;  = 0.140. We desire the 90th percentile: 30 + 1.28(0.14) = 3.179

40.  = 43;  = 4.5

a. P(X < 40) = = P(Z < –0.667) = .2514

P(X > 60) = = P(Z > 3.778)  0

b. 43 + (–0.67)(4.5) = 39.985

41. P(damage) = P(X < 100) = = P(Z < –3.33) = .0004

P(at least one among five is damaged) = 1 – P(none damaged)

= 1 – (.9996)5 = 1 – .998 = .002

152
 Chapter 4: Continuous Random Variables and Probability Distributions

42.

a. P(X  100) = = P(Z  2) = (2.00) = .9772

b. P(X  80) = = P(Z  0) = (0.00) = .5

c. P(65  X  100) = = P(–1.50  Z  2)

= (2.00) – (–1.50) = .9772 – .0668 = .9104

d. P(70  X) = P(–1.00  Z) = 1 – (–1.00) = .8413

e. P(85  X  95) = P(.50  Z  1.50) = (1.50) – (.50) = .2417

f. P(|X – 80|  10) = P(–10  X – 80  10) = P(70  X  90)

P(–1.00  Z  1.00) = .6826

43. Since 1.28 is the 90th z percentile (z.1 = 1.28) and –1.645 is the 5th z percentile (z.05 = 1.645),
the given information implies that  + (1.28) = 10.256 and  + (–1.645) = 9.671, from
which (–2.925) = –.585,  = .2000, and  = 10.

44.
a. By symmetry, P(–1.72  Z  –.55) = P(.55  Z  1.72) = (1.72) – (.55)

b. P(–1.72  Z  .55) = (.55) – (–1.72) = (.55) – [1 – (1.72)]

No, symmetry of the Z curve about 0.

45. With  = .500 inches, the acceptable range for the diameter is between .496 and .504 inches,
so unacceptable bearings will have diameters smaller than .496 or larger than .504. The new
distribution has  = .499 and  =.002. P(X < .496 or X >.504) =

= Φ(–1.5) + [1

– Φ(2.5)] = .073. 7.3% of the bearings will be unacceptable.

46.

a. P( – 1.5  X   + 1.5) = P(–1.5  Z  1.5) = (1.50) – (–1.50) = .8664

b. P( X <  – 2.5 or X >  + 2.5) = 1 – P( – 2.5  X   + 2.5)

= 1 – P(–2.5  Z  2.5) = 1 – .9876 = .0124

c. P( – 2  X   –  or  +   X   + 2) = P(within 2 sd’s) – P(within 1 sd) =

P( – 2  X   + 2) – P( –   X   + )
= .9544 – .6826 = .2718

153
 Chapter 4: Continuous Random Variables and Probability Distributions

47. P(|X – |  ) = P(X   –  or X   + ) = 1 – P( –   X   + )

= 1 – P(–1  Z  1) = .3174.
Similarly, P(|X – |  2) = 1 – P(–2  Z  2) = .0456 and P(|X – |  3) = .0026.
These are considerably less than the bounds 1, .25, and .11 given by Chebyshev.

48.
a. P(67  X  75) = P(–1.00  Z  1.67) = .7938

b. P(70 – c  X  70 + c) =

c. 10P(a single one is acceptable) = 7.938

d. p = P(X < 73.84) = P(Z < 1.28) = .9, so P(Y  8) = B(8;10,.9) = .264

49. X N(3432, 482)

a.

b.

c. We will use the conversion 1 lb = 454 g, then 7 lbs = 3178 grams, and we wish to find

d. We need the top .0005 and the bottom .0005 of the distribution. Using the Z table,
both .9995 and .0005 have multiple z values, so we will use a middle value, ±3.295.
Then 3432±(482)3.295 = 1844 and 5020, or the most extreme .1% of all birth weights
are less than 1844 g and more than 5020 g.

e. Converting to lbs yields mean 7.5595 and sd 1.0608. Then

This yields the same

answer as in part c.

154
 Chapter 4: Continuous Random Variables and Probability Distributions

50. We use a Normal approximation to the Binomial distribution: X  b(x;1000,.03) ≈

N(30,5.394)

a.

b. 5% of 1000 = 50:

51. The stated condition implies that 99% of the area under the normal curve with  = 12 and  =
3.5 is to the left of c – 1, so c – 1 is the 99th percentile of the distribution. Thus c – 1 =  +
(2.33) = 20.155, and c = 21.155.

52.
a. P(20 – .5  X  30 + .5) = P(19.5  X  30.5) = P(–1.1  Z  1.1) = .7286

b. P(at most 30) = P(X  30 + .5) = P(Z  1.1) = .8643.

P(less than 30) = P(X < 30 – .5) = P(Z < .9) = .8159

53. p = .5  μ = 12.5 & σ2 = 6.25; p = .6  μ = 15 & σ2 = 6; p = .8  μ = 20 and σ2 = 4

a.
p P(15 ≤ X ≤ 20) P(14.5 ≤ Normal ≤ 20.5)
.5 = .212 = P(.80  Z  3.20) = .2112
.6 = .577 = P(–.20  Z  2.24)
= .5668
.8 = .573 = P(–2.75  Z  .25)
= .5957

b.
p P(X ≤ 15) P(Normal ≤ 15.5)
.5 = .885 = P(Z  1.20) = .8849
.6 = .575 = P(Z  .20) = .5793
.8 = .017 = P(Z  –2.25)
= .0122

c.
p P(X ≥ 20) P(Normal ≥ 19.5)
.5 = .002 = P(Z ≥ 2.80) = .0026
.6 = .029 = P(Z ≥ 1.84) = .0329
.8 = .617 = P(Z ≥ –0.25)
= .5987

54.
a. P(X > .25) = P(Z > –.83) = 1 – .2033 = .7967
155
 Chapter 4: Continuous Random Variables and Probability Distributions

b. P(X  .10) = (–3.33) = .0004

c. We want the value of the distribution, c, that is the 95th percentile (5% of the values are
higher). The 95th percentile of the standard normal distribution = 1.645. So c = .30 +
(1.645)(.06) = .3987. The largest 5% of all concentration values are above .3987 mg/cm 3.

55. n = 500, p = .75   = 375, σ = 9.68246

a. P(360  X  400) = P(359.5  X  400.5) = P(–1.60  Z  2.58) = .9409

b. P(X < 400) = P(X  399.5) = P(Z  2.53) = .9943

56. P(X   + [(100p)th percentile for std normal])

= P(Z  […]) = p as desired

156
 Chapter 4: Continuous Random Variables and Probability Distributions

57.

a. Fy(y) = P(Y  y) = P(aX + b  y) = (for a > 0).

Now differentiate with respect to y to obtain

fy(y) = so Y is normal with mean a + b

and variance a22.

b. Normal, mean , variance = 12.96

58.

a. P(Z  1) 

b. P(Z > 3) 

c. P(Z > 4)  , so

P(–4 < Z < 4)  1 – 2(.0000317) = .999937

d. P(Z > 5) 

157
 Chapter 4: Continuous Random Variables and Probability Distributions

Section 4.4

59. Mean = implies  = .00004

a. P(X > 20,000) = 1 – P(X  20,000) = 1 – F(20,000; .00004)

P(X  30,000) = F(30,000; .00004)

P(20,000  X  30,000) = .699 – .551 = .148

b. , so P(X >  + 2) = P( x > 75,000) =

1 – F(75,000;.00004) = .05.
Similarly, P(X >  + 3) = P( x > 100,000) = .018

60.
a. P(X  100 ) =
P(X  200 ) =
P(100  X  200) = P(X  200 ) – P(X  100 ) = .9375 – .7499 = .1876

b.  = ,  = 72.15
P(X >  + 2) = P(X > 72.15 + 2(72.15)) = P(X > 216.45) =

c. .5 = P(X  )

158
 Chapter 4: Continuous Random Variables and Probability Distributions

61.
a. E(X) =

b.

c. P(X  4 ) =

d. P(2  X  5) =

62. Clearly E(X) = 0 by symmetry, so V(X) = E(X2) = =

= . Solving = (40.9)2 yields λ = 0.034577

a. P(|X – 0| ≤ 40.9) = = = 1 – e–40.9 λ = .75688

63.
a. If a customer’s calls are typically short, the first calling plan makes more sense. If a
customer’s calls are somewhat longer, then the second plan makes more sense, viz. 99¢ is
less than 20min(10¢/min) = $2 for the first 20 minutes under the first (flat–rate) plan.

b. h1(X) = 10X, while h2(X) = 99 for X ≤ 20 and 99 + 10(X – 20) for X > 20. With μ = 1/λ
for the exponential distribution, it’s obvious that E[h1(X)] = 10E[X] = 10μ. On the other
hand,
E[h2(X)] = 99 + = 99 + = 99 + 10μe–20/μ.
When μ = 10, E[h1(X)] = 100¢ = $1.00 while E[h2(X)] = 99 + 100e–2 ≈ $1.13.
When μ = 15, E[h1(X)] = 150¢ = $1.50 while E[h2(X)] = 99 + 150e–4/3 ≈ $1.39.
As predicted, the first plan is better when expected call length is lower, and the second
plan is better when expected call length is somewhat higher.

64. With xp = (100p)th percentile, p = F(xp) = 1 – ,

. For p = .5, x.5 = .

65.
a. P(X  5) = F(5;7) = .238

b. P(X < 5) = P(X  5) = .238

c. P(X > 8) = 1 – P(X  8) = 1 – F(8;7) = .313

159
 Chapter 4: Continuous Random Variables and Probability Distributions

d. P( 3  X  8 ) = F(8;7) – F(3;7) = .653

e. P( 3 < X < 8 ) =.653

f. P(X < 4 or X > 6) = 1 – P(4  X  6 ) = 1 – [F(6;7) – F(4;7)] = .713

66.
a.  = 20, 2 = 80   = 20, 2 = 80   = ,=5

b. P(X  24) = = F(6;5) = .715

c. P(20  X  40) = F(10;5) – F(5;5) = .411

67.  = 24, 2 = 144   = 24, 2 = 144   = 6,  = 4

a. P(12  X  24) = F(4;4) – F(2;4) = .424

b. P(X  24) = F(4;4) = .567, so while the mean is 24, the median is less than 24, since
P(X  ) = .5. This is a result of the positive skew of the gamma distribution.

c. We want a value for which F(x;4) = .99. In table A.4, we see F(10;4) = .990. So with  =
6, the 99th percentile = 6(10) = 60.

d. We want a value for which F(t;4)=.995. In the table, F(11;4)=.995, so t = 6(11)=66. At

66 weeks, only .5% of all transistors would still be operating.

68.
a. E(X) =  = for  = .5, n = 10, E(X) = 20

b. P(X  30) = = F(15;10) = .930

c. P(X  t) = P(at least n events in time t) = P( Y  n) when Y  Poisson with parameter t

. Thus P(X  t) = 1 – P( Y < n) = 1 – P( Y  n – 1)

69.
a. {X  t} = A1  A2  A3  A4  A5

b. P(X  t) =P( A1 )  P( A2 )  P( A3 )  P( A4 )  P( A5 ) = , so Fx(t) =

P(X  t) = 1 – , fx(t) = for t  0. Thus X also has an exponential
distribution , but with parameter  = .05.

160
 Chapter 4: Continuous Random Variables and Probability Distributions

c. By the same reasoning, P(X  t) = 1 – , so X has an exponential distribution with

parameter n.

70.
a. (6) = 5! = 120

b.

c. F(4;5) = .371 from row 4, column 5 of Table A.4

d. F(5;4) = .735

e. F(0;4) = P(X  0; = 4) = 0

71. {X2  y} =

a. FY(y) = P(X2  y) = , so fY(y) =

. We recognize
this as the chi–squared p.d.f. with  = 1.

161
 Chapter 4: Continuous Random Variables and Probability Distributions

Section 4.5

72.

a. E(X) = ,

Var(X) =

b. P(X  6) =

c. P(1.5  X  6) =

73.
a. P(X  250) = F(250;2.5, 200) =
P(X < 250) = P(X  250)  .8257
P(X > 300) = 1 – F(300; 2.5, 200) =

b. P(100  X  250) = F(250;2.5, 200) – F(100;2.5, 200)  .8257 – .162 = .6637

c. The median is requested. The equation F( ) = .5 reduces to

.5 = , i.e., ln(.5)  , so = (.6931).4(200) = 172.727.

74.
a. For x > 3.5, F(x) = P( X  x) = P(X – 3.5  x – 3.5) = 1 –

b. E(X – 3.5) = = 1.329 so E(X) = 4.829

Var(X) = Var(X – 3.5) =

c. P(X > 5) = 1 – P(X  5) =

d. P(5  X  8) =

162
 Chapter 4: Continuous Random Variables and Probability Distributions

75. Using the substitution y = and dy = dx , =

by definition of the gamma function

76.
a. P(X ≤ 0.5) = F(0.5) = 1 – exp[– (0.5/β)α] = .3099

b. Using a computer, = Γ(1.55) = 0.889 and = Γ(2.10) =

1.047. From these we find μ = (.863)(0.889) = 0.785 and σ2 = (.863)2{1.047 – [0.889]2} =

.1911, or σ = 0.437. Hence, P(X > μ + σ) = P(X > 1.222) = 1 – F(1.222) = exp[–
(1.222/β)α] = .1524

c. F(x) = ½  ½ = 1 – exp[– (x/β)α]  exp[– (x/β)α] = ½  (x/β)α = ln 2  x = β(ln 2)1/α

= .863(ln 2)1/1.817 = .7054

d. Using the same math as part c, η(p) = β(–ln(1 – p))1/α = .863(–ln(1 – p))1/1.817

77.
a.

b.

c.

78.
a. 

b. = –2.25 ln(.5) = 1.5596  = 4.75

c. P = F(xp) = 1 –  (xp/) = –ln(1 – p)  xp = [ –ln(1–p)]1/

d. The desired value of t is the 90th percentile (since 90% will not be refused and 10% will
be). From c, the 90th percentile of the distribution of X – 3.5 is 1.5[ –ln(.1)]1/2 = 2.27661,
so t = 3.5 + 2.2761 = 5.7761

163
 Chapter 4: Continuous Random Variables and Probability Distributions

79.
a. E(X) = = 68.0335; V(X) = ;
x = 122.0949

b. P(50  X  250) =

P(Z  1.68) – P(Z  .34) = .9535 – .6331 = .3204.

c. P(X  68.0335) = = P(Z  .60) = .7257. The lognormal

distribution is not a symmetric distribution.

80.

a. E(X) = , V(X) =

b. f(x) = for 0  X  1,

so P(X  .2) =

c. P(.2  X  .4) =

d. E(1 – X) = 1 – E(X) = 1 –

81.
a. E(X) = ; Var(X) =

b. P(X > 125) = 1 – P(X  125) =

c. P(110  X  125)

d. =

e. P(any particular one has X > 125) = .9573  expected # = 10(.9573) = 9.573

f. We wish the 5th percentile, which is

164
 Chapter 4: Continuous Random Variables and Probability Distributions

82.

a. .5 = F( )= , (where refers to the lognormal distribution and  and 

to the normal distribution). Since the median of the standard normal distribution is 0,

, so ln( )=  = . For the power distribution, =

b. 1 –  = (z) = P(Z  z) =

= , so the 100(1 – )th percentile is . For the power distribution,

th
the 95 percentile is

83. The point of symmetry must be , so we require that , i.e.,

, which in turn implies that   .

84.
a. E(X) = ; Var(X) = , x = 11.20

b. P(X  10) = P(ln(X)  2.3026) = P(Z  .45) = .6736

P(5  X  10) = P(1.6094  ln(X) 2.3026)
= P(–.32  Z  .45) = .6736 – .3745 = .2991

85.

a. E(X) =

b. E[(1 – X)m] =

For m = 1, E(1 – X) = .

165
 Chapter 4: Continuous Random Variables and Probability Distributions

86.

a. E(Y) = 10 ; Var(Y) =

, after some algebra.

b. P(8  Y  12) = = F(.6;3,3) – F(.4; 3,3).

The standard density function here is 30y2(1 – y)2,

so P(8  Y  12) = .

c. We expect it to snap at 10, so P( Y < 8 or Y > 12) = 1 – P(8  X  12)

= 1 – .365 = .665.

Section 4.6

87. The given probability plot is quite linear, and thus it is quite plausible that the tension
distribution is normal.

88. The Weibull plot uses ln(observations) and the extreme value percentiles of the p i values
given; i.e., η(p) = ln[–ln(1–p)]. The accompanying probability plot appears sufficiently
straight to lead us to agree with the argument that the distribution of fracture toughness in
concrete specimens could well be modeled by a Weibull distribution.

166
 Chapter 4: Continuous Random Variables and Probability Distributions

89. The z percentile values are as follows: –1.86, –1.32, –1.01, –0.78, –0.58, –0.40, –0.24,–0.08,
0.08, 0.24, 0.40, 0.58, 0.78, 1.01, 1.30, and 1.86. The accompanying probability plot is
reasonably straight, and thus it would be reasonable to use estimating methods that assume a
normal population distribution.

90. The z percentiles and observations are below, along with the probability plot. The plot is quite
straight except for the point corresponding to the largest observation. This observation is
clearly much larger than what would be expected in a normal random sample. Because of this
outlier, it would be inadvisable to analyze the data using any inferential method that depended
on assuming a normal population distribution.

Percentile Observation
–1.645 152.7
–1.040 172.0
–0.670 172.5
–0.390 173.3
–0.130 193.0
0.130 204.7
0.390 216.5
0.670 234.9
1.040 262.6
1.645 422.6

167
 Chapter 4: Continuous Random Variables and Probability Distributions

91. The (z percentile, observation) pairs are (–1.66, .736), (–1.32, .863), (–1.01, .865),
(–.78, .913), (–.58, .915), (–.40, .937), (–.24, .983), (–.08, 1.007), (.08, 1.011), (.24, 1.064),
(.40, 1.109), (.58, 1.132), (.78, 1.140), (1.01, 1.153), (1.32, 1.253), (1.86, 1.394). The
accompanying probability plot is straight, suggesting that an assumption of population
normality is plausible.

92.
a. The 10 largest z percentiles are 1.96, 1.44, 1.15, .93, .76, .60, .45, .32, .19 and .06; the
remaining 10 are the negatives of these values. The accompanying normal probability
plot is reasonably straight. An assumption of population distribution normality is
plausible.

168
 Chapter 4: Continuous Random Variables and Probability Distributions

b. For a Weibull probability plot, the natural logs of the observations are plotted against
extreme value percentiles; these percentiles are –3.68, –2.55, –2.01, –1.65, –1.37, –1.13,
–.93, –.76, –.59, –.44, –.30, –.16, –.02, .12, .26, .40, .56, .73, .95, and 1.31. The
accompanying probability plot is roughly as straight as the one for checking normality (a
plot of ln(x) versus the z percentiles, appropriate for checking the plausibility of a
lognormal distribution, is also reasonably straight – any of 3 different families of
population distributions seems plausible.)

93. The pattern in the plot (below, generated by Minitab) is reasonably linear. By visual
inspection alone, it is plausible that strength is normally distributed.

169
 Chapter 4: Continuous Random Variables and Probability Distributions

94.
a. The plot of the original (untransformed) data appears somewhat curved.

b. The square root transformation results in a very straight plot. It is reasonable that this
distribution is normally distributed.

2.0

1.5
sqrt

1.0

0.5
-2 -1 0 1 2
z %iles

c. The cube root transformation also results in a very straight plot. It is very reasonable that
the distribution is normally distributed.

170
 Chapter 4: Continuous Random Variables and Probability Distributions

95. To check for plausibility of a lognormal population distribution for the rainfall data of
Exercise 81 in Chapter 1, take the natural logs and construct a normal probability plot. This
plot and a normal probability plot for the original data appear below. Clearly the log
transformation gives quite a straight plot, so lognormality is plausible. The curvature in the
plot for the original data implies a positively skewed population distribution – like the
lognormal distribution.

171
 Chapter 4: Continuous Random Variables and Probability Distributions

96. We use the data (table below) to create the desired plot. This half–normal plot reveals some
extreme values, without which the distribution may appear to be normal.

ordered absolute
values (w's) probabilities z values
0.89 0.525 0.063
1.15 0.575 0.19
1.27 0.625 0.32
1.44 0.675 0.454
2.34 0.725 0.6
3.78 0.775 0.755
3.96 0.825 0.935
12.38 0.875 1.15
30.84 0.925 1.44
43.4 0.975 1.96

172
 Chapter 4: Continuous Random Variables and Probability Distributions

97. The (100p)th percentile (p) for the exponential distribution with  = 1 satisfies F((p)) = 1 –
exp[–(p)] = p, i.e., (p) = –ln(1 – p). With n = 16, we need (p) for p = .
These are .032, .398, .170, .247, .330, .421, .521, .633, .758, .901, 1.068, 1.269, 1.520, 1.856,
2.367, 3.466. this plot exhibits substantial curvature, casting doubt on the assumption of an
exponential population distribution. Because  is a scale parameter (as is  for the normal
family),  = 1 can be used to assess the plausibility of the entire exponential family.

173
 Chapter 4: Continuous Random Variables and Probability Distributions

Supplementary Exercises

98.
a. Let S = defective. Then p = P(S) = .05; n = 250   = np = 12.5,  = 3.446. The random
variable X = the number of defectives in the batch of 250. X ~ Binomial. Since np =
12.5  10, and nq = 237.5  10, we can use the normal approximation.

P(Xbin  25)  1

b. P(Xbin = 10)  P(Xnorm  10.5) – P(Xnorm  9.5)

=

174
 Chapter 4: Continuous Random Variables and Probability Distributions

99.

a. For 0  y  25, F(y) = . Thus

F(y) =

b. P(Y  4) = F(4) = .259, P(Y > 6) = 1 – F(6) = .5

P(4  X  6) = F(6) – F(4) = .5 – .259 = .241

c. E(Y) =

E(Y2) = , so V(Y) = 43.2 – 36 = 7.2

d. P(Y < 4 or Y > 8) = 1 – P(4  X  8) = .518

e. the shorter segment has length min(Y, 12 – Y) so

E[min(Y, 12 – Y)] =

100.
a. E(cX) = cE(X) =

b. E[c(1 – .5eax)] =

101.

a. By differentiation, f(x) =

175
 Chapter 4: Continuous Random Variables and Probability Distributions

1.0

0.8

0.6

f(x)
0.4

0.2

0.0

0.0 0.5 1.0 1.5 2.0 2.5

x

b. P(.5  X  2) = F(2) – F(.5) =

c. E(X) =

102.  = 40 V;  = 1.5 V

a. P(39 < X < 42) =

= (1.33) – (–.67) = .9082 – .2514 = .6568

b. We desire the 85th percentile: 40 + (1.04)(1.5) = 41.56

c. P(X > 42) = 1 – P(X  42) = 1 = 1 – (1.33) = .0918

Let D represent the number of diodes out of 4 with voltage exceeding 42.

P(D  1 ) = 1 – P(D = 0) = =1 – .6803 = .3197

a. P(10  X  20) =

b. P(X  10) = P(10  X  25) =

c. For 0  X  25, F(x) = . F(x)=0 for x < 0 and = 1 for x > 25.

d. E(X) = ; Var(X) = , so σx = 7.22

176
 Chapter 4: Continuous Random Variables and Probability Distributions

103.  = 137.2 oz.;  = 1.6 oz

a. P(X > 135) = 1 = 1 – (–1.38) = 1 – .0838 = .9162

b. With Y = the number among ten that contain more than 135 oz, Y ~ Bin(10, .9162). So,
P(Y  8) = b(8; 10, .9162) + b(9; 10, .9162) + b(10; 10, .9162) =.9549
c.  = 137.2;

104.  = 40 V;  = 1.5 V

a. P(39 < X < 42) =

= (1.33) – (–.67) = .9082 – .2514 = .6568

b. We desire the 85th percentile: 40 + (1.04)(1.5) = 41.56

c. P(X > 42) = 1 – P(X  42) = 1 = 1 – (1.33) = .0918

Let D represent the number of diodes out of 4 with voltage exceeding 42.

P(D  1 ) = 1 – P(D = 0) = =1 – .6803 = .3197

105.

a. A lognormal distribution, since is a normal r.v.

b.

c.

177
 Chapter 4: Continuous Random Variables and Probability Distributions

106.
a. F(X) = 0 for x < 1 and = 1 for x > 3. For 1  x  3,

b. P(X  2.5) = F(2.5) = 1.5(1 – .4) = .9; P(1.5  x  2.5) =

F(2.5) – F(1.5) = .4

c. E(X) =

d. E(X2) = , so V(X) = E(X2) – [E(X)]2 = .284,

 =.553

e. h(x) =

so E[h(X)] =

107.
a.
0.5

0.4

0.3
f(x)

0.2

0.1

0.0

-1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5

x

b. F(x) = 0 for x < –1 or = 1 for x > 2. For –1  x  2,

c. The median is 0 iff F(0) = .5. Since F(0) = , this is not the case. Because < .5, the
median must be greater than 0.

d. Y is a binomial r.v. with n = 10 and p = P(X > 1) = 1 – F(1) =

178
 Chapter 4: Continuous Random Variables and Probability Distributions

108.
a. E(X) = = 1.075, = 1.075

b. P(3.0 < X) = 1 – P(X  3.0) = 1 – F(3.0) = e–.93(3.0) = .0614

P(1.0  X  3.0) = F(3.0) – F(1.0) = .333

c. The 90th percentile is requested; denoting it by c, we have

.9 = F(c) = 1 – e–(.93)c, whence c =

109.

a. P(X  150) = ,

where exp(u) = eu. P(X  300) = ,

and P(150  X  300) = .828 – .368 = .460.

b. The desired value c is the 90th percentile, so c satisfies

.9 = . Taking the natural log of each side twice in

succession yields ln[–ln(.9)] = , so c = 90(2.250367) + 150 = 352.53.

c. f(x) = F(X) =

d. We wish the value of x for which f(x) is a maximum; this is the same as the value of x for

which ln[f(x)] is a maximum. The equation of gives

, so , which implies that x = . Thus the mode is .

e. E(X) = .5772 +  = 201.95, whereas the mode is 150 and the median is
–(90)ln[–ln(.5)] + 150 = 182.99. The distribution is positively skewed.

110.

a. Clearly f(x)  0. The c.d.f. is , for x > 0,

( F(x) = 0 for x  0.)

179
Chapter 4: Continuous Random Variables and Probability Distributions

Since F() = f(x) is a legitimate pdf.

b. See above

c. P(2  X  5) = F(5) – F(2) =

d.

e. E(salvage value) =

180
 Chapter 4: Continuous Random Variables and Probability Distributions

111.
a. From a graph of f(x; , ) or by differentiation, x* = .

b. No; the density function has constant height for A  X  B.

c. F(x;) is largest for x = 0 (the derivative at 0 does not exist since f is not continuous
there) so x* = 0.

d.

e. From d

112.
a.

0.10

0.08

0.06
f(x)

0.04

0.02

0.00
-2 -1 0 1 2
x

b. For x < 0, F(x) = .

For x  0, F(x) = .

c. P(X < 0) = F(0) = , P(X < 2) = F(2) = 1 – .5e–.4 = .665,

P(–1  X  2) – F(2) – F(–1) = .256, 1 – (–2  X  2) = .670

181
 Chapter 4: Continuous Random Variables and Probability Distributions

113.
a. Clearly f(x; 1, 2, p)  0 for all x, and

=
= p + (1 – p) = 1

b. For x > 0, F(x; 1, 2, p) =

c. E(X) =

d. E(X2) = , so Var(X) =

e. For an exponential r.v., CV = . For X hyperexponential,

CV = =

= [2r – 1]1/2 where r = . But straightforward algebra shows that r >

1 provided , so that CV > 1.

f. , , so and CV = if n > 1.

182
 Chapter 4: Continuous Random Variables and Probability Distributions

114.

a. where we must have  > 1.

b. For x  5, F(x) = .

c. E(X) = , provided  > 2.

d.

, the cdf of an exponential r.v. with parameter  – 1.

115.

a. P(X > 100) = 1

b. P(50 < X < 80) =

= (–1.5) – (–3.29) = .1271 – .0005 = .1266.

c. a = 5th percentile = 96 + (–1.645)(14) = 72.97.

b = 95th percentile = 96 + (1.645)(14) = 119.03. The interval (72.97, 119.03) contains the
central 90% of all grain sizes.

183
 Chapter 4: Continuous Random Variables and Probability Distributions

116.
a. The Weibull pdf exists for x ≥ 0, but only a section is graphed below.
0.020

0.015

0.010

f(x)
0.005

0.000

150 160 170 180 190

x

b. P(X > 175) = 1 – F(175; 9, 180) =

P(150  X  175) = F(175; 9, 180) – F(150; 9, 180)
= .5398 – .1762 = .3636

c. P(at least one) = 1 – P(none) = 1 – (1 – .3636)2 = .5950

d. We want the 10th percentile: .10 = F( x; 9, 180) = . A small bit of algebra

leads us to x = 140.178. Thus 10% of all tensile strengths will be less than 140.178 MPa.

117. F(y) = P(Y  y) = P(Z +   y) = . Now

differentiate with respect to y to obtain a normal pdf with parameters  and .

118.

a. FY(y) = P(Y  y) = P(60X  y) = Thus fY(y) =

, which shows that Y has a gamma distribution with

parameters  and 60.

b. With c replacing 60 in a, the same argument shows that cX has a gamma distribution
with parameters  and c.

184
 Chapter 4: Continuous Random Variables and Probability Distributions

119.
a. Y = –ln(X)  x = e–y = k(y), so k(y) = –e–y. Thus since f(x) = 1,
g(y) = 1  | –e–y | = e–y for 0 < y < , so y has an exponential distribution with parameter
 = 1.

b. y = Z +   y = h(z) = Z +   z = k(y) = and k(y) = , from which the

result follows easily.

c. y = h(x) = cx  x = k(y) = and k(y) = , from which the result follows easily.

120.
a. If we let and , then we can manipulate f(v) as follows:

which is in the Weibull family of distributions.

b. ; cdf: , so

121.
a. Assuming independence, P(all 3 births occur on March 11) =

b.

c. Let X = deviation from due date. XN(0, 19.88). Then the baby due on March 15 was 4
days early. P(x = –4) ≈ P(–4.5 < x < –3.5)

Similarly, the baby due on April 1 was 21 days early, and P(x = –21) ≈

The baby due on April 4 was 24 days early, and P(x = –24) ≈ .0097

Again, assuming independence, P( all 3 births occurred on March 11) =

d. To calculate the probability of the three births happening on any day, we could make
similar calculations as in part c for each possible day, and then add the probabilities.

185
 Chapter 4: Continuous Random Variables and Probability Distributions

122.

a. F(x) = and F(x) = , so r(x) = , a constant; this is consistent

with the memoryless property of the exponential distribution.

b. r(x) = ; for  > 1 this is increasing, while for  < 1 it is a decreasing function.

c. ln(1 – F(x)) = ,

f(x) = 0x

123.

a. FX(x) =

since FU(u) = u (U is uniform on [0, 1]). Thus X has an

exponential distribution with parameter .

b. By taking successive random numbers u1, u2, u3, …and computing ,

… we obtain a sequence of values generated from an exponential distribution with
parameter  = 10.

124.
a. E(g(X))  E[g() + g()(X – )] = E(g()) + g()E(X – ), but E(X) –  = 0 and
E(g()) = g() ( since g() is constant), giving E(g(X))  g().
V(g(X))  V[g() + g()(X – )] = V[g()(X – )] = (g())2V(X – ) =
(g())2V(X).

b. , so

125. g() + g()(X – )  g(X) implies that E[g() + g()(X – )] = E(g()) = g()  E(g(X)),
i.e. that g(E(X))  E(g(X)).

186
 Chapter 4: Continuous Random Variables and Probability Distributions

126. For y > 0, . Now

take the cdf of X (Weibull), replace x by , and then differentiate with respect to y to

obtain the desired result fY(y).

127.

a. E(X) = 150 + (850 – 150) = 710; V(X) =  σX ≈ 84.423

P(|X – 710| ≤ 84.423) = P(625.577 ≤ X ≤ 794.423) =

= .684. The computation of this

integral requires a calculator or computer.

b. P(X > 750) = = .376. Again, the

computation of the requested integral requires a calculator or computer.

128.
a. For w < 0, F(w) = 0. F(0) = P(V ≤ vd) = 1 – exp(–λvd), since V is exponential. For w > 0,
F(w) = P(W ≤ w) = P(k(V–vd) ≤ w) = P(V ≤ vd + w/k) = 1 – exp(–λ[vd+w/k]). This can be
written more compactly as

F(w) = 1 – exp(–λ[vd+w/k]) for w ≥ 0 and F(w) = 0 otherwise

b. For w > 0, f(w) = F′(w) = exp(–λ[vd+w/k]). Since the only other possible value of W is
zero, which would not contribute to the expected value, we can compute E(W) as the
appropriate integral:

E(W) = =

187