LUBS267001

©UNIVERSITY OF LEEDS
January 2014

Examination for the degree of

BA and BSc

STATISTICS FOR BUSINESS AND ECONOMICS 2

ANSWERS
QUESTION ANSWER QUESTION ANSWER
1 A 31 B
2 D 32 A
3 C 33 E
4 D 34 A
5 D 35 A
6 A 36 C
7 C 37 A
8 E 38 D
9 B 39 C
10 A 40 B
11 C 41 E
12 E 42 E
13 C 43 D
14 E 44 A
15 A 45 B
16 E 46 E
17 B 47 D
18 C 48 E
19 A 49 B
20 D 50 C
21 C 51 A
22 C 52 E
23 E 53 C
24 B 54 B
25 D 55 A
26 E 56 C
27 E 57 B
28 D 58 E
29 E 59 B
30 C 60 A

2
1. The continuous random variable X , the time in minutes that elapses between any two
defects, has the following probability density function (pdf ):

f ( x ) = 0.069 e −0.069 x , x>0

which is the equation of a (negative) exponential function.

Since e = 2.718... > 0 , the expression e − 0.069 x is also positive for any value x. Therefore

f (x ) > 0 ∀ x, x > 0

i.e. f (x ) is positive.

First derivative of f (x ) with respect to x:

df ( x ) de g ( x )
f ' (x ) = = 0.069 (− 0.069 e − 0.069 x ) = − 01 − 0.069 x
23 ⋅ e123
.069 2
recall : = g ' (x )e g ( x )
dx dx
+ +

Therefore

f ' (x ) < 0 ∀ x, x > 0

i.e. f (x ) is decreasing.

Second derivative of f ( x ) with respect to x:

d 2 f (x )
f ' ' (x ) =
dx 2
( )
= −0.069 2 − 0.069 e − 0.069 x = 01
.069
2 3
3 − 0.069 x
⋅ e123
+ +

Therefore

f ' ' (x ) > 0 ∀ x, x > 0

i.e. f (x ) is convex.

3
 f (x )
x = 0 ⇒ f (x ) = 0.069 e 0 = 0.069 (1) = 0.069
0.069

x → ∞ ⇒ f ( x ) → 0.069 e − ∞
1
→ 0.069 → 0
e∞

0 x

ANSWER: A

2. The continuous random variable X , the time in minutes that elapses between any two
defects, follows a (negative) exponential distribution, whose pdf has the general form

f ( x ) = θ e −θ x , x > 0, θ > 0

In our case:

f ( x ) = 0.069 e −0.069 x , x>0

therefore

θ = 0.069

The expected value (mean) of the exponential distribution of X is given by

E(X ) =
1
θ

So, in our case, the expected value (mean) of X in minutes is

E(X ) =
1 1
= = 14.493
θ 0.069

ANSWER: D

3. The variance of the exponential distribution of X is given by

var( X ) =
1
θ2

therefore

4
 SD ( X ) = var ( X ) =
1 1
=
θ 2
θ

So, in our case, the standard deviation of X in minutes is

SD( X ) =
1 1
= = 14.493
θ 0.069

ANSWER: C

4. The mean occurrence rate is the mean rate at which defects occur, i.e. the mean
number of defects per unit of time:

1
θ=
E(X )

The random variable X is the time in minutes that elapses between any two defects.
So, in our case

1
θ= = 0.069
E(X )

is the mean occurrence rate of defects per minute.

Therefore, the mean occurrence rate of defects per hour (60 minutes) is

0.069 × 60 = 4.140

ANSWER: D

5. We use the property of the pdf f ( x ) :

b
P(a < X < b ) = ∫ f ( x ) dx
a

In our case, the probability that the time that elapses between any two defects will be
less than 15 minutes, i.e. between 0 and 15 minutes since x > 0 , is

15 15
P ( X < 15) = P (0 < X < 15) = ∫ θ e dx =
−θ x
∫ 0.069 e
− 0.069 x
dx =
x =0 x =0

5
 15 15
⎡ 1 ⎤ 1
= 0.069 ∫ e − 0.069 x
dx = 0.069 ⎢− e − 0.069 x ⎥ = recall : ∫ e c x dx = ec x
x =0 ⎣ 0.069 ⎦0 c

[
= − e − 0.069 x ]
15
0
[ ]
= − e − 0.069 (15 ) − e − 0.069 (0 ) = − ( e − 1.035 − e 0 ) =

= −(0.355 − 1) = 1 − 0.355 = 0.645

(using a calculator to evaluate e −1.035 = 0.355 ).

ANSWER: D

6. The probability that the time that elapses between any two defects will be more than 5
minutes, but less than 15 minutes, is

P (5 < X < 15) = P ( X < 15) − P ( X < 5)

We know from answers 5 that P ( X < 15) = 0.645 . We need to find P ( X < 5) , the
probability that the time that elapses between any two defects will be less than 5
minutes, i.e. between 0 and 5 minutes since x > 0 :

5 5
P ( X < 5) = P (0 < X < 5) = ∫ θ e dx =
−θ x
∫ 0.069 e
− 0.069 x
dx =
x =0 x =0

5 5
⎡ 1 ⎤
= 0.069 ∫ e − 0.069 x
dx = 0.069 ⎢− e − 0.069 x ⎥ =
x =0 ⎣ 0.069 ⎦0

[
= − e − 0.069 x ] 5
0
[ ]
= − e − 0.069 (5 ) − e − 0.069 (0 ) = − ( e −0.345 − e 0 ) =

= −(0.708 − 1) = 1 − 0.708 = 0.292

(using a calculator to evaluate e −0.345 = 0.708 ).

Therefore

P (5 < X < 15) = P ( X < 15) − P ( X < 5) = 0.645 − 0.292 = 0.353

ANSWER: A

7. We can use the following property of a joint probability distribution:

∑ ∑ f ( x, y ) = 1
x y

6
 i.e. the sum of all probabilities is equal to 1.

In our case

⎛ 1 2 ⎞
f ( x, y ) = c ⎜1 −
1 2
x + y− x y ⎟, x = 0, 1, 2, 3, 4, y = 0, 1, 2, 3
⎝ 20 20 ⎠

So we have

4 3
⎛ 1 1 2 ⎞
∑ ∑ c ⎜⎝1 − 20 x
x =0 y =0
2
+ y−
20
x y⎟ = 1
⎠

4 3
⎛ 1 2 1 2 ⎞ 4 3
⎛ 1 2 1 2 ⎞
∑ ∑
x =0 y =0
c ⎜ 1
⎝ 20
− x + y −
20
x y ⎟
⎠
= c ∑ ∑ ⎜1 −
x =0 y =0 ⎝ 20
x + y−
20
x y⎟ =
⎠

⎛ 4 3 1 4 3 4 3
1 4 3 ⎞
= c ⎜⎜ ∑∑1 − ∑ ∑ x 2 + ∑ ∑ y − ∑ x 2 ∑ y ⎟⎟ =
⎝ x =0 y =0 20 x =0 y =0 x =0 y =0 20 x =0 y =0 ⎠

⎡4 ⎤
= c ⎢∑ 4(1) − ∑ 4 (x 2 ) + ∑ 6 − ∑ x 2 (6 )⎥ =
1 4 4
1 4
⎣ x =0 20 x =0 x =0 20 x =0 ⎦

⎡4 1 4 4
3 4 ⎤
= c ⎢∑ 4 − ∑ x 2 + ∑ 6 − ∑ x 2 ⎥ =
⎣ x =0 5 x =0 x =0 10 x =0 ⎦

⎡ ⎤
= c ⎢5(4 ) − (30 ) + 5(6 ) − (30 )⎥ = c (20 − 6 + 30 − 9 ) = 35 c
1 3
⎣ 5 10 ⎦

1
∴ 35 c = 1 ⇒ c =
35

The joint probability distribution of X and Y is therefore:

1 ⎛ 1 2 ⎞
f ( x, y ) =
1 2
⎜1 − x + y− x y ⎟, x = 0, 1, 2, 3, 4, y = 0, 1, 2, 3 .
35 ⎝ 20 20 ⎠

ANSWER: C

8. If X and Y are discrete random variables and f ( x, y ) is their joint probability

distribution, the general definition of marginal distribution of X is

f ( x ) = ∑ f ( x, y ) for x = x1 , x 2 , ..., x n
y

7
 In our case

⎛ 1 2 ⎞
f ( x, y ) = c ⎜1 −
1 2
x + y− x y ⎟, x = 0, 1, 2, 3, 4, y = 0, 1, 2, 3
⎝ 20 20 ⎠

So we have

3
⎛ 1 2 ⎞
f ( x ) = ∑ c ⎜1 −
1 2
x + y− x y⎟
y =0 ⎝ 20 20 ⎠

ANSWER: E

3
⎛ 1 2 ⎞
f ( x ) = ∑ c ⎜1 −
1 2
9. x + y− x y⎟ =
y =0 ⎝ 20 20 ⎠

3
1 ⎛ 1 2 1 2 ⎞
=∑ ⎜1 − x + y− x y⎟ =
y =0 35 ⎝ 20 20 ⎠

1⎛ 3 1 3 3
1 2 3 ⎞
= ⎜ ∑1 − ∑ x 2 + ∑ y − x ∑ y⎟ =
35 ⎜⎝ y =0 20 y =0 y =0 20 y =0 ⎟⎠

4 (1) − 4(x 2 ) + 6 −
1 ⎡ 1 2 ⎤ 1⎛ 6 2⎞
x (6 )⎥ = ⎜ 4 −
1 4 2
= ⎢ x +6− x ⎟=
35 ⎣ 20 20 ⎦ 35 ⎝ 20 20 ⎠

1⎛ 10 2 ⎞ 1 ⎛ 2 2⎞ 1⎛ 1 ⎞
= ⎜10 − x ⎟ = ⎜2 − x ⎟ = ⎜2 − x2 ⎟
35 ⎝ 20 ⎠ 7 ⎝ 20 ⎠ 7 ⎝ 10 ⎠

Therefore, the marginal distribution of X is

1⎛ 1 ⎞
f (x ) = ⎜ 2 − x 2 ⎟ , x = 0, 1, 2, 3, 4
7⎝ 10 ⎠

ANSWER: B

10. If X and Y are discrete random variables and f ( x, y ) is their joint probability
distribution, the general definition of marginal distribution of Y is

f ( y ) = ∑ f ( x, y ) for y = y1 , y 2 , ..., y m
x

In our case

8
 ⎛ 1 2 ⎞
f ( x, y ) = c ⎜1 −
1 2
x + y− x y ⎟, x = 0, 1, 2, 3, 4, y = 0, 1, 2, 3
⎝ 20 20 ⎠

So we have

4
⎛ 1 2 ⎞
f ( y ) = ∑ c ⎜1 −
1 2
x + y− x y⎟
x =0 ⎝ 20 20 ⎠

ANSWER: A

4
⎛ 1 2 ⎞
f ( y ) = ∑ c ⎜1 −
1 2
11. x + y− x y⎟ =
x =0 ⎝ 20 20 ⎠

4
1 ⎛ 1 2 1 2 ⎞
=∑ ⎜1 − x + y− x y⎟ =
x = 0 35 ⎝ 20 20 ⎠

1⎛ 4 1 4 4
1 4 2⎞
= ⎜ ∑1 − ∑ x 2 + ∑ y − y∑ x ⎟ =
35 ⎝ x =0 20 x =0 x =0 20 x =0 ⎟⎠

1 ⎡ ⎤ 1⎛ 3 ⎞
= 5 (1) −
1
(30 ) + 5( y ) −
1
y (30 ) = ⎜ 5 −
3
+ 5 y − y⎟ =
35 ⎢⎣ 20 20 ⎥⎦ 35 ⎝ 2 2 ⎠

1 ⎛7 7 ⎞ 1⎛1 1 ⎞ 1
= ⎜ + y ⎟ = ⎜ + y ⎟ = (1 + y )
35 ⎝ 2 2 ⎠ 5 ⎝ 2 2 ⎠ 10

Therefore, the marginal distribution of Y is

f (y) = (1 + y ),
1
y = 0, 1, 2, 3
10

ANSWER: C

12. The formula for the required conditional distribution is

1 ⎛ 1 2 ⎞ 1 ⎛ 4 2⎞
x (3)⎟
1 2
⎜1 − x + 3− ⎜4 − x ⎟
f (x, Y = 3) 35 ⎝ 20
f (x Y = 3) =
20 ⎠ 35 ⎝ 20 ⎠
= = =
f (Y = 3) 1
(1 + 3) 4
10 10

10 1 ⎛ 4 2⎞ 5 1 ⎛ 4 2⎞ 1 1 ⎛ 2 2⎞
= ⋅ ⎜4 − x ⎟ = ⋅ ⎜4 − x ⎟ = ⋅ ⎜2 − x ⎟=
4 35 ⎝ 20 ⎠ 2 35 ⎝ 20 ⎠ 1 7 ⎝ 20 ⎠

9
 1⎛ 1 2⎞
= ⎜2 − x ⎟
7 ⎝ 10 ⎠

Therefore, the conditional probability distribution of X , the daily in-store sales, when
Y = 3 – i.e. when the bookshop sells 3 textbooks online – is

1⎛ 1 2⎞
f (x Y = 3) = ⎜ 2 − x ⎟, x = 0, 1, 2, 3, 4
7 ⎝ 10 ⎠

ANSWER: E

13. The random variables X and Y are independent if

f ( x ) f ( y ) = f ( x, y ) ∀ x, y over the whole sets of their possible values

ANSWER: C

14. For the marginal distributions of X and Y we have:

1⎛ 1 ⎞
f (x ) = ⎜ 2 − x 2 ⎟ , x = 0, 1, 2, 3, 4
7⎝ 10 ⎠

f (y) = (1 + y ),
1
y = 0, 1, 2, 3
10

and the joint probability distribution of X and Y is:

1 ⎛ 1 2 ⎞
f ( x, y ) =
1 2
⎜1 − x + y− x y ⎟, x = 0, 1, 2, 3, 4, y = 0, 1, 2, 3
35 ⎝ 20 20 ⎠

So in our case we have

⎡1 ⎛ 1 ⎞⎤ ⎡ 1 ⎤ 1 ⎛ ⎞
f ( x ) f ( y ) = ⎢ ⎜ 2 − x 2 ⎟⎥ ⋅ ⎢ (1 + y )⎥ = ⎜ 2 + 2 y − x 2 − x 2 y ⎟ =
1 1
⎣ 7 ⎝ 10 ⎠⎦ ⎣10 ⎦ 70 ⎝ 10 10 ⎠

2 ⎛ 1 2 1 2 ⎞ 1⎛ 1 2 ⎞
x y ⎟ = f ( x, y )
1 2
= ⎜1 + y − x − x y ⎟ = ⎜1 − x + y−
70 ⎝ 20 20 ⎠ 35 ⎝ 20 20 ⎠

Thus f ( x ) f ( y ) = f ( x, y ) and therefore X and Y , the daily in-store and online sales,
are independent. In terms of conditional probability, this also implies:

f (x y ) = f ( x ) ∀ x, y

10
 which we saw to be the case in answer 12.

ANSWER: E

15. If X and Y are continuous random variables and f ( x, y ) is their joint probability
distribution, the general definition of marginal distribution of X is

d
f (x ) = ∫ f (x, y ) dy for a< x<b
y =c

In our case, f ( x, y ) = xy (x + y ), 1 < x < 2 , 0 < y < 2 . So we have

3
26

(x y + xy 2 )dy
2 2
f (x ) = xy (x + y ) dy =
3 3 2
∫
y =0
26 ∫
y =0
26

ANSWER: A

(x y + xy 2 )dy =
2
f (x ) =
3 2
16. ∫
y =0
26

2
3 ⎡ 2 y2 y3 ⎤ 3 ⎡⎛ 2 4 ⎤
( ) 8⎞
2

⎜ x + x ⎟ − (0 )⎥ =
3
= ∫ + = + =
2 2
x y xy dy ⎢ x x ⎥ ⎢
3 ⎦ 0 26 ⎣⎝ 2 3⎠
26 y =0 26 ⎣ 2 ⎦

3 ⎛8 2⎞ 3 ⎛4 2⎞
= ⎜ x + 2x ⎟ = ⎜ x + x ⎟
26 ⎝ 3 ⎠ 13 ⎝ 3 ⎠

Therefore, the marginal distribution of X is

3 ⎛4 2⎞
f (x ) = ⎜ x + x ⎟, 1< x < 2
13 ⎝ 3 ⎠

ANSWER: E

2
17. E (X ) = ∫ x f (x ) dx =
x =1

11
 2
3 ⎛4 ⎞ ⎛4 2 3⎞ 3 ⎡4 x3 x4 ⎤
2 2
3
= ∫
x =1
x ⋅ ⎜ x + x 2 ⎟ dx =
13 ⎝ 3 ⎠ 13 ∫
x =1
⎜
⎝3
x + x ⎟
⎠
dx = ⋅ + ⎥ =
13 ⎢⎣ 3 3 4 ⎦1

3 ⎡⎛ 4 8 16 ⎞ ⎛ 4 1 1 ⎞⎤ 3 ⎡⎛ 32 16 ⎞ ⎛ 4 1 ⎞⎤
= ⎜ ⋅ + ⎟−⎜ ⋅ + ⎟ = ⎜ + ⎟ − ⎜ + ⎟⎥ =
13 ⎢⎣⎝ 3 3 4 ⎠ ⎝ 3 3 4 ⎠⎥⎦ 13 ⎢⎣⎝ 9 4 ⎠ ⎝ 9 4 ⎠⎦

3 ⎡ 32 16 4 1 ⎤ 3 ⎛ 28 15 ⎞ 3 ⎛ 28 ⋅ 4 + 15 ⋅ 9 ⎞
= + − − ⎥= ⎜ + ⎟= ⎜ ⎟=
13 ⎢⎣ 9 4 9 4 ⎦ 13 ⎝ 9 4 ⎠ 13 ⎝ 36 ⎠

1 ⎛ 112 + 135 ⎞ 1 ⎛ 247 ⎞ 1 ⎛ 19 ⎞ 19

= ⎜ ⎟= ⎜ ⎟= ⎜ ⎟= = 1.5833
13 ⎝ 12 ⎠ 13 ⎝ 12 ⎠ 1 ⎝ 12 ⎠ 12

Since X , the excess amount on the car policy, is expressed in £ ’00, this means that
the expected excess amount on the car policy is £ (1.5833 × 100) = £ 158.33 .

ANSWER: B

( )= ∫
2
18. E X 2
x 2 f ( x ) dx =
x =1

2
3 ⎛4 ⎞ ⎛4 3 4⎞ 3 ⎡4 x 4 x5 ⎤
2 2
3
= ∫ x ⋅ ⎜ x + x 2 ⎟ dx = ∫ ⎜ + ⎟ = ⋅ + ⎥ =
2

13 ⎢⎣ 3 4
x x dx
x =1
13 ⎝ 3 ⎠ 13 x =1 ⎝3 ⎠ 5 ⎦1

3 ⎡⎛ 4 16 32 ⎞ ⎛ 4 1 1 ⎞⎤ 3 ⎡⎛ 64 32 ⎞ ⎛ 4 1 ⎞⎤
= ⎜ ⋅ + ⎟−⎜ ⋅ + ⎟ = ⎜ + ⎟−⎜ + ⎟ =
13 ⎢⎣⎝ 3 4 5 ⎠ ⎝ 3 4 5 ⎠⎥⎦ 13 ⎢⎣⎝ 12 5 ⎠ ⎝ 12 5 ⎠⎥⎦

3 ⎡ 64 32 4 1 ⎤ 3 ⎛ 60 31 ⎞ 3 ⎛ 31 ⎞
= + − − = ⎜ + ⎟ = ⎜ 5 + ⎟=
13 ⎢⎣ 12 5 12 5 ⎥⎦ 13 ⎝ 12 5 ⎠ 13 ⎝ 5⎠

3 ⎛ 5 ⋅ 5 + 31 ⎞ 3 ⎛ 25 + 31 ⎞ 3 ⎛ 56 ⎞ 168
= ⎜ ⎟= ⎜ ⎟= ⎜ ⎟= = 2.5846
13 ⎝ 5 ⎠ 13 ⎝ 5 ⎠ 13 ⎝ 5 ⎠ 65

ANSWER: C

19. ( )
var( X ) = E X 2 − [E ( X )] =
2

= 2.5846 − 1.58332 = 2.5846 − 2.5069 = 0.0777

Therefore

12
 SD ( X ) = var ( X ) = 0.0777 = 0.2787

Since X , the excess amount on the car policy, is expressed in £ ’00, this means that
the standard deviation of the excess amount on the car policy is
£ (0.2787 × 100) = £ 27.87 .

ANSWER: A

20. If X and Y are continuous random variables and f ( x, y ) is their joint probability
distribution, the general definition of marginal distribution of Y is

b
f (y) = ∫ f (x, y )dx for c< y<d
x =a

In our case, f ( x, y ) = xy (x + y ), 1 < x < 2 , 0 < y < 2 . So we have

3
26

(x y + xy 2 )dx
2 2
f (y) = xy ( x + y ) dx = ∫
3 3 2
∫
x =1
26 x =1
26

ANSWER: D

( )
2
f (y) =
3 2
21. ∫
x =1
26
x y + xy 2 dx =

2
3 ⎡ x3 x2 2 ⎤
( )
2
3
26 x∫=1
= x y + xy dx =
2 2
y+ y ⎥ =
26 ⎢⎣ 3 2 ⎦1

3 ⎡⎛ 8 4 2⎞ ⎛1 1 2 ⎞⎤ 3 ⎛ 7 3 2⎞
= ⎢⎜ 3 y + 2 y ⎟ − ⎜ 3 y + 2 y ⎟⎥ = 26 ⎜ 3 y + 2 y ⎟
26 ⎣⎝ ⎠ ⎝ ⎠⎦ ⎝ ⎠

Therefore, the marginal distribution of Y is

3 ⎛7 3 2⎞
f (y) = ⎜ y + y ⎟, 0< y<2
26 ⎝ 3 2 ⎠

ANSWER: C

13
 2
22. E (Y ) = ∫ y f ( y ) dy =
y =0

2
3 2⎞ ⎛7 2 3 3⎞ 3 ⎡7 y3 3 y 4 ⎤
2 2
3 ⎛7 3
= ∫
y =0
y ⋅ ⎜ y + y ⎟ dy =
26 ⎝ 3 2 ⎠ 26 y∫=0 ⎝ 3
⎜ y + y ⎟ dy =
2 ⎠
⋅ + ⋅
26 ⎢⎣ 3 3 2 4 ⎥⎦ 0
=

3 ⎡⎛ 7 8 3 16 ⎞ ⎤ 3 ⎛ 56 48 ⎞ 3 ⎛ 56 ⎞ 3 ⎛ 28 ⎞
= ⎢⎜ ⋅ + ⋅ ⎟ − (0 )⎥ = ⎜ + ⎟ = ⎜ + 6 ⎟ = ⎜ + 3⎟ =
26 ⎣⎝ 3 3 2 4 ⎠ ⎦ 26 ⎝ 9 8 ⎠ 26 ⎝ 9 ⎠ 13 ⎝ 9 ⎠

3 ⎛ 28 + 3 ⋅ 9 ⎞ 1 ⎛ 28 + 27 ⎞ 55
= ⎜ ⎟= ⎜ ⎟= = 1.4103
13 ⎝ 9 ⎠ 13 ⎝ 3 ⎠ 39

Since Y , the excess amount on the home policy, is expressed in £ ’00, this means that
the expected excess amount on the home policy is £ (1.4103 × 100) = £ 141.03 .

ANSWER: C

E (Y 2 ) =
2
23. ∫ y 2 f ( y ) dy =
y =0

2
3 ⎛7 3 ⎞ ⎛7 3 3 4⎞ 3 ⎡7 y 4 3 y5 ⎤
2 2
3
= ∫ y ⋅ ⎜ y + y 2 ⎟ dy =
26 y∫=0 ⎝ 3
⎜ + ⎟ = ⋅ + ⋅ =
2

26 ⎢⎣ 3 4 2 5 ⎥⎦ 0
y y dy
y =0
26 ⎝ 3 2 ⎠ 2 ⎠

3 ⎡⎛ 7 16 3 32 ⎞ ⎤ 3 ⎛7 3 16 ⎞ 3 ⎛ 28 48 ⎞
= ⎢⎜ ⋅ + ⋅ ⎟ − (0 )⎥ = ⎜ ⋅ 4 + ⋅ ⎟ = ⎜ + ⎟ =
26 ⎣⎝ 3 4 2 5 ⎠ ⎦ 26 ⎝ 3 1 5 ⎠ 26 ⎝ 3 5⎠

3 ⎛ 14 24 ⎞ 3 ⎛ 14 ⋅ 5 + 24 ⋅ 3 ⎞ 1 ⎛ 70 + 72 ⎞ 1 ⎛ 142 ⎞
= ⎜ + ⎟= ⎜ ⎟= ⎜ ⎟= ⎜ ⎟=
13 ⎝ 3 5 ⎠ 13 ⎝ 15 ⎠ 13 ⎝ 5 ⎠ 13 ⎝ 5 ⎠

142
= = 2.1846
65

ANSWER: E

var (Y ) = E (Y 2 ) − [E (Y )] =
2
24.

= 2.1846 − 1.41032 = 2.1846 − 1.9889 = 0.1958

Therefore

14
 SD (Y ) = var (Y ) = 0.1958 = 0.4425

Since Y , the excess amount on the home policy, is expressed in £ ’00, this means that
the standard deviation of the excess amount on the home policy is
£ (0.4425 × 100) = £ 44.25 .

ANSWER: B

25. If X and Y are continuous random variables and f ( x, y ) is their joint probability
distribution, the general definition of expected value of the product of X and Y is

b d
E ( XY ) = ∫ ∫ xy f (x, y ) dy dx
x = a y =c

In our case, f ( x, y ) = xy (x + y ), 1 < x < 2 , 0 < y < 2 . So we have

3
26

2 2
E ( XY ) = xy ( x + y ) dy dx =
3
∫ ∫
x =1 y = 0
xy ⋅
26

(x y + x 2 y 3 )dy dx
2 2
3 3 2
= ∫ ∫
x =1 y = 0
26

ANSWER: D

2
⎡ 3 y3 2 y ⎤
(x y + x 2 y 3 )dy dx =
2 2 2 4
E ( XY ) =
3 3 2 3
26. ∫ ∫
x =1 y = 0
26 26 x∫=1 ⎢ 3
⎣
x + x
4 ⎥⎦ 0
dx =

⎡⎛ 3 8 2 16 ⎞ ⎤ ⎛ 2 8 3⎞
2 2

⎢⎜ x 3 + x 4 ⎟ − (0 )⎥ dx = 26 ∫
3 3
26 x∫=1
= ⎜ 4 x + x ⎟ dx =
⎣⎝ ⎠ ⎦ x =1 ⎝ 3 ⎠

2
3 ⎛ 2 4 3⎞
2
3 ⎡ x3 4 x4 ⎤ 3 ⎡⎛ 8 4 16 ⎞ ⎛ 1 4 1 ⎞⎤
= ∫ ⎜ 2 x + x ⎟ dx =
13 x =1 ⎝ 3 ⎠ 13 ⎢2 ⋅ 3 + 3 ⋅ 4 ⎥ = 13 ⎢⎜ 2 ⋅ 3 + 3 ⋅ 4 ⎟ − ⎜ 2 ⋅ 3 + 3 ⋅ 4 ⎟⎥ =
⎣ ⎦1 ⎣⎝ ⎠ ⎝ ⎠⎦

⎡⎛ 16 16 ⎞ ⎛ 2 1 ⎞⎤ 1 ⎡⎛ 16 16 ⎞ ⎛ 2 1 ⎞⎤ 1
⎢⎜ 1 + 1 ⎟ − ⎜ 1 + 1 ⎟⎥ = 13 (16 + 16 − 2 − 1) =
3
= ⎢⎜ 3 + 3 ⎟ − ⎜ 3 + 3 ⎟⎥ = 13
13 ⎣⎝ ⎠ ⎝ ⎠⎦ ⎣⎝ ⎠ ⎝ ⎠⎦

=
1
(29) = 2.2308
13

15
 ANSWER: E

27. cov( X , Y ) = E ( XY ) − E ( X )E (Y ) =

= 2.2308 − 1.5833 (1.4103) = 2.2308 − 2.2329 = −0.0021

ANSWER: E

cov( X , Y )
28. ρ X ,Y = =
SD( X ) SD(Y )

− 0.0021 − 0.0021
= = = −0.0173
0.2787 ⋅ 0.4425 0.1233

ANSWER: D

29. We can apply the formula for the expected value of the sum of two functions of X
and Y :

E [g ( X ) + h(Y ) ] = E [g ( X ) ] + E [h(Y ) ]

In our case the annual premium for a combined car and home policy is

P = 4.15 − 0.40 X − 0.30Y

so g ( X ) = 4.15 − 0.40 X and h(Y ) = −0.30Y .

We know from answers 17 and 22 that E ( X ) = 1.5833 and E (Y ) = 1.4103 . So for

the expected value of the annual premium P we have:

E (P ) = E (4.15 − 0.40 X − 0.30 Y ) =

= E (4.15 − 0.40 X ) + E (− 0.30Y ) =

= 4.15 − 0.40 E ( X ) − 0.30 E (Y ) =

= 4.15 − 0.40 (1.5833) − 0.30 (1.4103) =

= 4.15 − 0.6333 − 0.4231 = 3.0936

Since P , the annual premium, is expressed in £ ’00, this means that the expected
annual premium is £ (3.0936 × 100) = £ 309.36 .

16
 ANSWER: E

30. If (a + bX ) is a linear combination of X , (c + dY ) is a linear combination of Y , and

P is their sum, i.e. P = (a + bX ) + (c + dY ) – where a, b, c and d are constants – the
variance of P , var (P ) , is given by the formula for the variance of linear
combinations of two random variables:

var (P ) = b 2 var ( X ) + d 2 var (Y ) + 2 b d cov( X , Y )

Since P = 4.15 − 0.40 X − 0.30Y , in our case a = 4.15 , b = −0.40 , c = 0 and

d = −0.30 . We know from answers 19, 24 and 27 that var ( X ) = 0.0777 ,
var (Y ) = 0.1958 and cov ( X , Y ) = −0.0021 . So we have:

var (P ) = (− 0.40 ) var ( X ) + (− 0.30 ) var (Y ) + 2 (− 0.40 )(− 0.30 ) cov( X , Y ) =

2 2

= 0.16 (0.0777 ) + 0.09 (0.1958) + 0.24(− 0.0021) =

= 0.0124 + 0.0176 − 0.0005 = 0.0295

Therefore

SD (P ) = var (P ) = 0.0295 = 0.1719

Since P , the annual premium, is expressed in £ ’00, this means that the standard
deviation of the annual premium is £ (0.1719 × 100) = £ 17.19 .

ANSWER: C

31. For the test of the hypothesis that the population mean amount paid annually into a
private pension fund per person by professional women and professional men is the
same, against the alternative hypothesis that the population mean amount paid
annually into a private pension fund per person by professional women is smaller than
(not as much as) that paid by professional men, the null and alternative hypotheses
are:

H 0 : µw = µm H1 : µw < µm

that is

H 0 : µw − µm = 0 H1 : µw − µm < 0

17
 ANSWER: B

32. For the large sample of professional women ( nw = 187 ) taken from a finite population
( N w = 935 ) without replacement, the sampling ratio is:

n w 187
= = 0.20 > 0.10 (10%)
N w 935

⎛ N − nw ⎞
Therefore we need to use the finite population correction of the variance: ⎜⎜ w ⎟.
− ⎟
⎝ wN 1 ⎠

For the large sample of professional men ( nm = 276 ) taken from a large population
without replacement, we do not need to use the finite population correction of the
variance.

The two samples are independent. Sampling theory establishes that the sampling
distribution of the difference between the two sample means (X w − X m ) is:

⎛ σ w2 ⎛ N w − n w ⎞ σ m2 ⎞
⎜
( X w − X m ) ~ N µw − µm , ⎜ ⎟ ⎟
⎜ ⎜ N −1 ⎟ + n ⎟
⎝ nw ⎝ w ⎠ m ⎠

The population standard deviations σ w and σ m are known. Therefore the test statistic
is given by

( X w − X m ) − ( µ w − µ m )0
σ w2 ⎛ N w − n w ⎞ σ m2
⎜ ⎟+
n w ⎜⎝ N w − 1 ⎟⎠ nm

ANSWER: A

33. It is known that, for professional women, the sample size is nw = 187 , the finite
population size is N w = 935 , the sample mean amount paid annually into a private
pension fund per person is X w = 3352 , and its population standard deviation is
σ w = 1100 ; for professional men, the sample size is nm = 276 , the sample mean
amount paid annually into a private pension fund per person is X m = 3727 , and its
population standard deviation is σ m = 1700 . It is also known that the difference
between population means under the null hypothesis is ( µ w − µ m ) 0 = 0 . So the value
of the test statistic is:

18
 ( X w − X m ) − ( µ w − µ m )0 (3352 − 3727) − (0)
= =
σ w2 ⎛ N w − n w ⎞ σ m2 1100 2 ⎛ 935 − 187 ⎞ 1700 2
⎜ ⎟+ ⎜ ⎟+
n w ⎜⎝ N w − 1 ⎟⎠ nm 187 ⎝ 935 − 1 ⎠ 276

− 375
= =
1210000 ⎛ 748 ⎞ 2890000
⎜ ⎟+
187 ⎝ 934 ⎠ 276

− 375
= =
6470.5882(0.8009 ) + 10471.0145

− 375 − 375
= = = −2.9973
15653.0273 125.1121

ANSWER: E

34. Since the sampling distribution of the difference between the two sample means
(X w − X m ) is normal:
⎛ σ2 ⎛ N w − n w ⎞ σ m2 ⎞
( X w − X m ) ~ N ⎜ µw − µm , w ⎜ ⎟ ⎟
⎜ ⎜ N −1 ⎟ + n ⎟
⎝ nw ⎝ w ⎠ m ⎠

the test statistic follows the standard normal distribution:

( X w − X m ) − ( µ w − µ m )0
= Z ~ N (0, 1)
σ w2 ⎛ N w − n w ⎞ σ m2
⎜ ⎟+
n w ⎜⎝ N w − 1 ⎟⎠ nm

ANSWER: A

35. The test statistic follows the standard normal distribution:

Z ~ N (0, 1)

Since the alternative hypothesis takes the form H 1 : µ w − µ m < 0 , the test is one-
tailed (left-hand side tail).

The required significance level is α = 1% .

Therefore, the critical value for the test is:

19
 z 0.99 = −2.3263

1
1%
99%

0 Z
− 2.3263

ANSWER: A

36.

1
1%
99%

0 Z
− 2.3263

Reject Do not
H0 reject H 0

• if Z ≤ −2.3263 , reject H 0
• if Z > −2.3263 , do not reject H 0

ANSWER: C

37. Since the test statistic Z = −2.9973 < −2.3263 , we reject H 0 : µ w = µ m at the 1%
significance level in favour of the alternative H 1 : µ w < µ m . There is significant
evidence (at the 1% level) that the population mean amount paid annually into a
private pension fund per person by professional women ( µ w ) is smaller than (not as
much as) that paid by professional men ( µ m ).

ANSWER: A

38. If X 1 = 48 is the number of morning shoppers who are men, X 2 = 192 the number of
afternoon shoppers who are men, n1 = 400 the size of the sample of morning

20
 shoppers, and n2 = 480 the size of the sample of afternoon shoppers, sampling theory
X
establishes that P1 = 1 , the sample proportion of morning shoppers who are men,
n1
X
and P2 = 2 , the sample proportion of afternoon shoppers who are men, are unbiased
n2
point estimators of the corresponding population proportions π 1 and π 2 , i.e.
E (P1 ) = π 1 and E (P2 ) = π 2 . Therefore:

X1 48 X 2 192
P1 = = = 0.12 and P2 = = = 0.40
n1 400 n 2 480

are unbiased point estimators of π 1 and π 2 .

ANSWER: D

39. For the large sample of morning shoppers ( n1 = 400 ) taken from a finite population
( N 1 = 1740 ) without replacement, the sampling ratio is:

n1 400
= = 0.23 > 0.10 (10%)
N 1 1740

Therefore we need to use the finite population correction of the variance.

For the large sample of afternoon shoppers ( n2 = 480 ) taken from a finite population
( N 2 = 2860 ) without replacement, the sampling ratio is:

n2 480
= = 0.17 > 0.10 (10%)
N 2 2860

Therefore we need to use the finite population correction of the variance.

The two samples are independent. Sampling theory establishes that the sampling
distribution of the difference between the sample proportion of morning shoppers who
are men and the sample proportion of afternoon shoppers who are men, (P1 − P2 ) , is:

⎛ π (1 − π 1 ) ⎛ N 1 − n1 ⎞ π 2 (1 − π 2 ) ⎛ N 2 − n2 ⎞ ⎞⎟
( P1 − P2 ) ~ N ⎜ π 1 − π 2 , 1 ⎜
⎜ N −1 ⎟ +
⎟ ⎜
⎜ N −1 ⎟⎟
⎟
⎜
⎝ n1 ⎝ 1 ⎠ n 2 ⎝ 2 ⎠⎠

ANSWER: C

21
40. It is known that, for morning shoppers, the sample size (large) is n1 = 400 and the
finite population size N 1 = 1740 ; for afternoon shoppers, the sample size (large) is
n2 = 480 and the finite population size N 2 = 2860 .

The population proportions π 1 and π 2 are unknown. So we use the sample

proportions P1 and P2 from answer 38, the unbiased point estimators of π 1 and π 2 , in
estimating the standard error of the difference between sample proportions (P1 − P2 ) ,
SE (P1 − P2 ) .

From the sampling distribution of the difference between sample proportions

(P1 − P2 ) in answer 39, the standard error SE (P1 − P2 ) is:

π 1 (1 − π 1 ) ⎛ N 1 − n1 ⎞ π 2 (1 − π 2 ) ⎛ N 2 − n 2 ⎞
SE (P1 − P2 ) = ⎜
⎜ N −1
⎟+
⎟
⎜ ⎟
⎜ N −1 ⎟ =
n1 ⎝ 1 ⎠ n2 ⎝ 2 ⎠

P1 (1 − P1 ) ⎛ N 1 − n1 ⎞ P2 (1 − P2 ) ⎛ N 2 − n2 ⎞
= ⎜ ⎟ ⎜ ⎟
⎜ N −1 ⎟ + ⎜ N −1 ⎟ =
n1 ⎝ 1 ⎠ n 2 ⎝ 2 ⎠

0.12(1 − 0.12) ⎛ 1740 − 400 ⎞ 0.40(1 − 0.40) ⎛ 2860 − 480 ⎞

= ⎜ ⎟+ ⎜ ⎟=
400 ⎝ 1740 − 1 ⎠ 480 ⎝ 2860 − 1 ⎠

0.12(0.88) ⎛ 1340 ⎞ 0.40(0.60) ⎛ 2380 ⎞

= ⎜ ⎟+ ⎜ ⎟=
400 ⎝ 1739 ⎠ 480 ⎝ 2859 ⎠

=
0.1056
(0.7706 ) + 0.24 (0.8325) = 0.0002 + 0.0004 =
400 480

= 0.0006 = 0.0249

ANSWER: B

41. The sampling distribution of the difference between the two sample proportions
(P1 − P2 ) is:
⎛ π (1 − π 1 ) ⎛ N 1 − n1 ⎞ π 2 (1 − π 2 ) ⎛ N 2 − n2 ⎞ ⎞⎟
( P1 − P2 ) ~ N ⎜ π 1 − π 2 , 1 ⎜
⎜ N −1 ⎟ +
⎟ ⎜
⎜ N −1 ⎟⎟
⎟
⎜
⎝ n1 ⎝ 1 ⎠ n 2 ⎝ 2 ⎠⎠

Therefore the 95% confidence interval for the difference between the population
proportion of morning shoppers who are men and the population proportion of
afternoon shoppers who are men, (π 1 − π 2 ) , is given by

22
 (π 1 − π 2 ) = (P1 − P2 ) ± z 0.025 ⋅ SE (P1 − P2 )

ANSWER: E

42. It is known from answer 38 that P1 = 0.12 and P2 = 0.40 . We also know from answer
40 that the standard error of the difference between sample proportions (P1 − P2 ) is:

SE (P1 − P2 ) = 0.0249

So the 95% confidence interval for the difference between population proportions
(π 1 − π 2 ) is:
(π 1 − π 2 ) = (P1 − P2 ) ± z 0.025 ⋅ SE (P1 − P2 )

= (0.12 − 0.40 ) ± 1.9600 ⋅ 0.0249 = (− 0.3288, − 0.2312 )

ANSWER: E

43. The sampling distribution of the difference between the two sample proportions
(P1 − P2 ) is:
⎛ π (1 − π 1 ) ⎛ N 1 − n1 ⎞ π 2 (1 − π 2 ) ⎛ N 2 − n2 ⎞ ⎞⎟
( P1 − P2 ) ~ N ⎜ π 1 − π 2 , 1 ⎜
⎜ N −1 ⎟ +
⎟ ⎜
⎜ N −1 ⎟⎟
⎟
⎜
⎝ n1 ⎝ 1 ⎠ n2 ⎝ 2 ⎠⎠

Therefore the 99% confidence interval for the difference between the population
proportion of morning shoppers who are men and the population proportion of
afternoon shoppers who are men, (π 1 − π 2 ) , is given by

(π 1 − π 2 ) = (P1 − P2 ) ± z 0.005 ⋅ SE (P1 − P2 )

ANSWER: D

44. It is known from answer 38 that P1 = 0.12 and P2 = 0.40 . We also know from answer
40 that the standard error of the difference between sample proportions (P1 − P2 ) is:

SE (P1 − P2 ) = 0.0249

So the 99% confidence interval for the difference between population proportions
(π 1 − π 2 ) is:

23
 (π 1 − π 2 ) = (P1 − P2 ) ± z 0.005 ⋅ SE (P1 − P2 )

= (0.12 − 0.40 ) ± 2.5758 ⋅ 0.0249 = (− 0.3441, − 0.2159 )

ANSWER: A

45. The difference between the population proportion of morning shoppers who are men
and the population proportion of afternoon shoppers who are men, (π 1 − π 2 ) , is some
value between − 0.3288 and − 0.2312 with a 95% probability and some value
between − 0.3441 and − 0.2159 with a 99% probability. Since these intervals do not
include the value 0 , both confidence intervals obtained do not include the possibility
that the difference between population proportions is equal to 0 : (π 1 − π 2 ) ≠ 0 . In
other words, both confidence intervals obtained do not include the possibility that the
proportion of men is the same in the two populations of morning and afternoon
shoppers: π 1 ≠ π 2 . There is therefore clear evidence (with both 95% and 99%
confidence) in favour of the claim that the proportion of men is different in the two
populations.

ANSWER: B

46. For the test of the hypothesis that the population variance in prices of flats is the same
in the two cities, against the alternative hypothesis that it is different, the null and
alternative hypotheses are:

H 0 : σ E2 = σ C2 H 1 : σ E2 ≠ σ C2

ANSWER: E

47. It is assumed that the random variables X E (price of flats in Exeter) and X C (price of
flats in Cardiff) are normally distributed. The two samples are small ( n E = 21 and
nC = 16 ). The object of our inference is the equality of the two population variances
σ E2 and σ C2 .

Following the convention that the larger sample variance is placed in the numerator
of the test statistic ratio, the test statistic will be:

s E2 / σ E2
• either if s E2 > sC2
sC2 / σ C2

24
 sC2 / σ C2
• or if sC2 > s E2
s E2 / σ E2

In our case we have:

s E2 = 2300 2

sC2 = 1750 2

Therefore, since s E2 > sC2 , the test statistic is given by

s E2 / σ E2
sC2 / σ C2

ANSWER: D

48. The sample variance of prices of flats in Exeter is s E2 = 2300 2 and the sample
variance of prices of flats in Cardiff is sC2 = 1750 2 . Under the null hypothesis H 0 , we
have: σ E2 = σ C2 . So the value of the test statistic is:

s E2 / σ E2 s E2 2300 2 5290000
= = = = 1.727
sC2 / σ C2 sC2 1750 2 3062500

ANSWER: E

49. It is assumed that the random variables X E (price of flats in Exeter) and X C (price of
flats in Cardiff) are normally distributed and that the two samples are small ( n E = 21
and nC = 16 ) and independent.

Therefore the test statistic follows the F distribution:

s E2 / σ E2 s E2
= = F ~ F distribution with ν 1 = n E − 1 = 21 − 1 = 20 ,
sC2 / σ C2 sC2
ν 2 = nC − 1 = 16 − 1 = 15
degrees of freedom

ANSWER: B

50. The test statistic follows the F distribution:

25
 F ~ F distribution with ν 1 = 20 , ν 2 = 15 degrees of freedom

Since the alternative hypothesis takes the form H 1 : σ M2 ≠ σ F2 , the test is two-tailed.

The required significance level for the test is α = 1% .

Therefore, the critical value for the test is:

F0.005, 20, 15 = 3.883

0.5%

0 F20, 15
3.883

ANSWER: C

51.

0.5%

0 F20, 15
3.883

Do not Reject
reject H 0 H0

• if F ≥ 3.883 , reject H 0
• if F < 3.883 , do not reject H 0

26
 ANSWER: A

52. Since the test statistic F = 1.727 < 3.883 , we do not reject H 0 : σ E2 = σ C2 at the 1%
significance level in favour of the alternative H 1 : σ E2 ≠ σ C2 . There is no significant
evidence (at the 1% level) that the variance in prices of flats is different in the two
cities.

ANSWER: E

53. It is known that, for Exeter, the sample variance in prices of flats is s E2 = 2300 2 and
the sample size n E = 21 ; for Cardiff, the sample variance in prices of flats is
sC2 = 1750 2 and the sample size nC = 16 . So the value of the pooled estimate of the
common population variance is:

(nE − 1) s E2 + (nC − 1) sC2 (21 − 1) 2300 2 + (16 − 1) 1750 2

s 2p = = =
n E + nC − 2 21 + 16 − 2

=
(20) 5290000 + (15) 3062500 = 4335357.1429
35

ANSWER: C

54. For the test of the hypothesis that the population mean price of flats is the same in the
two cities, against the alternative hypothesis that it is different, the null and alternative
hypotheses are:

H 0 : µ E = µC H 1 : µ E ≠ µC

that is

H 0 : µ E − µC = 0 H 1 : µ E − µC ≠ 0

ANSWER: B

55. It is assumed that the random variables X E (price of flats in Exeter) and X C (price of
flats in Cardiff) are normally distributed. We know from the previous test that the
prices of flats in the two cities have a common population variance, i.e.,
σ E2 = σ C2 = σ 2 . For two small samples ( n E = 21 and nC = 16 ), sampling theory
establishes that the sampling distribution of the difference between the two sample
means ( X E − X C ) is:

27
 ⎛ σ2 σ2⎞
(X E
− X C ) ~ N ⎜⎜ µ E − µ C , + ⎟
⎟
⎝ n E
n C ⎠

The common population variance σ E2 = σ C2 = σ 2 is unknown and we replace it with its

sample estimate, the pooled estimate of the common population variance s 2p .

Therefore the test statistic is given by

(X − X C ) − (µ E − µ C )0 (X − X C ) − (µ E − µ C )0
E
= E

s 2p s 2p 1
+
1
+ sp
n E nC
nE nC

ANSWER: A

56. It is known that, for Exeter, the sample mean price of flats is X E = 116900 and the
sample size n E = 21 ; for Cardiff, the sample mean price of flats is X C = 114000 and
the sample size nC = 16 . From answer 53 we know that the pooled estimate of the
common population variance is s 2p = 4335357.1429 . So the value of the test statistic
is:

(X − X C ) − (µ E − µ C )0 (116900 − 114000) − (0 ) 2900

E
= = = 4.1971
1 1 1 1 37
sp + 4335357.1429 + 2082.1520
n E nC 21 16 336

ANSWER: C

57. It is assumed that the random variables X E (price of flats in Exeter) and X C (price of
flats in Cardiff) are normally distributed. The common population variance
σ E2 = σ C2 = σ 2 is unknown and we replace it with the pooled estimate s 2p . The two
samples are small ( n E = 21 and nC = 16 ).

Therefore the test statistic follows the t distribution:

(X − X C ) − (µ E − µ C )0
E
= t ~ t distribution with ν = n E + nC − 2 = 21 + 16 − 2 = 35
1 1
sp + degrees of freedom
n E nC

28
 ANSWER: B

58. The test statistic follows the t distribution:

t ~ t distribution with ν = 35 degrees of freedom

Since the alternative hypothesis takes the form H 1 : µ E − µ C ≠ 0 , the test is two-
tailed.

The required significance level for the test is α = 1% .

Therefore, the critical value for the test is:

t 0.005, 35 = 2.7238

0.5% 99% 0.5%

0 t35
− 2.7238 2.7238

ANSWER: E

59.

0.5% 99% 0.5%

0 t35
− 2.7238 2.7238

Reject Do not Reject

H0 reject H 0 H0

• if t ≥ 2.7238 , reject H 0
• if t < 2.7238 , do not reject H 0

ANSWER: B

29
60. Since the test statistic t = 4.1971 > 2.7238 , we reject H 0 : µ E = µ C at the 1%
significance level in favour of the alternative H 1 : µ E ≠ µ C . There is significant
evidence (at the 1% level) that the population mean prices of flats are different in the
two cities.

ANSWER: A

30