Chapter 8 The Normal Distribution

8 THE NORMAL
DISTRIBUTION
Objectives
After studying this chapter you should

appreciate the wide variety of circumstances in which the normal

distribution can be used;

be able to use tables of the normal distribution to solve problems;

be able to use the normal distribution as an approximation to other

distributions in appropriate circumstances.

Statistical tables are available in many books and can also be found
online.
You should note that the accuracy of your solutions will depend on the
tables (or graphic calculator) you are using.

8.0

Introduction

The tallest accurately recorded human being was Robert Wadlow in the
USA. On his death at the age of 22 he was 272 cm (8 feet 11.1 inches)
tall. If you were an architect and you had to design doorways in a
building you would clearly not make them all 9 feet high - most ceilings
are lower than this!
What height should the ceilings be?
In 1980 the Government commissioned a survey, carried out on 10 000
adults in Great Britain. They found that the average height was 167.3 cm
with SD (standard deviation) 9.1 cm. You cannot make a door size that
everyone can fit through but what height of door would 95% of people get
through without stooping? This chapter should help you find the answer.

Activity 1

Data collection

There are many sets of data you could collect from people in your group,
such as heights, weights, length of time breath can be held, etc. However,
you will need about 100 results to do this activity properly so here are a
few suggestions where large quantities of data can be collected quickly.
1.

Lengths of leaves
Evergreen bushes such as laurel are useful - though make sure all the
leaves are from the same year's growth.

151

Chapter 8 The Normal Distribution

2.

Weights of crisp packets

Borrow a box of crisps from a canteen and weigh each packet
accurately on a balance such as any Science laboratory would
have.

3.

Pieces of string
Look at 10 cm on a ruler and then take a ball of string and try
to cut 100 lengths of 10 cm by guessing. Measure the lengths
of all the pieces in mm.

4.

Weights of apples
If anyone has apple trees in their garden they are bound to
have large quantities in the autumn.

5.

Size of pebbles on a beach

Geographers often look at these to study the movement of
beaches. Use a pair of calipers then measure on a ruler.

6.

Game of bowls
Make a line with a piece of rope on the grass about 20 metres
away. Let everyone have several goes at trying to land a tennis
ball on the line. Measure how far each ball is from the line.

Try at least two of these activities. You will need about 100 results
in all. To look at the data it would help to have a data handling
package on a computer.

8.1

Looking at your data

The data shown on the opposite page gives the length from top to
tail (in millimetres) of a large group of frogs. This has been run
through a computer package so you can see some useful facts about
the data.
In the computer analysis you will see that most of the frogs are
close to the mean value, with fewer at the extremes. This 'bellshaped' pattern of distribution is typical of data which follows a
normal distribution. To obtain a perfectly shaped and symmetrical
distribution you would need to measure thousands of frogs.
Does your data follow a 'bell shaped' pattern?
You may notice that median mean mode, as might be
expected for a symmetrical distribution. From the analysis of data
you also see that the mean is 90.9 mm and the standard deviation is
11.7 mm. Now look at how much of the data is close to the mean,
i.e. within one standard deviation of it. From the stem and leaf
table you can see that 74 frogs have a length within one standard
deviation above the mean and 59 within a SD below the mean.
Altogether, 133 frogs are + or one SD from the mean, which
is 66.5%.
152

Chapter 8 The Normal Distribution

Frog Data

The data below show the length from top to tail in millimetres of a large group of frogs.
83
91
73
87
103
65
95
80
84
90
91
113
85
91
76
83
80
72
82

69
100
103
96
79
97
96
98
84
85
91
86
87
117
94
104
98
78
97

50

Frogs

97
77
96
87
104
86
92
96
97
82
107
93
97
91
104
91
101
108

53
110
72
72
105
97
95
94
95
81
102
77
92
87
80
91
95
91

*
* *

89
68
71
79
91
111
100
101
108
97
89
100
107
118
77
96
96
91

*
* * * *
* *
* * * * * * *
* * * * * * * * *

60

95
118
99
102
82
98
90
85
94
79
85
96
102
96
94
88
102
91

*
*
*
*

*
*
*
*
*
*

*
*
*
*
*

*
*
*
*
*
*
*
* * *
* * *

105
87
121
98
102
92
91
113
79
90
98
90
104
89
84
115
78
110

*
*
*
*
*
*
*
*

*
*
*
*
*
*

*
*
*
*
*

75

*
*
*
*
*

*
*
*
*

*
*
*
*
*
*
*

80
97
81
97
75
74
95
96
81
90
84
97
94
88
78
96
97
86

*
*
*
*
*
*
*
*
*
*
*
*

*
*
*
*
*
*
*
*
* * *
* * *

*
*
*
*
*
*
*
*
*

*
*
*
*
*
*
*
*
*
*
*

76
78
104
88
95
88
75
103
92
94
91
109
93
111
73
74
80
101

*
*
*
*
*
*
*
*
*
*

*
*
*
*
*
*
* *

*
*
*
*
* * * *
* * * *
* * * *

90

117
100
68
87
90
84
70
98
85
98
90
102
75
120
92
88
87
81

*
*
*
* *
* *

74
95
89
86
62
80
84
95
87
73
86
84
96
92
81
86
82
97

* *
* *
* * * * *

*
*

* *
* *

105

* *

120

MEAN

MEDIAN

TRMEAN

STDEV

MIN

MAX

Q1

Q3

200

90.905

91.000

90.822

11.701

53.000

121.000

83.250

97.000

Stem and leaf of frogs

Leaf Unit = 1.0
1
1
2
6
17
33
57
85
(3 4)
81
41
20
13
7
2

5
5
6
6
7
7
8
8
9
9
10
10
11
11
12

N = 200
3
2
5
0
5
0
5
0
5
0
5
0
5
0

8
1
5
0
5
0
5
0
5
0
7
1

8
2
5
0
5
0
5
0
7
1
7

9
2
6
0
5
0
5
0
7
1
8

2
6
0
5
0
5
1
8
3
8

3
7
0
6
0
5
1
8
3

3
7
1
6
0
5
1
9

3
7
1
6
1
5
2

4
8
1
6
1
5
2

4
8
1
6
1
6
2

4
8
1
6
1
6
2

8
2
7
1
6
2

9
2
7
1
6
2

9
2
7
1
6
3

9
2
7
1
6
3

9
3
7
1
6
3

3
7
1
6
4

4
7
1
6
4

4
7
1
6
4

4
8
1
6
4

4 4 4 4
8 8 8 8 9 9 9 9
2 2 2 2 2 2 3 3 4 4 4 4 4 4
7 7 7 7 7 7 7 7 7 7 7 7 8 8 8 8 8 8 8 9
4

153

Chapter 8 The Normal Distribution

Activity 2
Apply the same techniques to your own sets of data (i.e. draw up
frequency tables or histograms and calculate means and SDs) and
calculate the percentage which lie within one SD of the mean. If
the data is normally distributed then this should be about 68%.
Similarly, you could look for the amount of data within 2 SDs ,
3 SDs, etc. The table below gives approximately the percentages
to expect.

34%
14%
2%
3

SDs

Distances from mean in

terms of standard deviation
in one direction
Proportion of area in
the above range

0 1 12

34 %

2 3

over 3

2%

negligible

14 %

Note that very few items of data fall beyond three SDs from the
mean.

What is clearly useful is that no matter what size the numbers are,
if data are normally distributed, the proportions within so many
SDs from the mean are always the same.

Example
IQ test scores, and the results of many other standard tests, are
designed to be normally distributed with mean 100 and standard
deviation 15.
Therefore statements such as the following can be made:
'68% of all people should achieve an IQ score between
85 and 115.'
'Only 2% of people should have an IQ score less than 70.'
'Only 1 in a 1000 people have an IQ greater than 145.'

Exercise 8A
The survey mentioned in the introduction also showed
that the average height of 16-19 year olds was
approximately 169 cm with SD 9 cm.
1. Assuming the data follows a normal distribution,
find:
(a) the percentage of sixth formers taller than
187 cm;

154

(b) the percentage of sixth formers smaller than

160 cm;
(c) in a sixth form of 300, the number of
students smaller than 151 cm.
(Note these are not truly normal, as the pattern for girls
and boys is different.)

Chapter 8 The Normal Distribution

8.2

The p.d.f. of the normal

If you could work in only whole numbers of SDs, the number of

problems that could be solved would be limited. To calculate
the proportions or probabilities of lying within so many SDs of
the mean, you need to know the p.d.f. This was first discovered
by the famous German mathematician, Gauss (1777-1855) and
this is why the normal distribution is sometimes called the
Gaussian distribution.
It is given by the formula
f (z) =

1
2

12 z

f (z)
0.4
0.3
0.2

.
0.1

z is called the standard normal variate and represents a normal

distribution with mean 0 and SD 1. The graph of the function is
shown opposite.
Note that the function f ( z ) has no value for which it is zero, i.e.
it is possible, though very unlikely, to have very large or very
small values occurring.

-3

-2

-1

.00

.02

.01

.03

.04

.05

0 . 0 . 5 0 0 0 0 . 5 0 3 9 9 . 5 0 7 9 8 . 5 11 9 7 . 5 1 5 9 5 . 5 1 9 9 x
0.1 .53983 .54380 .54776 .55172 .55567 .5596x
0.2 .57926 .58317 .58706 .59095 .59483 .5987x

In order to find the probabilities of all possible SDs from the

mean you would have to integrate the function between the
values. This is a tedious task involving integration by parts and
to avoid this, tables of the function are commonly used.

0.3 .61791 .62172 .62552 .62930 .63307 .6368x

0.4 .65542 .65910 .66276 .66640 .67003 .6736x
0.5 .69146 .69497 .69847 .70194 .70540 .7088x
0.6 .72575 .72907 .73237 .73565 .73891 .742xx
0 . 7 . 7 5 8 0 4 . 7 6 11 5 . 7 6 4 2 4 . 7 6 7 3 0 . 7 7 0 3 5 . 7 7 x x x

(z) = P ( Z < z)

0.8 .78814 .79103 .79389 .79673 .79955

0.9 .81594 .81859 .82121 .82381 .826hx
1.0 .84134 .84375 .84614 .848hh

f (z) d z .

(1.0 )

1.1 .86433 .86650 .86864

1.2 .88493 .88686 .88hhh
1.3 .90320 .90490 .9hhhh

Here has been used to denote the cumulative probability.

1.4 .91924 .92073

1.5 .93319 .93448

For positive z, the function gives you the probability of being

less than z SDs above the mean.

1.6 .94520 .94630

For example, (1.0 ) = 0.84313 , therefore 84.13% of the

distribution is less than one SD above the mean.

1.9 .97128 .9hhhh

Tables usually give the area to the left of z and only for values
above zero. This is because symmetry enables you to calculate
all other values.

(1.5)

1.7 .95543 .95637

1.8 .96407 .9648h

84.13%

Example
What is the probability of being less than 1.5 SDs below the
mean i.e. ( 1.5) ?

155

Chapter 8 The Normal Distribution

Solution
From tables,
( +1.5) =

0.93319

and by symmetry,
( 1.5) = 1 0.93319 = 0.06681

1.5

1.5

i.e. about 6.7%.

A random variable, Z, which has this p.d.f. is denoted by
Z ~ N ( 0,1)

showing that it is a normal distribution with mean 0 and standard

deviation 1.
This is often referred to as the standardised normal distribution.

Example
If Z ~ N ( 0,1) , find
(a) P( Z > 1.2 )
(b) P( 2.0 < Z < 2.0 )
(c) P( 1.2 < Z < 1.0 )
Solution
(a) P( Z > 1.2 ) = 1 (1.2 )

= 1 0.88493

(from tables)

= 0.11507

(b) P( 2.0 < Z < 2.0 )

1.2

= P( Z < 2.0 ) P( Z < 2.0 )

= (2.0 ) P( Z > 2.0 )

= (2.0 ) (1 P( Z < 2.0 ))

= 2( 2.0 ) 1
= 2 0.97725 1

= 0.9545

(c) P( 1.2 < Z < 1.0 )

= P( Z < 1.0 ) P( Z < 1.2 )
= P( Z < 1.0 ) P( Z > 1.2 )

= (1.0 ) (1 (1.2 ))
156

1.2

Chapter 8 The Normal Distribution

= 0.84134 (1 0.88493)
= 0. 72627

You can also use the tables to find the value of a when P( Z > a ) is
a given value and Z ~ N ( 0,1) . This is illustrated in the next
example.

Example

0.90

If Z ~ N ( 0,1) , find a such that

(a) P( Z < a ) = 0.90
(b) P( Z > a ) = 0.25

Solution
(a) Here ( a ) = 0.90 , and from the tables
a 1.28

0.25

(b) Here ( a ) = 1 0.25 = 0. 75 and from the tables

a 0.67

Exercise 8B
If Z ~ N ( 0,1) , find
1. P( Z > 0.82 )
2. P( Z < 0.82 )
3. P( Z > 0.82 )
4. P( Z < 0.82 )
5. P( 0.82 < Z < 0.82 )

8.3

6. P( 1 < Z < 1)
7. P( 1 < Z < 1.5)
8. P( 0 < Z < 2.5)
9. P( Z < 1.96 )
10. P( 1.96 < Z < 1.96 )

Transformation of normal
p.d.f.s

The method needed to transform any normal variable to the

standardised variable is illustrated in the example below.

Example
Eggs laid by a particular chicken are known to have lengths
normally distributed, with mean 6 cm and standard deviation
1.4 cm. What is the probability of:
(a) finding an egg bigger than 8 cm in length;
(b) finding an egg smaller than 5 cm in length?
157

Chapter 8 The Normal Distribution

Solution
(a)

The number of SDs that 8 is above the mean is given by

z =

86
1. 4

= 1. 429 ,

but

(1. 43)

so

P ( X > 8) = 1 0.92364 = 0.07646 .

z =

(b)

= 0.92364 (from tables)

56

= 0. 7143 ,

1. 4

but

( 0. 7143) = 0. 7625 (from tables using

interpolation),

so

P ( X < 5) = 1 0. 7625 = 0.2375 .

and variance , this is denoted by

2

X ~ N ,

)
(

So in the last example, X ~ N 6, 1. 4 .

To use normal tables, the transformation
Z =

is used. This ensures that Z has mean 0 and standard deviation

1, and the tables are then valid.
Using the UK data on heights in Section 8.0, the z value for
Robert Wadlow's height is
z =

272 167.3
9.1

11.5 .

So his height is 11.5 SDs above the mean. The most accurate
tables show that 6 SDs is only exceeded with a probability of
10

10 , so it is extremely unlikely that a taller person will ever

appear!

158

Note that using interpolation from

tables is not necessary for the
AEB examination, but it is good
practice to use it to improve
accuracy.

Note that in order to find the probability you need to establish

whether you need the area greater than a half or less than a half.
Drawing a diagram will help.
When a variable X follows a normal distribution, with mean

Chapter 8 The Normal Distribution

Example
If X ~ N ( 4, 9) , find
(a) P( X > 6 )
(b) P( X > 1)
Solution
Now Z =

X X 4
=
,

(a) Hence
P( X > 6 ) = 1 P( X < 6 )

6 4
= 1
3
= 1 ( 0.67)
= 1 0. 74857

= 0.25143

(b) P( X > 1) = P( X < 7) (by symmetry)

7 4
=
3
= (1)
= 0.84134

Exercise 8C
1. If X ~ N ( 200,625) , find
(a) P( X > 250 )

(b) P(175 < X < 225)

(c) P( X < 275)

2. If X ~ N ( 6, 4 ) , find
(a) P( X > 8)

(b) P( 4 < X < 8)

(c) P( 5 < X < 9 )

3. If X ~ ( 10,36 ) , find
(a) P( X < 0 )

(b) P( 12 < X < 8)

(c) P( 15 < X < 0 )

3 1

For the banks to burst a flow of 8700 m s

required. What is the probability of this
happening in a particular year?

is

6. IQs are designed to be N (100, 225) . To join

Mensa an IQ of 138 is required. What
percentage of the population are eligible to join?
A psychologist claims that any child with an IQ
of 150+ is 'gifted'. How many 'gifted' children
would you expect to find in a school of 1800
pupils?
7. Rainfall in a particular area has been found to be

N 850, 100 mm over the years. What is the

probability of rainfall exceeding 1000 mm?

4. Components in a personal stereo are normally

8. In a verbal reasoning test on different ethnic
distributed with a mean life of 2400 hours with SD
groups, one group was found to have scores
300 hours. It is estimated that the average user
2
distributed N 98.42, 15.31 . Those with a score
listens for about 1000 hours in one year. What is
the probability that a component lasts for more than
less than 80 were deemed to be in need of help.
three years.
What percentage of the overall group were in
need of help?
5. The maximum flow of a river in Africa during the
'rainy season' was recorded over a number of years
and found to be distributed

N 6300, 1900

)m s

3 1

159

Chapter 8 The Normal Distribution

8.4

More complicated examples

The following examples illustrate some of the many uses and

applications of the normal distribution.

Example
A machine produces bolts which are N ( 4, 0.09) , where
measurements are in mm. Bolts are measured accurately and any
which are smaller than 3.5 mm or bigger than 4.4 mm are rejected.
Out of a batch of 500 bolts how many would be acceptable?
Solution

( 4. 4 4 )
P ( X < 4. 4 ) =
(1.33) = 0.90824
0.3
(3.5 4 )
P ( X < 3.5) =
( 1.67) = 0.04746 .
0.3
P (3.5 < X < 4. 4 ) 0.90824 0.04746

Hence

= 0.86078 .
The number of acceptable items is therefore

0.86078 500 = 430.39 430 (to nearest whole number).)

Example
IQ tests are measured on a scale which is N (100, 225) . A woman
wants to form an 'Eggheads Society' which only admits people with
the top 1% of IQ scores. What would she have to set as the cut-off
point in the test to allow this to happen?
Solution
From tables you need to find z such that ( z ) = 0.99 .
This is most easily carried out using a 'percentage points of the
normal distribution' table, which gives the values directly.
Now

1 ( 0.99) = 2.3263

which is an alternative way of saying that

(2.3263) = 0.99 .
(Check this using the usual tables.)

160

3.5

4.4

Chapter 8 The Normal Distribution

This means that

x 100
= 2.3263 .
225

x = 100 + 2.3263 15

Hence

= 134.8945 134.9 .

Example
A manufacturer does not know the mean and SD of the diameters
of ball bearings he is producing. However, a sieving system
rejects all bearings larger than 2.4 cm and those under 1.8 cm in
diameter. Out of 1000 ball bearings 8% are rejected as too small
and 5.5% as too big. What is the mean and standard deviation of
the ball bearings produced?
Solution
Assume a normal distribution of

(1 0.08)

= 1. 4 ;

0.08

so 1.8 is 1.4 standard deviations below mean.

1.8

0.055
2.4

1 (1 0.055) = 1.6 ,

Also

so 2.4 is 1.6 standard deviations above the mean.

This can be written as two simultaneous equations and solved:

+ 1.6 = 2. 4
1. 4 = 1.8 .
Subtracting,

3.0 = 0.6

= 0.2

Using the first equation,

+ (1.6 0.2 ) = 2. 4

= 2. 4 (1.6 0.2 )

= 2.08 .

So diameters are distributed N 2.08, 0.2 .

161

Chapter 8 The Normal Distribution

Exercise 8D
1. Bags of sugar are sold as 1 kg. To ensure bags
are not sold underweight the machine is set to
put a mean weight of 1004 g in each bag. The
manufacturer claims that the process works to a
standard deviation of 2.4. What proportion of
bags are underweight?
2. Parts for a machine are acceptable within the
'tolerance' limits of 20.5 to 20.6 mm. From
previous tests it is known that the machine

produces parts to N 20.56, ( 0.02 )

).

4. Sacks of potatoes are packed by an automatic

loader with mean weight 114lb. In a test it was
found that 10% of bags were over 116 lb. Use
this to find the SD of the process. If the machine
is now adjusted to a mean weight of 113 lb, what
% are now over 116 lb if the SD remains
unaltered?
5. In a soap making process it was found that 6 23 %
of bars produced weighed less than 90.50 g and
4% weighed more than 100.25 g.
(a) Find the mean and the SD of the process.

Out of a batch of 1000 parts how many would be

expected to be rejected?
3. Buoyancy aids in watersports are tested by
adding increasing weights until they sink. A
club has two sets of buoyancy aids. One set is
two years old, and should support weights
according to N ( 6.0, 0.64 ) kg; the other set is
five years old and should support weights of

N ( 4.5, 1.0 ) kg. All the aids are tested and any
which are unable to support at least 5 kg are
thrown out.

(b) What % of the bars would you expect to

weigh less than 88 g?
6. A light bulb manufacturer finds that 5% of his
bulbs last more than 500 hours. An improvement in the process meant that the mean lifetime
was increased by 50 hours . In a new test, 20%
of bulbs now lasted longer than 500 hours.
Find the mean and standard deviation of the
original process.

(a) If there are 24 two-year-old aids, how many

are still usable?
(b) If there are 32 five-year-old aids how many
are still usable?

8.5

Using the normal as an

approximation to other
distributions

In earlier chapters you looked at discrete distributions such as

the binomial. Let us suppose that the probability of someone
buying the Daily Sin newspaper in a particular town is 0.4.
Consider these problems:
(a) What is the probability that in a row of six houses all six
buy the Sin?
(b) Of 25 customers who come into a shop what is the
probability of 10 or more buying the Sin?
(c) Two hundred people live on an estate. What is the
probability that 100 or more buy the Sin?
In part (a) you would probably use the binomial distribution and
a calculator to find ( 0. 4 ) and in (b) you would probably use
tables to save on calculation. However, in part (c) there is a
6

162

Chapter 8 The Normal Distribution

problem. Tables do not go beyond 50; you could use a Poisson

approximation, but p is not really small and this would still
involve enormous calculations. Imagine a probability histogram
with 200 columns it would look almost continuous! You will
already know that for p approximately half, you get a
symmetrical bell shaped graph. In fact you can use the normal
distribution as an approximation in such cases.
You know that for a binomial distribution

= n p = 200 0. 4 = 80
o = n p (1 p ) = 200 0. 4 0.6 = 48
2
= 6.93 .
2

and

A slight adjustment needs to be made since the 100 column

actually goes from 99.5 to 100.5. To include 100 you need to
find P ( x > 99.5) . This is sometimes called a continuity
correction factor.
So

P(100 or more buy Sin)

= 1

99 100

99.5 80
6.93

= 1 ( 2.81)

= 1 0.99752

(from tables)

= 0.00248.

In the same way you can use the normal distribution to

approximate for the Poisson.

Example
Customers arrive at a garage at an average rate of 2 per five
minute period. What is the probability that less than 15 arrive in
a one hour period?
Solution

= = 24 per hour , so = 4.9 .

2

Hence

14.5 24
P ( less than 15 in an hour ) =

4.9

( 1.94 )

= 1 0.97381

= 0.02619
(Note that 14.5 was used since less than 15 is required.)

163

Chapter 8 The Normal Distribution

Knowing when to use the normal distribution is important.

Remember that it is only an approximation and if a simple
calculation or tables will give the answer, this should be used.
You may have access to a computer package which can draw
histograms of binomial and Poisson distributions for different
n, p and , and overlay a normal distribution. The following
diagrams show this for different cases.
Binomial distributions
0.2

0.14

n = 30
p = 0.8

n = 50
p = 0.2

0.12

0.15

0.1
0.08

0.1

0.06
0.04

0.05

0.02

0
0

0.12
0.1

24

10

0.35

n = 50
p = 0.5

n = 10
p = 0.2

0.3
0.25

0.08

0.2

0.06

0.15

0.04

0.1

0.02

0.05

25

Poisson distributions
0.2

0.12

=4

= 15

0.1

0.15

0.08

0.1

0.06
0.04

0.05

0.02

4
0.14

=9

0.12
0.1
0.08
0.06
0.04
0.02
0

164

16

Chapter 8 The Normal Distribution

Activity 3
Check that the diagrams illustrate that
(a) for a binomial distribution, if p is close to 0.5, the normal is
a good approximation even for quite small n. However, if p
is small or large, then a larger value of n will be required
for the approximation to be good;
(If n > 30 , np > 5 , nq > 5 , then this is generally regarded
as a satisfactory set of circumstances to use a normal
approximation.)
(b) for a Poisson distribution, the larger n is the better the
approximation.
( > 20 is usually regarded as a necessary condition to use
a normal approximation.)

To summarise, including the use of the Poisson to approximate

to the binomial,
Distribution

X ~ B( n, p )

X ~ B( n, p )

Conditions for using

Approximating distribution
approximation

n large (say >50)

and p small (say <0.1)
p close to

X ~ Po( n p )

1
and n >10
2

or
p moving away from

X ~ Po( )

1
and
2

X ~ N ( n p, n p q )

n > 30

(q = 1 p)

> 20 (say)

X ~ N ( , )

Example
If X ~ B(20, 0. 4 ) , find P(6 X 10 ) .
Also find approximations to this probability by using the
(a) normal distribution
(b) Poisson distribution.

165

Chapter 8 The Normal Distribution

Solution
P( X = 6 ) = 20C6 ( 0.6 )

14

Similarly

( 0. 4 )6 = 0.1244

P( X = 7) = 0.1659
P( X = 8) = 0.1797
P( X = 9) = 0.1597
P( X = 10 ) = 0.1171
P(6 X 10 ) = 0. 747 to 3 decimal places.

Hence

(a) Using a normal distribution,

X ~ N ( n p, n p q ) where n p = 20 0. 4 = 8

n p q = 20 0. 4 0.6 = 4.8

and
So
X ~ N (8, 4.8)

and P(6 X 10 ) P(5.5 < X < 10.5) .

With Z =

X 8
,
4.8

10.5 8
5.5 8
P( 5.5 < X < 10.5) =

4.8
4.8
= (1.141) ( 1.141)
= 2(1.141) 1
2 0.87286 1
= 0. 746 to 3 decimal places.
(Note that this is very close to the value found above.)

(b) Using a Poisson distribution

=np=8
So
X ~ Po(8) and P( X = x ) = e 8

8x
x!

This gives
P( X = 6 ) = e 8
Similarly

86
= 0.1221
6!

P( X = 7) = 0.1396
P( X = 8) = 0.1396

166

5.5

10.5

Chapter 8 The Normal Distribution

P( X = 9) = 0.1241
P( X = 10 ) = 0.0993

Thus

P(6 X 10 ) = 0.625 to 3 decimal places.

(This is a poor approximation, since you should have n > 50 and

1
p<
to use a Poisson approximation.)
10
Finally you should now be in a position to decide which of the
distributions to use in order to model a situation.

Example
Answer the following questions using, in each case, tables of the
binomial, Poisson or normal distribution according to which you
think is most appropriate.
(a) Cars pass a point on a busy city centre road at an average
rate of 7 per five second interval. What is the probability
that in a particular five second interval the number of cars
passing will be
(i) 7 or less
(ii) exactly 7?
(b) Weather records show that for a certain airport during the
winter months an average of one day in 25 is foggy enough
to prevent landings. What is the probability that in a period
of seven winter days landings are prevented on
(i) 2 or more days?
(ii) no days?
(c) The working lives of a particular brand of electric light
bulb are distributed with mean 1200 hours and standard
deviation 200 hours. What is the probability of a bulb
lasting more than 1150 hours?
(AEB)
Solution
(a) The Poisson distribution is suitable here since the question
concerns a random event that can occur 0, 1, 2, ... times.
The mean value is x = 7, giving, from tables,
(i) P( 7 or less) = 0.5987
(ii) P( 7) = P( 7 or less) P(6 or less)

= 0.5987 0. 4497
= 0.149 .

167

Chapter 8 The Normal Distribution

(b) The binomial distribution is a suitable distribution with

1
n = 7 and p = . Using tables,
25
(i) P(2 or more ) = 1 P(1 or less)
= 1 0.9706
= 0.0294.

24
(ii) P( no days) =
25

0. 7514.

(c) The normal distribution is the model to use here, although

the 'working lives' are not necessarily normal; so assume X,
the working life, is distributed

X ~ N 1200, 200 2

and

P( X > 1150 ) = 1 P( X 1150 )

1150 1200

200
= 1 ( 0.25)

= 1

= ( 0.25)
= 0.59871.

Exercise 8E
1. The probability of someone smoking is about 0.4.
What is the probability that:
(a) in a group of 50 people more than half of
them smoke;
(b) in a group of 150, less than 50 of them
smoke?
2. It is known nationally that support for the Story
party is 32% from election results. In a survey
carried out on 200 voters what is the probability
that more than 80 of them are Story supporters?
3. A manufacturer knows from experience that his
machines produce defects at a rate of 5%. In a
day's production of 500 items 40 defects are
produced. The Production Manager says this is
not surprising. Is there evidence to support this?

168

4. Tickets for a concert are sold according to a

Poisson distribution with mean 30 per day. What
are the probabilities that:
(a) less than 20 tickets are sold in one day;
(b) all 180 tickets are sold in a five day working
week?
5. Parts for a washing machine are known to have a
weekly demand according to a Poisson
distribution mean 20. How many parts should be
stocked to ensure that a shop only runs out of
parts on 1 in 20 weeks?

Chapter 8 The Normal Distribution

8.6

A very important application

of the normal

Most modern calculators have a random number generator. The

numbers produced generally follow a rectangular distribution in
the range 0.000 to 0.999. These should therefore have mean 0.5,
variance 0.083 ( = 0.289) . (See Section 7.6 )

Activity 4
Generate 10 random numbers and put them straight into the
statistical function of your calculator. Write down x, the mean
of your sample.
Repeat this 20 times and write down the means of the samples
(remember to clear the statistical memories each time).
Plot these twenty results on normal probability paper and find
the mean and SD of the sample means.

You should find that the twenty values are roughly normal, with
mean, not suprisingly, 0.5 and SD 0.1. The SD has been
decreased by a factor equivalent to the square root of the size of
the sample, i.e. 10 = 3.16 .
This is the basis of a very important theorem, called the Central
Limit Theorem. This says that, irrespective of the original
distribution, sample means are normally distributed about the
original distribution mean with 'standard error' equal to

n
being the original SD and n the sample size. This will be
explained in more detail in the next chapter.

169

Chapter 8 The Normal Distribution

8.7

Miscellaneous Exercises

1. The masses of plums from a certain orchard have

mean 24g and standard deviation 5g. The plums
are graded small, medium or large. All plums
over 28g in mass are regarded as large and the
rest equally divided between small and medium.
Assuming a normal distribution find:
(a) the proportion of plums graded large;
(b) the upper limit of the masses of the plums in
the small grade.
(AEB)
2. A student is doing a project on the hire of videos
from a local shop. She finds that the daily
demand for videos is approximately normal, with
mean 50 and SD 10.

(b) If Henri visits his office in this way 5 days

each week and if his arrival times at the
station each day are independent, what is the
probability that he misses his train once, and
only once, in a given week?
(c) Henri visits his office 46 weeks every year.
Assuming that there are no absences during
this time, what is the probability that he
misses his train less than 35 times in the
year?
(AEB)
6. The weights of pieces of home made fudge are
normally distributed with mean 34 g and
standard deviation 5 g.

(a) What is the probability of more than 65

videos being hired on a particular day?

(a) What is the probability that a piece selected

at random weighs more than 40g?

(b) The shop is considering stopping the hire as

it is uneconomical and decides that if demand
is less than 40 on more than 3 days out of the
next 7 it will do this.
How likely is this to happen?

(b) For some purposes it is necessary to grade

the pieces as small, medium or large. It is
decided to grade all pieces weighing over
40 g as large and to grade the heavier half of
the remainder as medium. The rest will be
graded as small. What is the upper limit of
the small grade?
(AEB)

(c) The student reckons that with a wider range

of videos, demand would increase by 25% on
average with no effect on the SD.
What is the probability of more than 65
videos being hired if this happens?
3. A Dungeons & Dragons player is suspicious of a
new die he has bought. He rolls the die 200
times and says he will throw it away if he gets
more than 40 sixes. What is the probability of
this happening with a fair die?
A friend who is a Statistics student suggests that
it would be better not to use 40 but to take a
figure which a fair die would only exceed 5% of
the time. What would this figure be?
4. In a survey of heights it was found that, of males
in the 16 - 19 year old age group, 25% were taller
than 178.8 cm and 10% were smaller than 165.4.
Use this information to find the mean and SD of
the distribution assuming it to be normal.
What is the likelihood of a male in this age group
being more than 183 cm (6 feet) tall?
5. Henri de Lade regularly travels from his home in
the suburbs to his office in Paris. He always
tries to catch the same train, the 08.05, from his
local station. He walks to the station from his
home in such a way that his arrival times form a
normal distribution with mean 08.00 hours and
SD 6 minutes.
(a) Assuming that his train always leaves on
time, what is the probability that, on any
given day, Henri misses his train?

170

7. Yuk Ping belongs to an athletics club. In javelin

throwing competitions her throws are normally
distributed with mean 41.0 m and standard
deviation 2.0 m.
(a) What is the probability of her throwing
between 40 m and 46 m?
(b) What distance will be exceeded by 60% of
her throws?
Gwen belongs to the same club. In competitions
85% of her javelin throws exceed 35 m and 70%
exceed 37.5 m. Her throws are normally
distributed.
(c) Find the mean and standard deviation of
Gwen's throws, each correct to two
significant figures.
(d) The club has to choose one of these two
athletes to enter a major competition. In
order to qualify for the final round it is
necessary to achieve a throw of at least 48 m
in the preliminary rounds. Which athlete
should be chosen and why?
(AEB)
8. Describe the main features of a normal
distribution.
A company has two machines cutting cylindrical
corks for wine bottles. The diameters of corks
produced by each machine are normally
distributed. The specification requires corks
with diameters between 2.91 cm and 3.12 cm.
Corks cut on Machine A have diameters with a
mean 3.03 cm and standard deviation 0.05 cm.

Chapter 8 The Normal Distribution

Calculate the percentage of corks cut on this

machine that
(a) are rejected as undersize.
(b) meet the specification.
Machine B cuts corks with a mean diameter of
3.01 cm of which 1.7% are rejected as oversize.
Calculate the standard deviation of the diameters
of corks cut on Machine B.
Which machine, if either, do you consider to be
the better? Explain.
(AEB)
9. In parts (a) and (b) of this question use the
binomial, Poisson or normal distribution
according to which you think is the most
appropriate. In each case draw attention to any
feature of the data which supports or casts doubt
on the suitability of the model you have chosen.
Indicate, where appropriate, that you are using
one distribution as an approximation to another.

(a) A technician looks after a large number of

machines on a night shift. She has to make
frequent minor adjustments. The necessity
for these occurs at random at a constant
average rate of 8 per hour. What is the
probability that
(i) in a particular hour she will have to
make 5 or fewer adjustments;
(ii)in an eight hour shift she will have to
make 70 or more adjustments?
(b) A number of neighbouring allotment tenants
bought a large quantity of courgette seeds
which they shared between them. Overall
15% failed to germinate. What is the
probability that a tenant who planted 20 seeds
would have
(i) 5 or more failing to germinate;
(ii) at least 17 germinating?

(AEB)

171

Chapter 8 The Normal Distribution

172