KENYATTA UNIVERSITY

SCHOOL OF PURE AND APPLIED SCIENCE

DEPARTMENT: MATHEMATICS AND ACTUARIAL SCIENCE

SST 111: INTRODUCTION TO PROBABILITY AND STATISTICS

WRITTEN BY: KIPKOECH RUTO AUGUSTINE

VETTED BY:

COURTESY OF ROSE NJERU

 LESSON ONE
DEFINITION OF STATISTICS, DATA COLLECTION AND
DATA PRESENTATION TECHNIQUES
1.1 : Introduction

Statistics is used in various fields like medicine, biology, sociology and others.
1.1 Lesson Learning Outcomes
By the end of this lesson the learner will be able to:
i. Define statistics
ii. State types of statistics
iii. State types of data and sources of data
iv. List various methods of representing data
1.2 : Definition of Statistics
Statistics is the science of data. This involves collecting, classifying, summarizing,
organizing, analyzing, and interpreting numerical information.
1.3 : Types of Statistics
There are two types of statistics namely descriptive and inferential statistics.
Descriptive statistics uses numerical and graphical methods to look for patterns in
a data set, and to present that information in a convenient form; while inferential
statistics uses sample data to make estimates, decisions, predictions, or other
generalizations about a larger set of data.
1.3.1 : Fundamental Elements of Statistics

(a) A population is a set of units (usually, people, objects, transactions, or events)

that we are interested in studying.
(b) A variable is a characteristic or property of an individual population unit.
(c) A sample is a subset of the units of a population.

COURTESY OF ROSE NJERU

 (d) A statistical inference is an estimate, prediction, or some generalizations about
a population based on information contained in a sample.
(e) A measure of reliability is a statement (usually associated with a statistical
inference.
1.4 : Types of Data
There are two types of data namely quantitative and qualitative data. Quantitative
data are measurements that are recorded on a naturally occurring numerical scale;
while qualitative data are measurements that cannot be measured on a natural
numerical scale; they can only be classified into one of a group of categories.
1.4.1 : Data Collection

Generally, you can obtain data in four different ways:

(a) Data from a published source
(b) Data from a designed experiment
(c) Data from a survey
(d) Data collected observationally.
1.4.2 : Methods of Presenting Data

We can use frequency distribution, histogram, frequency polygon, cumulative

frequency, graphs, pictograms etc.
1.4.2.1 : Frequency distribution (Frequency table)

A frequency distribution is simply a table in which the data is grouped into classes
and the number of cases which fall in each class are referred to as frequencies.
Example 1.1
In a survey of 35 families in a certain village, the number of children per family was
recorded and the following data were obtained:

1 0 2 3 4 5 6
7 2 3 4 0 2 5
8 4 5 12 6 3 2
7 6 5 3 3 7 8
COURTESY OF ROSE NJERU
 9 7 9 4 5 4 3
Represent the data in the form of a discrete frequency distribution.
Solution
Frequency distribution of the number of children: Table 1
Tally marks Frequency

0 // 2
1 / 1
2 //// 4
3 //// / 6
4 //// 5
5 //// 5
6 /// 3
7 //// 4
8 // 2
9 // 2
10 - 0
11 - 0
12 / 1
Total 35

Suppose we have the following table

Table 2
Masses of 100 female students at a certain school
Mass (Kg.) Number of students
50 – 52 5
53 – 55 18
56 – 58 42
59 – 61 27
62 - 64 8
Total 100

COURTESY OF ROSE NJERU

 This type of representation of frequencies is called a grouped frequency
distribution. In the above table 2 we have 5 groups viz 50 – 52, 53 -55, 56 – 58,
59 – 61, 62 – 64. The groups are also known as classes.
A symbol defining a class such as 50 – 52 in the above table 2 is called a class
interval. The numbers 50 and 52 are called class limits. The figures on the left side
of the classes are called lower class limits while figures on the right are known as
upper limits.
In the above table 2 the true dividing line between the first and second class is
52.5. These numbers 49.5, 52.5, 55.5, 57.5, 61.5, 64.5 are called class boundaries
or true class limits.
The class width or class size of any class interval, is the difference between the
upper and the lower class boundaries of the class. In table 2 the sizes of classes ae
the same that is 3 (52.2 – 49.5=3).
The class mark of a class interval is the mid-point of the class interval e.g. the last
class has lower boundary 61.5 and upper boundary 64.5. Thus, the class mark for
this class is 63 i.e. (61.5 + 64.5) /2.

Example 1.2
Prepare a frequency distribution for the following observations. Take classes as
15 – 24, 25 – 34, 35 – 44, 45 – 54, 55 – 64, 65 -74, 75 -84 .

Data

15 45 40 42 50 60 62 68 70 42
75 75 80 81 25 26 31 32 78 45
31 45 42 43 55 56 78 80 81 62
60 62 58 69 70 45 50 56 72 58
75 62 62 65 60 70 35 37 40 55

COURTESY OF ROSE NJERU

 Solution
Frequency distribution
Marks Tally marks Frequency
15 – 24 / 1
25 – 34 //// 5
35 – 44 //// /// 8
45 – 54 //// / 6
55 – 64 //// //// //// 14
65 – 74 //// // 7
75 - 84 //// //// 9
Total 50

1.4.2.2 : Histogram

A histogram or frequency histogram consists of a set of rectangles

f
15

10

14.5 24.5 34.5 44.5 54.5 64.5 74.5 84.5 marks

1.4.2.3 : Frequency polygon

A frequency polygon is a line graph of class frequency plotted against class mark. It
can be obtained by connecting midpoints of the tops of rectangles in the histogram.

COURTESY OF ROSE NJERU

 1.4.2.4 : Cumulative Frequency

The total frequency of all values less than the upper class boundary of a given class
interval is called the cumulative frequency up to and including the class interval.
Let us consider table 2
Mass(Kg) 49.5 52.5 55.5 58.5 61.5 64.5
Cumulative 0 5 23 65 92 100
frequency

1.6: Assessment
1. Construct a frequency table for the following data of 50 marks obtained by 50
students in TCU 201 examination. Take classes 5 -15, 15 – 25, etc.

64 26 56 85 42 38 10 55 72 65
15 47 59 62 52 54 48 62 81 31
44 54 50 52 66 38 77 88 17 58
15 25 49 51 64 68 53 50 72 58
50 61 70 80 90 05 16 51 61 52
2. Draw a histogram for the following data

COURTESY OF ROSE NJERU

 Variable 100 - 110 - 120 - 130 - 140 – 150 - 160 -
110 120 130 140 150 160 170
Frequency 11 28 36 49 33 20 8

References
1. Statistical Methods by S.P. Gupta
2. Applied Statistics and Probability for Engineers by D.C. Runger

COURTESY OF ROSE NJERU

 LESSON TWO
MEASURES OF CENTRAL TENDENCY
2.1 : Introduction

One of main objectives of statistical analysis is to obtain one single value that
describes the characteristic of the entire mass of unwieldy data. Such a value is
called the central value or average value. In this lesson we will consider some of
central values which are commonly used.
2.2 : Lesson Learning Outcomes

By the end of this lesson the learner will be able to:

i. State measures of location
ii. Compute measures of central tendency
iii. State uses of measures of location

2.3 : Measures of Location

The various types of measures of location or measures of central tendency, in

common use, are:
i. Arithmetic mean or simply the mean
ii. Geometric mean
iii. Harmonic mean
iv. Median and Quartiles
v. Mode
Notation
We shall denote 𝑥𝑖 (𝑖 = 1,2,3, … , 𝑛) as any of the n values 𝑥1, 𝑥2, … , 𝑥𝑛 assumed
by a variable X. Therefore, 𝑥𝑖 represents the ith observation. We shall use the
symbol ∑𝑛 𝑥 to denote the sum of all the 𝑥 `𝑠 from i=1 to i=n, that is
𝑖=1 𝑖 𝑖
𝑛 𝑥𝑖 = 𝑥1 + 𝑥2 + … + 𝑥𝑛.
∑𝑖=1

COURTESY OF ROSE NJERU

 2.3.1 : Mean

The arithmetic mean or simply the mean of a set of n numbers 𝑥1, 𝑥2, … , 𝑥𝑛 is
denoted by 𝑥 and is defined as
𝑛
1 1
𝑥= (𝑥1 + 𝑥2 + … + 𝑥𝑛) = ∑ 𝑥𝑖
𝑛 𝑛
𝑖=1
Therefore, to compute the mean of a simple series e.g. 3, 6, 9, 13, 15 we simply
add the items and divide by the number of items. Thus,
1
𝑥 = (3 + 6 + 9 + 13 + 15) = 9.2
5
If the numbers 𝑥1, 𝑥2, … , 𝑥𝑘 occur with frequencies 𝑓1, 𝑓2, … , 𝑓𝑘 respectively, the
mean is computed as follows:
𝑓1𝑥1 + 𝑓2𝑥2 + … + 𝑓𝑘𝑥𝑘 𝑘 𝑓𝑖𝑥𝑖
∑𝑖=1 𝑘 𝑓𝑖𝑥𝑖
∑𝑖=1
𝑥= = 𝑘 =
𝑓1 + 𝑓2 + ⋯+ 𝑓𝑘 ∑𝑖=1 𝑓𝑖 𝑛
Where ∑𝑘𝑖=1 𝑓𝑖 = 𝑛.
Example 2.1
Calculate the mean of the frequency distribution shown below
Mark 𝑥𝑖 22 24 25 27 30 31 34 36
(out of
50)
Number 𝑓𝑖 2 5 5 6 7 4 5 6
of
students

Solution
(22×2)+(24×5)+⋯+(36×6) 1171
Mean mark (𝑥 )= = = 29.3
2+5+⋯+6 40

The arithmetic mean of a grouped frequency distribution is computed by the

above formula after the classes have been replaced by their mid-values (class
marks).

COURTESY OF ROSE NJERU

 Example 2.2 Given the following frequency distribution, calculate the mean
Monthly 13-17 18-22 23-27 28-32 33-37 38-42 43-47 49-52 53-57
wages
($)
No. of 2 22 19 14 3 4 6 1 1
workers

Solution
Mid-value No. of workers Product
x f fx
15 2 30
20 22 440
25 19 475
30 14 420
35 3 105
40 4 160
45 6 270
50 1 50
55 1 55
Total ∑9 𝑓𝑖 =72 ∑9 𝑓𝑖𝑥𝑖 =2,005
𝑖=1 𝑖=1

𝑘
Hence 𝑥 = ∑𝑖=1 𝑓𝑖𝑥𝑖 = 2005 = 27.8
𝑛 72

2.3.1.1 : Assumed mean method for computing mean

Suppose 𝑥1, 𝑥2, … , 𝑥𝑖, … , 𝑥𝑛 are the observed values whose mean is required; the
we may proceed as follows:
Let A be the assumed mean, and let 𝑑𝑖 = 𝑥𝑖 − 𝐴 (𝑖 = 1, 2, … , 𝑛) , then
𝑥𝑖 = 𝐴 + 𝑑𝑖
Thus, ∑𝑛 𝑥𝑖 = ∑𝑛 (𝐴 + 𝑑𝑖)
𝑖=1 𝑖=1

Dividing through by n we have

COURTESY OF ROSE NJERU

 𝑛 𝑛
1 1
∑ 𝑥𝑖 = ∑(𝐴 + 𝑑𝑖)
𝑛 𝑛
𝑖=1 𝑖=1

∴𝑥 = 𝐴+𝑑
1
Where 𝑑 = ∑𝑛 𝑑 .
𝑛 𝑖=1 𝑖

Therefore, the mean 𝑥 is equal to the assumed mean plus the mean of deviations
from the assumed mean.
Example 2.3
Calculate the mean of the values 307, 320, 325, 341, 315, 319.
Solution
Let 𝐴 = 320. Then
𝑑1 = 307 − 320 = −13, 𝑑2 = 320 − 320 = 0, 𝑑3 = 325 − 320 = 5,
𝑑4 = 341 − 320 = 21, 𝑑5 = 315 − 320 = −5, 𝑑6 = 319 − 320 = −1.
∴𝑥 =𝐴+𝑑
1
= 320 + [(−13) + 0 + 5 + 21 + (−5) + (−1)] = 321.17
6

Suppose 𝑥1, 𝑥2, … , 𝑥𝑘 occur with frequencies 𝑓1, 𝑓2, … , 𝑓𝑘 respectively, then using
the assumed A, we compute the actual mean as follows:
Let 𝑑𝑖 = 𝑥𝑖 − 𝐴 (𝑖 = 1, 2, … , 𝑘), then 𝑥𝑖 = 𝐴 + 𝑑𝑖. Thus

𝑓𝑖𝑥𝑖 = 𝑓𝑖𝐴 + 𝑓𝑖𝑑𝑖 and

𝑘
∑𝑖=1 𝑓𝑖𝑥𝑖 = ∑𝑘𝑖=1 (𝑓𝑖𝐴 + 𝑓𝑖𝑑𝑖 )
𝑘
= 𝐴 ∑ 𝑖=1 𝑓𝑖 + ∑𝑖=1
𝑘 𝑓𝑑
𝑖 𝑖

Dividing both sides by ∑𝑘𝑖=1 𝑓𝑖 we obtain

∑𝑘 𝑓𝑖𝑑𝑖
𝑥 = 𝐴 + 𝑑 , where 𝑑 = 𝑖=1
∑𝑘 𝑓𝑖
𝑖=1

11
For grouped data we have to compute the class marks. If 𝑥𝑖 is the class mark and
𝑓𝑖 is the class frequency, then
𝑘
∑𝑖=1 𝑓𝑖𝑥𝑖
𝑥 = ∑𝑘 𝑓𝑖
if there are k classes.
𝑖=1

Suppose our assumed mean A coincides with some class mark. Then
𝑘
∑ 𝑓𝑖𝑑𝑖
𝑥= 𝐴 + 𝑖=1
∑𝑘 𝑓𝑖 .
𝑖=1

If all the class intervals have equal size c say, then 𝑥 may be expressed in the form
𝑐 ∑𝑘 𝑓𝑖𝑢𝑖 𝑘 𝑓𝑢
𝑥 = 𝐴+ 𝑖=1 = 𝐴 + 𝑐 𝑢, where 𝑑𝑖 = 𝑐 𝑢𝑖 and 𝑢= 𝑖 =∑1 𝑖 𝑖 . This
∑𝑘
𝑘
𝑖=1 𝑓𝑖
∑𝑖=1 𝑓𝑖

method is known as the coding method for computing the mean.

Example 2.4 Using

i. the direct method,
ii. assumed mean method, A=22.5,
iii. the coding method,
compute the mean for the following grouped data

Class 5-8 9-12 13- 17- 21- 25- 29- 33- 37- 41-
16 20 24 28 32 36 40 44
f 3 9 11 10 20 16 15 6 8 2

12
 Solution
To calculate 𝑥 using the three methods we make a table as shown below
Class f X fx d u fd fu
5–8 3 6.5 19.5 -16.0 -4 -48.0 -12
9 – 12 9 10.5 94.5 -12.0 -3 -108.0 -27
13 – 16 11 14.5 159.5 -8.0 -2 -88.0 -22
17 – 20 10 18.5 185.0 -4.0 -1 -40.0 -10
21 – 24 20 22.5 450.0 0 0 0 0
25 – 28 16 26.5 424.0 4.0 1 64.0 16
29 - 32 15 30.5 457.5 8.0 2 120.0 30
33 – 36 6 34.5 207.0 12.0 3 72.0 18
37 – 40 8 38.5 308.0 16.0 4 128.0 32
41 - 44 2 42.5 85.0 20.0 5 40.0 10
Total 100 2390.0 140.0 35

C = 4, A = 22.5. Thus
∑𝑘 𝑓𝑖𝑥𝑖 2390.0
i. 𝑖=𝑘1𝑥 = =
∑𝑖=1 𝑓𝑖 100

𝑘
∑𝑖=1 140.0
𝑓𝑖𝑑𝑖 = 22.5 + = 23.9
ii. 𝑥 =𝐴+ 100
𝑘
∑𝑖=1 𝑓𝑖

𝑘 35
iii. 𝑥 = 𝐴 + 𝑐 ∑ 𝑖=1 𝑓𝑖𝑢𝑖 = 22.5 + 4 ( ) = 23.9
𝑘 𝑓𝑖 100
∑𝑖=1

NB: The sum of the deviations of the values 𝑥𝑖 from their mean 𝑥 is zero.
Proof
Let 𝑑1 = 𝑥1 − 𝑥 , 𝑑2 = 𝑥2 − 𝑥, … , 𝑑𝑛 = 𝑥𝑛 − 𝑥 be the deviations of
𝑥1, 𝑥2, … , 𝑥𝑖, … , 𝑥𝑛 from their mean 𝑥. Then
Sum of deviations= ∑𝑛 𝑑𝑖 = ∑𝑛 (𝑥𝑖 − 𝑥)
𝑖=1 𝑖=1
𝑛
= ∑ 𝑖=1 𝑥𝑖 − 𝑛𝑥
13
 ∑𝑛 ∑𝑛 𝑥𝑖 ∑𝑛 ∑𝑛
= 𝑖=1 𝑥𝑖 −𝑛( 𝑖=1
)= 𝑖=1 𝑥𝑖 − 𝑖=1 𝑥𝑖 =0
𝑛

2.4 : Assessment

1. Compute the mean for the following grouped data using the three methods
Class 63-67 68-72 73-77 78-82 83-87 88-92 93-97 98-102
f 2 5 12 17 14 6 3 1

2. Find the missing frequency, if the arithmetic mean is 34 marks,

of the data given below
Marks 0 - 10 10 - 20 20 - 30 30 - 40 40 - 50 50 - 60
No. of 5 15 20 - 20 10
students

References
1. Statistical Methods by S.P. Gupta
2. Applied Statistics and Probability for Engineers by D.C. Runger

14
 LESSON THREE
MEASURES OF CENTRAL TENDENCY
3.1 : Introduction

In this lesson will discuss the other measures of location or measures of central
tendency namely geometric mean, harmonic mean, median and mode of given
data.
3.2 : Lesson Learning Outcomes

By the end of this lesson the learner will be able to:

i. Compute geometric mean, harmonic mean, median and mode of given
data
ii. Determine median and mode using graphical method.

3.3 : Geometric Mean

The Geometric mean G of a set of 𝑛 non-zero positive numbers 𝑥1, 𝑥2, … , 𝑥𝑛 is

the nth root of the product of the numbers. i.e.

𝐺 = 𝑛√𝑥1 𝑥2 … 𝑥𝑛
Example 3.1 Compute the geometric mean of the following numbers 3,9,27.
Solution
3
G = √3 × 9 × 27 = 9
If the k non-zero positive variate-values 𝑥1, 𝑥2, … , 𝑥𝑘 occur with frequencies
𝑓1, 𝑓2, … , 𝑓𝑘 respectively, then geometric mean G, is given by
𝑛 𝑓1 𝑓2 𝑓𝑘 𝑘
𝐺 = √𝑥1 𝑥2 … 𝑥𝑘 , where 𝑛 = ∑𝑖=1 𝑓𝑖
𝟏
= [𝑥𝑓1𝑥𝑓2 … 𝑥𝑓𝑘]𝒏
1 2 𝑘

Therefore,

15
 1
log 𝐺 = [𝑥𝑓1𝑥𝑓2 … 𝑥𝑓𝑘]
𝑛 1 2 𝑘
1
= [𝑓 log 𝑥 + 𝑓 log 𝑥 + ⋯ + 𝑓 log 𝑥 ]
𝑛 1 1 2 2 𝑘 𝑘

Example 3.2
Compute the geometric mean of the following numbers
X 3 9 27
f 2 2 2

Solution
6
G = √32 × 92 × 272 = 9 or
1
𝐺 = [2 log 3 + 2 log 9 + 2 log 27]
6
1 1
= [2 log 3 + 4 log 3 + 6 log 3] = [12 log 3]
6 6

= 2log3
= log32
⇒ 𝐺 = 32 = 9.
3.4 : Harmonic Mean

The harmonic mean of the variate-values is the reciprocal of their arithmetic

mean of their reciprocals. That is the harmonic mean, H, of a set of n numbers
𝑥1, 𝑥2, … , 𝑥𝑛 is given by
1
𝐻=
1 𝑛 1
∑
𝑛 𝑖=1 𝑥𝑖
𝒏
= ∑𝑛 1
𝑖=1 𝑥𝑖

Example 3.3
Compute the harmonic mean of the following numbers 3, 9, 27
Solution

16
 3
𝐻 = 1 1 1 = 6.23
3 + 9 + 27
The harmonic mean, H, of k non-zero different variate-values
𝑥1, 𝑥2, … , 𝑥𝑘 occurring with frequencies 𝑓1, 𝑓2, … , 𝑓𝑘 respectively is given by
𝑘
∑𝑖=1 𝑓𝑖
𝐻= 𝑘 𝑓𝑖
∑𝑖=1 𝑥𝑖

NB: The relation between arithmetic mean 𝑥, geometric mean 𝐺 and the
harmonic mean 𝐻 is
𝐻 ≤𝐺 ≤𝑥
The equality signs hold only if all the numbers 𝑥1, 𝑥2, … , 𝑥𝑛 are identical.
Example 3.4
The set of numbers 3, 9, 27 has 𝑥 = 13.0, 𝐺 = 9.0, 𝐻 = 6.23.
∴𝐻<𝐺<𝑥
3.5 : Median, Quartiles, Deciles and Percentiles

3.5.1 : Median

The median of a set of n numbers arranged in order of magnitude is the middle

value if n is odd or the arithmetic mean of the two middle values if n is even. For
example, the set
i. 4,7,7,7,9,10,10,12,15 has median 9
8+10
ii. 6,7,8,10,11,12 has median 9 i.e. ( )=9
2

If the data is grouped then the median is obtained by interpolation given by

𝑛
−(∑ 𝑓)1
Median= 𝐿1 + (2 )𝑐
𝑓𝑚𝑒𝑑𝑖𝑎𝑛

Where,
𝐿1 =lower class boundary of the median class (the class containing the
Median)

17
 n=number of items in the data (total frequency)
(∑ 𝑓)1 = sum of frequencies of all classes lower than the median class
𝑓𝑚𝑒𝑑𝑖𝑎𝑛 =frequency of median class
C= size of median class interval
Example 3.5 Calculate the median for the following frequency distribution of the
number of marks obtained by 49 students in a class.
Marks 5 - 10 10 - 15 15 - 20 20 - 25 25 - 30 30 - 35 35 - 40 40 - 45
group
No. of 5 6 25 10 5 4 2 2
students

Solution
Class Frequency Cumulative frequency
(f) (cf)
5 – 10 5 5
10 – 15 6 11
15 – 20 15 26
20 – 25 10 36
25 – 30 5 41
30 – 35 4 45
35 – 40 2 47
40 – 45 2 49

n 49 1
= = 24
2 2 2
Therefore, the median class is 15 – 20
Thus, 𝐿1 = 15, (∑ 𝑓)1 = 11 (i.e. 5+6=11), 𝑓𝑚𝑒𝑑𝑖𝑎𝑛 = 15, c=5.
𝑛 49
−(∑ 𝑓)1 −11
Median= 𝐿1 + (2 ) 𝑐 = 15 + (2 ) 5 = 19.5 marks
𝑓𝑚𝑒𝑑𝑖𝑎𝑛 15

18
3.5.2 : Quartiles, Deciles and Percentiles

Quartiles are those values which divide the total frequency into four equal parts;
deciles and percentiles divide into ten and one hundred equal parts respectively.
Quartiles are denoted by 𝑄1, 𝑄2 and 𝑄3, that is the first, second and third
quartiles respectively, the value 𝑄2 being equal to the median. The first quartile
𝑄1 is often referred to as the lower quartile and the third quartile 𝑄3 is referred to
as the upper quartile.
Example 3.6
The set 2,3,3,4,4,4,5,5,6,7,7,8 has
4+5 3+4 6+7
𝑄 = = 4.5 , 𝑄 = = 3.5, 𝑄 = = 6.5
2 2 1 2 3 2

The formulae for computing the lower and upper quartiles of grouped data is the
n n 3n
same as that of median except instead of we have and for the lower and
2 4 4
upper quartiles respectively. Thus
𝑛
𝑄 =𝐿 + (4 − (∑ 𝑓)𝑄1 ) 𝑐
1 𝑄1 1
𝑓𝑄1

and

3𝑛
𝑄 =𝐿 + ( 4 − (∑ 𝑓)𝑄3 )𝑐
3 𝑄3 3
𝑓𝑄3

Where 𝐿𝑄1 =lower class boundary of the class containing the first quartile,
𝑛 = total frequency,
(∑ 𝑓)𝑄1 =sum of frequencies of all classes lower than the first quartile class,
𝑓𝑄1 =frequency of the first quartile class,
𝑐1 =size of first quartile class interval,
𝐿𝑄3 = lower class boundary of the class containing the third quartile,

19
 (∑ 𝑓)𝑄3 =sum of frequencies of all classes lower than the third quartile class,
𝑓𝑄3 =frequency of the third quartile class,
𝑐3 =size of third quartile class interval.
Example 3.7
Compute the lower and upper quartiles for the following frequency distribution.
Class 5.5- 10.5- 15.5- 20.5- 25.5- 30.5- 35.5- 40.5-
9.5 14.5 19.5 24.5 29.5 34.5 39.5 44.5
Frequency 5 6 15 10 5 4 2 2

Solution
Class Frequency (f) Cumulative frequency(cf)
5.5 – 9.5 5 5
10.5 – 14.5 6 11
15.5 – 19.5 15 26
20.5 – 24.5 10 36
25.5 – 29.5 5 41
30.5 – 34.5 4 45
35.5 – 39.5 2 47
40.5 – 44.5 2 49
Total 49

𝑛 49 1
To compute 𝑄1 we have 4 = 4 = 12 4 . Thus 𝑄1 lies in the class 15.5 – 19.5.
Therefore, 𝐿𝑄1 = 15, 𝑛 = 49, (∑ 𝑓)𝑄1 = 11, 𝑓𝑄1 = 15, 𝑐1 = 5.
Hence
𝑛
− (∑ 𝑓)𝑄1
𝑄1 = 𝐿𝑄1 + (4 ) 𝑐1
𝑓 𝑄1

49 − 11
= 15 + ( 4 ) 5 = 15.41
15

20
 3𝑛 3(49) 3
To compute 𝑄3 we have 4 = 4 = 36 4 . Thus 𝑄3 lies in the class 25.5 – 29.5.
Therefore, 𝐿𝑄3 = 25, 𝑛 = 49, (∑ 𝑓)𝑄3 = 36, 𝑓𝑄1 = 5, 𝑐1 = 5.
Hence
3𝑛
𝑄 =𝐿 + ( 4 − (∑ 𝑓)𝑄3 ) 𝑐
3 𝑄3 3
𝑓𝑄1

3(49) − 36
= 25 + ( 4 ) 5 = 25.75.
5

Deciles are denoted 𝐷1, 𝐷2, … , 𝐷8, 𝐷9 and percentiles are denoted by
𝑃1, 𝑃2, … , 𝑃98, 𝑃99 . For grouped data we compute deciles and percentiles using
the following formulae:
𝑖𝑛
− (∑ 𝑓)𝐷 𝑖
𝐷𝑖 = 𝐿𝐷𝑖 +(10
𝑓𝐷𝑖 ) 𝑐𝐷𝑖 , 𝑖 = 1,2, … , 9

and

𝑖𝑛
− (∑ 𝑓)𝑃 𝑖
𝑃𝑖 = 𝐿𝑃𝑖 + (100
𝑓𝑃𝑖 ) 𝑐𝑃𝑖 , 𝑖 = 1,2, … ,99

Quartiles, Deciles and Percentiles for grouped data can be obtained using
graphical method.
3.6 : Mode

The mode or modal value of the distribution is that value of the variable for which
the frequency is maximum.
Example 3.8
The set
i. 2, 2, 5, 8, 8, 8 has mode 8,

21
 ii. 3, 5, 6, 8, 9, 10 has no mode,
iii. 14, 16, 12.2, 12.2, 12, 17, 17, 18, 20 has two modes 12.2 and 17 and is
known as bimodal.
For a frequency distribution, the mode is computed by the following formula:
𝑓𝑚−𝑓1
Mode= 𝐿 + ( )𝑐 ,
2𝑓𝑚−𝑓1−𝑓2

Where, 𝐿 =lower class boundary of the modal class (the class interval having
Maximum frequency),
𝑓𝑚 = the maximum frequency (frequency of the modal class),
𝑓1 = the frequency of the class preceding the modal class,
𝑓2 = the frequency of the class following the modal class,
𝑐 = size of the modal class interval.
Example 3.9
Find the mode of the distribution given below:

Marks 5 - 10 10 - 15 15 – 20 - 25 25 - 30 30 - 35 35 - 40 40 - 45
group 20
No. of 5 6 15 10 5 4 2 2
students

Solution
Clearly the class interval 15 – 20 has maximum frequency, hence it is the modal
class.
Therefore, 𝐿 = 15, 𝑓1 = 6, 𝑓2 = 10, 𝑓𝑚 =15, 𝑐 = 5. Hence
𝑓𝑚−𝑓1
Mode= 𝐿 + ( )𝑐
2𝑓𝑚−𝑓1−𝑓2
15−6
= 15 + ( ) 5 = 18.21
2(10)−6−10

22
3.7 : Assessment

1. Marks obtained by 80 students are given below:

Marks 0 - 10 10 – 20 20 - 30 30 - 40 40 - 50 50 - 60
Frequency 3 9 15 30 18 5
a) Compute the median, quartiles and mode for the above frequency
distribution.
b) Draw a cumulative frequency curve, hence determine the median and
quartiles.

3. The median and mode of the following wage distribution are known to be
$ 33.5 and $ 34 respectively. Three frequency values from the table are
however missing. Find these missing values.
Wages 0 - 10 10 -20 20-30 30-40 40-50 50-60 60-70 Total
($)
No. of 4 16 ? ? ? 6 4 230
Workers

References
1. Statistical Methods by S.P. Gupta
2. Applied Statistics and Probability for Engineers by D.C. Runger

23
 LESSON FOUR
MEASURES OF DISPERSION
4.1 : Introduction

The measures of central tendency are not sufficient measures to reveal the shape
of the distribution of data set. The measures that show the spread of a data set
are called the measures of dispersion. The main measures of dispersion are range,
inter quartile range, standard deviation, variance and coefficient of variation. In
this lesson we will discuss these measures of dispersion.
4.2 : Lesson Learning Outcomes

By the end of this lesson the learner will be able to:

i. State measures of dispersion
ii. Compute measures of dispersion
4.3 : Range

The range is the difference between two extreme values of a distribution. If 𝑥𝑔

and 𝑥𝑙 are the greatest and the smallest observations respectively in a
distribution, then the range of the distribution is 𝑥𝑔 − 𝑥𝑙 .
The range is the simplest and also the least reliable measure of dispersion. This is
due to the fact that, it is based only on two extreme values.
Example 4.1
Find the range for the following data: 20, 21, 15, 10, 7, 4, 51, 49
Solution
The range for the above data is 𝑥𝑔 − 𝑥𝑙 = 51 − 4 = 47.
4.4 : Quartile Deviation (semi- interquartile range)

The quartile deviation or semi-interquartile range 𝑄 is given by

1
𝑄 = (𝑄 − 𝑄 ) ,
2 3 1

Where 𝑄1 and 𝑄3 are the first and third quartiles of the distribution respectively.

24
Interquartile range= 𝑄3 − 𝑄1.
Quartile deviation is definitely a better measure of dispersion than the range as it
makes use of 50% of the data. Since it ignores the rest 50% of the data, it cannot
be regarded as a reliable measure of dispersion.
4.5 : Standard Deviation

The standard deviation of a set of n numbers 𝑥1, 𝑥2, … , 𝑥𝑛 is usually denoted by s

and is the positive root of the arithmetic mean of the squares of deviations of the
given values from their arithmetic mean.
Thus, the standard deviation of n observations 𝑥1, 𝑥2, … , 𝑥𝑛 is given by
1
𝑠 = √ ∑𝑛 (𝑥 − 𝑥)2
𝑛 𝑖=1 𝑖

Example 4.2
Compute the standard deviation of the following set of numbers 120, 60, 70, 30,
150, 100, 180, 50.
Solution
1
𝑥 = (120 + 60 + ⋯ + 50) = 95
8
𝑛
1
∴𝑠=√ ∑(𝑥𝑖 − 𝑥)2
𝑛
𝑖=1

1
𝑠 = √ [(120 − 95)2 + (60 − 95)2 + ⋯ + (50 − 95)2]
8

= √2375 = 48.7

Suppose 𝑥1, 𝑥2, … , 𝑥𝑘 occur with frequencies 𝑓1, 𝑓2, … , 𝑓𝑘 respectively, the
standard deviation s can be expressed as
1
𝑠 = √ ∑𝑘 𝑓 (𝑥 − 𝑥)2 , where 𝑛 = ∑𝑘 𝑓
𝑛 𝑖=1 𝑖 𝑖 𝑖=1 𝑖

25
This formula is also applicable to grouped data if we identify 𝑥𝑖 as the class mark
of the ith class and 𝑓𝑖 as the corresponding class frequency.
4.6 : Variance

The square of the standard deviation is called the variance and is denoted by 𝑠2.
Thus, if 𝑥1, 𝑥2, … , 𝑥𝑛 is a set of n numbers, then their variance is given by
1
𝑠2 = ∑𝑛 (𝑥 − 𝑥)2 .
𝑛 𝑖=1 𝑖

If 𝑥1, 𝑥2, … , 𝑥𝑘 occur with frequencies 𝑓1, 𝑓2, … , 𝑓𝑘 respectively, their variance is
given by
1
𝑠2 = ∑𝑘 𝑓 (𝑥 − 𝑥)2 , where 𝑛 = ∑𝑘 𝑓
𝑛 𝑖=1 𝑖 𝑖 𝑖=1 𝑖

This formula is also applicable to grouped data if we identify 𝑥𝑖 as the class mark
of the ith class and 𝑓𝑖 as the corresponding class frequency.
Let us consider different formula for computing variance.
The variance of a set of n numbers is given by
1
𝑠2 = ∑𝑛 (𝑥 − 𝑥)2
𝑛 𝑖=1 𝑖
1
= ∑𝑛 𝑥2 − 𝑥2
𝑛 𝑖=1 𝑖

Example 4.3 Find the variance and standard deviation of the following set of
numbers: 12, 6, 7, 9, 15, 13, 18, 11.
Solution
To find the variance of these numbers we first find the mean of the
numbers i.e. 𝑥 .Thus
1
𝑥 = (12 + 6 + 7 + 9 + 15 + 13 + 18 + 11) = 11.375
8
Therefore, 𝑥2 = 129.391
Next, we find the mean of the squares of numbers i.e.

26
 1
1 ∑𝑛 𝑥2 = [122 + 62 + ⋯ + 182 + 112] = 143.625.
𝑛 𝑖 𝑖 8
1 1
Finally, 𝑠2 = ∑𝑛 𝑥2 − 𝑥2 = ∑8 𝑥2 − 𝑥2 = 143.625 − 129.391 = 14.2
𝑛 𝑖 𝑖 8 𝑖=1 𝑖

Hence the standard deviation 𝑠 = √14.2 = 3.8

If 𝑥1, 𝑥2, … , 𝑥𝑘 occur with frequencies 𝑓1, 𝑓2, … , 𝑓𝑘 respectively, their variance is
given by
1
𝑠2 = ∑𝑘 𝑓 (𝑥 − 𝑥)2 , where 𝑛 = ∑𝑘 𝑓
𝑛 𝑖=1 𝑖 𝑖 𝑖=1 𝑖
1
= ∑𝑘 𝑓 𝑥2 − 𝑥2
𝑛 𝑖=1 𝑖 𝑖

Example 4.4
The marks obtained by 30 students in statistics quiz marked out of 20 were as
shown below:

Mark 8 9 10 11 12 14 15 17 18 20
x
Number 2 3 4 4 5 3 3 3 2 1
of
students
f
Find the variance and standard deviation of the marks.

27
 Solution
Mark x Frequency f fx fx2
8 2 16 128
9 3 27 243
10 4 40 400
11 4 44 484
12 5 60 720
14 3 42 588
15 3 45 675
17 3 51 867
18 2 36 648
20 1 20 400
Total 30 381 5153

1 5153
𝑠2 = ∑𝑘 𝑓 𝑥 2 − 𝑥2 = − 381 2 = 10.48.
( )
𝑛 𝑖=1 𝑖 𝑖 30 30

Hence 𝑠 = √10.48 = 3.24

4.6.1 : Variance of Combined series

Suppose we have two series of numbers i.e.

𝑥11, 𝑥12, … , 𝑥1𝑛1 and 𝑥21, 𝑥22, … , 𝑥2𝑛2
The first series has 𝑛1 numbers, and the second series has 𝑛2 numbers.
Let the mean of the first series be 𝑥1 and the mean of the second series be 𝑥2 .
Let the variance of the first series be 𝑠21 and the variance of the second series be
𝑠22 .
If 𝑥 denotes the mean of the combined series, then
𝑛1𝑥1+𝑛2𝑥2
𝑥= .
𝑛1+𝑛2
Suppose 𝑠2 is the variance of the combined series. Then

28
 𝑛1 𝑛2
1
𝑠2 = [∑(𝑥1𝑖 − 𝑥)2 + ∑(𝑥2𝑖 − 𝑥)2 ]
𝑛1 + 𝑛2 𝑖=1 𝑖=1

But,
∑ 𝑛1 (𝑥 − 𝑥)2 = ∑ 𝑛1 [(𝑥 − 𝑥 ) + (𝑥 − 𝑥)]2
𝑖=1 1𝑖 𝑖=1 1𝑖 1 1
= ∑ 𝑛1 (𝑥 − 𝑥 )2 + 2(𝑥 − 𝑥) ∑ 𝑛1 (𝑥 − 𝑥 ) + 𝑛 (𝑥 − 𝑥)2
𝑖=1 1𝑖 1 1 𝑖=1 1𝑖 1 1 1
= 𝑛 {𝑠2 + (𝑥 − 𝑥)2} since ∑ 𝑛1 (𝑥 −𝑥 )=0
1 1 1 𝑖=1 1𝑖 1

Similarly,
∑𝑛2 (𝑥 − 𝑥) = 𝑛 {𝑠 + (𝑥 − 𝑥) }
2 2 2
𝑖=1 2𝑖 2 2 2

Hence,
1
𝑠2 = [𝑛 {𝑠2 + (𝑥 − 𝑥)2} + 𝑛 {𝑠2 + (𝑥 − 𝑥)2}]
1 1 1 2 2 2
𝑛1 + 𝑛2
If there are k series then the combined variance (pooled variance) of the series is
given by
1
𝑠2 = [𝑛 {𝑠2 + (𝑥 − 𝑥)2} + 𝑛 {𝑠2 + (𝑥 − 𝑥)2} + ⋯ +
1 1 1 2 2 2
𝑛1+𝑛2+⋯+𝑛𝑘
𝑛𝑘{𝑠2𝑘 + (𝑥𝑘 − 𝑥)2}]
where,
𝑛1𝑥1+𝑛2𝑥2+⋯+𝑛𝑘𝑘
𝑥=
𝑛1+𝑛2+⋯+𝑛𝑘
Example 4.5
Analysis of monthly wages paid to the workers in two firms A and B belonging to
the same industry gives the following results:
Firm A Firm B
Number of workers 500 600
Average monthly wage $ 186 $ 175
Variance of distribution 81 100
a) Which firm has a larger wage bill?
b) Calculate,
29
 i. the average monthly wage, and
ii. the variance and hence standard deviation of the distribution of
wages in the firms A and B taken together.
Solution
a) Wage bill for firm A= 𝑛1𝑥1 = $ (186 × 500) = $ 93,000
Wage bill for firm B= 𝑛2𝑥2 = $ (175 × 600) = $ 105,000.
Therefore, firm B has a larger wage bill.
b) 𝑛1 = 500, 𝑛2 = 600, 𝑥1 = $ 186, 𝑥2 = 175, 𝑠2 = 81, 𝑠2 = 100
1 2

i. The pooled (or combined) average monthly wage is

𝑛1𝑥1+𝑛2𝑥2 93,000+105,000
𝑥= =𝑥= = $ 180
𝑛1+𝑛2 500+600

ii. The pooled (or combined) variance is given by

1
𝑠2 = [500{81 + (186 − 180)2} + 600{100 + (175 − 180)2}]
500 + 600
= 121.36.

Hence standard deviation 𝑠 = √121.36 = 11.02

4.7 : Coefficient of Variation

The standard deviation is an absolute measure of dispersion. The corresponding

relative measure is known as the coefficient of variation. The coefficient of
variation (C.V) is defined by the formula
𝑆𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛
𝐶. 𝑉 = × 100%
𝐴𝑟𝑖𝑡ℎ𝑚𝑒𝑡𝑖𝑐 𝑚𝑒𝑎𝑛
𝑠
= × 100%
𝑥
This measure is used in problems where we want to compare the variability of
two or more series. The series(group) for which the C.V is greater is said to be
more variable or conversely less consistent or less homogeneous. On the other

30
hand, the series for which the C.V is small is said to be less variable or more
consistent or more homogeneous.

4.8 : Assessment

1. Compute the mean, range, variance and standard deviation of the

following set of numbers: 6, 10, 15, 25, 30, 32, 40, 46.
2. Calculate the mean, variance and standard deviation for the
following set of data.
𝑥 0−3 4−7 8 − 11 12 − 15 16 − 19 20 − 23
𝑓 7 4 18 12 7 6

3. The following data give the mean and standard deviation of 3

subgroups.
Subgroup No. of workers Average wages Standard
($) deviation ()$
A 50 61.0 8.0
B 100 70.8 9.0
C 120 80.5 10.0

Calculate,
i. the mean and standard deviation of the whole group,
ii. the C.V for each group and the C.V for the whole group.

References
1. Statistical Methods by S.P. Gupta
2. Applied Statistics and Probability for Engineers by D.C. Runger

31
 LESSON FIVE
MOMENTS
Moments are popularly used to describe the characteristics of a distribution and
are measured about a point.
Suppose 𝑥1, 𝑥2, … , 𝑥𝑛 are the n values assumed by the variable X, then we define
the following quantity
𝑥1𝑟+𝑥𝑟2+⋯+𝑥𝑛𝑟
𝑥𝑟 = 𝑛
As the 𝑟𝑡ℎ moment about zero. For different values of r we obtain different
moments. When r=1 we obtain the first moment about zero. That is
1 𝑛
𝑥1+𝑥2+⋯+𝑥𝑛 = ∑ 𝑥 =𝑥
𝑥1 = 𝑛 𝑛 𝑖=1 𝑖
The second moment about zero is obtained when we put r=2. That is

𝑥2 = 𝑥21+𝑥22+⋯+𝑥2𝑛 = 1 ∑𝑛 𝑥2
𝑛 𝑖=1 𝑖 𝑛
The 𝑟𝑡ℎ moment about the mean 𝑥 is given by
1
𝑚 = ∑𝑛 (𝑥 − 𝑥)𝑟
𝑟 𝑛 𝑖=1 𝑖
If r=1, then
1
𝑚 = ∑𝑛 (𝑥 − 𝑥)1 = 0.
1 𝑛 𝑖=1 𝑖

If r=2, then
1
𝑚 = ∑𝑛 (𝑥 − 𝑥)2 = 𝑠2,
2 𝑛 𝑖=1 𝑖
Which is the variance of the series 𝑥1, 𝑥2, … , 𝑥𝑛 .
The 𝑟𝑡ℎ moment about any value A is defined as
1
𝑚′ = ∑𝑛 (𝑥 − 𝐴)𝑟 = 1 ∑𝑛 𝑑𝑟,
𝑟 𝑛 𝑖=1 𝑖 𝑛 𝑖=1 𝑖

32
where 𝑑𝑖 = 𝑥𝑖 − 𝐴.

5.1 : Moments for Grouped Data

Suppose 𝑥1, 𝑥2, … , 𝑥𝑘 occur with frequencies 𝑓1, 𝑓2, … , 𝑓𝑘 respectively, then the
𝑟𝑡ℎ moment about zero is given by
𝑓1𝑥𝑟+𝑓2𝑥𝑟+⋯+𝑓𝑘𝑥𝑟
𝑥𝑟 = 1 2 𝑘
𝑓1+𝑓2+⋯+𝑓𝑘
1
= ∑𝑘 𝑓 𝑥𝑟
𝑛 𝑖=1 𝑖 𝑖

Where,
𝑛 = ∑𝑘 𝑓𝑖
𝑖=1
Therefore the 𝑟𝑡ℎ moment about the mean is given by
1
𝑚 = ∑𝑘 𝑓 (𝑥 − 𝑥)𝑟
𝑟 𝑛 𝑖=1 𝑖 𝑖
Where,
𝑘
𝑛 = ∑ 𝑖=1 𝑓𝑖
And the 𝑟𝑡ℎ moment about any value A is given by
1 1
𝑚′ = ∑𝑘 𝑓 (𝑥 − 𝐴)𝑟 = ∑𝑘 𝑓 𝑑𝑟,
𝑟 𝑛 𝑖=1 𝑖 𝑖 𝑛 𝑖=1 𝑖 𝑖
where 𝑛 = ∑𝑖=1
𝑘 𝑓𝑖 and 𝑑𝑖 = 𝑥𝑖 − 𝐴.
If 𝑥𝑖′𝑠 are class marks of grouped data with class intervals of equal sizes and A is
one of the class marks then each 𝑑𝑖 can be written as 𝑐𝑢𝑖, where c is the class
interval width (class size). Hence in this case the 𝑟𝑡ℎ moment about A can be
expressed as
1
𝑚′ = 1 ∑𝑘 𝑓 (𝑥 − 𝐴)𝑟 = ∑𝑘 𝑓 𝑑𝑟
𝑟 𝑛 𝑖=1 𝑖 𝑖 𝑛 𝑖=1 𝑖 𝑖
= 1 ∑𝑘 𝑓 (𝑐𝑢 )𝑟
𝑛 𝑖=1 𝑖 𝑖

33
 𝑐𝑟
= ∑𝑘 𝑓𝑢 𝑟

𝑛 𝑖=1 𝑖 𝑖
where 𝑛 = ∑𝑖=1
𝑘 𝑓𝑖 and 𝑑𝑖 = 𝑥𝑖 − 𝐴.

5.2 : Relations between Moments

The first moment about any value A is given by
1
𝑚′ = ∑𝑘 𝑓 (𝑥 − 𝐴)
1 𝑛 𝑖=1 𝑖 𝑖
1
= ∑𝑘 𝑓 𝑥 − 𝐴 = 𝑥 − 𝐴.
𝑛 𝑖=1 𝑖 𝑖
The first moment about the mean is given by
1
𝑚 = ∑𝑘 𝑓 (𝑥 − 𝑥) = 0.
1 𝑛 𝑖=1 𝑖 𝑖
The second moment about any value A is given by
1
𝑚′ = ∑𝑘 𝑓 (𝑥 − 𝐴)2.
2 𝑛 𝑖=1 𝑖 𝑖
The second moment about the mean is given by
1
𝑚 = ∑𝑘 𝑓 (𝑥 − 𝑥)2
2 𝑛 𝑖=1 𝑖 𝑖
1 𝑘
= ∑ 𝑓 {(𝑥 − 𝐴) − (𝑥 − 𝐴)}2
𝑛 𝑖=1 𝑖 𝑖

We know that (𝑎 − 𝑏)2 = 𝑎2 − 2𝑎𝑏 + 𝑏2

Let 𝑥𝑖 − 𝐴 = 𝑎 and 𝑥 − 𝐴 = 𝑏. Therefore,

{(𝑥𝑖 − 𝐴) − (𝑥 − 𝐴)}2 = (𝑥𝑖 − 𝐴)2 − 2(𝑥𝑖 − 𝐴)(𝑥 − 𝐴) +

(𝑥 − 𝐴)2
Thus,
1
𝒎 = ∑𝑘 𝑓 { (𝑥 − 𝐴)2 − 2(𝑥 − 𝐴)(𝑥 − 𝐴) + ( 𝑥 − 𝐴)2}
𝟐 𝑛 𝑖=1 𝑖 𝑖 𝑖

34
 1 1
= ∑𝑘 𝑓 (𝑥 − 𝐴)2 − 2 (𝑥 − 𝐴) ∑𝑘 𝑓 (𝑥 − 𝐴) +
𝑛 𝑖=1 𝑖 𝑖 𝑛 𝑖=1 𝑖 𝑖
(𝑥 − 𝐴)2
= 𝒎′ − 𝟐(𝑥 − 𝐴)2+(𝑥 − 𝐴)2 = 𝑚′ − (𝑥 − 𝐴)2
𝟐 2
= 𝑚′ − (𝑚′ )2
2 1
Hence 𝒎𝟐 = 𝒎′ − (𝒎′ )𝟐.
𝟐 𝟏
The third moment about any value A is given by
1
𝑚′ = ∑𝑘 𝑓 (𝑥 − 𝐴)3.
3 𝑛 𝑖=1 𝑖 𝑖
The third moment about the mean is given by
1
𝑚 = ∑𝑘 𝑓 (𝑥 − 𝑥)3
3 𝑛 𝑖=1 𝑖 𝑖
1 𝑘
= ∑ 𝑓 {(𝑥 − 𝐴) − (𝑥 − 𝐴)}3
𝑛 𝑖=1 𝑖 𝑖
We know that

(𝑎 − 𝑏)3 = 𝑎3 − 3𝑎2𝑏 + 3𝑎𝑏2 − 𝑏3

Let 𝑥𝑖 − 𝐴 = 𝑎 and 𝑥 − 𝐴 = 𝑏. Therefore,
{(𝑥𝑖 − 𝐴) − (𝑥 − 𝐴)}3 = (𝑥𝑖 − 𝐴)3 − 3(𝑥𝑖 − 𝐴)2(𝑥 − 𝐴) +
3(𝑥𝑖 − 𝐴)(𝑥 − 𝐴)2 − (𝑥 − 𝐴)3
Therefore,
𝑘 𝑘
𝑚 =1 ∑ 𝑓 (𝑥 − 𝐴)3 − 3( 𝑥 − 𝐴)1 ∑ 𝑓 (𝑥 − 𝐴)2 +
3 𝑖 𝑖 𝑖 𝑖
𝑛 𝑛
𝑖=1 𝑖=1
1 𝑘
3( 𝑥 − 𝐴)2 ∑ 𝑓 (𝑥 − 𝐴) − (𝑥 − 𝐴)3
𝑖 𝑖
𝑛
𝑖=1

35
 = 𝑚′ − 3𝑚′ 𝑚′ + 3(𝑚′ )2𝑚′ − (𝑚′ )3
3 1 2 1 1 1
= 𝑚′ − 3𝑚′ 𝑚′ + 3(𝑚′ )3 − (𝑚′ )3
3 1 2 1 1
Hence 𝒎𝟑 = 𝒎′ − 𝟑𝒎′ 𝒎′ + 𝟐(𝒎′ )𝟑
𝟑 𝟏 𝟐 𝟏
Similarly, the fourth moment about any value A is given by
1
𝑚′ = ∑𝑘 𝑓 (𝑥 − 𝐴)4,
4 𝑛 𝑖=1 𝑖 𝑖
and the fourth moment about the mean is given by
1
𝑚 = ∑𝑘 𝑓 (𝑥 − 𝑥)4
4 𝑛 𝑖=1 𝑖 𝑖
1 𝑘
= ∑ 𝑓 {(𝑥 − 𝐴) − (𝑥 − 𝐴)}4
𝑛 𝑖=1 𝑖 𝑖

But (𝑎 − 𝑏)4 = 𝑎4 − 4𝑎3𝑏 + 6𝑎2𝑏2 − 4𝑎𝑏3 + 𝑏4.

Letting 𝑥𝑖 − 𝐴 = 𝑎 and 𝑥 − 𝐴 = 𝑏 we have
{(𝑥𝑖 − 𝐴) − (𝑥 − 𝐴)}4 = (𝑥𝑖 − 𝐴)4 − 4(𝑥𝑖 − 𝐴)3(𝑥 − 𝐴) +
6(𝑥𝑖 − 𝐴)2(𝑥 − 𝐴)2
−4(𝑥𝑖 − 𝐴)(𝑥 − 𝐴)3 + (𝑥 − 𝐴)4
𝑘
1 − 𝐴)4 − 4(𝑥 − 𝐴)3(𝑥 − 𝐴)
∴ 𝑚 = ∑ 𝑓 {(𝑥
4 𝑖 𝑖 𝑖
𝑛
𝑖=1
+ 6(𝑥𝑖 − 𝐴)2(𝑥 − 𝐴)2 − 4(𝑥𝑖 − 𝐴)(𝑥 − 𝐴)3
1 (𝑥
+ − 𝐴)4}
= ∑𝑘 𝑓 (𝑥 − 𝐴)4 − 4(𝑥 − 𝐴) 1 𝑘 𝑓 (𝑥 − 𝐴)3
𝑖=1 𝑖 𝑖
∑𝑖=1 𝑖 𝑖
𝑛 𝑛
1
+6(𝑥 − 𝐴)2 ∑𝑘 𝑓 (𝑥 − 𝐴)2
𝑛 𝑖=1 𝑖 𝑖

36
 1
−4(𝑥 − 𝐴)3 ∑𝑘 𝑓 (𝑥 − 𝐴) + (𝑥 − 𝐴)4
𝑛 𝑖=1 𝑖 𝑖
= 𝑚′ − 4𝑚′ 𝑚′ + 6(𝑚′ )2𝑚′ − 4(𝑚′ )3𝑚′ + (𝑚′ )4
4 1 3 1 2 1 1 1
Hence,
𝒎𝟒 = 𝒎′ − 𝟒𝒎′ 𝒎′ + 𝟔(𝒎′ )𝟐𝒎′ − 𝟑(𝒎′ )𝟒 .
𝟒 𝟏 𝟑 𝟏 𝟐 𝟏
Derive similar results for 𝑚5 and 𝑚6.

Example 5.1
Calculate the first four moments about the mean for the following distribution
x 0 1 2 3 4 5 6 7 8
f 1 8 28 56 70 56 28 8 1

Solution
Let A=4. Thus, we have to compute the moments about A=4.
x f d=x - 4 fd 𝑓𝑑2 𝑓𝑑3 𝑓𝑑4
0 1 -4 -4 16 -64 256
1 8 -3 -24 72 -216 648
2 28 -2 -56 112 -224 448
3 56 -1 -56 56 -56 56
4 70 0 0 0 0 0
5 56 1 56 56 56 56
6 28 2 56 112 224 448
7 8 3 24 72 216 648
8 1 4 4 16 65 256
Total 256 0 0 512 0 2,816

Therefore,
1 0
𝑚′ = ∑𝑘 𝑓 𝑑 = =0
1 𝑛 𝑖=1 𝑖 𝑖 256

37
 1 512
𝑚′ = ∑𝑘 𝑓 𝑑2 = =2
2 𝑛 𝑖=1 𝑖 𝑖 256
1 0
𝑚′ = ∑𝑘 𝑓 𝑑3 = =0
𝑖 𝑖
3 1 𝑘𝑖=1
𝑛 256
𝑚′ = ∑ 4 2816 = 11.
=
4 𝑛 𝑖=1 𝑓𝑖 𝑑𝑖 256
Therefore, the first four moments about the mean are:

𝑚1 = 0.
𝑚2 = 𝑚′ − (𝑚′ )2 = 2 − 0 = 2.
2 1
𝑚3 = 𝑚′ − 3𝑚′ 𝑚′ + 2(𝑚′ )3 = 0 − 3(0)(2) + 2(0)
3 1 2 1

= 0.
𝑚4 = 𝑚′ − 4𝑚′ 𝑚′ + 6(𝑚′ )2𝑚′ − 3(𝑚′ )4
4 1 3 1 2 1

= 11 − 4(0)(0) + 6(0)(2) − 3(0) = 11.

Example 5.2:
The first four moments of a distribution about a value 5 of a variable are 2, 20, 40
and 50. Show that the mean of the distribution is 7. Also find the first four
moments about the mean.

Solution
𝐴 = 5, 𝑚′ = 2, 𝑚′ = 20, 𝑚′ = 40, 𝑚′ = 50
1 2 3 4
1 𝑘
𝑚′ = ∑ 𝑓 (𝑥 − 𝐴)
1 𝑖 𝑖
𝑛
𝑖=1
1
= ∑𝑘 𝑓 𝑥 − 𝐴 = 𝑥 − 𝐴.
𝑛 𝑖=1 𝑖 𝑖

∴ 𝑥 − 𝐴 = 𝑚′1 ⇒ 𝑥 − 5 = 2 ⇒ 𝑥 = 7.

38
 1
𝑚 = ∑𝑘 𝑓 (𝑥 − 𝑥) = 0.
1 𝑛 𝑖=1 𝑖 𝑖
𝑚2 = 𝑚′ − (𝑚′ )2 = 20 − 22 = 16.
2 1
𝑚3 = 𝑚′ − 3𝑚′ 𝑚′ + 2(𝑚′ )3
3 1 2 1

= 40 − 3(2)(20) + 2(2)3 = −64.

𝑚4 = 𝑚′ − 4𝑚′ 𝑚′ + 6(𝑚′ )2𝑚′ − 3(𝑚′ )4
4 1 3 1 2 1

= 50 − 4(2)(40) + 6(22)(20) − 3(24) = 162.

Assessment
1. From the following data, calculate the first four moments about the mean.
Wage (per Number of
day) Kshs workers
3 -7 4
8 -12 10
13 -17 20
18 -22 36
23 -27 16
28 -32 12
33-37 2

2. The arithmetic mean of a distribution is 5. The second and the third

moments about the mean are 20 and 140 respectively. Find the third
moment of the distribution about 10.
3. The first three moments of a distribution about the value 2 of a variable are
1, 16 and -40. Show that the mean is 3, the variance is 15 and 𝑚3 = −86.
Also show that the first three moments about A=0 are 3, 24 and 76.
4. If the first four moments of a set of numbers about 3 are equal to -2, 10, -
25 and 50. Calculate the corresponding moments about
a) The mean
b) The number 5
c) Zero .

39
 LESSON SIX
SKEWNESS AND KURTOSIS
6.1 Skewness
Skewness refers to lack of symmetry. A skewed distribution is a frequency
distribution that is asymmetric (not symmetric). When the longer tail of a
distribution extends to the right, it is said to be skewed to the right and when the
longer tail of a distribution extends to the left, it said to be skewed to the left.
When a distribution is skewed to the right, it contains a large number of relatively
low scores and a few extremely high scores. Generally, the mode is less than the
median which in turn is less than the mean.
Skewed to the right (Positive Skewness)
frequency

mode median mean variable

When a distribution is skewed to the left, it contains a large number of relatively
high scores and a few extremely low scores. Generally, the mean is less than the
median which in turn is less than the mode.

40
 Skewed to the left (Negative Skewness)
frequency

Mean median mode variable

For skewed distributions, the mean is different from the median which in turn is
different from the mode.
Variation tells us about the amount of spread while skewness tells us about the
direction of spread.

6.1.1 : Measures of Skewness

6.1.1.1 : Karl Pearson’s Coefficient of Skewness
Karl Pearson’s first coefficient of skewness is given by
𝑚𝑒𝑎𝑛−𝑚𝑜𝑑𝑒 𝑥−𝑚𝑜𝑑𝑒
𝑆𝐾𝑃 = 𝑠𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛
= … ............ (i)
𝑠
Karl Pearson’s second coefficient of skewness is given by
3(𝑚𝑒𝑎𝑛−𝑚𝑒𝑑𝑖𝑎𝑛) 3(𝑥−𝑚𝑒𝑑𝑖𝑎𝑛)
𝑆𝐾𝑃 = 𝑠𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛
= …………(ii)
𝑠

41
Expression (ii) is particularly useful for distributions with more than one mode.
Theoretically the value of the coefficient of skewness given by these formulae (i)
and (ii) can lie between -3 and +3. However, in practice the value of this
coefficient usually lies between -1 and +1.

6.1.1.2 : Bowley’s Coefficient of Skewness

This is also known as quartile coefficient of skewness and is based on quartiles.
This is given by
𝑄3+𝑄1−2𝑄2
𝑆𝐾𝐵 =
𝑄3−𝑄1
𝑄3+𝑄1−2 𝑚𝑒𝑑𝑖𝑎𝑛
=
𝑄3−𝑄1
It varies between -1 and +1. It is particularly useful when we have open-end
distributions and where extreme values are present.

6.1.1.3 : Percentile Coefficient of Skewness

This was devised by Kelly and is defined as
𝑃90+𝑃10−2𝑃50
𝑆𝐾𝐾 = 𝑃90−𝑃10
𝑃90+𝑃10−2 𝑚𝑒𝑑𝑖𝑎𝑛
= 𝑃90−𝑃10

𝐷9 + 𝐷1 − 2 𝑚𝑒𝑑𝑖𝑎𝑛
=
𝐷9 − 𝐷1
6.1.1.3 : Moment Coefficient of Skewness
The moment coefficient of skewness denoted by 𝛽1 is defined by
𝑚2
𝛽1 = 𝑚332,

where 𝑚2 is the second moment about the mean and 𝑚3 is the third moment
about the mean.

42
For perfectly symmetrical curves such as the normal curve, 𝛽1 is zero.
Measures of coefficient of skewness are used mainly for making comparisons
between two or more distributions. We can talk of slight skewness, moderate
skewness or marked skewness.
Example 6.1 The following data relate to the profits of 1,000 companies in
thousands Kenya shillings.
Profit 100- 120 – 140 - 160 - 180 - 200 - 220 -
(in 119 139 159 179 199 219 239
thousand
shillings)
Number of 17 53 199 194 327 208 2
companies

Calculate:
i) the Karl Pearson’s first coefficient of skewness and comment on its
value,
ii) Bowley’s coefficient of skewness,
iii) moment coefficient of skewness.
Solutions
Profit Class f 𝑑 𝑢 𝑓𝑢 𝑓𝑢2 𝑓𝑑 𝑓𝑑2 𝑓𝑑3
mark = 𝑥 − 169.5 𝑑
=
x 20
100- 109.5 17 -60 -3 -51 153 -1020 61200 -3672000
119
120- 129.5 53 -40 -2 -106 212 -2120 84800 -3392000
139
140- 149.5 199 -20 -1 -199 199 -3980 79600 -1592000
159
160- 169.5 194 0 0 0 0 0 0 0
179
180- 189.5 327 20 1 327 327 6540 130800 2616000
199
200- 209.5 208 40 2 416 832 8320 332800 13312000
219

43
220- 229.5 2 60 3 6 18 120 7200 432000
239
Total 1,000 393 1,741 7860 565600 7704000
𝑚𝑒𝑎𝑛−𝑚𝑜𝑑𝑒
𝑥−𝑚𝑜𝑑𝑒
i) 𝑆𝐾𝑃 = 𝑠𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛 =
𝑠
1 393
mean (𝑥) = 𝐴 + ( ∑7 𝑓 𝑢 ) 𝑐 = 169.5 + ( ) 20 = 177.36
𝑛 𝑖=1 𝑖 𝑖 1000
The modal class is 180 – 199.

𝑓𝑚−𝑓1
Mode= 𝐿 + ( )𝑐
2𝑓𝑚−𝑓1−𝑓2
327−194 133×20
= 179.5 + ( ) 20 = 179.5 +
2×327−194−208 252

= 190.06.
1 1
𝑠 = √𝑐2{ ∑7 𝑓 𝑢2 − ( ∑7 𝑓𝑢 2}

𝑛 𝑖=1 𝑖 𝑖 𝑛 𝑖=1 𝑖 𝑖)

393 2 2
2 1741 ) } = √400 × 1.5866 = 25.2.
= √20 { −(
1000 1000

Hence,
𝑚𝑒𝑎𝑛−𝑚𝑜𝑑𝑒
𝑥−𝑚𝑜𝑑𝑒
𝑆𝐾𝑃 = 𝑠𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛 =
𝑠

177.36 − 190.06
= = −0.504
25.2
This is considered moderate skewness.

𝑄3+𝑄1−2𝑄2
ii) 𝑆𝐾𝐵 =
𝑄3−𝑄1
𝑄3+𝑄1−2 𝑚𝑒𝑑𝑖𝑎𝑛
=
𝑄3−𝑄1

44
 Class 99.5 119.5 139.5 159.5 179.5 199.5 219.5 239.5
boundaries
Cf 0 17 70 269 463 790 998 1000

1000 = 250 ⇒ 𝑄 lies in the class interval 140 – 159. Therefore,

4 1

𝑛
4 − (∑ 𝑓)𝑄1 250 − 70
𝑄1 = 𝐿𝑄1 +( ) 𝑐1 = 139.5 + ( ) 20
𝑓𝑄1 199
= 139.5 + 18.09 = 157.59.
1000 = 500 ⇒ 𝑄 (median) lies in the class interval 180 – 199. Therefore,
2 2

𝑛
−(∑ 𝑓)1
Median= 𝐿 + (2 ) 𝑐 = 179.5 + (500 −463) 20
1 𝑓𝑚𝑒𝑑𝑖𝑎𝑛 327
= 179.93
3 (1000) = 750 ⇒ 𝑄 (median) lies in the class interval 180 – 199.
4 3
Therefore,

3𝑛 − (∑ 𝑓)𝑄 750 − 463

+( 4 3
𝑄3 = 𝐿𝑄3 ) 𝑐3 𝑄3 = 179.5 + ( ) 20
𝑓𝑄1 327
= 197.05.
Hence,

𝑄3+𝑄1−2𝑄2
𝑆𝐾𝐵 =
𝑄3−𝑄1
𝑄3+𝑄1−2 𝑚𝑒𝑑𝑖𝑎𝑛
=
𝑄3−𝑄1
197.05+157.59−2(179.93)
= = −0.1323.
197.05−157.59

45
 𝑚 32
iii) 𝛽1 = 𝑚32
1
𝑚′ = ∑𝑘 𝑓𝑑 7860 = 7.86
1
=
1 𝑘𝑖=1 𝑖 𝑖
𝑛 1000
𝑚′ = ∑ 2 565600 = 565.6
𝑓 𝑑 = 1000
2 𝑛1 𝑖=1 𝑖 𝑖
𝑚′ = ∑𝑘 3 7704000 = 7704
3 𝑛 𝑖=1 𝑓𝑖 𝑑𝑖 = 1000

𝑚2 = 𝑚′ − (𝑚′ )2 = 565.6 − 7.862 = 503.8

2 1

𝑚3 = 𝑚′ − 3𝑚′ 𝑚′ + 2(𝑚′ )3
3 1 2 1
= 7704 − 3(7,86)(565.6) + 2(7.86)2
= −5509.3.
Hence,

2 (−5509,3)2
𝑚3
𝛽1 = = = 0.2374.
𝑚32 503.83

6.2 : Kurtosis
Kurtosis is the degree of peaked-ness of a distribution, usually taken relative to a
normal distribution. If a curve is more peaked than the normal curve, it is called
leptokurtic; if it is more flat-topped than the normal curve it is called platykurtic
or flat topped. The normal curve is known as mesokurtic. The diagram below
illustrates the three different curves.

46
 Measures of Kurtosis

Leptokurtic

Mesokurtic
Platykurtic

6.2.1 : Measures of Kurtosis

Kurtosis is measured by the coefficient 𝛽2 or its derivative 𝛾2 given by

4 𝑚
𝛽2 = 𝑚2 , 𝛾2 = 𝛽2 − 3,
2

Where 𝑚4 is the fourth moment about the mean and 𝑚2 is the second moment
about the mean. For the normal curve (mesokurtic) 𝛽2 = 3 𝑖. 𝑒 𝛾2 = 0 . For a
curve that is more flat-topped than the normal curve (platykurtic curve) 𝛽2 <
3 𝑖. 𝑒 𝛾2 < 0 and for a curve more peaked than the normal curve (leptokurtic
curve) 𝛽2 > 3 𝑖. 𝑒 𝛾2 > 0.

47
 LESSON SEVEN
REGRESSION AND CORRELATION ANALYSIS
7.1 : Introduction

In this lesson we will discuss regression and correlation. Correlation analysis deals
with the association between two or more variables; while regression analysis
attempts to establish the nature of the relationship between variables.
7.2 : Lesson Learning Outcomes

By the end of this lesson the learner will be able to:

i. determine regression lines of two variables,
ii. find correlation coefficient between two variates,
iii. compute Spearman’s rank correlation coefficient.
7.3 : Regression

For studying the relationship between two variables X and Y, the simplest
regression equation is of the form
𝑦 = 𝑎 + 𝑏𝑥
Where a and b are constants. This equation specifies the linear regression of Y on
X. The constant b which is the slope of the above regression line, is known as the
coefficient of regression of Y on X.
7.3.1 : The least squares regression line of Y on X

The regression line of Y on X is given by the equation

𝑦 = 𝑎 + 𝑏𝑥
The constants a and b are determined by using normal equations. In this case the
normal equations are
∑ 𝑦 = 𝑛𝑎 + 𝑏 ∑ 𝑥 and ∑ 𝑥𝑦 = 𝑎 ∑ 𝑥 + 𝑏 ∑ 𝑥2.

48
7.3.2 : The least squares regression line of X on Y

The regression line of X on Y is given by the equation

𝑥 = 𝑐 + 𝑑𝑥
The constants c and d are determined by using normal equations. In this case the
normal equations are
∑ 𝑥 = 𝑛𝑐 + 𝑑 ∑ 𝑥 and ∑ 𝑥𝑦 = 𝑐 ∑ 𝑦 + 𝑑 ∑ 𝑦2.
Example 7.1
The following table shows the distribution of marks in calculus and statistics of
eight students in a certain university.
Calculus(X) 55 44 33 47 42 65 51 63
Statistics(Y) 43 52 35 38 51 43 48 50

i. Fit simple regression lines of Y on X and X on Y.

ii. Estimate the score in statistics for a student who scored 50 in calculus.
iii. Estimate the score in calculus for a student who scored 50 in statistics.
Solution

X Y XY X2 Y2
55 43 2,365 3,025 1,849
44 52 2,288 1,936 2,704
33 35 1,155 1,089 1,225
47 38 1,786 2,209 1,444
42 51 2,142 1,764 2,601
65 43 2,785 4,225 1,849
51 48 2,448 2,601 2,304
63 50 3,150 3,969 2,500
∑ 400 ∑ 360 ∑ 18,129 ∑ 20,818 ∑ 16,476

i. The regression line of Y on X is

𝑌 = 𝑎 + 𝑏𝑋
The normal equations are:

49
 ∑ 𝑌 = 𝑛𝑎 + 𝑏 ∑ 𝑋 ⇒ 360 = 8𝑎 + 400𝑏 ………………………………… (1)
and
∑ 𝑋𝑌 = 𝑎 ∑ 𝑋 + 𝑏 ∑ 𝑋2 ⇒ 18,129 = 400𝑎 + 20,818𝑏 ……………. (2)

Therefore, solving equations (1) and (2) simultaneously we have

(1) × 50 ⇒ 400𝑎 + 20,000𝑏 = 18,000 ............................................. (3)
(2) × 1 ⇒ 400𝑎 + 20,818𝑏 = 18,129.............................................. (4)
(3) − (4) we have −818𝑏 = −129
∴ 𝑏 = 0.1577 and from equation (1) we have 𝑎 = 37.115. Hence the
regression line of Y on X is
𝑌 = 37.115 + 0.1577𝑋

The regression line of X on Y is

𝑋 = 𝑐+𝑑𝑌
The normal equations are:
∑ 𝑋 = 𝑛𝑐 + 𝑑 ∑ 𝑌 ⇒ 400 = 8𝑐 + 360𝑑 ………………………………… (1)
and
∑ 𝑋𝑌 = 𝑐 ∑ 𝑌 + 𝑑 ∑ 𝑌2 ⇒ 18,129 = 360𝑐 + 16.476d ............... (2)

Therefore, solving equations (1) and (2) simultaneously we have

(1) × 45 ⇒ 360𝑐 + 16,200𝑏 = 18,000 ............................................. (3)
(2) × 1 ⇒ 360𝑐 + 16,476𝑑 = 18,129 ............................................... (4)
(3) − (4) we have −276𝑑 = −129
∴ 𝑑 = 0.4674 and from equation (1) we have 𝑐 = 28.967. Hence the
regression line of Y on X is
𝑋 = 28.967 + 0.4674𝑌

ii. The required score in statistics for a student who scored 50 in calculus is
𝑌 = 37.115 + 0.1577𝑋 = 37.115 + 0.1577(50) = 45 .

iii. The required score in calculus for a student who scored 50 in statistics is

𝑋 = 28.967 + 0.4674𝑌 = 28.967 + 0.4674(50) = 52.

50
7.4 : Correlation

In bivariate distributions there are two variates x and y. If the change in one
affects the change in other, the variables are said to be correlated, otherwise they
are said to be uncorrelated. If, increase (or decrease) in one, results in increase
(or decrease) in the other, the correlation is said to be positive otherwise
negative.

As a measure of degree of linear relationship between the variates, coefficient of

correlation is defined by
𝐶𝑜𝑣(𝑥, 𝑦)
𝑟𝑥𝑦 =
√[𝑉𝑎𝑟(𝑥) × 𝑉𝑎𝑟(𝑦)]

51
Where 𝐶𝑜𝑣(𝑥, 𝑦) denotes the covariance between variates x and y, and is given
by
1
𝐶𝑜𝑣(𝑥, 𝑦) = ∑𝑛 (𝑥 − 𝑥) (𝑦 − 𝑦) . Where 𝑥 and 𝑦are the means of x and y
𝑛 𝑖=1 𝑖 𝑖
respectively. Thus,

1 ∑𝑛 (𝑥 − 𝑥) (𝑦 − 𝑦)
𝑟𝑥𝑦 = 𝑛 𝑖=1 𝑖 𝑖
1 𝑛 1
√{[ ∑ (𝑥 − 𝑥)2] [ ∑𝑛 (𝑦 − 𝑦2) ]}
𝑛 𝑖=1 𝑖 𝑛 𝑖=1 𝑖

1 𝑛
𝑛 𝑥𝑖𝑦𝑖−
∑𝑖=1 ∑ 𝑥𝑖 ∑𝑛 𝑦 𝑖
𝑛 𝑖=1 𝑖=1
= √{[∑𝑛 2 1 2 1 2
∑𝑛 ∑𝑛 𝑦 2− ( ∑𝑛 𝑦 ) ]}
𝑖=1 𝑥 𝑖 − (
𝑛 𝑖=1
𝑥 𝑖 ) ][ 𝑖=1 𝑖 𝑛 𝑖=1 𝑖

This formula is referred to as product-moment (or Karl Pearson’s correlation

coefficient) formula for linear correlation.
7.4.1 : Properties of Correlation Coefficient

i. The correlation coefficient is numerically independent of origin

and scale.
ii. For two independent variates correlation coefficient is zero.
iii. The correlation coefficient cannot numerically exceed unity i.e.
|𝑟𝑥𝑦| ≤ 1 𝑜𝑟 − 1 ≤ 𝑟𝑥𝑦 ≤ 1.

Example 7.2
The ages (x) and systolic blood pressure (y) of 12 students are given below.
Ages in 56 42 72 36 63 47 55 49 38 42 68 60
years(x)
Blood 147 125 160 118 149 128 150 145 115 140 152 155
pressure(y)
Calculate the correlation coefficient between x and y and comment on your
solution.

52
 Solution
Let 𝑢 = 𝑥 − 52 and 𝑣 = 𝑦 − 140. Then we have the table of values as below.
𝑥 𝑦 𝑢 𝑣 𝑢2 𝑣2 𝑢𝑣
56 147 4 7 16 49 28
42 125 -10 -15 100 225 150
72 160 20 20 400 400 400
36 118 -16 -22 256 484 352
63 149 11 9 121 81 99
47 128 -5 -12 25 144 60
55 150 3 10 9 100 30
49 145 -3 5 9 25 -15
38 115 -14 -25 196 625 350
42 140 -10 0 100 0 0
68 152 16 12 256 144 192
60 155 8 15 64 225 120
4 4 1,552 2,502 1,766

1
∑(𝑢𝑣) − (∑ 𝑢)(∑ 𝑣)
𝑛
𝑟𝑥𝑦 = 𝑟𝑢𝑣 = 1
√{[∑ 𝑢2 − (∑ 𝑢)2][∑ 𝑣2 − 1 (∑ 𝑣)2]}
𝑛 𝑛
1
1766− (4)(4)
= 1
12
1
= 0.896
)2
√{[1552− ( 4 ][2502− (4)2]}
12 12

The ages and blood pressures of the students are positively correlated.
7.4.2 : Rank Correlation

7.4.2.1 : Non-repeated Ranks

Let n be the number of individuals which are ranked according to two characters
A and B. Let 𝑥 and 𝑦 be the ranks with respect to A and B respectively. Assuming
that the ranks are not repeated in either series, both 𝑥 and 𝑦 take the same
values 1, 2, …, n.

53
Then,
𝑛(𝑛+1)
∑𝑥 =∑𝑦 = 1 + 2 + 3 + ⋯+ 𝑛 =
2
and
𝑛(𝑛+1)(2𝑛+1)
∑ 𝑥2 = ∑ 𝑦2 = 12 + 22 + 32 + ⋯ + 𝑛2 =
6
1 ∑𝑥 2
𝑉𝑎𝑟(𝑥) = 𝑉𝑎𝑟(𝑦) = ∑ 𝑥2 − ( )
𝑛 𝑛
1 𝑛(𝑛+1) 2
= {𝑛(𝑛+1)(2𝑛+1) } − ( )
𝑛 6 2𝑛
2
(𝑛+1)(2𝑛+1) (𝑛+1)
=
6
− 4
2(𝑛+1)(2𝑛+1)−3(𝑛+1)2
=
12
𝑛2−1
=
12
Let d be the difference between the ranks i.e. 𝑑 = 𝑥 − 𝑦

⟹ ∑ 𝑑2 = ∑(𝑥 − 𝑦)2

= ∑{(𝑥 − 𝑥) − (𝑦 − 𝑦)}2 since 𝑥 = 𝑦

= ∑{(𝑥 − 𝑥)2 + (𝑦 − 𝑦)2 − 2(𝑥 − 𝑥)(𝑦 − 𝑦)}
= ∑(𝑥 − 𝑥)2 + ∑(𝑦 − 𝑦)2 − 2 ∑(𝑥 − 𝑥)(𝑦 − 𝑦)
Therefore, dividing both sides by 𝑛 we have
1 1 2 1 2 1
∑ 𝑑2 = ∑(𝑥 − 𝑥) + ∑(𝑦 − 𝑦) − (2) ∑(𝑥 − 𝑥)(𝑦 − 𝑦)
𝑛 𝑛 𝑛 𝑛
= 𝑉𝑎𝑟(𝑥) + 𝑉𝑎𝑟(𝑦) − 2𝐶𝑜𝑣(𝑥, 𝑦)
𝑛2−1 𝑛2−1
= + − 2𝐶𝑜𝑣(𝑥, 𝑦)
12 12

54
 𝑛2 − 1 1
∴ 2𝐶𝑜𝑣(𝑥, 𝑦) = − ∑ 𝑑2
6 𝑛
𝑛2 − 1 1
∴ 𝐶𝑜𝑣(𝑥, 𝑦) = − ∑ 𝑑2
12 2𝑛
Thus, the rank correlation coefficient between 𝑥 and 𝑦 is given by
𝐶𝑜𝑣(𝑥,𝑦)
𝑟=
√𝑉𝑎𝑟(𝑥) 𝑉𝑎𝑟(𝑦)
𝑛2−1
− 1 ∑ 𝑑2
12 2𝑛
𝑟= 𝑛2−1 𝑛2−1
√
( 12 )( )
6 ∑12𝑑2
𝑟=1−
𝑛(𝑛2−1)
.
This is also called Spearman’s rank correlation coefficient.
Example 7.3
Compute Spearman’s rank correlation coefficient from the following data and
interpret your result.
𝑥 33 61 20 19 40
𝑦 26 36 65 25 35

Solution
𝑥 𝑦 𝑅𝑥 𝑅𝑦 𝑑 = 𝑅𝑥 − 𝑅𝑦 𝑑2
33 26 3 4 -1 1
61 36 1 2 -1 1
20 65 4 1 3 9
19 25 5 5 0 0
40 35 2 3 -1 1
Total 12

55
Therefore, the Spearman’s rank correlation
2 coefficient is given by
6∑𝑑
=1− 6(12)
𝑟𝑠𝑝 =1− = 0.4
𝑛(𝑛2−1) 5(52−1)

Hence the ranks are positively correlated.

7.4.2.2 : Repeated Ranks

If any two or more individuals are bracketed equal in any classification with
respect to characteristic A and B, or if there is more than one item with the same
value in the series, then the Spearman’s formula for calculating the rank
correlation coefficient breaks down since in this case each of the variable X and Y
does not assume the values 1, 2, …, n and consequently 𝑥 ≠ 𝑦.
𝑚(𝑚2−1)
Therefore, we add the factor to ∑ 𝑑2, where m is the number of times an
12
item is repeated. This correction factor is to be added for each repeated value.
Thus, the rank correlation formula becomes
( 2 )
𝑚 𝑚 −1
6[∑ 𝑑2+∑ ]
𝑟 =1− 12

𝑠𝑝 𝑛(𝑛2−1)
Example 7.4
Obtain the rank correlation coefficient for the following data.
X 68 64 75 50 64 80 75 40 55 64
Y 62 58 68 45 81 60 68 48 50 70

Solution
In the X-series we see that the value 75 occurs 2 times. The common rank given to
these values is 2.5 which is the average of 2 and 3, the ranks which these values
would have taken if they were different. The next value 68, gets the next rank
which is 4. Again, we see that value 64 occurs thrice. The common rank given to it
is 6 which is the average of 5, 6 and 7. Similarly in the Y-series value 68 occurs
twice and its common rank is 3.5 which is the average of 3 and 4. As a result of

56
these common rankings, the formula for 𝑟𝑠𝑝 has to be corrected. To ∑ 𝑑2 we add
2−1)
∑ 𝑚(𝑚 for each value repeated, where m is the number of times a value
12
occurs. In the X-series the correction is to be applied twice, once for the value 75
which occurs twice (m =2) and then for the value 64 which occurs thrice (m = 3).
The total correction for the X-series is
1 5
2(4−1) 3(9−1) = +2 =
+
12 12 2 2
2(4−1) 1
Similarly, this correction for the Y-series is = , as the value 68 occurs
12 2
twice.
Thus,
𝑚(𝑚2−1) 5 1
6[∑ 𝑑2+∑ 12 ] 6[72+ + ]
𝑟𝑠𝑝 = 1 − =1− 2 2 = 0.545
𝑛(𝑛2−1) 10(100−1)

7.5 : Assessment

1. Obtain the equations of lines of regression lines and find an estimate of

Y which should correspond on average to X = 6.2.
X 1 2 3 4 5 6 7 8 9
Y 9 8 10 12 11 13 14 16 15

2. The coefficient of rank correlation is 0.8. If the sum of squares of the

difference in ranks is 33, find the number of individuals.
3. Calculate correlation coefficient and Spearman’s rank correlation
coefficient for the following data:

X 5 15 10 20 25 40
y 21 14 28 7 35 42

4. The table below shows the respective heights of 12 fathers and their
oldest sons in inches.

57
 Father(x) 63 63 67 64 68 62 70 66 68 67 69 71
Son(y) 68 66 68 65 69 66 68 65 71 67 68 70

Find the Spearman’s rank correlation coefficient and comment on

Your result.

References
1. Statistical Methods by S.P. Gupta
2. Applied Statistics and Probability for Engineers by D.C. Runger

58
 LESSON EIGHT
MUTUALLY EXCLUSIVE AND EXHAUSTIVE EVENTS;
AND PROBABILITY
8.1 : Introduction

In this lesson we will discuss mutually exclusive and exhaustive events, and also
probability.
8.2 : Lesson Learning Outcomes

By the end of this lesson the learner will be able to:

i. define mutually exclusive and exhaustive events,
ii. define probability of an event,
iii. find probability of an event when sampling with and without
replacement.
8.3 : Mutually Exclusive and Exhaustive Events

Two events A and B are said to be mutually exclusive if they are disjoint i.e.
if𝐴 ∩ 𝐵 = 𝜙. In other words, A and B are mutually exclusive if they cannot occur
simultaneously.
If two events A and B satisfy 𝐴 𝖴 𝐵 = 𝑆, then they are said to be exhaustive,
because the events taken together exhaust the sample space S.
In general, events 𝐸1, 𝐸2, … , 𝐸𝑘 all in the same sample space S, are said to be
mutually exclusive and exhaustive if they satisfy
𝐸𝑖 ∩ 𝐸𝑗 = 𝜙 for 𝑖 ≠ 𝑗 , 𝑖, 𝑗 = 1, 2, … , 𝑘
and
𝐸1 𝖴 𝐸2 𝖴 … 𝖴 𝐸𝑘 = 𝑆.
Example 8.1 A fair coin is tossed twice. A and B are two events such that
A= {observe at least one head}
B= {observe two heads}

59
 𝑆 = {𝐻𝐻, 𝐻𝑇, 𝑇𝐻, 𝑇𝑇} , 𝐴 = {𝐻𝐻, 𝐻𝑇, 𝑇𝐻} , 𝐵 = {𝑇𝑇}
𝐴 ∩ 𝐵 = 𝜙 and 𝐴 𝖴 𝐵 = {𝐻𝐻, 𝐻𝑇, 𝑇𝐻, 𝑇𝑇} = 𝑆.
Therefore, the events A and B are mutually exclusive and exhaustive.
8.4 : Definition of Probability

Suppose that the outcomes of a random experiment are represented by a sample

space S, then probability is a real valued function P, defined on the events of S,
which satisfies the following axioms:
i. For every event A, 0 ≤ 𝑃(𝐴) ≤ 1
ii. 𝑃(𝑆) = 1
iii. If A and B are mutually exclusive events, then
𝑃(𝐴 𝖴 𝐵) = 𝑃(𝐴) + 𝑃(𝐵)
iv. If 𝐴1, 𝐴2, … , 𝐴𝑛 is a sequence of mutually exclusive events, then
𝑃(𝐴1 𝖴 𝐴2 𝖴 … 𝖴 𝐴𝑛) = 𝑃(𝐴1) + 𝑃(𝐴2) + ⋯ + 𝑃(𝐴𝑛)
This axiom is the additive rule of probability.
Let us now consider some theorems of probability.
8.4.1 : Theorem 1

If 𝜙 is the empty set, the 𝑃(𝜙) = 0.

8.4.2 : Theorem 2

If 𝐴𝑐 is the complement of an event A, then 𝑃(𝐴𝑐) = 1 − 𝑃(𝐴)

8.4.3 : Theorem 3

If 𝐴 ⊂ 𝐵, then 𝑃(𝐴) ≤ 𝑃(𝐵)

8.4.4 : Theorem 4

If A and B are any two events in the same sample space S, then
𝑃(𝐴\𝐵) = 𝑃(𝐴) − 𝑃(𝐴 ∩ 𝐵)
8.4.5 : Theorem 5

If A and B are any two events in the same sample space S, then

60
 𝑃(𝐴 𝖴 𝐵) = 𝑃(𝐴) + 𝑃(𝐵) − 𝑃(𝐴 ∩ 𝐵)
For any events A, B and C in the same sample space S,
𝑃(𝐴 𝖴 𝐵 𝖴 𝐶) = 𝑃(𝐴) + 𝑃(𝐵) + 𝑃(𝐶) − 𝑃(𝐴 ∩ 𝐵) − 𝑃(𝐴 ∩ 𝐶) − 𝑃(𝐵 ∩ 𝐶) + 𝑃(𝐴 ∩ 𝐵 ∩ 𝐶).

8.4.6 : De Morgan’s Laws

If A and B are any events in the same sample space S, then

i. (𝐴 𝖴 𝐵)𝑐 = 𝐴𝑐 ∩ 𝐵𝑐
ii. (𝐴 ∩ 𝐵)𝑐 = 𝐴𝑐 𝖴 𝐵𝑐
Example 8.2
3 1 1
Let A and B be events with 𝑃(𝐴) = , 𝑃(𝐵) = and 𝑃(𝐴 ∩ 𝐵) = . Find
8 2 4

i. 𝑃(𝐴𝑐) and 𝑃(𝐵𝑐)

ii. 𝑃(𝐴 𝖴 𝐵)
iii. 𝑃(𝐴𝑐 ∩ 𝐵𝑐)
iv. 𝑃(𝐴𝑐 𝖴 𝐵𝑐)
v. 𝑃(𝐴 ∩ 𝐵𝑐)
vi. 𝑃(𝐴𝑐 ∩ 𝐵)

Solution
3 5 1 1
i. 𝑃(𝐴𝑐) = 1 − 𝑃(𝐴) = 1 − = and 𝑃(𝐵𝑐) = 1 − 𝑃(𝐵) = 1 − =
8 8 2 2
ii. 3 1 1 5
𝑃(𝐴 𝖴 𝐵) = 𝑃(𝐴) + 𝑃(𝐵) − 𝑃(𝐴 ∩ 𝐵) = + − =
8 2 4 8
iii. Using, De Morgan’s law we have
5 3
𝑃(𝐴𝑐 ∩ 𝐵𝑐) = 𝑃((𝐴 𝖴 𝐵)𝑐) = 1 − 𝑃(𝐴 𝖴 𝐵) = 1 − =
8 8
iv. Using, De Morgan’s law we have
1 3
𝑃(𝐴𝑐 𝖴 𝐵𝑐) = 𝑃[(𝐴 ∩ 𝐵)𝑐] = 1 − 𝑃(𝐴 ∩ 𝐵) = 1 − =
4 4
v. 3 1 1
𝑃(𝐴 ∩ 𝐵𝑐) = 𝑃(𝐴\𝐵) = 𝑃(𝐴) − 𝑃(𝐴 ∩ 𝐵) = − =
8 4 8
vi. 1 1 1
𝑃(𝐴𝑐 ∩ 𝐵) = 𝑃(𝐵\𝐴) = 𝑃(𝐵) − 𝑃(𝐴 ∩ 𝐵) = − =
2 4 4

61
 Example 8.3
If the probability that a student A will fail a certain statistics examination is 0.3,
the probability that student B will fail the examination is 0.15, and the
probability that both students A and B will fail the examination is 0.08. What is
the probability that
i. at least one of these two students will fail the examination?
ii. Neither student A nor student B will fail?

Solution
𝑃(𝐴) = 0.3 , 𝑃(𝐵) = 0.15 . 𝑃(𝐴 ∩ 𝐵) = 0.08. Hence
i. 𝑃(𝐴 𝖴 𝐵) = 𝑃(𝐴) + 𝑃(𝐵) − 𝑃(𝐴 ∩ 𝐵)
= 0.3 + 0.15 − 0.08 = 0.37
ii. 𝑃((𝐴 𝖴 𝐵)𝑐) = 1 − 𝑃(𝐴 𝖴 𝐵) = 1 − 0.37 = 0.63

Example 8.4 A fair coin is tossed three times and events A and B are defined as
follows:
A = {At least one head is observed}
B = {The number of heads observed is odd}
a) List the sample points in the events 𝐴, 𝐵, 𝐴 𝖴 𝐵, 𝐴𝑐, 𝑎𝑛𝑑 𝐴 ∩ 𝐵.
b) Find 𝑃(𝐴), 𝑃(𝐵), 𝑃(𝐴 𝖴 𝐵), 𝑃(𝐴𝑐), 𝑎𝑛𝑑 𝑃(𝐴 ∩ 𝐵).
c) Are the events A and B mutually exclusive?
Solution
a) 𝑆 = {𝐻𝐻𝐻, 𝐻𝐻𝑇, 𝐻𝑇𝐻, 𝑇𝐻𝐻, 𝐻𝑇𝑇, 𝑇𝐻𝑇, 𝑇𝑇𝐻, 𝑇𝑇𝑇} ,
𝐴 = {𝐻𝐻𝐻, 𝐻𝐻𝑇, 𝐻𝑇𝐻, 𝑇𝐻𝐻, 𝐻𝑇𝑇, 𝑇𝐻𝑇, 𝑇𝑇ℎ},
𝐵 = {𝐻𝐻𝐻, 𝐻𝑇𝑇, 𝑇𝐻𝑇, 𝑇𝑇𝐻},
𝐴 𝖴 𝐵 = {𝐻𝐻𝐻, 𝐻𝐻𝑇, 𝐻𝑇𝐻, 𝑇𝐻𝐻, 𝐻𝑇𝑇, 𝑇𝐻𝑇, 𝑇𝑇𝐻, 𝑇𝑇𝑇},
𝐴𝑐 = {𝑇𝑇𝑇}
𝐴 ∩ 𝐵 = {𝐻𝐻𝐻, 𝐻𝑇𝑇, 𝑇𝐻𝑇, 𝑇𝑇𝐻]

62
 4 7
b) (𝐴) = 7 , 𝑃(𝐵) = = 1 , 𝑃(𝐴 𝖴 𝐵) = 7 , 𝑃(𝐴𝑐) = 1 − 𝑃(𝐴) = 1 − =
8 8 2 8 8
4 1
1 , 𝑎𝑛𝑑 𝑃( 𝐴 ∩ 𝐵) = = .
8 8 2
c) 7 1
𝑃(𝐴) + 𝑃(𝐵) = + ≠, 𝑃(𝐴 𝖴 𝐵) ⇒ A and B are not mutually
8 2
exclusive.
8.4.7 : Random Sampling

8.4.7.1 : Sampling without replacement

This the case of drawing items which are equally likely from a population without
replacing them
Example 8.5
A bag contains 5 red balls and 3 black balls. If 3 balls are drawn without
replacement, what is the probability that
a) no black balls will be selected,
b) exactly one red ball will be selected,
c) at least one red ball will be selected,
d) at most one black ball will be selected?

63
 Solution

3
Tree diagram R
6

R- red ball, B- black ball R

B
6

R
R

B B
7

R
R

B
B

R
B B

a) 𝑃(𝑅𝑅𝑅) = 5 4 3 5
( 8) ( 7) ( 6) = 56

3 2 3 5 2 5
b) 𝑃(𝑅𝐵𝐵) + 𝑃(𝐵𝑅𝐵) + 𝑃(𝐵𝐵𝑅) = ( ) ( ) ( ) + ( ) ( ) ( ) +
8 7 6 8 7 6 3 2 5
( )( )( )
8 7 6
𝟏𝟓
=
𝟓𝟔
55 3 2 1
c) 1 − 𝑃(𝐵𝐵𝐵) = 1 − ( × × ) = 56
8 7 6

5
d) 𝑃(𝑅𝑅𝐵) + 𝑃(𝑅𝐵𝑅) + 𝑃(𝐵𝑅𝑅) + 𝑃(𝑅𝑅𝑅) = × 4 × 3 + 5 × 3 × 4 +
8 7 6 8 7 6

64
 3 5 4 5 4 3 5
× × + × × =
8 7 6 8 7 6 7

8.4.7.2 : Sampling with replacement

Example 8.6
Three bolts and three nuts are put in a box. If two parts are selected one at a time
at random from the box with replacement, find the probability that one is a bolt
and one is a nut.
Solution
Tree diagram
B B- bolt, N- nut
B N
B

N
N
3 3 3 3 1
𝑃(𝐵𝑁) + 𝑃𝑃(𝑁𝐵) = 6 × 6 + × =
6 6 2

7.6: Assessment
1. Three light bulbs are chosen from 15 bulbs of which 5 are defective. Find the
probability that
i. non is defective
iii. exactly one is defective
iv. at least one is defective.

65
References
1. Statistical Methods by S.P. Gupta
2. Applied Statistics and Probability for Engineers by D.C.
Runger

66
 LESSON NINE
CONDITIONAL PROBABILITY, INDEPENDENT EVENTS
AND BAYES’ THEOREM
9.1 : Introduction

In this lesson we will discuss conditional probability, independent events and

Bayes’ theorem and its application.
9.2 : Lesson Learning Outcomes

By the end of this lesson the learner will be able to:

i. determine the conditional probability of an event given another one has
occurred,
ii. state when are events independent,
iii. apply Bayes’ theorem.
9.3 : Conditional Probability

Let, E be an arbitrary event in a sample space S with 𝑃(𝐸) > 0. The probability
that an event A occurs once E has occurred or, in other words, the conditional
probability of A given E, written as 𝑃(𝐴/𝐸), is defined as follows:
𝑃(𝐴 ∩ 𝐸)
𝑃(𝐴 ∕ 𝐸) =
𝑃(𝐸)
If S is a finite equiprobable space and |𝐴| denotes the number of elements in an
event A, then
|𝐴∩𝐸| |𝐸|
𝑃(𝐴 ∩ 𝐸) = , 𝑃(𝐸) = ,
|𝑆| |𝑆|
𝑃(𝐴∩𝐸) |𝐴∩𝐸|
and also 𝑃(𝐴⁄𝐸) = = .
𝑃(𝐸) |𝐸|

Example 9.1 Suppose that a family with two children is selected at random from a
residential estate. Assume that each child has an equal chance of being a boy as

67
of being a girl. Calculate the conditional probability that both children are girls,
given that
i. the older child is a girl,
ii. at least one of the children is a girl.

Solution
b-boy, g-girl
𝑆 = {𝑏𝑏, 𝑏𝑔, 𝑔𝑏, 𝑔𝑔}
Let 𝐸1 be the event that the older child is girl. Let 𝐸2 be the event that at least one
of the children is a girl and let 𝐸3 be the event that both children are girls. Then
𝐸1 = {𝑔𝑏. 𝑔𝑔} , 𝐸2 = {𝑏𝑔, 𝑔𝑏, 𝑔𝑔} , 𝐸3 = {𝑔𝑔}
𝑃(𝐸 ∩𝐸 )
i. 𝑃(𝐸 ⁄𝐸 ) = 3 1 , but 𝐸 ∩ 𝐸 = {𝑔𝑔}
3 1 𝑃(𝐸1) 3 1
𝑃(𝐸3 ∩ 𝐸1) 41 1
∴ 𝑃(𝐸3 ⁄𝐸1 ) =
𝑃(𝐸 ) = 2 = 2
1
4
𝑃(𝐸3∩𝐸2)
ii. 𝑃(𝐸 ⁄𝐸 ) = , but 𝐸 ∩ 𝐸 = {𝑔𝑔}
3 2 𝑃(𝐸2) 3 2
𝑃(𝐸3 ∩ 𝐸2) 41 1
∴ 𝑃(𝐸3 ⁄𝐸2 ) =
𝑃(𝐸 ) = 3 = 3
2
4

1
Example 9.2 Let A and B be events with 𝑃( 𝐴) = 1 , 𝑃( 𝐵) = 1 and 𝑃( 𝐴 𝖴 𝐵) = .
3 4 2

Find
i. 𝑃(𝐴⁄𝐵),
ii. 𝑃(𝐵⁄𝐴),
iii. 𝑃(𝐴 ∕ 𝐵𝑐).

68
 Solution
𝑃(𝐴∩𝐵)
i. 𝑃(𝐴⁄𝐵) = , but
𝑃(𝐵)

𝑃(𝐴 𝖴 𝐵) = 𝑃(𝐴) + 𝑃(𝐵) − 𝑃(𝐴 ∩ 𝐵)

⇒ 𝑃(𝐴 ∩ 𝐵) = 𝑃(𝐴) + 𝑃(𝐵) − 𝑃(𝐴 𝖴 𝐵)
1 1 1 1
= + − =
3 4 2 12
1
1
∴ 𝑃(𝐴⁄𝐵) = 12 =
1 3
4
1
𝑃(𝐵∩𝐴) 1
ii. 𝑃(𝐵⁄𝐴) = = 12
1 =4
𝑃(𝐴)
3

𝑃(𝐴∩𝐵𝑐) 𝑃(𝐴\𝐵) 𝑃(𝐴)−𝑃(𝐴∩𝐵)

iii. 𝑃(𝐴⁄𝐵𝑐) = 𝑐
= =
𝑃(𝐵 ) 𝑃{𝐵𝑐) 1−𝑃(𝐵)

1 1
− 12 1
= 3
1−
1 =3
4

9.3.1 : Multiplication Rule for Conditional Probability

𝑃(𝐸1 ∕ 𝐸2 ) = 𝑃(𝐸1 ∩ 𝐸2)

𝑃(𝐸 )
2

If both sides of this equation, is multiplied by 𝑃(𝐸2) we obtain the formula

𝑃(𝐸1 ∩ 𝐸2 = 𝑃(𝐸2)𝑃(𝐸1⁄𝐸2) ,
Which is called the general multiplication rule for calculating the probability that
the two events will occur.
For any events 𝐸1, 𝐸2, … , 𝐸𝑛

69
 𝑃(𝐸1 ∩ 𝐸2 ∩ … ∩ 𝐸𝑛)
= 𝑃(𝐸1)𝑃(𝐸2⁄𝐸1)𝑃(𝐸3⁄𝐸1 ∩ 𝐸2) … 𝑃(𝐸𝑛 ∕ 𝐸1 ∩ 𝐸2 ∩ … ∩ 𝐸𝑛−1)
Example 9.3 In a certain small town, the probability that a woman attends a
family planning clinic is 0.4 and the probability that her husband attends the clinic
is 0.1. The probability that a husband attends a clinic given that the wife does is
0.08. Calculate the probability that
i. both wife and husband will attend the clinic,
ii. the wife will attend the clinic given that the husband does,
iii. one of the two persons will attend a clinic.

Solution
Let A denotes the event that the wife attends a clinic and B denotes the event
that the husband attends the same clinic. Then
𝑃(𝐴) = 0.4, 𝑃(𝐵) = 0.1 and 𝑃(𝐵⁄𝐴) = 0.08.
i. 𝑃(𝐴 ∩ 𝐵) = 𝑃(𝐴)𝑃(𝐵⁄𝐴) = (0.4)(0.08) = 0.032

𝑃(𝐴∩𝐵) 𝑜.032
ii. 𝑃(𝐴⁄𝐵) = = = 0.32
𝑃(𝐵) 0.1

iii. 𝑃(𝐴 𝖴 𝐵) = 𝑃(𝐴) + 𝑃(𝐵) − 𝑃(𝐴 ∩ 𝐵) = 0.4 + 0.1 − 0.032 = 0.468.

9.3.2 : Independence of Events

Let 𝐸1 and 𝐸2 be any two events in the sample space S. Assume that 𝑃(𝐸1) > 0
and 𝑃(𝐸2) > 0. Then we have the following definition:
The events 𝐸1 and 𝐸2 are said to be stochastically or statistically independent if
𝑃(𝐸1⁄𝐸2) = 𝑃(𝐸1) or 𝑃(𝐸2⁄𝐸1) = 𝑃(𝐸2)
If two events 𝐸1 and 𝐸2 are not independent then they are said to be dependent.
The general multiplication rule is

𝑃(𝐸1 ∩ 𝐸2 = 𝑃(𝐸2)𝑃(𝐸1⁄𝐸2),

70
It follows that the events 𝐸1 and 𝐸2 are independent if and only if
𝑃(𝐸1 ∩ 𝐸2 = 𝑃(𝐸1)𝑃(𝐸2)
Example 9.4
Two fair dice are tossed, and the following events are defined:
A = {Sum of the numbers showing is odd}
B = {Sum of the numbers showing is 9, 11, 12}
Are events A and B independent? Why?
Solution
𝑆 = {(1,1), (1,2), (1,3), … , (6,6)}
There are 36 sample elements in S.
(1,2), (1,4), (1,6), (2,1), (2,3), (2,5), (3,2), (3,4), (3,6),
𝐴={ }
(4,1), (4,3), (4,5), (5,2), (5,4), (5,6), ( 6,1), (6,3), (6,5)

𝐵 = {(3,6), (4,5), (5,4), (5,6), (6,3), (6,5), (6,6)}

𝐴 ∩ 𝐵 = {(3,6), (4,5), (5,4), (5,6), (6,3), (6,5)}.
Thus, there are 18, 7, 6 sample elements in A, B and 𝐴 ∩ 𝐵 respectively.
Therefore,
18 1 7 6 1
𝑃(𝐴) = = , 𝑃(𝐵) = and 𝑃(𝐴 ∩ 𝐵) = = .
36 2 36 36 6
1 7 7
𝑃(𝐴). 𝑃(𝐵) = ( ) ( ) = ⇒ 𝑃(𝐴). 𝑃(𝐵) ≠ 𝑃(𝐴 ∩ 𝐵)
2 36 72
Thus, the events A and B are not independent.
9.3.3 : Bayes’ Theorem

Suppose 𝐸1, 𝐸2, … , 𝐸𝑘 are k mutually exclusive events in the sample space S such
that 𝐸1 𝖴 𝐸2 𝖴 … 𝖴 𝐸𝑘 = 𝑆 and 𝑃(𝐸𝑖) > 0 for 𝑖 = 1, 2, … , 𝑘. Then some arbitrary
event B, which is associated with 𝐸𝑖′𝑠 such that 𝑃(𝐵) > 0, we can find out the
probabilities 𝑃(𝐵⁄𝐸1), 𝑃(𝐵⁄𝐸2), … , 𝑃(𝐵⁄𝐸𝑘) .

71
In Bayes’ approach we want to find out the posterior probability of an event 𝐸𝑖
given that B has occurred i.e. 𝑃(𝐸𝑖)
𝑃(𝐸𝑖∩𝐵)
𝑃(𝐸𝑖 ∕ 𝐵) = 𝑃(𝐵)
if and only if 𝑃(𝐵) > 0

We know that B is also a set in S and hence 𝐵 = 𝐵 ∩ 𝑆 or in other form

𝐵 = 𝐵 ∩ (𝐸1 𝖴 𝐸2 𝖴 … 𝖴 𝐸𝑘) and 𝐸𝑖′𝑠 are disjoint
Or 𝐵 = (𝐵 ∩ 𝐸1) 𝖴 (𝐵 ∩ 𝐸2) 𝖴 … 𝖴 (𝐵 ∩ 𝐸𝑘).
Thus,
𝑃(𝐵) = ∑𝑘𝑖=1 𝑃(𝐵⁄𝐸𝑖)𝑃(𝐵)
𝑃(𝐵 ∩ 𝐸𝑖) = 𝑃(𝐸𝑖⁄𝐵)𝑃(𝐵)
Also
𝑃(𝐵 ∩ 𝐸𝑖) = 𝑃(𝐵⁄𝐸𝑖)𝑃(𝐸𝑖)
Therefore,
𝑃(𝐵) = ∑𝑘𝑖=1 𝑃(𝐵⁄𝐸𝑖)𝑃(𝐸𝑖)
Hence,
𝑃(𝐸 ∕ 𝐵) = 𝑃(𝐵⁄𝐸𝑖)𝑃(𝐸𝑖)
𝑖 𝑘
∑𝑖=1 𝑃(𝐵⁄𝐸𝑖)𝑃(𝐸𝑖)

This formula is known as Bayes’ formula.

Example 9.5
In a city there are 3 stores each having 20 pieces of an item. Let these stores be
denoted by 𝑆1, 𝑆2 and 𝑆3. The stores 𝑆1, 𝑆2 and 𝑆3 have 10%, 20% and 30%
defective items, respectively. A costumer first chooses a store randomly and then
selects an item randomly from the store. Find the probability that
a) the selected item is defective,
b) the selected item is defective has come from
i. store 𝑆1 ,
ii. store 𝑆2 ,
iii. store 𝑆3 .

72
 Solution
Let 𝐸𝑖 denote the event that the store 𝑆𝑖 (𝑖 = 1, 2,3) is selected. Then,
1 1
𝑃(𝐸 ) = 𝑖. 𝑒. 𝑃(𝐸 ) = 𝑃(𝐸 ) = 𝑃(3) = .
𝑖 3 1 2 3

Let B denotes the event that a selected item is defective. Therefore,

𝑃(𝐵⁄𝐸1) = 10% = 0.1, 𝑃(𝐵⁄𝐸2) = 20% = 0.2, 𝑃(𝐵⁄𝐸3) = 30% = 0.3 .
a) 𝑃(𝐵) = 𝑃(𝐵⁄𝐸1)𝑃(𝐸1) + 𝑃(𝐵⁄𝐸2)𝑃(𝐸2) + 𝑃(𝐵⁄𝐸3)𝑃(𝐸3)
1 1 1 1
= (0.1) ( ) + (0.2) ( ) + (0.3) ( ) =
3 3 3 5
b)
𝑃(𝐵⁄𝐸1)𝑃(𝐸1)
i. 𝑃(𝐸1 ⁄𝐵) = 𝑃(𝐵⁄𝐸
1)𝑃(𝐸1)+𝑃(𝐵⁄𝐸2)𝑃(𝐸2)+𝑃(𝐵⁄𝐸3)𝑃(𝐸3)
1
(0.1) ( )
3 1
= 1 =
6
5

𝑃(𝐵⁄𝐸2)𝑃(𝐸2)
ii. 𝑃(𝐸2 ⁄𝐵) =
𝑃(𝐵⁄𝐸1)𝑃(𝐸1)+𝑃(𝐵⁄𝐸2)𝑃(𝐸2)+𝑃(𝐵⁄𝐸3)𝑃(𝐸3)

1
(0.2) ( )
3 1
= 1 =
3
5

𝑃(𝐵⁄𝐸3)𝑃(𝐸3)
iii. 𝑃(𝐸3 ⁄𝐵) =
𝑃(𝐵⁄𝐸1)𝑃(𝐸1)+𝑃(𝐵⁄𝐸2)𝑃(𝐸2)+𝑃(𝐵⁄𝐸3)𝑃(𝐸3)
1
(0.3) ( )
3 1
= 1 =
2
5
Check: A defective selected item belongs to one of the stores. Hence

73
 ∑3 𝑃( 𝐸 ⁄𝐵) = 1 + 1 + 1 = 1.
𝑖=1 3 6 3 2

9.4 : Assessment

1. For two events A and B, 𝑃(𝐴) = 0.4, 𝑃(𝐵) = 0.2 and 𝑃(𝐴 𝖴 𝐵) =
0.52. Find 𝑃(𝐴 ∩ 𝐵), 𝑃(𝐴⁄𝐵) and 𝑃(𝐵⁄𝐴).
2. Suppose that two balanced dice are tossed. Let A denote the event of
an odd total, B the event of a ‘1’ on the first die, and C the event of a
total of seven.
i. Are A and B independent?
ii. Are A and C independent?
iii. Are B and C independent?
3. A gift shop employs three persons to wrap the gifts. Helen who
wraps 45% of the gifts fails to remove the price tag on 2% of the
items. Jane who wraps 35% of the items fails to remove the price tag
on 5% of the items she wraps. Mary who wraps the remaining items
fails 3% of her wrapping.
a) Find the probability that a gift wrapped at this store has
a price tag on it.
b) A gift wrapped in this store was found to have a price
tag on it later on. What is the probability that this packet
was wrapped by
i. Helen,
ii. Jane.
iii. Mary.

References
3. Statistical Methods by S.P. Gupta
4. Applied Statistics and Probability for Engineers by D.C.
Runger

74
75