STANDARD LECTURE NOTES

FOR
STATISTICS FOR DIPLOMA

ke
o.
.c
es
ot
cn

QUANTITATIVE
ne

TECHNIQUES
.k
w
w
w

1
QUANTITATIVE TECHNIQUES
Course Outline

1. Introduction to Quantitative Techniques

2. Fundamentals of Mathematics and Statistics.
3. Data Collection
4. Measures Of Central Tendency
5. Measures Of Variation /Depression
6. Correlation Analysis
7. Regression Analysis
8. Index Number
9. Time Series Analysis
10. Probability
11. Sampling
12. Test of Hypothesis
13. Linear Programming

ke
o.
.c
es
ot
cn
ne
.k
w
w
w

2
 QUANTITATIVE TECHNIQUES

GENERAL OBJECTIVES

At the end of this course unit, the trainee should be able to;-
 broaden his/her knowledge in mathematical application;
 understand and appreciate the role of quantitative methods in
decision making;
 collect and organize statistical data for management;
 analyze quantitative data for management decision making;
 Apply quantitative methods in solving business problems.

Introduction
Def; quantitative techniques are those techniques which provides the
decision maker with a systematic and powerful means of analysis and help,
based on quantitative data in exploring policies for achieving pre-
determined goals Involves the use of numbers, symbols and other

ke
mathematical expressions.
They are essentially helpful in supplementing to judgment and intuition.

o.
These techniques evaluate planning factors of alternatives as and when
.c
they arise rather than prescribe courses of action. They are particularly
es
relevant to problems of complex business enterprises.
Classification of Q.T
ot
cn

a) Statistical Techniques
Are those techniques which are used in conducting the statistical inquiry
ne

concerning a certain phenomenon? They include statistical methods

.k

beginning from the collection of data till the task of interpretation of the
data collected.
w
w

b) Programming Techniques
w

Are the model building techniques used by decision maker?

Quantitative techniques

Statistical techniques programming techniques

Statistical techniques Programming techniques

Methods of collecting data Linear programming
Classification and tabulation of Decision theory
collected data Theory of games
Probability theory and sampling Simulation
Correlation and regression analysis a) monte carlo technique

3
Index numbers b) system simulation
Time series analysis queuing (waiting line) theory
Interpretation and extrapolation inventory planning
Survey techniques and methodology network analysis/ PERT
Ratio analysis integrated production model
Statistical quality control others;
Analysis of variance non-linear programming
Statistical inference and the theory of replacement
interpretation quadratic programming
Theory of attributes. Parametric programming etc.

QT and Business management

Production Management
 Selecting building site fro a plant, scheduling and controlling its
development and designing its layout

ke
 Locating within the plant and controlling movement required
materials and finished goods inventories.

o.
 Scheduling and sequencing production by adequate preventive
.c
maintenance with optimum number of operatives by proper
es
allocation of machines
 Calculating optimum product mix.
ot

Personnel Management
cn

 Optimum manpower planning

ne

 No of persons to be maintained on permanent or full time roll

 The no. of persons to be kept in work pool intended fro meeting the
.k

absenteeism.
w

 Optimum manner of sequencing and routing of personnel to a variety

w

of jobs
w

 Studying personnel recruiting procedures, accidents rates and labor

turnover
Market Management
 Where distribution and warehousing should be located the size ,
quantity to be stocked and choice of customers.
 Optimum allocation of sales budget to direct selling of promotional
expenses.
 Choice of different media of advertising and bidding strategies.
 Financial management
 Finding long range capital requirement as well as how to generate
theses requirements
 Determining optimum replacement policies
 Working out a profit plan for the firm
 Developing capital investment plans
 Estimating credit and investment risks.

4
Limitation of Q.T s
1. the inherent limitation concerning mathematical expressions
2. high costs involved in the use of QTs
3. They do not take into consideration the intangible factors i.e. non-
measurable human factors.
4. Quantitative techniques are just the tools of analysis and not the
complete decision making process.

Role of QT in business and industry

1. they provide a tool for scientific analysis
these techniques provides executives with a more precise description of the
cause and effect relationship and risks underlying the business operations in
measurable terms and this eliminates the conventional
intuitive and subjective basis on which management used to formulate their
decisions.
2. they provide solution for various business problems
Are used in the field of production, procurement, marketing, finance and

ke
other allied fields. Problems like how best can managers and executive4s
allocate available resources to various products so that in a given time the

o.
profits are maximized or costs are minimized. Is it possible for an
.c
enterprise to arrange the time and quantity of orders of its stocks such that
es
the overall profit with given resources is maximized?
3. they enable proper development of resources
ot

E.g. programmed evaluation and review techniques (PERT) enables us to

cn

determine earliest and the latest time fro each of he events and activities
and thereby helps in the identification of the critical path
ne

All these helps in deployment of resources from one activity to another to

.k

enable the project completion on time.

4. They help in minimizing waiting and servicing time.
w

The queuing theory helps management in minimizing the total waiting of

w

servicing costs. It also analyses the feasibility of adding facilities and

w

thereby helping to take correct and profitable decision.

5. They enable management to decide when to buy and how much to
buy.
The main objective of inventory planning is to achieve balance between the
costs of holding stock and benefits of holding stock. Helps in determining
when to buy and how much to buy.
6. They assist in choosing an optimum strategy.
In a competitive situation game theory helps to determine optimum
strategy which maximizes profits or minimizes loses but adopting optimum
strategy.
7. they render great help in optimum resource allocation
8. they facilitate the process of decision making
9. Through various QTs management can know the reactions of the
integrated business systems.

5
 CHAPTER ONE

FUNDAMENTALS OF MATHEMATICS AND STATISTICS

Specific Objectives
At the end of this topic, the trainee should be able to:
 Form and solve algebraic equations.
 Apply the various techniques of counting to solving management
decision problems;
 Applying set theory to business decision problems;
 Derive and apply the binomial theorem to business problems;
 Evaluate mathematical series.

ALGEBRAIC EQUATIONS

Algebra
Algebra is a branch of mathematics in which, instead of using numbers,

ke
we use letters to represent numbers.

o.
We all know that 2+3=5.
.c
es
Suppose, though, that we substitute letters for the first two numbers, so
that:
ot
cn

2=a
ne

3=b
.k

We can then write:

w
w

a+b=5
w

All that has happened is that we have replaced the numbers with
letters. However, a number is a specific quantity – e.g., 5 is more than
4, but less than 6 – whereas a letter can be used to represent any
number. Thus in the above expression, ‗a‘ could be 4 and ‗b‘ could be 1.
We only know that they are 2 and 3 respectively because we defined
them as such before.

The main consequence of this is that algebra uses general expression

and gives general results, whereas arithmetic (using numbers) uses
definite numbers and gives definite results. Arithmetic is specific
whereas algebra is general.

6
Equations
An equation is an expression with an equal sign (=)
Equations are classified into two main groups‘ linear equations and non
linear equations. Examples of linear equations are
x + 13 = 15
7x + 6 = 0

Non linear equations in the variable x are equations in which x appears in

the second or higher degrees. They include quadratic and cubic equations
amongst others. For example
5x2 + 3x + 7 = 0 (quadratic equation)

ke
2x3 + 4x2 + 3x + 8 = 0 (cubic equation)

o.
The solution of equations or the values of the variables for which the
.c
equations hold is called the roots of the equation or the solution set.
es
Solution of linear equations.
ot
cn

Supposing M, N, and P are expressions that may or may not involve

variables, then the following constitute some rules which will be useful in
ne

the solution of linear equations

.k

Rule 1: Additional rule

w

If M = N then M + P = N + P
w

Rule 2: Subtraction rule

w

If M = N, Then M – P = N – P
Rule 3: multiplication rule
If M = N and P ≠ O then M x P = N x P
Rule 4: Division rule
If P x M = N and P ≠ O
And N/P = Q Q being a raterial number then
M = N/P

Example
i. Solve 3x + 4 = - 8
y
ii. Solve =-4
3
Solutions
7
 i. 3x + 4 = –8
3x + 4 – 4 = – 8 – 4 (by subtraction rule)
3x = – 12 (simplifying)
3x 12

3 3 (by division rule)

x=–4 (simplifying)
y
ii. 3  4  3
3
y = –12 (simplifying)

ke
Solutions of inequalities
The solutions sets of inequalities frequently contain many elements. In a

o.
number of cases they contain infinite elements.
.c
es
Example
ot

Solve and graph the following inequalities

cn

x – 2 > 2; x  w (where x is a subset of w)

ne

Solution
.k

x – 2 > 2 so x – 2 + 2 > 2 + 2
w
w

Thus, x>4
w

The solution set is infinite, being all the elements in w greater than 4

0 1 2 3 4 5 6 7 8 9 10 11

Example
Solve and graph
3x – 7 < - 13;
Solution

8
3x - 7 < -13

 3x - 7 + 7 < -13 + 7

 3x < -6

3x -6
<
3 3

x < -2

….. R Line

ke
-4 -3 -2 -1 0 2 3 4

o.
.c
es
Rules for solving linear inequalities
Suppose M, M1, N, N1 and P are expressions that may or may not involve
ot

variables, then the corresponding rules for solving inequalities will be:
cn

Rule 1: Addition rule

ne

If M > N and M1> N1

.k

Then M + P > N + P and

w
w
w

M1 + P >N1+ P
Rule 2: Subtraction Rule
If M < N and M1 ≥N1
Then M – P < N – P and
M1 – P ≥N1– P
Rule 3: Multiplication rule
If M ≥N and M1 > N1 and P≠ 0
Then MP ≥NP; M1P > N1P
M (-P) ≤ N (-P) and M1 (-P) < N1(-P)
Rule 4: Division
If M > N and M1< N1 and P≠ 0
9
 Then M/P > N/P: M1/P < N1/P
M/ (-P) < N/ (-P): and M1/ (-P) > N1/ (-P)
Rule 5: Inversion Rule
If M/P ≤ N/Q where P, Q ≠ 0
M1/P > N1/Q
Then P/M ≥ Q/N and P/M1 < Q/N1
Note: The rules for solving equations are the same as those for solving
equations with one exception; when both sides of an equation is multiplied
or divided by a negative number, the inequality symbol must be reversed
(see rule 3 & Rule 4 above).

Example
Solve and graph the following:

ke
i. 7 – 2x > - 11 ;

o.
ii. –5x + 4 ≤ 2x – 10 ;
iii. –3 ≤ 2x + 1 < 7 ; .c
es
Solutions
ot

i. 7 - 2x > -11
cn
ne

-2x > -18 (subtraction rule)

.k
w

-2x -18
< (bydivision rule)
w

-2 -2
w

x<9

line Q
-3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11

10
 ii. -5x + 4  2x - 10

-7x + 4  -10 (by subtraction rule)

-7x  -14 (by subtraction rule)

x2 (by division rule)

Q
line -4 -3 -2 -1 0 1 2 3 4 5

iii. -3  2x + 1 < 7

ke
o.
-4  2x < 6 (by substraction rule)
.c
es
-2  x < 3 ( by division rule)
ot
cn

Q
ne

line -4 -3 -2 -1 0 1 2 3 4 5
.k
w
w

Linear inequalities in two variables: relations

w

An expression of the form

Y ≥ 2x – 1
Is technically called a relation. It corresponds to a function, but different
from it in that, corresponding to each value of the independent variable x,
there is more than one value of the dependent variable y
Relations can be successfully presented graphically and are of major
importance in linear programming.

Solutions of linear simultaneous equations.

Two or more equations will form a system of linear simultaneous equations

if such equations be linear in the same two or more variables.

11
For instance, the following systems of the two equations is simultaneous in
the two variables x and y.
2x + 6y = 23
4x + 7y = 10
The solution of a system of linear simultaneous equations is a set of values
of the variables which simultaneously satisfy all the equations of the
system.

Solution techniques
a) The graphical technique
The graphical technique of solving a system of linear equations consists of
drawing the graphs of the equations of the system on the same rectangular
coordinate system. The coordinates of the point of intersection of the
equations of the system would then be the solution.

ke
Example

o.
10

9 .c
es
.
8
ot

.
7
cn

.
6
.
ne

5
. (2,4)
4
.k

.
3
w

.
w

2 x + 2y = 10
.
w

2x + y = 8
1

-1 1 2 3 4 5 6 7 8 9 10 11 12 13

The above figure illustrates:

Solution by graphical method of two equations
2x + y = 8
x + 2y = 10
The system has a unique solution (2, 4) represented by the point of
intersection of the two equations.

b) The elimination technique

12
This method requires that each variable be eliminated in turn by making
the absolute value of its coefficients equal in the equations of the system
and then adding or subtracting the equations. Making the absolute values of
the coefficients equal necessitates the multiplication of each equation by
an appropriate numerical factor.
Consider the system of two equations (i) and (ii) below
2x – 3y = 8 …….. ........................... (i).
3x + 4y = -5 …….. .......................... (ii).
Step 1
Multiply (i) by 3
6x – 9y = 24 …… .................................... (iii).
Multiply (ii) By 2

ke
6x + 8y = - 10 …… .................................. (iv).

o.
Subtract (iii) from (iv).
.c
17y = -34 ……........................................ (v).
es
 y = -2
ot
cn

Step 2
ne

Multiply (i) by 4
.k

8x – 12y = 32 ……. ................................. (vi)

w
w

Multiply (ii) by 3
w

9x + 12y = -15 ….. ................................. (vii)

Add (vi) to (vii)
17x = 17 ……......................................... (viii)
 x=1
Thus x = 1, y = -2 i.e. {1,-2}

c) The substitution technique

To illustrate this technique, consider the system of two equations (i) and
(ii) reproduced below
... 2x – 3y = 8 …….. (i).
... 3x + 4y = -5 …… (ii).
13
The solution of this system can be obtained by
a) Solving one of the equations for one variable in terms of the other
variable;
b) Substituting this value into the other equation(s) thereby obtaining
an equation with one unknown only
c) Solving this equation for its single variable finally
d) Substituting this value into any one of the two original equations so
as to obtain the value of the second variable

Step 1
Solve equation (i) for variable x in terms of y
2x – 3y = 8
x= 4 + 3/2 y (iii)
Step 2

ke
Substitute this value of x into equation (ii). And obtain an equation in y only

o.
3x + 4y = -5
3 (4 + 3/2 y) + 4y = -5 .c
es
8 ½ y = - 17 ……. (iv)
ot

Step 3
cn

Solve the equation (iv). For y

ne

8½y = -17
.k

y = -2
w

Step 4
w
w

Substitute this value of y into equation (i) or (iii) and obtain the value of x
2x – 3y = 8
2x – 3(-2) = 8
x=1

Example
Solve the following by substitution method
2x + y = 8
3x – 2y = -2
Solution
Solve the first equation for y

14
 y = 8 – 2x
Substitute this value of y into the second equation and solve for x
3x – 2y = -2
3x – 2 (8-2x) = -2
x=2
Substitute this value of x into either the first or the second original
equation and solve for y
2x + y = 8
(2) (2) + y = 8
y=4

TECHNIQUES OF COUNTING

ke
o.
Permutations
This is an order arrangement of items in which the order must be strictly
observed
.c
es
ot

Example
cn

Let x, y and z be any three items. Arrange these in all possible

ne

permutations
.k

1st 2nd 3rd

w

X Y Z
w

X Z Y
w

Y X Z Six different permutations

Y Z X
Z Y X
Z X Y

NB: The above 6 permutations are the maximum one can ever obtain in a
situation where there are only 3 items but if the number of items exceeds 3
then determining the no. of permutations by outlining as done above may
be cumbersome. Therefore we use a special formula to determine such
permutations. The formula is given below

The number of permutations of ‗r‘ items taken from a sample of ‗n‘ items
n!
may be provided as nPr = where;! = factorial
n - r !
e.g.

15
 3 3!
i. P3 =
3  3!
3 21
= note; 0! = 1
0!
6
= =6
1

5 5!
ii. P3 =
5 - 3!
5 4  3 21
=
1 2
= 60

7 7!
iii. P5 =
7 - 5!
765 4  3 21
=

ke
21
5040

o.
=
2
= 2520 .c
es
Example
There are 6 contestants for the post of chairman secretary and treasurer.
ot

These positions can be filled by any of the 6. Find the possible no. of ways
cn

in which the 3 positions may be filled.

Solution
ne

Chairman Secretary Treasurer

6 5 4
.k

Therefore the no of ways of filing the three positions is 6 x 5 x 4 = 120

w

6 6!
w

P3 =
6 - 3!
w

65 4  3 21
=
3 21
720
=
6
= 120

Combinations

A combination is a group of times in which order is not important.

For a combination to hold at any given time it must comprise of the same
items but if a new item is added to the group or removed from the group
then we have a new combination

Example

16
3 items x, y and z will have 6 different permutations but only one
combination.
The following formula is usually used to determine the no. of combinations
in a given situation.
n!
n
Cr 
r ! n  r !
Example
8!
i. 8
C7 
7! 8  7 !
8! 8  7!
 
7!1! 1 7!
=8

6!
ii. C4 
6

4! 6  4  !

ke
6! 6  5  4!

o.
 
4!2! 4! 2 1
.c
es
= 15
ot

8!
C3 
cn

iii. 8

3!5!
ne

8  7  6  5!

3  2  1 5!
.k

= 56
w
w
w

Application

Example
There is a committee to be selected comprising of 5 people from a group of
5 men and 6 women. If the selection is randomly done. Find the possibility
of having the following possibilities (combinations)
i. Three men and two women
ii. At least one man and at least one woman must be in the
committee
iii. One particular man and one particular woman must not be in the
committee (one man four women)

Solution
i. The committee size = 5 people
The group size = 5m + 6w

17
 ∴ assuming no restrictions the committee can be selected in 11C5
The committee has to consist of 3m & 2w
∴ these may be selected as follows.
5
C3 × 6C2
P(committee 3m and 2w)
5
C  6C2
 311 note that this formula can be fed directly to
C5
your scientific calculator and attain a solution.

5! 6!

 3!2! 4!2!
11!
5!6!

5  4  3  2  1 6  5  4! 5  4  3  2 1 6!
  

ke
3  2  1 2  1 2  1 4! 11 10  9  8  7  6!
27
=

o.
77
.c
es
ii. P(at least one man and at least one woman must be in the
committee)
ot

The no. of possible combinations of selecting the committee without

cn

any woman = 5C5

The probability of having a committee of five men only
ne

5
C 1
 11 5 
.k

C5 462
w

The probability of having a committee of five women only

w
w

6!
6
C 5!1!
 11 5 
C5 11!
5!6!

6  5! 5!6!
 
5!1! 1110  9  8  7  6!

1
=
77
∴ P (at least one man and at least one woman)
= 1 – {P (no man) + P (no woman)}
1 1 
=1–   
 77 462

18
 =1–
6  1
462
7
=1–
462
455
=
465
iii. P(one particular man and one particular woman must not be in
the committee would be determined as follows
The group size = 5m + 6w
Committee size = 5 people

Actual groups size from which to

Select the committee = 4m + 5w
Committee = 1m + 4w

The committee may be selected in 9C5

The one man may be selected in 4C1 ways

ke
The four women may be selected in 5C4 ways
∴ P (committee of 4w1man).

o.
C  4C1
5
 49 .c
es
C5
ot

5! 4!

cn

 4!1! 1!3!
ne

9!
4!5!
.k
w

5  4! 4  3! 4!5!
  
w

1!4! 1 3! 9  8  7  6  5!
w

10
=
63

SETS THEORY
Introduction
Sets and set theory
A set is a collection of distinct objects. We may consider all the ocean in
the world to be a set with the objects being whales, sea plants, sharks,
octopus etc, similarly all the fresh water lakes in Africa can form a set.
Supposing A to be a set
A = {4, 6, 8, 13}

19
The objects in the set, that is, the integers 4, 6, 8 and 13 are referred to as
the members or elements of the set. The elements of a set can be listed in
any order. For example,
A = {4, 6, 8, 13} = {8, 4, 13, 6}
Sets are always precisely defined. Each element occurs once and only once
in a set.

The notion  is used to indicate membership of a set. ∉ represents non

membership. However, in order to represent the fact that one set is a
subject of another set, we use the notion  . A set ―S‖ is a subject of
another set ―T‖ if every element in ―S‖ is a member of ―T‖
Example
If A = {4, 6, 8, 13} then

i) 4  {4, 6, 8, 13} or 4  A; 16 ∉ A

ke
ii) {4, 8}  A; {5, 7}  A; A  A
Methods of set representation

o.
Capital letters are normally used to represent sets. However, there are two
different methods for representing members of a set:.c
es
i. The descriptive method and
ot

ii. The enumerative method

cn

The descriptive method involves the description of members of the set in

ne

such a way that one can determine the elements of the set without
difficulty.
.k

The enumerative method requires that one writes out all the members of
w

the set within the curly brackets.

w
w

For example, the set of numbers 0, 1, 2, 3, 4, 5, 6 and 7 can be

represented ass follows
P = {0, 1, 2, 3, 4, 5, 6, 7} , enumerative method
P = {X/x = 0, 1, 2…7} descriptive method
Or
P = {x/0 ≤ x ≤7} where x is an integer.

Types of sets
a) Finite and infinite sets
A set can be classified as a finite or infinite set, depending upon the
number of elements it has. A finite set has a finite number of elements
whereas an infinite set has an infinite number of elements.

20
For example, set P below has ten elements and is therefore a finite set. Set
S, on the other hand, is an infinite set since it has an infinite number of
elements.

P = {2, 4, 6…20}
S = {1, 3, 5…}
b) Universal set
The term refers to the set that contains all the elements that an analyst
wishes to study.
The notation U or ξ is generally used to denote universal sets
c) The null set or empty set
This is a set which contains no elements. It is normally designated by a

ke
Greek letter Ø, or { }.

o.
The sets Ø and {Ø} are not the same thing since the former has no elements
in it, while the later has one element in it, namely zero
.c
es
d) Equal or equivalent sets
Two sets C and D are said to be equal if every member of set C belongs to D
ot

and every member of set D also belongs to C

cn

e) Complement of a set
ne

The complement of set A is written as A΄. This set contains all those
elements of universal set which are not in A
.k

f) Intersection and union

w

B  C Denotes the intersection of B and C. it is the set containing all those

w

elements, which belong to both B and C

w

If B = {5, 8, 11, 20, 25} and C = {1, 3, 5, 7, 9, 11, 13}

Then B  C = {5, 11}
B  C = {1, 3, 5, 7, 8, 9, 11, 13, 20 25}

Set Operations and Laws

A simple way of representing sets and relations between sets is by means of
the Venn diagram. Venn diagram consists of a rectangle that represents the
universal set. Subjects of the universal set are represented by circles drawn
within the rectangle, or the universe.
Suppose that the universal set is designated by U and the sets A, B and C
are subject of U.The Venn diagram below can be used to illustrate the sets
as follows

21
 U
B
A
C

Venn diagram below representing the intersection of set A and B or A  B =

C is illustrated as follows

Intersection of sets

ke
o.
A C B
.c
es
ot
cn

Example:
You are given the universal set
ne

T = {1, 2, 3, 4, 5, 6, 7, 8}
.k

And the following subjects of the universal set:

w

A = {3, 4, 5, 6,}
w
w

B = {1, 3, 4, 7, 8}
Determine the intersection of A and B
Solution
The intersection of A and B is the subject of T, containing elements that
belong to both A and B

A  B = {3, 4, 5, 6,}  {1, 3, 4, 7, and 8}

= {3, 4}

22
 Or

T
U
5 3 1
A 6 4 7 B
8

Example
Consider the following universal set T and its subjects C, D and E
T = {0, 2, 4, 6, 8, 10, 12}
C = {4, 8,}
D = {10, 2, 0}

ke
E = {0}
Find

o.
i) DE
ii) C DE .c
es
Solution
ot

ii) D  E = {10, 2, 0}  {0} = {0}

cn
ne

T
.k
w

E
D
w
w

D  E = Shaded area
ii) C  D  E = {4, 8}  {10, 2, 0}  {0} = { } = Ø

23
 T

D
C

4; 8 E 0
2; 10

Mutually exclusive or disjointed sets

Two sets are said to be disjointed or mutually exclusive if they have no

elements in common. Sets P and R below are disjointed

ke
o.
T .c
es
ot

P R
cn
ne
.k
w

Disjointed sets are represented by a null set in this case

w

PR = Ø
w

The union of sets

Venn diagram representing the union of sets A and B or A  B = Shaded area
is illustrated below;-

AA
 B = Shaded area B

24
Example
Consider the universal set T and its subsets A, B and C below:
T = {a, b, c, d e, f}
A = {a, d}
B = {b, c, f}
C = {a, c, e, f}

Find
i) A B
ii) AC
iii) B C
iv) A B C

Solution
i) A  B = {a, d}  {b, c, f} = {a, b, c, d, f}

ke
ii) A  C = {a, d}  {a, c, e, f} = {a, c, d, e, f}
B  C = {b, c, f}  {a, c, e, f} = {a, b, c, e, f}

o.
iii)
A  B  C = {a, d}  {b, c, f }  {a, c, e, f} = {a, b, c, d, e, f} =
iv)
T
.c
es
ot
cn

Complement of a set
Venn diagram representing the complement of a set say A represented by A1
ne

is illustrated below.
.k
w

U
w
w

A1

A1= shaded area

Example
For the universal set T = {1, 2, 3, 4, 5} and its subset A ={2, 3} and B ={5, }

Find
ii) A1

25
 iii) (A1)1
iv) (B1)1

Solution
i) A1 ={2, 3}1 = {1, 4, 5}
ii) (A1)1 =({2, 3}1)1 = {1, 4, 5}1={2, 3} = A
iii) (B1)1=({5}1)1 = {1, 2, 3, 4}1={5} = B

Laws of Set Algebra

From the following Venn diagram where T is the universal set and A its
subset, we can deduce a number of laws.

ke
A

o.
.c
es
i) AØ = A
ii) AT = T
ot

iii) AA = A
cn

iv) A  A= A
v) A  T= A
ne

vi) A  A1 = T
A  A1=Ø
.k

vii)
viii) (A1)1 = A
w
w
w

Applications

Example 1
Of the 20 girls in a form, 16 play hockey 12 play tennis and 4 play
basketball. Every girl plays at least one game and two plays all the three.
How many play two and only two games.
Solution

26
 N (ξ) = 20

n (T) = 12
n (H) = 16

y 12 – y –z – 2
16 – x – y – 2

2
x z

4 –x –z –2

n (B) = 4

ke
o.
H = Hockey T = Tennis B = Basketball
.c
es
Those who play two and only two games = x + y + z
ot

Using the diagram

cn

(14 – x – y) + (10 – y – z) + (2 – x – z) + 2 + x + y+ z = 20
ne

28 –x –y –z = 20
x+y+z=8
.k
w

Example 2
w

250 members of a certain society have voted to elect a new chairman. Each
w

member may vote for either one or two candidates. The candidate elected
is the one who polls most votes
Three candidates x, y z stood for election and when the votes were
counted, it was found that
- 59 voted for y only, 37 voted for z only
- 12 voted for x and y, 14 voted for x and z
- 147 voted for either x or y or both x and y but not for z
- 102 voted for y or z or both but not for x

Required
i) How may voters did not vote
ii) How many voters voted for x only
iii) Who won the elections

Solution

27
 N (ξ) = 250

X Tail end Y
P

14 Q

37

ke
o.
P + 12 + 59 = 147 giving P = 76
Q + 59 + 37 = 102 giving Q = 6
.c
es
ot

i) Those who did not vote

= 250 – (76 + 12 + 14 + 59 + 6 + 37)
cn

= 250 – 204 = 46
ne

ii) x = 76 + 12 + 14 = 102
.k

y = 12 + 59 + 6 = 77
w

z = 37 + 14 + 6 = 57
w
w

iii) X won the election

BINOMIAL THEOREM

PROGRESSION

28
 CHAPTER TWO

DATA COLLECTION AND PRESENTATION

Specific objectives
At the end of this topic the trainee should be able to:
 Discuss the basic consideration for data collection.
 Classify collected data into various categories.
 Tabulate collected data.
 Diagrammatically and graphically present data.

Introduction

a) Statistics
Statistics viewed as a subject is a process of collecting, tabulating and
analyzing numerical data upon which significant conclusions are drawn.

ke
Statistics may also be defined as numerical data, which has been, collected
from a given source and for a particular purpose e.g. population statistics

o.
from the ministry of planning, Agricultural statistics from the ministry of
Agriculture .c
es
Statistics may also refer to the values, which have been obtained from
statistical calculations e.g. the mean, mode, range e.t.c.
ot
cn

b) Application of statistics
ne

1. Quality Control
.k

Usually there is a quality control departments in every industry which is

charged with the responsibility of ensuring that the products made do meet
w

the customers standards e.g. the Kenya bureau of standards (KeBS) is one of
w

the national institutions which on behalf of the government inspects the

w

various products to ensure that they do meet the customers specification.

The KeBS together with other control department have developed quality
control charts. They use these charts to check whether the products are up
to standards or not.

2. Statistics may be used in making or ordering economic order quantities

(EOQ).
It is important for a business manager to realize that it is an economic cost
if one orders a large quantity of items which have to be stored for too long
before they are sold. This is because the large stock holds a lot of capital
which could otherwise be used in buying other items for sale.
It is also important to realize that the longer the items are stored in the
stores the more will be the storage costs
On the other hand if one orders a few items for sale he will incur relatively
low storage expenses but may not be able to satisfy all the clients. These

29
may lose their customers if the goods are out of stock. Therefore it is
advisable to work out the EOQ which will be sufficient for the clients in a
certain period before delivery.
The EOQ will also ensure that minimal costs are incurred in terms of storage

3. Forecasting
Statistics is very important for business managers when predicting the
future of a business for example if a given business situation involves a
dependent and independent variables one can develop an equation which
can be used to predict the output under certain given conditions.

4. Human resource management

Statistics may be used in efficient use of human resources for example we
may give questionnaires to workers to find out where the management is
weak
By compiling the statistics of those who were signing it may be found useful
to analyze such data to establish the causes of resignation thus whether it

ke
is due to frustration or by choice.

o.
BASICS FOR DATA COLLECTION
.c
es
Introduction
A statistical investigation involves a number of stages:
ot

 Definition of the problem or issue;

cn

 Collection of relevant data;

 Classification and analysis of the collected data;
ne

 Presentation of the results.

.k

Even before the collection of data starts, then, there are some important
points to consider when planning a statistical investigation.
w
w

Preliminary considerations.
w

It is important to be aware of these issues as they impact on the data which

is to be collected and analyzed.
 Exact definition of the problem
This is necessary in order to ensure that nothing important is omitted
from the enquiry, and effort is not wasted by collecting irrelevant
data. The problem as originally put to the statistician is often of a
very general type and it needs to be specified precisely before work
can begin.

 Definition of the units

The results must appear in comparable units for any analysis to be
valid. If the analysis is going to involve comparisons, then the data
must all be in the same units. It is no use just asking for ‗output‘
from several factories – some may give their answers in numbers of

30
 items, some in weigh of items, some in number of inspected batches
and so on.
 Scope of the enquiry.
No investigation should be got under way without defining the field
to be covered. Are we interested in all departments of our business,
or only some? Are we to concern ourselves with our own business
only, or with others of the same kind?

 Accuracy of the data

To what degree of accuracy is data to be recorded? For example, are
ages of individuals to be given to the nearest year or to the nearest
month or as the number of completed years? If some of the data is to
come from measurements, then the accuracy of the measuring
instrument will determine the accuracy of the results. The degree of
precision required in an estimate might affect the amount of data we
need to collect. In general, the more precisely we wish to estimate a
value, the more readings we need to take.

ke
Data sources and types

o.
Data constitute the foundation of statistical analysis and interpretation.
Data can be obtained from three sources namely; .c
es
 Primary source
 Secondary source
ot

 Internal records
cn
ne

Primary data
Primary data are measurement, observed and recorded as part of an
.k

originals study. There are two basic methods of obtaining primary data
w

namely;
w

 Questionnaires
 Observation
w

 Interviews
 Sampling
Primary data is data which is both original and has been obtained in order
to solve the specific problem in hand. Primary data is, therefore, raw data
and has to be classified and processed using appropriate statistical methods
in order to reach a solution to the problem.

Secondary sources
Secondary data can be obtained from journals, reports, and government
publications, publications of research organization, trade and professional
bodies.
Secondary data must be used with utmost care, before using secondary data
the investigator should examine the following;
1. Whether the data are suitable for the purpose of investigation.
2. whether the data are adequate for the purpose of investigation
31
 3. Whether the data are reliable.
Secondary data is any data other than primary data. Thus, it includes any
data which has been subject to the processes of classification or tabulation
or which has resulted from the application of statistical methods to primary
data, and all published statistics.

Internal data
Internal data refers to the measurement that are the by products of routine
businesses records keeping like accounting , finance, production personnel,
quality control , sales , R&D.
Since internal data originate within the business collecting the desired
information does not usually offer much difficult. The particular procedure
depends largely upon the nature of the facts being collected and the form
in which they exist.

Data collection methods and limitations

The methods usually available are as follows:

ke
 Questionnaires
 Observation

o.
 Interview
 Use of published statistics .c
es
ot

a. Questionnaire
cn

As the name suggest is distinguished by the fact that data are collected by
asking questions from people who are thought to have the desired
ne

information.
.k

A formal list of such questions is called a questionnaire

Questionnaire refers to a device for securing answers to questions by using
w

a form which the respondent fills in it.

w

b. Observation
w

The investigator observes the object or action in which he is interested.

Sometime individual makes the observation on other occasions mechanical
device observe and record the desired information.
Observation method does not automatically produce accurate data. Physical
difficulties n the observation situation on the part of the observer may
result in errors.

Classification of Data
Classification is grouping of related facts into different classes. Facts in one
class differ from another class with respect to some characteristics called a
basis of classification.
Sorting facts on one basis of classification and then another basis is called
cross-classification

Rules of Classification

32
 1. the number of classes should preferably be between 5 and 15
2. as far as possible one should avoid odd values of class intervals
3. the starting point i.e. the lower limit of the first class should either
be zero or 5 or multiple of 5
4. to ensure continuity and to get correct interval we should adopt
exclusive method of classification
5. Whenever possible all classes should be of the same size.

Types of classification

 Geographical
 Chronological
 Qualitative
 Quantitative

Geographical classification; in geographical classification data are classified

on the basis of geographical or location difference between the various

ke
items.
Chronological classification; when data are observed over a period of time,

o.
the type of classification is known as chronological classification.
.c
es
Qualitative classification; in qualitative classification data re classified on
the basis of some attributes or quality such as sex, color of hair, literacy,
ot

religion.
cn

QUALITATIVE ANALYSIS DATA COLLECTION LIMITATION OF STATISTICS.

ne

Despite the usefulness of statistics in many fields, impression should not be

.k

carried that statistics are like magical devices which always provide the
correct solution of problems . unless the data are properly collected and
w

critically interpreted there is every likelihood of drawing wrong

w

conclusion. Therefore it is necessary to know the limitation and the possible

w

misuse of statistics.
The following are the important limitation of the science of statistics:
- Statistics does not deal with isolated measurement.
- Statistics deals only with quantitative characteristic i.e quantitative
characteristic such as honesty , efficiency , intelligence, blindness
and deafness cannot be studied directly
- Statistics result are true only an average.
- Statistic is only a means not a end.
- Statistics can be misused.

Classification functions.
Classification of data is a function very similar to text of sorting letters in a
post office. It is well known that the letter collected in a post office are
sorted into different geographical basis i.e in accordance with their

33
destination as Nairobi , Mombasa , kampala etc .they are then put in
separate bags each containing letters with a common characteristic , or
having the same destination.
Classification of statistical data is comparable to the sorting operation, the
process of classification gives prominence to important information
gathered while dropping unnecessary details facilitates comparison and
enables a statistical treatment of the material collected.

Formation of frequency distribution.

Here we just count the number of times a particular value is repeated
which is called the frequency. In order to facilitate counting prepare a
column of ‗‘tally‘‘ in another column all possible values of variables from
the lowest to the highest. Then put a bar( vertical line ) opposite the
particular value to which it relates.
To facilitate counting blocks of five bars are prepared and some space is
left in between each block.
We finally count the number of blocks and bars corresponding to each value

ke
to each value of the variable and place it in the column and frequency.

o.
Example .
.c
Construct a frequency distribution from the following data.
es
23, 30,20,26,30,30,20,23,40,40,26,20,23,40,28,26,30,40,28,28,30.
ot
cn
ne
.k

class tally frequency

w

20 111 3
w

23 1111 4
w

26 111 3
28 111 3
30 1111 5
40 1111 4
total 22

Classification according to class interval.

The following technical terms are important when data are classified
according to class intervals:
(a) Class limit.these are the lowest and the highest values that can be
included in the class e.g take the class 20-40.the lowest value of this
class is 20 and the highest is 40.

34
 The two boundaries of a class is the value below which there can be
no value in that class, while the upper limit of a class is the value
above which no value can belong to that class.
The class 70- 89 .70 is the lower limit and 80 is the upper limit i.e in
this class there can be no value which is less than 70 or more than
89.

(b) Classs interval .it is the span of a class i.e the difference between
the upper limit and the lower limit is known as class interval , for
example in the class 20-40 , the class interval is 20 (i.e 40 – 20) the
size of the class interval is dertemined by the number of classes and
the total range in the data.
(c) Class frequency . this is the number of observation corresponding to
the particular class.it is also known s the frequency of that class or
the class frequency .if we add together the frequency of all
individual classes , we obtain the total frequency.
(d) Class mid – point.it is the value lying half way between the lower and

ke
the upper class limit of a class interval.
Mid point of a class is ascertained as follows:

o.
.c
es
Mid point of a class = upper limit of the class + lower limit of the
class
ot

2
cn

There are two methods of classifying the data according to class

interval namely
ne

- Exclusive method.
.k

- Inclusive method.
(a) Exclusive method.
w

When the class interval are so fixed that the upper limit of one class is
w

the lower limit of the next class, it is known as the exclusive method of
w

classification .this can be illustrated as follow.

Class interval No of items

10-20 5
20-30 3
30-40 4
40-50 6
50-60 2
60-70 1

It is clear that the exclusive method ensures continuity of data in as much

as the upper limit of one class is the lower limit of the next class.

35
Whenever this method is used it is always assumed that the upper limit is
exclusive i.e the observation exactly equals to the upper limit is not
included in that class.

(b) Inclusive method.

Under this method the upper limit of one class is included in that class
itself.this can be illustrated as below:

Class interval frequency

1-10 2
11-20 5
21-30 4
31-40 10
41-50 15
51-60 30
61-70 12

ke
71-80 3

o.
81-90 2
.c
es
Whenever inclusive method is used for equal class interval is obtained by
ot

taking the difference between the two upper limit or lower limit.
cn

DATA TABULATION.
ne

- a table is a systematic arrangements of statistical data in column and

rows.
.k

- Rows are horizontal arrangements whereas columns are vertical

w

ones.
w

- The purpose of a table is to simplify the presentation and to

w

facilitate comparisons.
- The simplication result from the clear cut and systematic
arrangements which enables the reader to quickly locate desired
information.

Parts of a table.
- The various parts of a table may vary from case to case depending
upon the given data, but a good table must contain atleast te
following parts:
(a) Table number
(b) Title of the table
(c) Caption
(d) Stub
(e) Body of the table
(f) Headnote
(g) Footnote
36
 (a) table number. each table should be numbered.
(b) Title of the table. every table must have a suitable title.
(c) Caption.it refers to the column heading .it represent what the
column represent.
(d) Stub .are the designation of the rows or row heading .they are at
the extreme left and perform the same function for the horizontal
rows or numbers I the table as the Column heading do for the
horizontal rows or numbers in the table as the column heading do for
the vertical column or numbers.
(e) Body of the table. It contains the numerical information .data
presented in the body arranged according to description are
classification of the captions and stubs.
(f) Head note. It is a brief explanatory statement applying to all or a
major part of the material in the table and is placed below the tittle
entered and closed in brackets.

ke
(g) Footnote .they are placed directly below the body of the table.

o.
DIAGRAMATIC AND GRAPHICAL PRESENTATION . .c
es
Diagrams.
Rules for construction of diagrams.
ot

- There must be the title of the diagram .

cn

- Proper proportion between width and height should be maintained.

- The scale selected should be appropriate
ne

- If necessary footnotes should be given at the bottom of the diagram.

.k

- Diagram should be absolutely neat and clean

- An index illustration different types of lines or different shades ,
w

column , should be give so that the reader can easily make out the
w

meaning of the diagram.

w

- Diagram should be as simple as possible.

Types of diagrams.
- One dimensional diagram e.g bar diagram.
- Two dimensitional diagram e.g squares .
- Picto grams and cartograms.

One dimensional diagrams.

- Bar diagrams are the most common type of diagram used in practice
, a bar is a thick line whose width is shown merely for attention.
- They are called one dimensional because it is only the lengh of the
bar that matters and not the width.

37
Merits of bar diagrams.
- They are readily understood even by those unaccustomed to reading
charts or those who are not chart minded.
- They posses the outstanding advantage that they are the simplest
and the easiest to make.
- When a large number of observation are to be compared they are the
only form that can be used effectively..

Points to mind when constructing bar diagrams.

- The width of the bars should be uniform throughout the diagram.
- The gap between one bar and another should be uniform throughout.
- Bars may either horizontal or vertical .the vertical bars should be
preferred because they give a better look an also facilitate
comparison.
- While constructing the bar diagram it is desirable to write the

ke
respective figure at the end of each bar so that the reader can know
the precise value without looking at the scale.

o.
.c
es
Types of bar diagrams.
(a) Simple bar charts.
ot

(b) Sub divided or component bar charts.

cn

(c) Multiple bar charts

(d) Percentage component barcharts.
ne
.k

(a) Simple bar charts.

w

It is used to represent only one variable i.e show s only totals.

w

However an important limitation of such diagrams is that they can prefer

w

only one classification or one category of data e.g for example while
presenting the population for the last five decade , one can only depict the
total population in the simple bar diagram and not its sex- wise
distribution.

(b) Component bar charts.

These diagrams are used to represent various parts of the total, while
constructing such charts, the various components in each bar should be
kept in the same order.
A common and helpful arrangement is that of presenting each bar in the
order of magnitude from the largest component at the base of the bar to
the smallest at the end.
To distinguish between the different component, it is useful to use
different shade or colours.
They usually show component totals and the totals.

38
 (c) multiple bar charts
in multiple bar chart two or more set of interrelated data are represented
.the technigue of drawing such a diagram is the same as that of simple bar
charts.
The only difference is that since more than one phenomenon is represented
,different shades , colours .,dots or crossing are used to distinguish
between the bars. they normally show the component totals only.

(d) Percentage component bar charts.

They are particularly useful in statistical work which require the portrayal
of relative changes in data.
When such diagrams are prepared , the length of the bars is kept equal to
100 and segment are cut in these bars to represent the component of an
aggregate.

Illustration.

ke
Using the data below , construct the following charts.
(a) Simple bar chart.

o.
(b) Component bar chart.
(c) Multiple bar charts. .c
es
(d) Percentage component bar chart.
ot
cn

Shows the sales of Xyz Ltd.

ne
.k

year biscuits bread cake total

w

1995 200 150 300 650

1996 150 200 250 600
w

1997 250 250 300 700

w

1998 300 300 200 750

1999 100 200 100 400
2000 200 150 50 400

Solution
(a) Component chart

39
40
b) simple chart
w
w
w
.k
ne
cn
ot
es
.c
o.
ke
 TWO DIMENSITIONAL DIAGRAMS.
As distinguished from one dimensional diagram in which only the length
of the bars is taken into account in two dimension diagram, the length
as well as the width of the bars is considered.
Thus the area of the bar represents the given data. Two dimensional
diagrams are also known as surface diagrams example of this are-
- Rectangles
- Squares
- Circles

Pie chart or diagram

This type of diagram enables as to show the partitioning of a total into
component parts.
In constructing a pie chart , the steps involved are:
- Prepare the data so that the various components values are in
percentage by applying

ke
% component = x x 360
Total

o.
.c
es
- Draw a circle of appropriate size with a compass.
- Measure points on the circle representing the size of each component
ot

with the help of a protractor.

cn
ne
.k
w
w
w

41
 Pie chart of the above.

ke
o.
.c
es
ot
cn

Quantitative classification; it refers to the classification of data according

to some characteristics that can be measured such as height, weight,
ne

income, sales.
.k
w

Tabulation of Data
w

One of the simplest and most revealing devices of summarizing data and
w

presenting them in meaningful fashion is the statistical tables.

Types of tabulation
Tabulation may be classified as:
1. simple tabulation
2. complex or matrix tabulation

Simple tabulation only one characteristic is shown hence this type of table
is also known as one –path table. This can be illustrated as follows:

42
Table1: Distribution of workers by workshop

Workshop Number Of Employees

A 600
B 360
C 660
D 840
E 540
Total 3,000
N/B: simple tabulation does not tell us very much – although it may be
enough for the question of the moment.

Complex tabulation has two or more characteristics are shown i.e. two or
more aspects of a problem are dealt with at the same time. It is some times
called two path tables. Such tables show two characteristics and are
formed when either the stub or the caption is divided into two coordinate

ke
parts.

o.
Age (in years) Employees total
Males females .c
es
Bellow 25 32 18 50
25-35 40 27 67
ot

35-45 25 18 43
cn

45-55 10 5 15
55 and above 5 - 5
ne

Total 112 68 180

.k
w

Presentation of data
w

After the data has been collected the next step is to present them in a
w

suitable form.

Charting data

One of the most convincing and appealing ways in which data may be
presented is through charts
A chart can take the shape of either a diagram or a graph.

Types of diagrams
 One dimensional diagram e.g. bar graphs
 Two dimensional diagram e.g. rectangles squares
 Pictograms and cartograms (circles)

One dimensional diagram

Bar diagrams is the most common type of diagram used in practice.

43
Merits of bar diagrams
 They‘re readily understood
 They are simplest and easiest to make
 They are effective if the large number of observation is to be
compared.
Example

Two dimensional diagrams

As distinguished from one dimensional diagrams in which only the length of
the bar is taken into account in two dimensional diagram the length as well
as the width of the bars is considered. Thus the area of the bar represents
the given data.

Pictograms
Also known as picture grams are very popularly used in presenting situation

ke
data. They are no abstract presentation such as lines or bars
When pictograms are used data are represented through a pictorial symbol

o.
that is carefully selected.
Illustration .c
es
The following table gives the production of tea in India by a leading
company.
ot
cn

Year Production ( million kgs)

2004 421
ne

2005 561
.k

2006 587
2007 645
w

2008 660
w
w

Solution for representing the above data by a pictogram we will use the
symbol of a star

Pictogram
Year Production of tea
2004 ◊◊◊◊
2005 ◊◊◊◊◊◊
2006 ◊◊◊◊◊◊
2007 ◊◊◊◊◊◊◊
2008 ◊◊◊◊◊◊◊

44
Merits of diagrams
 Pictograms have a greater attraction thus stimulate interest in the
information being presented
 Facts portrayed in pictorial form are generally remembered longer.
Limitations
 They are difficult to construct

Graphs
A large variety of graphs are used in practice. Graphs can be divided under
the following headings
 Time series.
 Z- Charts.
 Scatter graphs
 Semi- logarithmic graphs
 Graphs of frequency distribution

Graphs of time series

ke
When we observe the values of a variable or different point of time the
series forms is known as time series. The technique of graphic presentation

o.
is extremely helpful in analyzing change at different point of time.
.c
Graphs of time series can be constructed either on a natural scale or on a
es
ratio scale.
ot

Graphs of one variable

cn

When one variable is to be represented on x-axis measure time and ion y-

axis in that various points are plotted and joined by one straight line, the
ne

fluctuation of this line show the variation in the variable

.k

Illustration
Represent the following data by suitable graph
w

Year Imports(in million tones)

w

2003-04 1.5
w

2004-05 2.5
2005-06 2.0
2006-07 2.7
Solution
Draw a graph
Graphs of frequency distribution
A frequency distribution can be presented graphically in any of the
following ways;
Histogram
Frequency polygon
Smoothes frequency curve
Cumulative frequency curve
Histogram

45
Histograms are a set of vertical bars whose areas are proportional to the
frequencies represent3ed. Histogram is most widely used for graphical
presentation of a frequency distribution.
Illustration
Represent the following data by a histogram
Marks number of students
0-10 8
10-20 12
20-30 22
30-40 35
40-50 40
Solution
Draw a histogram
Frequency polygon
A frequency polygon is a graph of frequency distribution it has more than
four sides
It is particularly effective I comparing two or more frequency distribution.

ke
Illustration
The daily profits (in thousand rupees) of 100 shops are distributed as

o.
follows
Daily profit No of shops .c
es
0-50 12
50-100 18
ot

100-150 21
cn

150-200 20
200-250 17
ne
.k

Solution
w

The frequency polygon of the above data are shown bellow

w

Draw graph
w

Smoothed frequency curve

The following points should be kept in mind while smoothing a frequency
graph.
Only frequency distribution based on sample should be smoothed
Only continuous series should be smoothed
The total area under the curve should be equal to the area under the
original histogram or polygon.
Illustration
Draw a histogram frequency polygon and frequency curve representing the
following information

Length of service (in years) Number of employees

46
5-10 5
10-15 12
15-20 25
20-25 48
25-30 32
30-35 6
1
Histogram frequency polygon and curve
Draw
Cumulative frequency curve / ogives
Sometimes one needs to know the answers to questions like how May
workers of a factory earn more the Rs 1500 per month
There are two methods of constructing a cumulative frequency curve
namely;
The less than method
The more than method
Less than method; in the less than method we start with the upper limit of

ke
the classes and go on adding the frequencies. When these frequencies are
plotted we get a rising curve.

o.
More than method; in the more than method, we start with the upper limits
.c
of the classes and from the total frequencies we subtract the frequency of
es
each class. When the frequencies are plotted we get a declining curve.
given by less than method
ot

Yearly profit(Rs) Frequency

cn

Less than850 21
Less than 900 50
ne

Less than 950 69

.k

Less than 1000 108

Less than1050 151
w

Less than 1100 245

w

Less than 1150 318

w

Less than 1200 422

Less than 1250 467
Less than 1300 494
Less than 1350 542
Less than 1400 563
Less than 1450 575
Less than 1500 580

Ogive by more than method

Yearly Rs Frequency

47
More than 800 580
More than 850 559
More than 900 530
More than 950 511
More than 1000 472
More than 1050 429
More than 1100 335
More than 1150 262
More than 1200 194
More than 1250 158
More than 1300 113
More than 1350 86
More than 1400 38
More than 1450 17
More than 1500 5

Less than and more than ogive and median

ke
Draw
Limitation of charts

o.
 They can present only approximate values
.c
 They can appropriately represent only limited amount of information
es
 They are intended mostly to explain quantitative facts to the general
public.
ot

 They can be easily misinterpreted

cn
ne
.k
w
w
w

48
 CHAPTER THREE

MEASURES OF CENTRAL TENDENCY

Specific Objectives
At the end of this topic the trainee should be able to:
 Define the measures of central tendency;
 State the properties of the measure of central tendency;
 Determine the measures of central tendency.

Introduction
Measures of central tendency show the tendency of some central value
around which data tends to cluster. These are statistical values which tend
to occur at the centre of any well ordered set of data. Whenever these
measures occur they do not indicate the centre of that data

ke
Objectives of averaging
 To get one single value that describes the characteristics of the

o.
entire data.
 To facilitate comparison. .c
es
Properties of Good Measure
ot

 It should be easy to understand i.e. since statistical methods are

cn

designed to simplify complexity, it is described that an average be

such that can be readily understood, its use is bound to be very
ne

limited.
.k

 Should be simple to compute.

 Should be based on all observation.
w

 Should be rigidly defined, i.e. an average should be properly defined

w

so that it has one and only one interpretation.

w

 Should be capable of further algebraic treatment.

 Should have sampling stability.
 Should not be unduly affected by the presence of extreme values.

N/B. The following statistical terms are commonly used in statistical

calculations. They must therefore be clearly understood.

i) Class limits
These are numerical values which limits the extend of a given class i.e. all
the observations in a given class are expected to fall within the interval
which is bounded by the class limits e.g. 15 & 19 are class limits as in the
table of the example above.

49
ii) Class boundaries
These are statistical boundaries, which separate one class from the other.
They are usually determined by adding the lower class limit to the next
upper class limit and dividing by 2 e.g. in the above table the class
19 + 20
boundary between 19 and 20 is 19.5 which is = .
2

iii) Class mid points

These are very important values which mark the center of a given class.
They are obtained by adding together the two limits of a given class and
dividing the result by 2.

iv) Class interval/width

This is the difference between an upper class boundary and lower class

ke
boundary. The value usually measures the length of a given class.

o.
The following are important measures of central tendency
 Arithmetic mean .c
es
 Median
 Mode
ot

 Geometric mean
cn

 Harmonic
ne

Arithmetic Mean
.k

The most popular and widely used measure for representing the entire data
by one value is an average. Its value is contained by adding together all the
w

observation and dividing this total by the number of observation

w

These is commonly known as average or mean it is obtained by first of all

w

summing up the values given and by dividing the total value by the total no.
of observations.
X
I.e. mean =
n
Where x = no. of values
∑ = summation
n = no of observations
Example
The mean of 60, 80, 90, 120

50
 60 + 80 + 90 +120
4

350
=
4

= 87.5

The arithmetic mean is very useful because it represents the values of most
observations in the population.
The mean therefore describes the population quite well in terms of the
magnitudes attained by most of the members of the population

Computation of the mean from grouped Data i.e. in classes.

The following data was obtained from the manufacturers of electronic

ke
cells. A sample of electronic cells was taken and the life spans were
recorded as shown in the following table.

o.
.c
es
Life span hrs No. of cells Class MP(x) X–A=d fd
ot

(f)
cn

1600 – 1799 25 1699.5 -600 -15000

ne

1800 – 1999 32 1899.5 -400 -12800

.k

2000-2199 46 2099.5 -200 -9200

w

2200 – 2399 58 2299.5(A) 0 0

w
w

2400 – 2599 40 2499.5 200 8000

2600 – 2799 30 2699.5 400 12000
2800 – 2999 7 2899.5 600 4200
A = Assumed mean, this is an arbitrary number selected from the data, MP =
mid point

Arithmetic mean = assumed mean +

 fd 12800
= 2299.5 + 
 f 238
= 2299.5 +-53.78

= 2245.72 hours

Example 2 – (use of the coded method)

51
The following data was obtained from students who were registered in a
certain college.
The table shows the age distribution

Age (yrs) No. of Students mid points x-a = d D/c = u fu

(f) (x)
15 – 19 21 17 -15 -3 -63
20 – 24 35 22 -10 -2 -70
25 – 29 38 27 -5 -1 -38
30 – 34 49 32(A) 0 0 0
35 – 39 31 37 +5 + 31
40 – 44 19 42 +10 +2 38
193 -102

Required calculate the mean age of the students using the coded method

 fu

ke
Actual mean = A(assumed mean) + c
f

o.
102
.c
es
= 32 +  5
193
ot
cn

= 29.36 years
ne
.k

Merits of arithmetic mean

w

 It possess first six our seven characteristics of good average and no

w

other average possess such large number of characteristics

w

 It is unduly affected by the presence of extreme value.

 It utilizes all the observations given.
 It is a very useful statistic in terms of applications. It has several
applications in business management e.g. hypothesis testing, quality
control e.t.c.
 It is the best representative of a given set of data if such data was
obtained from a normal population.
 The a.m. can be determined accurately using mathematical
formulas.

Limitation
 It is difficult to compute mean without making assumption regarding
the size of the class- interval of the open end classer.

52
  If the data is not drawn from a ‗normal‘ population, then the a.m.
may give a wrong impression about the population.
 In some situations, the a.m. may give unrealistic values especially
when dealing with discrete variables e.g. when working out the
average no. of children in a no. of families. It may be found that the
average is 4.4 which is unrealistic in human beings.

The Median
This is a statistical value which is normally located at the center of a given
set of data which has been organized in the order of magnitude or size e.g.
consider the set 14, 17, 9, 8, 20, 32, 18, 14.5, 13. When the data is ordered
it will be 8, 9, 13, 14, 14.5, 17, 18, 20, and 32. The middle number/median
is 14.5. The importance of the median lies in the fact that it divides the
data into 2 equal halves. The no. of observations below and above the
median is equal. In order to determine the value of the median from
grouped data, the data is grouped the median may be determined by using
the following methods

ke
i. Graphical method using the cumulative frequency curve

o.
(ogive).
ii. The formula
.c
es
ot

Example
cn

Referring to the table in 105, determine the median using the methods
ne

above
.k

The graphical method

w
w
w

IQ No of resid UCB Cumulative Frequency

0 – 20 6 20 6
20 – 40 18 40 24
40 – 60 32 60 56
60 – 80 48 80 104
80 – 100 27 100 131
100 – 120 13 120 144
120 – 140 2 140 146
146

53
Xv
160
140
120
100
80
60
40
20
20 40 60 80 100 120 140 160

ke
o.
Value of the median
.c
es
n+1 146+1
ot

The position of the median = =

2 2
cn

Computation
ne

The formula used is

.k

Media= L +i⁄f{m-c}
w
w

Where L = Lower class boundary of the class containing the median

w

m = (n/2)th or (n+1/2)th
c= Cumulative frequency of the class before that containing the
median.
f = Frequency of the class containing the median

Media= L +i⁄f{m-c}
73.5 - 56
= 60 + × 20
48
= 60 + 7.29
= 67.29

54
Merits of median
 It is not influenced by the presence extreme values
 The sum of the deviation of observation from median is minimum
 It shows the centre of a given set of data
 Knowledge of the determination of the median may be extended to
determine the quartiles
 The median can easily be defined
 It can be obtained easily from the cumulative frequency curve
 It can be used in determining the degrees of skewness

Limitation
 Its value is not determined by each and every observation.
 Median is not capable of algebraic treatment.
 Median is less reliable.
 It is affected by sampling fluctuation.
 In some situations where the no. of observations is even, the value of

ke
the median obtained is usually imaginary.
 The computation of the median using the formulas is not well

o.
understood by most businessmen.
.c
 In business environment the median has got very few applications.
es
Quartiles, Deciles, Percentiles
ot

The procedure for computing quartiles deciles etc is the same as for median
cn

For grouped data the following formula are used for quartiles, deciles and
percentiles:
ne

Q1 = L + i/f (m–c)
.k
w

Where, Q1= First or lower quartile.

w

L= Lower class limit of the lower quartile group.

w

i = class interval of the lower quartile group.

f = frequency of the lower quartile group.
m = n/4
c = cumulative frequency of the group preceding the lower
quartile group.

Decile:
D2 = L +i/f (m-c)

Where, D2 = second decile

m = 2n/10

Percentile:

P40 = L + i/f (m-c)

55
Where, P40 = fortieth percentile.
m = 40n/100

Example
The profit earned by 100 companies during 2003-04 are given bellow
Profits (Kshs.) Number of companies
20-30 4
30-40 8
40-50 18
50-60 30
60-70 15
70-80 10
80-90 8
90-100 7

Calculation Q median, D4 and P80 and interpret the values.

ke
Solution

o.
Calculation of Q1, Q2, Q4, and P80
Profits (Kshs.) f .c cf
es
20-30 4 4
30-40 8 12
ot

40-50 18 30
cn

50-60 30 60
60-70 15 75
ne

70-80 10 85
.k

80-90 8 93
90-100 7 100
w
w
w

Q1 =size of N/4th observation = 100/4 =25th observation

Hence q1 lies in the class 40-50
Q1= L+ i/f (m-c)

= 40 +10/18(25-12)
= 40 + 7.22 = 47.22
= 47.22

D4 = L+ i/f (m-c)
= 50 +10/30 (40-30)
= 50 + 3.33
= 53.33

56
Thus 40 percent of the companies earn an annual profit of kshs.53.33 or
less

P80 = size of 80/100 the observation = 80 the observation

P80 lies in the class 70-80
P80 = L+ i/f (m-c)
= 70 + 10/10(80-75)
= 70 +5
= 75
This means that 80 percent of the companies earn an annual profit of
kshs.75 or less
20 percent of the companies earn an annual profit of more than kshs. 75

Mode
This is one of the measures of central tendency. The mode is defined as a
value within a frequency distribution which has the highest frequency.

ke
Sometimes a single value may not exist as such in which case we may refer
to the class with the highest frequency. Such a class is known as a modal

o.
class
.c
The mode is a very important statistical value in business activities quite
es
often business firms tend to stock specific items which are heavily on
demand e.g. footwear, clothes, construction materials (beams, wires, iron
ot

sheets e.t.c.
cn

The mode can easily be determined form ungrouped data by arranging the
figures given and determining the one with the highest frequency.
ne

When determining the values of the mode from the grouped data we may
.k

use the following methods;-

i. The graphical method which involves use of the histogram
w

ii. The computation method which involves use of formula

w
w

Example
In a social survey in which the main purpose was to establish the
intelligence quotient (IQ) of resident in a given area, the following results
were obtained as tabulated below:

IQ No. of residents Upper class bound CF

1 – 20 6 20 6
21 – 40 18 40 24
41 – 60 32 fo 60 56
61 – 80 48 f1 80 104
81 – 100 27 f2 100 131
101 – 120 13 120 144
121 – 140 2 140 146

57
Required
Calculate the modal value of the IQ‘s tabulated above using
i. The graphical method and
ii. Formula

Graphical method

50

40

ke
o.
30
.c
es
20
ot
cn
ne

10
.k
w

20 40 60 80 100 120 140

w

Value of the mode

w

Computation method
 f1  f 0 
Mode = L +   ×c
 2f1  f 0  f 2 
Where L = Lower class boundary of the class containing the mode
f0 = Frequency of the class below the modal class
f1 = Frequency of the class containing the mode
f2 = frequency of the class above the modal class
c = Class interval
  48 - 32  
Therefore Mode = 60.50 +   × 20
 2  48  - 32 - 27 

= 69.14
58
Merits of mode
 Mode is not affected by extreme values.
 At can be easily used to decide qualitative phenomenon.
 Mode happens to be meaningful as an average.
 It can be determined from incomplete data provided the
observations with the highest frequency are already known.
 The mode has several applications in business.
 The mode can be easily defined.
 It can be determined easily from a graph.

Limitation of mode
 Mode cannot always be computed
 If the data is quite large and ungrouped, determination of the mode
can be quite cumbersome
 Use of the formula to calculate the mode is unfamiliar to most

ke
business people

o.
 The mode may sometimes be non existent or there may be two
.c
modes for a given set of data. In such a case therefore a single mode
may not exist
es
ot
cn

Geometric mean
This is a measure of central tendency normally used to measure industrial
ne

growth rates. It is defined as the nth root of the product of ‗n‘ observations
.k

or values
w

- i.e. GM = n x1 × x 2 ×... × x n
w
w

Example
In 1995 five firms registered the following economic growth rates; 26%. 32%
41% 18% and 36%.
Required
Calculate the GM for the above values

GM  5
26  32  41  18  36

 15  Log 26  Log 32  Log 41  Log18  Log 26

No. Log
26 1.4150
32 1.5052

59
 41 1.6128
18 1.2533
36 1.5563
7.3446
Therefore Log of GM = 1/5 x 7.3446 = 1.46892
So GM = Antilog of 1.46892
= 29.43
Merits
i. It makes use of all the values given (except when x = 0 or
negative)
ii. It is the best measure for industrial growth rates

ke
Demerits
i. The determination of the GM by using logarithms is not familiar

o.
process to all those expected to use it e.g. managers
ii. .c
If the data contains zeros or –ve values, the GM ceases to exist
es
ot
cn

Harmonic mean
This is a measure of central tendency which is used to determine the
ne

average growth rates for natural economies. It is defined as the reciprocal

.k

of the average of the reciprocals of all the values given by HM.

w

1
HM 
w

1 ( 1  1  ... 1 )
n x1 x2 x3
w

Example
The economic growth rates of five countries were given as 20%, 15%, 25%,
18% and 5%
Calculate the harmonic mean

60
 1
The HM =
1 (1 +1 +1 +1 +1
5 20 15 25 10 5

1
=
0.2(0.05 + 0.07 + 0.04 + 0.10 + 0.2)

1
=
0.092

10.86%

Merits – same as the arithmetic mean

Demerits – same as the arithmetic mean

ke
Weighted mean

o.
- This is the mean which uses arbitrarily given weights
- .c
It is a useful measure especially where assessment is being done yet the
es
conditions prevailing are not the same. This is particularly true when
assessment of students is being done given that the subjects being taken
ot

have different levels of difficulties.

cn

Examples
ne

The following table shows that marks scored by a student doing section 3
.k

and 4 of CPA
w
w

Subject Scores (x) Weight (w) wx

w

STAD 65 50 3250
BF 63 40 2520
FA2 62 45 2340
LAW 80 35 2800
QT 69 55 3795
FA3 55 60 3300
w = 285 wx = 18005

Weighted mean

61
 Ewx
Ew

18005

285

 63.17%

ke
o.
.c
es
ot
cn
ne
.k
w
w
w

62
 CHAPTER FOUR

MEASUREMENT OF VARIATION /DISPERSION

Specific objectives
At the end of this topic, the trainee should be able to:-
 State the characteristics of a good measure of dispersion.
 Differentiate between the absolute and relative measures.
 Calculate and interpret the measures of dispersion.

Introduction
Measures of variation help us in studying the important characteristics of a
distribution. The measures of dispersion are very useful in statistical work
because they indicate whether the rest of the data are scattered around
the mean or away from the mean. If the data is approximately dispersed
around the mean then the measure of dispersion obtained will be small
therefore indicating that the mean is a good representative of the sample

ke
data. But on the other hand, if the figures are not closely located to the
mean then the measures of dispersion obtained will be relatively big

o.
indicating that the mean does not represent the data sufficiently.
.c
es
Significance of measuring variation or dispersion
i) to determine the reliability of an average
ot

ii) to serve as a basis for the control of the variability

cn

iii) to compare two or more series with regard to their variability

ne

iv) to facilitate the use of other statistical measures

.k
w

Property of a good measure of variation/dispersion.

w

A good measure of variation should possess as far as possible

w

i) It should be simple to understand

ii) It should be easy to compute
iii) It should be rigidly d3efined
iv) It should be based on every observation of the distribution
v) It should be amenable to further algebraic treatment
vi) It should have sampling stability
vii) It should not be unduly affected by extreme observation

Methods of studying variation

The following Aretha important methods of studying variation

 The range
 The Mean deviation
 The interquartile range or quartile deviation
 The standard deviation
63
  The Lorenz curve

Absolute and relative measures of variation

Measures of variation may be either absolute or relative. Absolute measures
of variation are expressed in the same statistical unit in which the original
data are given. Such shillings, kilograms, tones, etc
These values may be used to compare the variation in two or more than two
distributions provided the variables are expressed in the same units and
have almost the same average value.
In case the two sets of data are expressed in different units such as
manager‘s salary versus workers salary, the absolute measures of variation
are not comparable. In such cases measures of relative variation should be
used.
A measure of relative variation is the ratio of a measure of absolute
variation to an average. It is sometimes called a coefficient of variation,
because coefficient means a pure number that is independent of the unit of
measurement.

ke
a) The Range

o.
.c
The range is defined as the difference between the highest and the smallest
values in a frequency distribution. This measure is not very efficient
es
because it utilizes only 2 values in a given frequency distribution. However
ot

the smaller the value of the range, the less dispersed the observations are
from the arithmetic mean and vice versa
cn

The range is not commonly used in business management because 2 sets of

ne

data may yield the same range but end up having different interpretations
regarding the degree of dispersion.
.k

Range is the simplest method of studying variation. It is defined as the

w

difference between the value of the smallest observation and the value of
w

the largest observation

w

Range = largest value- smallest value

R= L-S

Coefficient of Range = L-S

L+S
Example
The following are the prices of shares of a company from Monday to
Saturday
days Prices
Monday 200
Tuesday 210
Wednesday 208
Thursday 160
Friday 220
64
Saturday 250

Calculate the range and coefficient of range.

Solution
Range =L-S
L=250
S= 160
Range =250-160
= 90
Coefficient of range = L-S = 250-160 = 90 = 0.219
L+S 250+ 160 410
Merits of range
 It is the simplest to understand and easiest to compute
 It takes minimum time to calculate the value of range

Limitation
 Range is not based on each and every observation of the distribution

ke
 It is subject to fluctuation of considerable magnitude from sample to

o.
sample
 Range cannot be computed in case of open –end distribution
.c
 Range cannot tell us anything about the character of the distribution
es
within two extreme observations
ot

Uses of range
1. Quality control; the objective of quality control is to keep a check on
cn

the quality of the products without 100% inspection control charts

ne

are prepared.
2. fluctuation in the shares price; range is useful in studying the
.k

variation in the prices in stocks and shares and other commodities

w

3. Weather forecast; the meteorological department does make use of

w

the range in determining the difference between the minimum

w

temperature and maximum temperature.

(b) Mean Deviation

This is a useful measure of dispersion because it makes use of all the values
given. The average deviation is obtained by calculating the absolute
deviation of each observation from median (or mean) and then averaging
this deviation by taking their arithmetic mean. The formula for average
deviation may be written as;
A.D = ∑ /x – med/
N
Incase deviation are taken from mean the formula shall be written as;
A.D x = ∑ /x - x/
N
Computation of average deviation
The formula for computing average deviation is
A.D (med) = ∑ /x – med/
65
 N
If average deviation is small the distribution is highly compact or uniform
since more than half of the cases are concentrated within a small range
around the mean.
The relative measure corresponding to the average deviation called the
coefficient of average deviation is obtained by dividing average by the
measure of central tendency used, i.e. mean or median.

Example 1
In a given exam the scores for 10 students were as follows

ke
Student Mark (x) xx

o.
A 60 1.8
B 45
.c
16.8
es
C 75 13.2
ot

D 70 8.2
cn

E 65 3.2
F 40 21.8
ne

G 69 7.2
H 64 2.2
.k

I 50 11.8
w

J 80 18.2
w

Total 618 104.4

w

Required
Determine the absolute mean deviation
618
Mean, x = = 61.8
10

 X-X 104.4
Therefore AMD = = = 10.44
N 10

Example 2
The following data was obtained from a given financial institution. The data
refers to the loans given out in 1996 to several firms

66
 Firms Amount of fx xx x x .f
(f) loan per firm
(x)
3 20000 60000 4157.9 12473.70
4 60000 240000 35842.1 143368.40
1 15000 15000 9157.9 9157.9
5 12000 60000 12157.9 60789.50
6 14000 84000 10157.9 60947.40
Σf = 19 Σfx = 286736.90
459000

Required
Calculate the mean deviation for the amount of items given

 fx 459, 000
X    24157.9

ke
f 19

o.
 AMD 
 X -X

286736.90 .c
es
f 19
ot
cn

 Shs 15, 091.40

ne

NB if the absolute mean deviation is relatively small it implies that the data
is more compact and therefore the arithmetic mean is a fair sample
.k

representative.
w
w
w

(b) The interquartile range or quartile deviation

This is a measure of dispersion which involves the use of quartile. A quartile
is a mark or a value which lies at the boundary of a division when any given
set of data is divided into four equal divisions
Each of such divisions normally carries 25% of all the observations
The semi interquartile range is a good measure of dispersion because it
shows how the rest of the data are generally spread around the mean
The quartiles normally used are three namely;
i. The lower quartile (first quartile Q1) this usually binds the
lower 25% of the data
ii. The median (second quartile Q2)
iii. The upper quartile (third quartile Q3)

Interquartile range = Q3-Q1

67
Very often the interquartile range is reduced to the form of the semi-
interquartile range or quartile deviation by dividing it by 2
Q.D = quartile deviation
Q.D = Q3-Q1/2
Quartile deviation gives the average amount by which the quartile differs
from the median. Quartile deviation is an absolute measure of variation.
The relative measure corresponding to the measure called the coefficient
of quartile deviation.
Coefficient of Q.D = Q3-Q1/Q3+Q1
The semi-interquartile range,

Q3 - Q1
SIR =
2

Example 1
The weights of 15 parcels recorded at the GPO were as follows:

ke
16.2, 17, 20, 25(Q1) 29, 32.2, 35.8, 36.8(Q2) 40, 41, 42, 44(Q3) 49, 52, 55
(in kgs)

o.
Required
.c
Determine the semi interquartile range for the above data
es
Q3  Q1 44 - 25 19
ot

SIR = = = = 8.5
2 2 2
cn
ne

Example 2 (Grouped Data)

.k

The following table shows the levels of retirement benefits given to a group
w

of workers in a given establishment.

w
w

Retirement benefits No of retirees UCB cf

£ ‗000 (f)
20 – 29 50 29.5 50
30 – 39 69 39.5 119
40 – 49 70 49.5 189
50 – 59 90 59.5 279
60 – 69 52 69.5 331
70 – 79 40 79.5 371
80 – 89 11 89.5 382

Required
i. Determine the semi interquartile range for the above data
ii. Determine the minimum value for the top ten per cent.(10%)
iii. Determine the maximum value for the lower 40% of the retirees

68
Solution
The lower quartile (Q1) lies on position
N +1 382 + 1
=
4 4

= 95.75

(95.75 - 50)
 the value of Q1 = 29.5 + x 10
69

= 29.5 + 6.63

ke
= £36.13

o.
The upper quartile (Q3) lies on position .c
es
 N + 1
ot

 3 
 4 
cn

 382 + 1 
=3  
ne

 4 
.k
w

= 287.25
w

 287.25 - 279 
∴ the value of Q3
w

= 59.5 + × 10
52

= 61.08
Q3 - Q1
The semi interquartile range =
2
61.08 - 36.13
=
2
= 12.475
= £12,475
ii. The top 10% is equivalent to the lower 90% of the retirees
The position corresponding to the lower 90%

69
 90
= (n + 1) = 0.9 (382 + 1)
100

= 0.9 x 383

= 344.7

∴ the benefits (value) corresponding to the minimum value for top

10%
 344.7 - 331
= 69.5 + x 10
40
= 72.925
= £ 72925

ke
iii. The lower 40% corresponds to position

o.
40
= (382 + 1)
100 .c
es
ot

= 153.20
cn

∴ Retirement benefits corresponding to its position

ne

153.2 - 119 
.k

= 39.5 + x 10
w

70
w

= 39.5 + 4.88
w

= 44.38
= £ 44380

The 10th – 90th percentile range

This is a measure of dispersion which uses percentile. A percentile is a
value which separates one division from the other when a given data is
divided into 100 equal divisions.
This measure of dispersion is very important when calculating the co-
efficient of skewness.

Example

70
Using the above data for retirees calculate the 10th - 90th percentile. The
tenth percentile 10th percentile lies on position
10
(382 + 1) = 0.1 x 383
100

= 38.3
∴ the value corresponding to the tenth percentile

(38.3 x 10)
= 19.5 +
50
= 19.5 + 7.66
= 27.16
th
The 90 percentile lies on position
90
(382 + 1) = 0.9 x 383

ke
100
= 344.7

o.
∴ the value corresponding to the 90th percentile
.c
es
 344.7 - 331
= 69.5 + x 10
ot

40
cn

= 69.5 + 3.425
ne

= 72.925
.k

∴ the required value of the 10th – 90th percentile = 72.925 – 27.16 =

w

45.765
w
w

Merits of quartile deviation

i) In certain respect it is superior to range as a measure of variation
ii) It has a special utility in measuring variation in cases of open –
ended distribution
iii) It has also useful in erratic or highly skewed distribution where
the other measure of variation would be warped by extreme
values. It is not affected by the presence of extreme values.

Limitation
i) Quartile deviation ignores 50% item as the value of quartile
deviation does not depend upon every observation it cannot be
regarded as a god method of measuring variation.
ii) It is not capable of mathematical manipulation

71
 iii) Its value sis very much affected by sampling fluctuation
iv) It is in fact not a measure of variation as it really does not show
the scatter around an average but rather a distance on scale.

(d) Standard deviation

The standard deviation measures the absolute dispersion (or variability of
distribution, the greater amount of dispersion or variability). The greater
the standard deviation the greater will be the magnitude of the deviations
of the value from their mean.
A small standard deviation means a high degree of uniformity of the
observation as well as homogeneity of a series.
A large standard deviation means just the opposite.
Difference between deviation and standard deviation
Both these measure of dispersion are based on each and very item of the
distribution. But they differ in the following respects.
 Algebraic signs are ignores while calculating mean deviation whereas
in the calculation of standard deviation signs are taken into account.

ke
 Mean deviation can be computed either from median or mean. The
standard deviation on the other hand is always computed from the

o.
arithmetic mean because the sum of the squares of the deviation of
items from arithmetic mean is least. .c
es
This is one of the most accurate measures of dispersion. It has the following
advantages;
ot

i. It utilizes all the values given

cn

ii. It makes use of both negative and positive values if they occur
iii. The standard deviation reflects an accurate impression of how
ne

much the sample data varies from the mean. This is because its
.k

suitability can also be tested using other statistical methods

w

Example
w
w

A sample comprises of the following observations; 14, 18, 17, 16, 25, 31
Determine the standard deviation of this sample
Observation.

x
 x  x  x  x
2

14 -6.1 37.21
18 -2.1 4.41
17 -3.1 9.61
16 -4.1 16.81
25 4.9 24.01
31 10.9 118.81
Total 121 210.56

72
 121
X  20.1
6

 
2
 xx 210.56
 Standard deviation,   
n 6
= 5.93

Alternative method
x X2
14 196
18 324
17 289
16 256

ke
25 625

o.
31 961
Total 121 .c
2651
es
2 2
x x 2651  121 

2
ot

    
 n   6 
cn

n 6
ne

= 5.93
.k

Example 2
w
w

The following table shows the part-time rate per hour of a given no. of
w

laborers in the month of June 1997.

Rate per hr (x) No. of laborers fx fx2

Shs (f)
230 7 1610 370300
400 6 2400 960000
350 2 700 245000
450 1 450 202500
200 8 1600 320000
150 11 1650 247500
Total 35 8410 2345300

Calculate the standard deviation from the above table showing how the
hourly payment were varying from the respective mean

73
  fx   fx 
2
2

∴ Standard deviation,   -
f   f 
 

2
2345300  8410 
= - 
35  35 

= 67008.6  577372

= 9271.4

= 96.29

Example 3 – Grouped data

In business statistical work we usually encounter a set of grouped data. In

ke
order to determine the standard deviation from such data, we use any of
the three following methods

o.
i. The long method
ii. The shorter method .c
es
iii. The coded method
The above methods are used in the following examples
ot
cn

Example 3.1
ne

The quality controller in a given firm had an accurate record of all the iron
bars produced in May 1997. The following data shows those records
.k
w

i. Using long method

w
w

Bar lengths No. of bars(f) Class mid fx fx2

(cm) point (x)
201 – 250 25 225.5 5637.5 1271256.25
251 – 300 36 275.5 9918 2732409
301 – 350 49 325.5 15949.5 5191562.25
351 – 400 80 375.5 30040 11280020
401 – 450 51 425.5 21700.5 9233562.75
451 – 500 42 475.5 19971 9496210.50
501 - 550 30 525.5 15765 8284507.50
313 118981.50 47489526

Calculate the standard deviation of the lengths of the bars

74
  fx   fx 
2
2

∴ Standard deviation, σ = -
f   f 
 

2
47489526  118981.50 
= - 
313  313 

= 84.99 cm

ii. Using the shorter method

Bar lengths No. of mid point x-A = d fd Fd2

(cm) bars(f) (x)
201 – 250 25 225.5 -150 -3750 562500
251 – 300 36 275.5 -100 -3600 360000
301 – 350 49 325.5 -50 -2450 122500

ke
351 – 400 80 375.5 (A) 0 0 0

o.
401 – 450 51 425.5 50 2550 127500
451 – 500 42 475.5 100 4200 420000
501 - 550 30 525.5
.c150 4500 675000
es
Total 313 1450 2267500
ot

Calculate the standard deviation using the shorter method quagmire

cn
ne

 fd   fd 
2
2

∴ Standard deviation, σ = 
f   f 
.k

 
w
w

2
2267500  1450 
w

=  
313  313 

= 7244.40  21.50

= 7222.90

= 84.99 cm

iii. Using coded method

Bar lengths (f) mid point (x) x-A = d d/c = u fu fu2

(cm)
201 – 250 25 225.5 -150 -3 -75 225
251 – 300 36 275.5 -100 -2 -72 144
75
 301 – 350 49 325.5 -50 -1 -49 49
351 – 400 80 375.5 (A) 0 0 0 0
401 – 450 51 425.5 50 1 51 51
451 – 500 42 475.5 100 2 84 168
501 - 550 30 525.5 150 3 90 270
313 29 907

C = 50 where c is an arbitrary number, try picking a

different figure say 45 the answer should be the same.
Standard deviation using the coded method. This is the most preferable
method among the three methods

 fu   fu 
2
2

  c -
  f 
f  

ke
o.
2
907  29 
 50   
313  313 .c
es
= 50 × 1.6997
ot
cn

= 84.99
ne

Variance
.k

Square of the standard deviation is called variance.

w
w

(e) Lorenz curve

w

The Lorenz curve derived by Max O. Lorenz, famous economic statistician,

is a graphic method of studying dispersion. This curve was used by him for
the first time to measure the distribution off wealth and income. Now the
curve is also used to study the distribution of profits, wages, turnover, etc.

Illustration
In the following table is given the number of companies belonging to two
areas A and B according to the amount of profits earned by them. Draw in
the same diagram their Lorenz curve and interpret them.
Profits earned shs (000s) Number of companies
Area A Area B
6 6 2
25 11 38
60 13 52
84 14 28

76
105 15 38
150 17 26
170 10 12
400 14 4

Solution

ke
o.
.c
es
ot
cn
ne

Relative measures of dispersion

.k

Def: A relative measure of dispersion is a statistical value which may be

w

used to compare variations in 2 or more samples.

w

The measures of dispersion are usually expressed as decimals or

w

percentages and usually they do not have any other units

Example
The average distance covered by vehicles in a motor rally may be given as
2000 km with a standard deviation of 5 km.
In another competition set of vehicles covered 3000 km with a standard
deviation of 10 kms
NB: The 2 standard deviations given above are referred to as absolute
measures of dispersion. These are actual deviations of the measurements
from their respective mean
However, these are not very useful when comparing dispersions among
samples.
Therefore the following measures of dispersion are usually employed in
order to assess the degree of dispersion.
i. Coefficient of mean deviation
77
 Mean deviation
=
mean
ii. Coefficient of quartile deviation
1 Q - Q 
 2
3 1

Q2

Where Q1 = first quartile

Q3 = third quartile

iii. Coefficient of standard deviation

Standard deviation
=
mean

ke
iv. Coefficient of variation

o.
standard deviation
= ×100
mean .c
es
Example (see information above)
ot

First group of cars: mean = 2000 kms

cn

Standard deviation = 5 kms

ne
.k

∴ C.O.V = 5 x 100
w

2000
w
w

= 0.25%

Second group of cars: mean = 3000 kms

Standard deviation = 10kms

∴ C.O.V = 10 x 100
3000

= 0.33%

Conclusion
Since the coefficient of variation is greater in the 2 nd group, than in the
first group we may conclude that the distances covered in the 1 st group are
much closer to the mean that in the 2nd group.
78
Example 2
In a given farm located in the UK the average salary of the employees is £
3500 with a standard deviation of £150
The same firm has a local branch in Kenya in which the average salaries are
Kshs 8500 with a standard deviation of Kshs.800
Determine the coefficient of variation in the 2 firms and briefly comment
on the degree of dispersion of the salaries in the 2 firms.
First firm in the UK
C.O.V = 150 x 100
3500

= 4.29%

Second firm in Kenya

ke
C.O.V = 800 x 100
8500

o.
= 9.4% .c
es
Conclusively, since 4.29% < 9.4% then the salaries offered by the firm in UK
ot

are much closer to the mean given them in the case to the local branch in
Kenya
cn
ne

COMBINED MEAN AND STANDARD DEVIATION

Sometimes we may need to combine 2 or more samples say A and B. It is
.k

therefore essential to know the new mean and the new standard deviation
w

of the combination of the samples.

w
w

Combined mean
Let m be the combined mean
Let x1 be the mean of first sample
Let x2 be the mean of the second sample
Let n1 be the size of the 1st sample
Let n2 be the size of the 2nd sample
Let s1 be the standard deviation of the 1st sample
Let s2 be the standard deviation of the 2nd sample
n1 x1  n2 x2
 combined mean 
n1  n2

79
 n1s12  n1  m  x1   n2 s22  n2  m  x2 
2 2

combined standard deviation 

n1  n2

Example
A sample of 40 electric batteries gives a mean life span of 600 hrs with a
standard deviation of 20 hours.
Another sample of 50 electric batteries gives a mean lifespan of 520 hours
with a standard deviation of 30 hours.
If these two samples were combined and used in a given project
simultaneously, determine the combined new mean for the larger sample
and hence determine the combined or pulled standard deviation.

Size x s
40(n1) 600 hrs(x1) 20hrs (s1)
50 (n1) 520 hrs (x2) 30 hrs (s2)

ke
o.
40  600   50  520  50, 000
   555.56
Combined mean
40  50 .c90
es
ot
cn
ne

Combined standard deviation

.k
w

40(202 )  40(555.56 - 660)2  50(30)2  50( 555.56 - 520)2


w

40  50
w

1600  78996.54  45000  63225.68


90

 47.52 hrs

SKEWNESS
- This is a concept which is commonly used in statistical decision
making. It refers to the degree in which a given frequency curve is
deviating away from the normal distribution
- There are 2 types of skew ness namely
i. Positive skew ness
ii. Negative skew ness

80
 1. Positive Skewness
- This is the tendency of a given frequency curve leaning towards
the left. In a positively skewed distribution, the long tail
extended to the right.

In this distribution one should note the following

i. The mean is usually bigger than the mode and median
ii. The median always occurs between the mode and mean
iii. There are more observations below the mean than above the
mean
This frequency distribution as represented in the skewed distribution curve
is characteristic of the age distributions in the developing countries

Frequenc Positively skewed Frequenc

y frequency curve y Negatively skewed
frequency curve

Normal distribution

ke
o.
Long tail
.c
Median
Mode

Mean

es

Median

Mode
Mean
ot
cn

2. Negative Skewness
ne

This is an asymmetrical curve in which the long tail extends to the left
.k

NB: This frequency curve for the age distribution is characteristic of the
w

age distribution in developed countries

w

- The mode is usually bigger than the mean and median

w

- The median usually occurs in between the mean and mode

- The no. of observations above the mean are usually more than
those below the mean (see the shaded region)

MEASURES OF SKEWNESS
- These are numerical values which assist in evaluating the degree
of deviation of a frequency distribution from the normal
distribution.

- Following are the commonly used measures of skew ness.

1. Coefficient Skewness
 mean - median 
= 3
Standard deviation
81
 2. Coefficient of skewness
mean - mode
=
Standard deviation
NB: These 2 coefficients above are also known as Pearsonian measures of
skewness.

3. Quartile Coefficient of skewness

Q3+ Q1- 2Q2
=
Q3+ Q1
Where Q1 = 1st quartile
Q2 = 2nd quartile
Q3 = 3rd quartile
NB: The Pearsonian coefficients of skewness usually range between –ve 3
and +ve 3. These are extreme value i.e. +ve 3 and –ve 3 which therefore
indicate that a given frequency is negatively skewed and the amount of
skewness is quite high.

ke
Similarly if the coefficient of skewness is +ve it can be concluded that the

o.
amount of skew ness of deviation from the normal distribution is quite high
.c
and also the degree of frequency distribution is positively skewed.
es
Example
ot

The following information was obtained from an NGO which was giving
cn

small loans to some small scale business enterprises in 1996. the loans are
in the form of thousands of Kshs.
ne

Loans Units Midpoints(x) x-a=d d/c= fu Fu2 UCB cf

.k

(f) u
w

46 – 50 32 48 -15 -3 -96 288 50.5 32

w

51 – 55 62 53 -10 -2 -124 248 55.5 94

w

56 – 60 97 58 -5 -1 -97 97 60.5 191

61 –65 120 63 (A) 0 0 0 0 0 0
66 –70 92 68 5 +1 92 92 70.5 403
71 –75 83 73 10 +2 166 332 75.5 486
76 – 80 52 78 15 +3 156 468 80.5 538
81 – 85 40 83 20 +4 160 640 85.5 57.8
86 – 90 21 88 25 +5 105 525 90.5 599
91 – 95 11 93 30 +6 66 396 95.5 610
Total 610 428 3086

Required
Using the Pearsonian measure of skew ness, calculate the coefficients of
skew ness and hence comment briefly on the nature of the distribution of
the loans.

82
 c   fu 
Arithmetic mean = Assumed mean +
f

 428 × 5 
= 63 +
610
= 66.51

It is very important to note that the method of obtaining arithmetic mean (or any other statistic) by misusing
assumed mean (A) from X and then dividing by c can be abit confusing, if this is the case then just use the
straight forward method of:

Arithmetic mean 
 f .x where x is the midpoint, the answers are the same.
f
 fu   fu 
2
2

The standard deviation =c× -

f   f 
 

ke
2
3086  428 

o.
=5 × - 
610  610 
.c
es
= 10.68
ot
cn

n +1
The Position of the median lies m =
2
ne

610 +1
= = 305.5
.k

2
 305.5 - 191
w

= 60.5 + ×5
w

120
w

114.4 
= 60.5 + ×5
120
Median = 65.27
Therefore the Pearsonian coefficient
 66.51- 65.27 
= 3
10.68
= 0.348

Comment

83
The coefficient of skewness obtained suggests that the frequency
distribution of the loans given was positively skewed
This is because the coefficient itself is positive. But the skewness is not
very high implying the degree of deviation of the frequency distribution
from the normal distribution is small

Example 2
Using the above data calculate the quartile coefficient of skewness
Q3+ Q1- 2Q2
Quartile coefficient of skewness =
Q3+ Q1

610 +1
The position of Q1 lies on = = 152.75
4

∴ actual value Q1 =55. 5 +

152.75 - 94   5  58.53
97

ke
 610 +1 = 458.25

o.
The position of Q3 lies on = 3
4
 458.25 - 403 .c
es
∴ actual value Q3 =70.55 +  5  73.83 × 5
83
ot
cn

 610 + 1
Q2 position: i.e. 2 = 305.5
ne

Actual Q2 value  60.5 

 305.5 -191  5  65.27
.k

120
w
w

The required coefficient of skew ness

w

73.83  58.53  2  65.27 

=  0.013
73.83  58.53

Conclusion
Same as above when the Pearsonian coefficient was used

84
KURTOSIS
- This is a concept, which refers to the degree of peaked ness of a
given frequency distribution. The degree is normally measured
with reference to normal distribution.
- The concept of kurtosis is very useful in decision making
processes i.e. if is a frequency distribution happens to have either
a higher peak or a lower peak, then it should not be used to make
statistical inferences.
- Generally there are 3 types of kurtosis namely;-
i. Leptokurtic
ii. Mesokurtic
iii. Platykurtic
Leptokurtic
a) A frequency distribution which is lepkurtic has generally a
higher peak than that of the normal distribution. The
coefficient of kurtosis when determined will be found to be
more than 3. thus frequency distributions with a value of more

ke
than 3 are definitely leptokurtic
b) Some frequency distributions when plotted may produce a

o.
curve similar to that of the normal distribution. Such
.c
frequency distributions are referred to as mesokurtic. The
es
degree of kurtosis is usually equal to 3
c) When the frequency curve contacted produces a peak which is
ot

lower that that of a normal distribution when such a curve is

cn

said to be platykurtic. The coefficient of such is usually less

than 3
ne

- It is necessary to calculate the numerical measure of kurtosis.

.k

The commonly used measure of kurtosis is the percentile

coefficient of kurtosis. This coefficient is normally determined
w

using the following equation

w

 Q3 - Q1
w

Percentile measure of kurtosis, K (Kappa) = 1

2
P90 - P10
Example
Refer to the table above for loans to small business firms/units
Required
Calculate the percentile coefficient of Kurtosis
90
P90 =  n +1 = 0.9  610 +1
100
= 0.9 (611)
= 549.9
The actual loan for a firm in this position
 549.9 - 538 
(549.9) = 80.5 + x 5 = 81.99
40
10
P10 = (n + 1) = 0.1 (611) = 61.1
100

85
The actual loan value given to the firm on this position is

50.5 +
 61.1  32  x 5 = 52.85
62
= 0.9 (611)
= 549.9
∴ Percentile measure of kurtosis

 Q3 - Q1
K (Kappa) = ½
P90 - P10

=½
 73.83 - 58.53 
81.99 - 52.85

= 0.26
Since 0.26 < 3, it can be concluded that the frequency distribution
exhibited by the distribution of loans is platykurtic

ke
Kurtosis is also measured by moment statistics, which utilize the exact

o.
value of each observation.

i. M1 the first moment = M1 =

 X = Mean M1 or M1 .c
es
n
ot

X 2
cn

M2 =
n
ne

X 3
.k

M3 =
n
w
w

X 4
w

M4 =
n

3. M2 second moment about the mean M2 or f2

M2 = M2 – M12
4. M3 third moment about the mean M3 (a measure of the absolute
skew ness)
M3 = M3 – 3M2M1 + 2M13
5. M4 fourth moment about the mean M4 (a measure of the absolute
Kurtosis)
M4 = M4– 4M3M1 + 6M2M12 + 3M14
An alternative formula
  x  m
4
f
M4 = Where m is mean
f

86
 M4
Moment coefficient of Kurtosis
S4

Example
Find the moment coefficient of the following distribution
X f
12 1
14 4
16 6
18 10
20 7
22 2

X F xf (x-m) (x-m)2 (x-m)2f (x-m)4f

12 1 12 -5.6 31.36 31.36 983.45
14 4 56 -3.6 12.96 51.84 671.85

ke
16 6 96 -1.6 2.56 15.36 39.32
18 10 180 .4 0.16 1.60 0.256

o.
20 7 140 2.4 5.76 40.32 232.24
22 2 44 4.4 .c
19.36 38.72 749.62
es
30 528 179.20 2,676.74
ot
cn

528
M = = 17.6
30
ne

179.20
σ2 = = 5.973
.k

30
σ4
w

= 35.677
w

  x  m
w

4
f 2, 676.74
M4 = = = 89.22
f 30

89.22
Moment coefficient of Kurtosis = = 2.5
35.677

Note Coefficient of kurtosis can also be found using the method of assumed
mean.

CHAPTER FIVE

CORRELATION AND REGRESSION

87
Specific Objectives
At the end of the topic the trainee should be able to:
 Draw the scatter diagram;
 Differentiate between the various forms of correlation;
 Determine the correlation coefficient and interpret;
 Determine the coefficient of determination and interpret;
 Apply the linear regression models.

Introduction
When the relationship is of a quantitative nature, the appropriate
statistical tool for discovering and measuring the relationship and
expressing it in a brief formula is known as a correlation. This is an
important statistical concept which refers to interrelationship or
association between variables. The purpose of studying correlation is for
one to be able to establish a relationship, plan and control the inputs
(independent variables) and the output (dependent variables).
In business one may be interested to establish whether there exists a

ke
relationship between the
i. Amount of fertilizer applied on a given farm and the resulting

o.
harvest
ii. .c
Amount of experience one has and the corresponding
es
performance
iii. Amount of money spent on advertisement and the expected
ot

incomes after sale of the goods/service

cn

There are two methods that measure the degree of correlation between
two variables these are denoted by R and r.
ne
.k

(a) Coefficient of correlation denoted by r, this provides a measure of

the strength of association between two variables one the dependent
w

variable the other the independent variable r can range between +1

w

and – 1 for perfect positive correlation and perfect negative

w

correlation respectively with zero indicating no relation i.e. for

perfect positive correlation y increase linearly with x increment.
(b) Rank correlation coefficient denoted by R is used to measure
association between two sets of ranked or ordered data. R can also
vary from +1, perfect positive rank correlation and -1 perfect
negative rank correlation where O or any number near zero
representing no correlation.

88
Significance of the study of correlation
 Most of the variation shows some kind of relationship between price
and supply income and expenditure. With the help of correlation
analysis the degree of relationship existing between the variable can
be measured.
 The value of one variable can be estimated once it has been known
that they closely related. It can be done with the help of regression
analysis.
 It contributes to the economic behavior, aids in locating the critically
important variable on which other depend.
 Nature has been found to be multiplying of interrelated force.
 It helps in determining the degree of relationship between two or
more variable.

Types of correlation
 Positive and negative correlation

ke
 Simple, partial and multiple
 Linear and non linear.

o.
Positive and negative correlation .c
es
If both the variable is varying in the same direction i.e. if one variable is
increasing the other on an average is also increasing or if one variable is
ot

decreasing the other on an average is also decreasing. Correlation is said to

cn

be positive. On the other hand if they are varying in the opposite directions
ne

i.e. as one variable is increasing the other is decreasing or vice versa,

correlation is said to be negative.
.k

Positive correlation negative correlation

w

X T X T
w

10 15 20 40
12 20 30 30
w

15 22 40 22
18 25 60 15
20 37 80 16

Simple partial and multiple correlations

The distribution between simple, partial and multiple correlations are
based upon the number of variable studied.
When only two variables are studied it is a problem of simple correlation.
When three or more variables are studied it is a problem of either

Multiple or partial correlation.

In multiple correlations three or more variables are studied simultaneously.

89
Linear and non linear correlation
The distinction between linear and non linear correlation is based upon the
constancy of the ratio of change between the variables. If the amount of
change in one variable tends to bear constant ratio to the amount of
change in the other variable then the correlation is said to be linear. If such
variable are plotted on a graph paper, all the plotted points would fall on a
straight line. Correlation would be non linear or curvilinear, If the amount
of change in one variable does not bear constant ratio to the amount of
change in the other variable.

Scatter diagram
The simplest device for studying correlation in two variables is a special
type of dot chart called dotogram or scatter diagram.
A scatter graph is a graph which comprises of points which have been
plotted but are not joined by line segments
The pattern of the points will definitely reveal the types of relationship
existing between variables

ke
The following sketch graphs will greatly assist in the interpretation of
scatter graphs.

o.
.c
es
Perfect positive correlation
y
ot

Dependant variable x
cn

x
ne

x
.k

x
w
w

x
w

x
x
x

Independent variable
NB: For the above pattern, it is referred to as perfect because the points
may easily be represented by a single line graph e.g. when measuring
relationship between volumes of sales and profits in a company, the more
the company sales the higher the profits.

90
 Perfect negative correlation
y x
Quantity sold x
X
x
x
x
x
x
x
10 20 X

ke
Price

o.
.c
This example considers volume of sale in relation to the price, the cheaper
es
the goods the bigger the sale.
ot

High positive correlation

cn

y
ne

Dependant variable xx
.k

xx
w

x
w

x
w

xx
xx
xx
xx
x
xxx
x
x

Independent variable

91
 High negative correlation
y
Quantity sold x
x
xx
x
xx
x
x
x
x
xx
x

price

ke
o.
.c
es
ot
cn

No correlation
ne

y
.k
w
w

600 x x x x x
w

x x x
400 x x x x x
x x x x
200 x x x x x
x x x x
0
10 20 30 40 50 x

92
h) Spurious Correlations
in some rare situations when plotting the data for x and y we may have
a group showing either positive correlation or –ve correlation but when
you analyze the data for x and y in normal life there may be no
convincing evidence that there is such a relationship. This implies
therefore that the relationship only exists in theory and hence it is
referred to as spurious or non sense e.g. when high pass rates of student
show high relation with increased accidents.

Correlation coefficient
These are numerical measures of the correlations existing between the
dependent and the independent variables. These are better measures of
correlation than scatter graphs (diagrams).The range for correlation
coefficients lies between +ve 1 and –ve 1. A correlation coefficient of +1
implies that there is perfect positive correlation. A value of –ve shows
that there is perfect negative correlation. A value of 0 implies no

ke
correlation at all.
The following chart will be found useful in interpreting correlation

o.
coefficients.
.c
es
__ 1.0} Perfect +ve correlation
ot

} High positive correlation

cn

__ 0.5 }
ne

} Low positive correlation

.k

__0 }
w

} Low negative correlation

w

__-0.5}
w

} High negative correlation

__-1.0} Perfect –ve correlation

There are usually two types of correlation coefficients normally used

namely;-

Merits of scatter diagram

 It is simple and non mathematical method of studying correlation
between the variables.

93
  It is not influenced by the size of extreme values whereas most of
the mathematical methods of findings correlation are influenced by
extreme values.
Limitation
 Exact degree of correlation can not be established between the
variable

Product Moment Coefficient (r)

It gives an indication of the strength of the linear relationship between two
variables.
n xy   x y
r=
n x 2    x   n y 2    y 
2 2

Note that this formula can be rearranged to have different outlooks but the
resultant is always the same.

ke
Example

o.
The following data was observed and it is required to establish if there
exists a relationship between the two. .c
es
X 15 24 25 30 35 40 45 65 70 75
ot

Y 60 45 50 35 42 46 28 20 22 15
cn

Solution
ne
.k

Compute the product moment coefficient of correlation (r)

X Y X2 Y2 XY
w

15 60 225 3,600 900

w

24 45 576 2,025 1,080

w

25 50 625 2,500 1,250

30 35 900 1,225 1,050
35 42 1,225 1,764 1,470
40 46 1,600 2,116 1,840
45 28 2,025 784 1,260
65 20 4,225 400 1,300
70 22 4,900 484 1,540
75 15 5,625 225 1,125
 X  424  Y  363  X  21,926  Y 2  15,123
2
 XY  12,815

n xy   x y
r=
n x 2    x   n y 2    y 
2 2

94
 10 12,815  424  363
r=
10  21,926  424   10 15,123  363 
2 2

25, 762
=  0.93
 39, 484   19, 461
The correlation coefficient thus indicates a strong negative linear
association between the two variables.

Interpretation of r – Problems in interpreting r values

NOTE:
 A high value of r (+0.9 or – 0.9) only shows a strong association between
the two variables but doesn‘t imply that there is a causal relationship
i.e. change in one variable causes change in the other it is possible to

ke
find two variables which produce a high calculated r yet they don‘t have
a causal relationship. This is known as spurious or nonsense correlation

o.
e.g. high pass rates in QT in Kenya and increased inflation in Asian
countries. .c
es
 Also note that a low correlation coefficient doesn‘t imply lack of
relation between variables but lack of linear relationship between the
ot

variables i.e. there could exist a curvilinear relation.

cn

 A further problem in interpretation arises from the fact that the r value
here measures the relationship between a single independent variable
ne

and dependent variable, where as a particular variable may be

dependent on several independent variables (e.g. crop yield may be
.k

dependent on fertilizer used, soil exhaustion, soil acidity level, season

w

of the year, type of seed etc.) in which case multiple correlation should
w

be used instead.
w

The Rank Correlation Coefficient (R)

Also known as the spearman rank correlation coefficient, its purpose is to
establish whether there is any form of association between two variables
where the variables arranged in a ranked form.
6 d 2
R=1-
n  n 2  1

Where d = difference between the pairs of ranked values.

n = numbers of pairs of rankings

Example

95
A group of 8 accountancy students are tested in Quantitative Techniques
and Law II. Their rankings in the two tests were.
Student Q. T. ranking Law II ranking d d2
A 2 3 -1 1
B 7 6 1 1
C 6 4 2 4
D 1 2 -1 1
E 4 5 -1 1
F 3 1 2 4
G 5 8 -3 9
H 8 7 1 1
d 2
 22

ke
d = Q. T. ranking – Law II ranking

o.
6 d 2 6  22
R=1-
n  n  1
2
 1
8  82  1
.c
es
ot

= 0.74
cn

Thus we conclude that there is a reasonable agreement between student‘s

performances in the two types of tests.
ne

NOTE: in this example, if we are given the actual marks then we find
r. R varies between +1 and -1.
.k

Tied Rankings
w

A slight adjustment to the formula is made if some students tie and have
w

the same ranking the adjustment is

w

t3  t
where t = number of tied rankings the adjusted formula
12
becomes

R=1-
6  d  
2 t 3 t
12

n  n  1
2

Example
Assume that in our previous example student E & F achieved equal marks in
Q. T. and were given joint 3rd place.

96
Solution
Student Q. T. ranking Law II d d2
ranking
A 2 3 -1 1
B 7 6 1 1
C 6 4 2 4
D 1 2 -1 1
E 3½ 5 -1 ½ 2¼
F 3½ 1 2½ 6¼
G 5 8 -3 9
H 8 7 1 1
 d 2  26 1 2

R = 1-
6  d  
2 t 3 t
12
= 1-

6 26 1 2  212 2
3
 since t  2
n  n  1
2
8  82  1

ke
= 0.68

o.
.c
NOTE: It is conventional to show the shared rankings as above, i.e. E, & F
take up the 3rd and 4th rank which are shared between the two as 3½
es
each.
ot

Coefficient of Determination
cn

This refers to the ratio of the explained variation to the total variation and
ne

is used to measure the strength of the linear relationship. The stronger the
linear relationship the closer the ratio will be to one.
.k
w

Coefficient determination = Explained variation

w

Total variation
w

Example (Rank Correlation Coefficient)

In a beauty competition 2 assessors were asked to rank the 10 contestants
using the professional assessment skills. The results obtained were given as
shown in the table below

Contestants 1st assessor 2nd assessor

A 6 5
B 1 3
C 3 4
D 7 6
E 8 7
F 2 1
G 4 8
H 5 2
97
 J 10 9
K 9 10

REQUIRED
Calculate the rank correlation coefficient and hence comment briefly on
the value obtained

d d2

A 6 5 1 1
B 1 3 -2 4
C 3 4 -1 1
D 7 6 1 1
E 8 7 1 1
F 2 1 1 1
G 4 8 -4 16

ke
H 5 2 3 9
J 10 9 +1 1

o.
K 9 10 -1 1
.c
Σd2 = 36
es
ot
cn
ne

∴ The rank correlation coefficient R

.k

6 d 2
w

R=1-
n  n 2  1
w
w

6  36
=1-
10 102  1
216
=1-
990
=1–0
Comment: since the correlation is 0.78 it implies that there is
high positive correlation between the ranks awarded to the
contestants. 0.78 > 0 and 0.78 > 0.5

Example

98
 Contestant 1st 2nd d d2
assessor assessor
A 1 2 -1 1
B 5 (5.5) 3 2.5 6.25
C 3 4 -1 1
D 2 1 1 1
E 4 5 -1 1
F 5 (5.5) 6.5 -1 1
G 7 6.5 -0.5 0.25
H 8 8 0 0
Σd2 = 11.25
Required: Complete the rank correlation coefficient
6 d 2 6 11.25
∴R= 1- =1-
n  n  1
2
8  63
67.5

ke
=1–
504

o.
= 1 – 0.13
= 0.87 .c
es
This implies high positive correlation
ot

Example (Rank Correlation Coefficient)

Sometimes numerical data which refers to the quantifiable variables may
cn

be given after which a rank correlation coefficient may be worked out.

ne

Is such a situation, the rank correlation coefficient will be determined after

the given variables have been converted into ranks. See the following
.k

example;
w
w

Candidates Math r Accounts r d d2

w

P 92 1 67 5 -4 16
Q 82 3 88 1 2 4
R 60 5(5.5) 58 7(7.5) -2 4
S 87 2 80 2 0 0
T 72 4 69 4 0 0
U 60 5(5.5) 77 3 -2.50 6.25
V 52 8 58 7(7.5) 0.5 0.25
W 50 9 60 6 3 9
X 47 10 32 10 0 0
Y 59 7 54 9 -2 4
Σd2 = 43.5
6 d 2
∴ Rank correlation r = 1-
n  n 2  1

99
 6  43.5 261
=1- =1–
10 10  1
2
990
= 0.74 (High positive correlation between mathematics
marks and accounts)

Example
(Product moment correlation)
The following data was obtained during a social survey conducted in a given
urban area regarding the annual income of given families and the
corresponding expenditures.

Family (x)Annual (y)Annual xy x2 Y2

income £ expenditure £
000 000
A 420 360 151200 176400 129600
B 380 390 148200 144400 152100

ke
C 520 510 265200 270400 260100
D 610 500 305000 372100 250000

o.
E 400 360 144000 160000 129600
F 320 290 92800 .c
102400 84100
es
G 280 250 70000 78400 62500
H 410 380 155800 168100 144400
ot

J 380 240 91200 144400 57600

cn

K 300 270 81000 90000 72900

Total 4020 3550 1504400 1706600 1342900
ne

Required
.k

Calculate the product moment correlation coefficient briefly comment on

w

the value obtained

The produce moment correlation
w

n xy   x y
w

r=
n x 2    x   n y 2    y 
2 2

Workings:
4020 3550
X = = 402 Y  355
10 10

10 1,504, 400    4020  3550 

r=
10 1, 706, 600   40202  10 1,342,900    3550 
2

= 0.89

100
Comment: The value obtained 0.89 suggests that the correlation between
annual income and annual expenditure is high and positive. This implies
that the more one earns the more one spends.

REGRESSION

BASIC CONCEPTS
This is a concept, which refers to the changes which occur in the dependent
variable as a result of changes occurring on the independent variable.
Knowledge of regression is particularly very useful in business statistics
where it is necessary to consider the corresponding changes on dependant
variables whenever independent variables change. It should be noted that
most business activities involve a dependent variable and either one or
more independent variable. Therefore knowledge of regression will enable
a business statistician to predict or estimate the expenditure value of a
dependant variable when given an independent variable e.g. consider the
above example for annual incomes and annual expenditures. Using the

ke
regression techniques one can be able to determine the estimated
expenditure of a given family if the annual income is known and vice versa.

o.
.c
Difference between correlation and regression analysis
es
There are two important points of difference between correlation and
regression analysis
ot

1. Correlation coefficient as a measure of degree of relationship

cn

between x and y. while regression analysis is to study the nature of

relationship between the variables.
ne

2. The cause and effect relation is clearly indicated through regression

.k

analysis while correlation is merely a tool of ascertaining the degree

of relationship between two variables.
w
w

The general equation used in simple regression analysis is as follows

w

y = a + bx
Where y = Dependant variable
a= Interception y axis (constant)
b = Slope on the y axis
x = Independent variable
i. The determination of the regression equation such as given
above is normally done by using a technique known as ―the
method of least squares‘.
Regression equation of y on x i.e. y = a + bx

101
y x x Line of best fit
x x
x x
x x
x x
x x

The following sets of equations normally known as normal equation are

ke
used to determine the equation of the above regression line when given a
set of data.

o.
Σy = an + bΣx
Σxy = aΣx + bΣx2 .c
es
Where Σy = Sum of y values
Σxy = sum of the product of x and y
ot

Σx = sum of x values
cn

Σx2= sum of the squares of the x values

a = The intercept on the y axis
ne

b = Slope gradient line of y on x

.k

NB: The above regression line is normally used in one way only i.e. it is
used to estimate the y values when the x values are given.
w

Regression line of x on y i.e. x = a + by

w

- The fact that regression lines can only be used in one way leads to
w

what is known as a regression paradox

- This means that the regression lines are not ordinary mathematical
line graphs which may be used to estimate the x and y
simultaneously
- Therefore one has to be careful when using regression lines as it
becomes necessary to develop an equation for x and y before doing
the estimation.
The following example will illustrate how regression lines are used

Example
An investment company advertised the sale of pieces of land at different
prices. The following table shows the pieces of land their acreage and costs

102
Piece of (x)Acreage (y) Cost £ xy x2
land Hectares 000
A 2.3 230 529 5.29
B 1.7 150 255 2.89
C 4.2 450 1890 17.64
D 3.3 310 1023 10.89
E 5.2 550 2860 27.04
F 6.0 590 3540 36
G 7.3 740 5402 53.29
H 8.4 850 7140 70.56
J 5.6 530 2969 31.36
2
Σx =44.0 Σy = 4400 Σxy= 25607 Σx = 254.96

Required
Determine the regression equations of
i. y on x and hence estimate the cost of a piece of land with 4.5

ke
hectares

o.
ii. Estimate the expected average if the piece of land costs £ 900,000
Σy = an + bΣxy
Σxy = a∑x + bΣx2 .c
es
ot

By substituting of the appropriate values in the above equations we have

4400 = 9a + 44b …….. (i)
cn

25607 = 44a + 254.96b …….. (ii)

ne

By multiplying equation …. (i) by 44 and equation …… (ii) by 9 we have

193600 = 396a + 1936b …….. (iii)
.k

230463 = 396a + 2294.64b …….. (iv)

w

By subtraction of equation …. (iii) from equation …… (iv) we have

w

36863 = 358.64b
w

102.78 = b
by substituting for b in …….. (i)
4400 = 9a + 44(102.78)
4400 – 4522.32 = 9a
–122.32 = 9a
-13.59 = a
Therefore the equation of the regression line of y on x is
Y = 13.59 + 102.78x
When the acreage (hectares) is 4.5 then the cost
(y) = -13.59 + (102.78 x 4.5)
= 448.92
= £ 448, 920
Note that
Where the regression equation is given by
y= a + bx
Where a is the intercept on the y axis and
103
 b is the slope of the line or regression coefficient
n is the sample size
then,

intercept a =
 y  b x
n
n xy   x y
Slope b =
n x 2    x 
2

Example
The calculations for our sample size n = 10 are given below. The linear
regression model is
y = a + bx
Table

Distance x Time y mins xy x2 y2

miles

ke
3.5 16 56.0 12.25 256

o.
2.4 13 31.0 5.76 169
4.9
4.2
19
18
93.1
75.6
.c24.01
17.64
361
324
es
3.0 12 36.0 9.0 144
ot

1.3 11 14.3 1.69 121

cn

1.0 8 8.0 1.0 64

3.0 14 42.0 9.0 196
ne

1.5 9 13.5 2.25 81

4.1 16 65.6 16.81 256
.k

Σx = 28.9 Σy = 136 Σxy = Σx2 = 99.41 Σy2= 1972

w

435.3
w
w

10  435.3  28.9 136 422.6

The Slope b = 
10  99.41  28.92 158.9

= 2.66

136   2.66  28.9 

and the intercept a =
10

= 5.91
We now insert these values in the linear model giving
y = 5.91 + 2.66x
or
Delivery time (mins) = 5.91 + 2.66 (delivery distance in miles)

104
The slope of the regression line is the estimated number of minutes per
mile needed for a delivery. The intercept is the estimated time to prepare
for the journey and to deliver the goods which is the time needed for each
journey other than the actual traveling time.

PREDICTION WITHIN THE RANGE OF SAMPLE DATA

We can use the linear regression model to predict the mean of dependant
variable for any given value of independent variable
For example if the sample model is given by
Time (min) = 5.91 + 2.66 (distance in miles)
Then if the distance is 4.0 miles then our estimated mean time is
Ý = 5.91 + 2.66 x 4.0 = 16.6 minutes

Multiple Linear Regression Models

There are situations in which there is more than one factor which influence
the dependent variable

ke
Example
Cost of production per week in a large department depends on several

o.
factors;
i. Total numbers of hours worked .c
es
ii. Raw material used during the week
iii. Total number of items produced during the week
ot

iv. Number of hours spent on repair and maintenance

cn

It is sensible to use all the identified factors to predict department costs

Scatter diagram will not give the relationship between the various factors
ne

and total costs

.k

The linear model for multiple linear regression if of the type; (which is the
line of best fit).
w

y = α + b1x1 +b2x2 +………… + bnxn

w

We assume that errors or residuals are negligible.

w

In order to choose between the models we examine the values of the

multiple correlation coefficient r and the standard deviation of the
residuals α.
A model which describes well the relationship between y and x‘s has
multiple correlation coefficient r close to ±1 and the value of α which is
small.

Example
Odino chemicals limited are aware that its power costs are semi variable
cost and over the last six months these costs have shown the following
relationship with a standard measure of output.

105
 Month Output (standard units) Total power costs £
000
1 12 6.2
2 18 8.0
3 19 8.6
4 20 10.4
5 24 10.2
6 30 12.4

Required
i. Using the method of least squares, determine an appropriate
linear relationship between total power costs and output
ii. If total power costs are related to both output and time (as
measured by the number of the month) the following least
squares regression equation is obtained
Power costs = 4.42 + (0.82) output + (0.10) month

ke
Where the regression coefficients (i.e. 0.82 and 0.10) have t

o.
values 2.64 and 0.60 respectively and coefficient of multiple
correlation amounts to 0.976
.c
Compare the relative merits of this fitted relationship with one
es
you determine in (a). Explain (without doing any further analysis)
how you might use the data to forecast total power costs in seven
ot

months.
cn

Solution
ne

a)
Output (x) Power costs (y) x2 y2 xy
.k

12 6.2 144 38.44 74.40

w

18 8.0 324 64.00 144.00

w

19 8.6 361 73.96 163.40

w

20 10.4 400 108.16 208.00

24 10.2 576 104.04 244.80
30 12.4 900 153.76 372.00
2
Σx = 123 Σy = 55.8 Σx = 2705 Σy2 = Σxy=
542.36 1,206.60
n xy   x y
b=
n x 2    x 
2

6 1206.6  123  55.8

=
6  2705  123
2

376.2
= = 0.342
1101
106
 1
a = (Σy – bΣx)
n

1
=  (55.8 – 0.342)  123
6

= 2.29
 (Power costs) = 2.29 + 0.342 (output)

b. For linear regression calculated above, the coefficient of correlation r is

r=
 6 1206.6   123  55.8
6  2705  123 123 6  542.36  55.8  55.8

376.2
=

ke
1101 140.52

o.
= 0.96
.c
es
This show a strong correlation between power cost and output. The
ot

multiple correlations when both output and time are considered at the
same time are 0.976.
cn

We observe that there has been very little increase in r which means that
ne

inclusion of time variable does not improve the correlation significantly

The value for time variable is only 0.60 which is insignificant as compared
.k

with a t value of 2.64 for the output variable

w

In fact, if we work out correlation between output and time, there will be a
w

high correlation. Hence there is no necessity of taking both the variables.

w

Inclusion of time does improve the correlation coefficient but by a very

small amount.
If we use the linear regression analysis and attempt to find the linear
relationship between output and time i.e.

Month Output
1 12
2 18
3 19
4 20
5 24
6 30
The value of b and a will turn out to be 3.11 and 9.6 i.e. relationship will
be of the form
107
 Output = 9.6 + 3.11 × month
For this equation forecast for 7th month will be
Output = 9.6 + 3.11 × 7
= 9.6 + 21.77
= 31.37 units
Using the equation, Power costs = 2.29 + 0.34 × output
= 2.29 + 0.34 × 31.37
= 2.29 + 10.67
= 12.96 i.e. £ 12,960

Non Linear Relationships

If the scatter diagram and the correlation coefficient do not indicate linear
relationship, then the relationship may be non – linear
Two such relationships are of peculiar interest
y  ab x and y  axb
Both of these can be reduced to linear model. Simple or multiple linear

ke
regression methods are then used to determine the values of the

o.
coefficients

i. Exponential model .c
es
y  ab x
ot

Take log of both sides

log y = log a + log bx
cn

log y = log a + xlog b

ne

Let log y = Y and log a = A and log b = B

.k

Thus we get Y = A + Bx. This is a linear regression model

w
w

ii. Geometric model

w

y  axb
Using the same technique as above
Log y = log a + blog x
Y = A + bX
Where Y = log y
A = log a
X = log x
Using linear regression technique (the method of least squares), it is
possible to calculate the value of a and b

108
 CHAPTER SIX
ELEMENTS OF PROBABLITY

SPECIFIC OBJECTIVES
At the end of the topic the trainee should be able to:
 Discuss the basic concepts of probability;
 Apply the techniques of counting and set theory;
 Apply the laws of probability;
 Apply probability distribution concepts to decision problems.

Introduction
Probability is the ratio of the number of favorable cases to the total
number of equal likely cases. Probability is a very popular concept in
business management. This is because it covers the risks which may be
involved in certain business situations. It is a fact that when a business
investment is being arranged, the outcome is usually uncertain. Therefore
the concept of probability may be used to describe the degree of

ke
uncertainty of a particular business outcome. Probability may therefore
be defied as the chances of a given event occurring. Numerically,

o.
probability values range between 0 and 1. a probability of 0 implies that
.c
the event cannot occur at all. A probability of 1 implies that the event will
es
certainly occur.
Therefore other events have their probabilities with values lying between 0
ot

and 1
cn

The formula used to determine probability is as follow

ne

r Favourable outcomes
Probability (x) = 
.k

n Total outcomes
w
w

Uses of Probability in Business

1. Business games of chance e.g. Raffles Lotteries e.t.c.
w

2. Insurance firms: this is usually done when a new client or property is

being insured. The company has to be certain about the chances of the
insured risks occurring.
3. Business decision making regarding viability of projects thus the projects
with a greater probability has greater chances.

Example
A bag contains 80 balls of which 20 are red, 25 are blue and 35 are white.
A ball is picked at random what is the probability that the ball picked is:
(i) Red ball
(ii) Black ball
(iii) Red or Blue ball

Solution

109
 Number of red balls in the bag
(i) Probability of a red ball =  PR 
Total number of balls in the bag
20 1
= 
80 4
Number of black balls in the bag
(ii) Probability of black ball =  PB 
Total number of balls
0
= 0
80
20 25 20 25
(iii) P(R or B) = or  
80 80 80 80
9
=
16
Note: in probability or is replaced by a plus (+) sign. See addition
rule.

Common terms

ke
Events: an event is a possible outcome of an experiment or a result of a

o.
trial or an observation.

Mutually exclusive events

.c
es
A set of events is said to be mutually exclusive if the occurrence of any one
ot

of the events precludes the occurrence of any of the other events e.g.
when tossing a coin, the events are a head or a tail these are said to be
cn

mutually exclusive since the occurrence of heads for instance implies that
ne

tails cannot and has not occurred.

It can be represented in Venn diagram as.
.k
w
w
w

E1 E2

E1  E2 = Ø

110
 Non-mutually exclusive events
(independent events)
E1 E2

E1  E2 ≠ Ø

Consider a survey in which a random sample of registered voters is

selected. For each voter selected their sex and political party affiliation
are noted. The events ―KANU‖ and ―woman‖ are not mutually exclusive
because the selection of KANU does not preclude the possibly that the voter
is also a woman.

ke
Independent Events

o.
Events are said to be independent when the occurrence of any of the events
does not affect the occurrence of the other(s). .c
es
E.g. the outcome of tossing a coin is independent of the outcome of
the preceding or succeeding toss.
ot
cn

Example
From a pack of playing cards what is the probability of;
ne

(i) Picking either a ‗Diamond‘ or a ‗Heart‘ → mutually exclusive

.k

(ii) Picking either a ‗Flower‘ or an ‗Ace‘ → independent events

w

Solution.
w

(i) P (Diamond or Heart)

w

= P (Diamond) + P (Heart)
13 13 26
=  
52 52 52
= 0.5

(ii) P (Flower or Ace)

= P (Flower) + P (Ace) – P (Flower and Ace)
13 4 1
=  
52 52 52
4
= = 0.31
52

Note: that the formula used incase of independent events is different

to the one of mutually exclusive.

111
There are mainly four schools of thought on probability
i) the classical to priors approach
ii) the relative frequency of empirical approach
iii) the axiomatic approach
iv) the personalistic approach

The classical approach

This school of thought assumes that all the possible outcomes of an
experiment are mutually exclusive and equally likely.
P (E) = number of outcomes favorable to the occurrence of event E = a
Total number of outcome a+
b

Relative axiomatic approach

The axiomatic approach theory of probability is an honest attempt at

ke
constructing a theory probability large tree from the inadequacies of both
the classical and empirical approaches.

o.
Personality approach .c
es
According to the personality or subjective concept the probability of an
event is the degree of confidence placed in the occurrence of an event by a
ot

particular individual based on the evidence available to him.

cn

Relative frequency approach

ne

The relative frequency theoreticians agree that the only valid procedure for
.k

determining event probabilities is through repetitive experiments

P (E) = a/n
w
w

Rules of Probability
w

(a) Additional Rule – This rule is used to calculate the probability of two
or more mutually exclusive events. In such circumstances the
probability of the separate events must be added.

Example
What is the probability of throwing a 3 or a 6 with a throw of a die?

Solution
P (throwing a 3 or a 6) = 1  1  1
6 6 3

(b) Multiplicative rule

This is used when there is a string of independent events for which
individual probability is known and it is required to know the overall
probability.
112
 Example
What is the probability of a 3 and a 6 with two throws of a die?

Solution
P (throwing a 3) and P (6)
= P (3) and P (6) = 1  1  1
6 6 36

Note: 1) In probability ‗and‘ is replaced by ‗x‘ – multiplication.

2) P(x) and P(y) ≠ P(x and y) note that these two are
different. The first implies P(x) happening and P(y),
but if the order of which happened first is unimportant
then we have p(x and y).

ke
In the example above:
P (3) and P (6) = 1

o.
36
But
.c
P (3 and 6) = P (3 followed by 6) or P (6 followed by 3)
es
= [P (3) P (6)] or [P (6) P (3)]
ot

= 1 1  1
36 36 18
cn

Conditional Probability
ne

This is the probability associated with combinations of events but given that
.k

some prior result has already been achieved with one of them.
w

It‘s expressed in the form of

w

P (x|y) = Probability of x given that y has already occurred.

w

P ( xy )
P (x|y) = → conditional probability formula.
P( y)

Example:
In a competitive examination, 30 candidates are to be selected. In all 600
candidates appear in a written test, and 100 will be called for the
interview.
(i) What is the probability that a person will be called for the interview?
(ii) Determine the probability of a person getting selected if he has been
called for the interview?
(iii) Probability that person is called for the interview and is selected?

Solution:
Let event A be that the person is called for the interview and event B that
he is selected.

113
 100
(i)  P (A) = = 1
600 6
30 3
(ii) P (B|A) = 
100 10
(iii) P (AB) = P(A) × P(B|A)
= 1 3  3  1
6 10 60 20

Example:
From past experience a machine is known to be set up correctly on 90% of
occasions. If the machine is set up correctly then 95% of good parts are
expected but if the machine is not set up correctly then the probability of a
good part is only 30%.
On a particular day the machine is set up and the first component produced
and found to be good. What is the probability that the machine is set up
correctly.

ke
Solution:

o.
This is displayed in the form of a probability tree or diagram as follows:
.c
es
CS GP
GP = 0.95 CS – Correct Setting
ot
cn

IS – Incorrect Setting
CS = 0.9 BP = 0.05
ne

CS BP GP – Good Product
.k

IS = 0.1 GP = 0.3 IS GP BP – Bad Product

w
w
w

BP = 0.7

IS BP

P (CSGP) = 0.9 × 0.95 = 0.855

P (CSBP) = 0.9 × 0.05 = 0.045
P (ISGP) = 0.1 × 0.3 = 0.03
P (ISBP) = 0.1 × 0.7 = 0.07
1.00
Probability of getting a good part (GP) = CSGP or ISGP
= CSGP + ISGP
= 0.855 + 0.03 = 0.885

114
Note: Good parts may be produced when the machine is correctly set up
and also when it‘s incorrectly setup. In 1000 trials, 855 occasions when its
correctly setup and good parts produced (CSGP) and 30 occasions when it‘s
incorrectly setup and good parts produced (ISGP).

Probability that the machine is correctly set up after getting a good part.
Number of favourableoutcomes P(CSGP) 0.855
=    0.966
Total possible outcomes P(GP) 0.885
Or
P(CSGP) 0.855
= P(CS|GP) =   0.966
P(GP) 0.885

Example
In a class of 100 students, 36 are male and studying accounting, 9 are male

ke
but not studying accounting, 42 are female and studying accounting, 13 are
female and are not studying accounting.

o.
Use these data to deduce probabilities concerning a student drawn at
random. .c
es
Solution:
ot
cn

Accounting A Not accounting Total

ne

A
Male M 36 9 45
.k

Female F 42 13 55
Total 78 22 100
w
w

45
w

P(M) =  0.45
100
55
P (F) =  0.55
100
78
P(A) =  0.78
100


P A =
22
100
 0.22

36
P(M and A) = P(A and M) = = 0.36
100
P(M and A ) = 0.09
P(F and A ) = 0.13
These probabilities can be express differently as;
P(M) = P(M and A) or P(M and A )

115
 = 0.36 + 0.09 = 0.45

P(F) = P(F and A) or P(F and A )

= 0.42 + 0.13 = 0.55

P(A) = P(A and M) + P(A and F) = 0.36 + 0.42 = 0.78


P A = P( A and M) + P( A and F) = 0.09 + 0.13 = 0.22

Now calculate the probability that a student is studying accounting given

that he is male.

This is a conditional probability given as P(A|M)

P(A and M) 0.36
P(A|M) =   0.80
P(M) 0.45

ke
From the formula above we get that,
P(A and M) = P(M) P(A|M) ……………….. (i)

o.
Note that P(A|M) ≠ P(M|A)
.c
es
PA and M 
Since P(M|A) = this is known as the Bayes‘ rule.
ot

P(A)
cn
ne
.k
w
w
w

Bayes’ rule/Theorem

This rule or theorem is given by

P(A|B) =
 
PA   P B A
P(B)
It‘s used frequently in decision making where information is given the in
form of conditional probabilities and the reverse of these probabilities must
be found.

Example
Analysis of questionnaire completed by holiday makers showed that 0.75
classified their holiday as good at Malindi. The probability of hot weather
in the resort is 0.6. If the probability of regarding holiday as good given hot

116
weather is 0.9, what is the probability that there was hot weather if a
holiday maker considers his holiday good?

Solution

P(A|B) =
 
PA   P B A
P(B)

Let H = hot weather

G = Good
P(G) = 0.75 P(H) = 0.6 and P(G|H) = 0.9 (Probability of regard
holiday as good given hot weather)

Now the question requires us to get

P(H|G) = Probability of (there was) hot weather given that the
holiday has been rated as good).

=

PH P G H  0.60.9

ke
P(G) 0.75
= 0.72.

o.
Application
.c
1. A machine comprises of 3 transformers A, B and C. The machine may
es
operate if at least 2 transformers are working. The probabilities of each
transformer working are given as shown below;
ot

P(A) = 0.6, P(B) = 0.5, P(C) = 0.7

cn

A mechanical engineer went to inspect the working conditions of those

transformers. Find the probabilities of having the following outcomes
ne

i. Only one transformer operating

.k

ii. Two transformers are operating

w

iii. All three transformers are operating

w

iv. None is operating

v. At least 2 are operating
w

vi. At most 2 are operating

Solution
P(A) =0.6 P( A ) = 0.4 P(B) = 0.5 P(~B) = 0.5
P(C) = 0.7 P( C ) = 0.3

i. P(only one transformer is operating) is given by the following

possibilities
1st 2nd 3rd
P (A B C) = 0.6 x 0.5 x 0.3 = 0.09
P (A B C) = 0.4 x 0.5 x 0.3 = 0.06
P (A B C) = 0.4 x 0.5 x 0.7 = 0.14
∴ P(Only one transformer working)
= 0.09 + 0.06 + 0.14 = 0.29
117
 ii. P(only two transformers are operating) is given by the following
possibilities.
1st 2nd 3rd
P (A B C) = 0.6 x 0.5 x 0.3 = 0.09
P (A B C) = 0.6 x 0.5 x 0.7 = 0.21
P (A B C) = 0.4 x 0.5 x 0.7 = 0.14
∴ P(Only two transformers are operating)
= 0.09 + 0.21 + 0.14 = 0.44
iii. P(all the three transformers are operating).

= P(A) x P(B) x P(C)

= 0.6 x 0.5 x 0.7

= 0.21

ke
iv. P(none of the transformers is operating).

o.
= P( A ) x P( B ) x P( C )
.c
es
= 0.4 x 0.5 x 0.3
ot

= 0.06
cn

v. P(at least 2 working).

ne

= P(exactly 2 working) + P(all three working)

.k
w

= 0.44 + 0.21
w

= 0.65
w

vi. P(at most 2 working).

= P(Zero working) + P(one working) + P(two working)
= 0.06 + 0.29 + 0.44
= 0.79

118
POISSON PROBABILITY DISTRIBUTION
This is a set of probabilities which is obtained for discrete events which are
described as being rare. Occasions similar to binominal distribution but
have very low probabilities and large sample size.

Examples of such events in business are as follows:

i. Telephone congestion at midnight
ii. Traffic jams at certain roads at 9 o‘clock at night
iii. Sales boom
iv. Attaining an age of 100 years (Centureon)
Poisson probabilities are frequently applied in business situations in order
to determine the numerical probabilities of such events occurring.
The formula used to determine such probabilities is as follows
e  x
P  x 

ke
x!

o.
Where x = No. of successes
.c
⋋ = mean no. of the successes in the sample (⋋ = np)
es
e = 2.718
ot

Example 1
cn

A manufacturer assures his customers that the probability of having

defective item is 0.005. A sample of 1000 items was inspected. Find the
ne

probabilities of having the following possible outcomes

i. Only one is defective
.k

ii. At most 2 defective

w

iii. More than 3 defective

w

eλ λ x
w

P(x) =
x!
(⋋ = np = 1000 × 0.005) = 5
i. P(only one is defective) = P(1) = P(x = 1)
2.7185  51 1
= Note that 2.718-5=
1! 2.7185
5
=
2.7185
5
=
148.33
= 0.0333

ii. P(at most 2 defective) = P(x ≤ 2)

= P(0) + P(1) + P(2)

119
 e 5 50 P(1) = 0.0337
P(x = 0) =
0!
= 2.718-5 2.7185 52
P(2) =
1 2!
=
2.718 5 25
=
=
1 2  148.336
148.336
= 0.00674 = 0.08427

P(x≤2) = 0.00674 + 0.0337 +

0.08427

= 0.012471

ke
iii. P(more than 3 defective) = P(x > 3)

o.
= 1 – P0  P1  P2  P3
.c
es
POISSON MATHEMATICAL PROPERTIES
ot

1. The mean or expected value = np = λ

Where; n = Sample Size
cn

p = Probability of success
ne

2. The variance = np = ⋋
.k

3. Standard deviation = np = 
w

Example
w

The probability of a rare disease striking a given population is 0.003. A

sample of 10000 was examined. Find the expected no. suffering from the
w

disease and hence determine the variance and the standard deviation for
the above problem

Solution
Sample size n = 10000
P(a person suffering from the disease) = 0.003 = p
∴ expected number of people suffering from the disease
Mean = λ = 10000 × 0.003
= 30
= np = ⋋
Variance = np = 30
Standard deviation = np =⋋
= 30
= 5.477
120
NORMAL PROBABILITY DISTRIBUTION
In a continuous distribution, the variable can take any value within a
specified range, e.g. 2.21 or 1.64 compared to the specific values taken by
a discrete variable e.g. 1 or 3. The probability is represented by the area
under the probability density curve between the given values.
The uniform distribution, the normal probability distribution and the
exponential distribution are examples of a continuous distribution
The normal distribution is a probability distribution which is used to
determine probabilities of continuous variables
Examples of continuous variables are
o Distances
o Times
o Weights
o Heights
o Capacity e.t.c
Usually continuous variables are those, which can be measured by using the
appropriate units of measurement.

ke
Following are the properties of the normal distribution:

o.
1. The total area under the curve is = 1 which is equivalent to
the maximum value of probability.c
es
ot

Normal probability
cn

Distribution curve
ne
.k

Line of symmetry
w
w
w

Tail end Tail end

Age (Yrs)
2. The line of symmetry divides the curve into two equal halves
3. The two ends of the normal distribution curve continuously
approach the horizontal axis but they never cross it
4. The values of the mean, mode and median are all equal
NB: The above distribution curve is referred to as normal probability
distribution curve because if a frequency distribution curve is plotted from
measurements of a given sample drawn from a normal population then a
graph similar to the normal curve must be obtained.

121
It should be noted that 68% of any population lies within one standard
deviation, ±1σ
95% lies within two standard deviations ±2σ
99% lies within three standard deviations ±3σ

Where σ = standard deviation

ke
o.
.c
es
ot

0 Z
STANDARDIZATION OF VARIABLES
cn

Before we use the normal distribution curve to determine probabilities of

ne

the continuous variables, we need to standardize the original units of

measurement, by using the following formula.
.k

 μ
Z=
w

σ
w

Where χ = Value to be standardized

w

Z = Standardization of x
µ = population mean
σ = Standard deviation

Example
A sample of students had a mean age of 35 years with a standard deviation
of 5 years. A student was randomly picked from a group of 200 students.
Find the probability that the age of the student turned out to be as follows
i. Lying between 35 and 40
ii. Lying between 30 and 40
iii. Lying between 25 and 30
iv. Lying beyond 45 yrs
v. Lying beyond 30 yrs
vi. Lying below 25 years

Solution
122
(i). The standardized value for 35 years
  35 - 35
Z= = =0
σ 5

The standardized value for 40 years

  40 - 35
Z= = = 1
σ 5

∴ the area between Z = 0 and Z = 1 is 0.3413 (These values are checked

from the normal tables see appendix)
The value from standard normal curve tables.
When z = 0, p = 0
And when z = 1, p = 0.3413
Now the area under this curve is the area between z = 1 and z = 0
= 0.3413 – 0 = 0.3413
∴ the probability age lying between 35 and 40 yrs is 0.3413

ke
(ii). 30 and 40 years
 

o.
30  35 5
Z= = = = -1
 
σ 5
40  35
.c 5
es
Z= = = 1
σ 5
ot
cn

∴ the area between Z = -1 and Z = 1 is

= 0.3413 (lying on the positive side of zero) + 0.3413 (lying on the
ne

negative side of zero)

.k

P = 0.6826
w

∴ the probability age lying between 30 and 40 yrs is 0.6826

w

(iii). 25 and 30 years

w

  25  35 10
Z= = = = -2
σ 5 5

  30  35
Z= = = -1
σ 5

∴ the area between Z = -2 and Z = -1

Probability area corresponding to Z = -2
= 0.4772 (the z value to check from the tables is 2)
Probability area corresponding to Z = -1
= 0.3413 (the z value for this case is 1)
∴ the probability that the age lies between 25 and 30 yrs
= 0.4772 – 0.3413 (The area under this curve)
P = 0.1359

123
iv). P(beyond 45 years) is determined as follow = P(x > 45)
  45  35  10
Z= = = =+2
σ 5 5

Probability corresponding to Z = 2 = 0.4772 = probability of between 35

and 45
∴ P(Age > 45yrs) = 0.5000 – 0.4772
= 0.0228

PRACTICE QUESTION

Question One

ke
The quality controller, Mr. Brooks, at Queensville Engineers has become
aware of the need for an acceptance sampling programme to check the

o.
quality of bought-in components. This is of particular importance for a
.c
problem the company is currently having with batches of pump shafts
es
bought in from a local supplier. Mr. Brooks proposes the following criteria
to assess whether or not to accept a large batch of pump shafts from this
ot

supplier.
cn

From each batch received take a random sample of 50 shafts, and accept
ne

that batch if no more than two defectives are found in the sample.
.k

Mr. Brooks needed to calculate the probability of accepting a batch P a,

w

when the proportion of defectives in the batch, p, is small (under 10%, say)
w
w

Required:
a) Explain why the Poisson distribution is appropriate to investigate this
situation.
b) Using the Poisson distribution, determine the probability of accepting
a batch Pa, containing p=2% defectives if the method is used.
Determine Pa, for p = 0%, 2%, 5%, 10%, 15%

Question Two
A woven cloth is liable to contain faults and is subjected to an inspection
procedure. Any fault has a probability of 0.7 that it will be detected by the
procedure, independent of whether any other fault is detected or not.

Required:
a) If a piece of cloth contains three faults, A, B and C,

124
 i) Calculate the probability that A and C are detected, but that B is
undetected;
ii) Calculate the probability that any two of A, B and C be detected,
the other fault being undetected;
iii) State the relationship between your answers to parts (i) and (ii)
and give reasons for this.

b) Suppose now that, in addition to the inspection procedure given above,

there is a secondary check which has a probability of 0.6 of detecting
each fault missed by the first inspection procedure. This probability of
0.6 applies independently to each and every fault undetected by the
first procedure.
i) Calculate the probability that a piece of cloth with one fault has
this fault undetected by both the inspection procedure and the
secondary check;
ii) Calculate the probability that a piece of cloth with two faults has
one of these faults detected by either the inspection procedure

ke
or the secondary check, and one fault undetected by both;
iii) Of the faults detected, what proportions are detected by the

o.
inspection procedure and what proportion by the secondary
check? .c
es
ot
cn

Question Three
ne

A company has three production sections S1, S2 and S3 which contribute 40%,
35% and 25%, respectively, to total output. The following percentages of
.k

faulty units have been observed:

w

S1 2% (0.02)
w

S2 3% (0.03)
w

S3 4% (0.04)

There is a final check before output is dispatched. Calculate the probability

that a unit found faulty at this check has come from section 1, S1

Question Four
Assuming a Binomial Distribution what is the probability of a salesman
making 0, 1, 2,3,4,5 or 6 sales in 6 visits if the probability of making a sale
on a visit is 0.3?

Do not use tables for this question.

Question Five

125
Records show that 60% of students pass their examinations at first attempt.
Using the normal approximation to the binomial, calculate the probability
that at least 65% of a group of 200 students will pass at the first attempt.

Question Six
A batch of 5000 electric lamps has a mean life of 1000 hours and a standard
deviation of 75 hours. Assume a normal distribution.

a) How many lamps will fail before 900 hours?

b) How many lamps will fail between 950 and 1000 hours?
c) What proportion of lamps will fail before 925 hours?
d) Given the same mean life, what would the standard deviation
have to be to ensure that no more than 20% of lamps fail before
916 hours?

ke
o.
.c
es
ot
cn
ne
.k
w
w
w

126
 CHAPTER SEVEN

SAMPLING

SPECIFIC OBJECTIVE
At the end of the topic the trainee should be able to:
 Discuss the reasons for sampling;
 Differentiate between sampling and census;
 Discuss the various types of sampling.

Introduction
Census method and sampling method can be used in sampling.
Sampling is the only tool which helps to know the characteristics of the
universe or population by examining only a small part of it.

Advantages of sampling method

i) It is cheaper to collect data as only a small part of the whole

ke
population is studied.
ii) The data are collected and analyzed more quickly. Thuds

o.
sampling saves a lot of time.
iii) .c
A good quality of labor with better supervision can be provided
es
since only a part of the whole population is to be studied.
iv) An investigation of small part of the population gives us more
ot

detailed information.
cn

Disadvantages of sampling
ne

 Non accurate
.k

 Reliability is low.
w

Size of the sample

w

An important decision that has to be taken while adopting a sampling

w

technique is about the size of the sample. However the following two
considerations may be kept in mind in determining the appropriate size of
the sample;
a) The size of the sample should increase as the variation in the
individual item increases.
b) The greater the degree of accuracy desired the larger should
be the sample size.

Sampling and non- sampling errors

A sample being only a part of a population cannot represent the population
no matter how carefully the sample is selected. This gives rise to the
difference between the value of sample statistics and the true value of the
corresponding population parameters. Such difference is called sampling
error for that sample. These errors can be avoided through proper selection

127
of questionnaires, following up the non-response, proper training of the
investigator, and correct manipulation of the collected information.

Sampling errors are of two types.

I. Biased errors
II. Unbiased errors

Biased errors
These errors arise from any bias in selecting estimation.

Unbiased errors
These errors arise due to chance difference between the member of
population included in the sample and those not included. Thus the total
sampling errors is made up of errors due to bias and the random sampling.

Causes of bias
I. Faulty process of selection

ke
II. Faulty work during the collection of information

o.
Faulty methods of analysis.
.c
Sampling bias means a systematic component of error which deprives
es
statistical results of its representatives. Bias is introduced by the following
methods of selection.
ot
cn

i) Deliberate selection
Bias originates from deliberate selection which is biased on personal
ne

judgment of what is representative.

.k

ii) substitution
w

Sometime sit becomes difficult to make contact with certain member or

w

information is not obtained from certain units then we substitute

w

members or units that are conveniently available such substitution

introduce bias.

iii) incomplete coverage

Bias also enters when we fail to cover the whole of he selected sample.

iv) haphazard selection

Haphazard human selection can also introduce bias as every human being
has a tendency away from randomness in his choice.

v) Inadequate interviewing
Bias also enters when the interviewing is hasty, incomplete and misleading.

Sampling distribution

128
A sampling distribution is defined as a probability distribution of the values
of a statistics such as mean, standard deviation, proportion computed from
all possible samples of the same size. Sampling distribution constitutes the
theoretical basis of statistical inference and are considerable important in
business decision making.

Methods of Sampling
a). Random or probability sampling methods
They include:
i. Simple random sampling
ii. Stratified sampling
iii. Systematic sampling
iv. Multi stage sampling
b). Non random probability sampling methods

ke
These consist of:

o.
i. Judgment sampling
ii. Quota sampling
.c
es
iii. Cluster sampling
ot
cn

Simple Random Sampling

This refers to the sampling technique in which each and every item of the
ne

population is given an equal chance of being included in the sample. Since

.k

selection of items in the sample depends entirely on chance, this method is

also called chance selection or representative sampling.
w

It is assumed that if the sample is chosen at random and if the size of the
w

sample is sufficiently large, it will represent all groups in the population

w

Random sampling is of 2 types; sampling with replacement and sampling

without replacement
Sampling is said to be with replacement when from a finite population a
sampling unit is drawn observed and then returned to the population before
another unit is drawn. The population in this case remains the same and a
sampling unit might be selected more than once
If on the other hand a sampling unit is chosen and not retuned to the
population after it has been observed the sampling is said to be without
replacement.
Random samples may be selected by the help of lottery method or table of
random numbers (such as tippet‘s table of random numbers, Fischer and
Yates numbers or Kendall and Babington Smith numbers.)

Stratified sampling

129
In this case the population is divided into groups in such a way that units
within each group are as similar as possible in a process called
stratification. The groups are called strata. Simple random samples from
each of the strata are collected and combined into a simple. This technique
of collecting a sample from a population is called stratified sampling.
Stratification may be by age, occupation income group e.t.c.

Systematic Sampling
This sampling is a part of simple random sampling in ascending or
descending orders. In systematic sampling a sample is drawn according to
some predetermined object. Suppose a population consists of 1000
Units, then every tenth, 20th or 50th item are selected. This method is very
easy and economical. It also saves a lot of time

Multistage sampling
This is similar to stratified sampling except division is done on
geographical/location basis, e.g. a country can be divided into provinces

ke
and then survey is done in 4 towns in each province. This helps to cut
traveling costs for a surveyor.

o.
Cluster Sampling .c
es
This is where a few geographical regions e.g. a location, town or village are
selected at random and say every single household or shop in that area is
ot

interviewed. This again cuts on costs.

cn

Judgment Sampling
ne

Here the interviewer selects whom to interview believing that their view is
.k

more fundamental since they might be directly affected e.g. to find out
effects of public transport one may chose to interview only people who
w

don‘t own cars and travel frequently to work.

w
w

Merits of random or probability sampling

i) Since the sample is objective and unbiased, it is defensive before
the supervisor.
ii) The size of the sample does not depend upon the expediency or
mere tradition but on demonstrable statistical method.
iii) Te degree of deviation from the parameter i.e. the statistical
measure based on population can be estimated.
iv) It provides a more accurate method of drawing conclusion.
The sample may be combined and evaluated even though accomplished by
different individual.

PRACTICE QUESTIONS
Question One

130
A firm purchases a very large quantity of metal off-cuts and wishes to know
the average weight of an off-cut. A random sample of 625 off-cuts is
weighed and it is found that the mean sample weight is 150 grams with a
sample standard deviation of 30 grams. What is the estimate of the
population mean and what is the standard error of the mean? What would
be the standard error if the sample size was 1225?

Question Two
A sample of 80 is drawn at random from a population of 800. The sample
standard deviation was found to be 6 grams.
- What is the finite population correction factor?
- What is the approximation of the correction factor?
- What is the standard error of the mean?

ke
o.
.c
es
ot
cn
ne
.k
w
w
w

131
 CHAPTER EIGHT

ESTIMATION AND TEST OF HYPOTHESIS

Specific Objectives
At the end of this topic the trainee should be able to:
 Define estimation
 Differentiate between the two types of estimation
 Dertemine the sampling distribution of a statistics
 Determine the confidence interval for a parameter;
 Design a simple hypothesis;
 Define errors in hypothesis testing;
 Test various hypothesis.

Introduction

Statistical inference

ke
It is the process of drawing conclusions about attributes of a population
based upon information contained in a sample (taken from the population).

o.
It is divided into estimation of parameters and testing of hypothesis.
.c
Symbols for statistic of population parameters are as follows.
es
Sample Population
ot

Statistic Parameter
cn

Arithmetic mean x µ
Standard deviation s σ
ne

Number of items n N
.k
w

Statistical estimation
w

Def: It is the procedure of using statistic to estimate a population

w

parameter. It is divided into point estimation (where an estimate of a

population parameter is given by a single number) and interval estimation
(where an estimate of a population is given by a range in which the
parameter may be considered to lie) e.g. a bus meant to take a class of 100
students (population N) for trip has a limit to the maximum weight of 600kg
of which it can carry, the teacher realizes he has to find out the weight of
the class but without enough time to weigh everyone he picks 25 students
selected at random (sample n = 25). These students are weighed and their
average weight recorded as 64kg ( X - mean of a sample) with a standard
deviation (s), now using this the teacher intends to estimate the average
weight of the whole class (µ – population mean) by using the statistical
parameters standard deviation (s), and mean of the sample ( x ).

132
Characteristic of a good estimator
(i) Unbiased: where the expected value of the statistic is equal to
the population parameter e.g. if the expected mean of a
sample is equal to the population mean
(ii) Consistency: where an estimator yields values more closely
approaching the population parameter as the sample increases
(iii) Efficiency: where the estimator has smaller variance on
repeated sampling.
(iv) Sufficiency: where an estimator uses all the information
available in the data concerning a parameter

Confidence Interval
The interval estimate or a ‗confidence interval‘ consists of a range (upper
confidences limit and lower confidence limit) within which we are
confident that a population parameter lays and we assign a probability that

ke
this interval contains the true population value
The confidence limits are the outer limits to a confidence interval.

o.
Confidence interval is the interval between the confidence limits. The
.c
higher the confidence level the greater the confidence interval. For
es
example
A normal distribution has the following characteristic
ot

i. Sample mean ± 1.960 σ includes 95% of the population

cn

ii. Sample mean ± 2.575 σ includes 99% of the population

ne
.k

Sampling distribution
w

LARGE SAMPLES
w

These are samples that contain a sample size greater than 30(i.e. n>30)
w

(a) Estimation of population mean

Here we assume that if we take a large sample from a population then the
mean of the population is very close to the mean of the sample
Steps to follow to estimate the population mean includes
i. Take a random sample of n items where (n>30)
ii. Compute sample mean ( X ) and standard deviation (S)
iii. Compute the standard error of the mean by using the following
formula
s
Sx =
n
Where S x = Standard error of mean
S = standard deviation of the sample
n = sample size
iv. Choose a confidence level e.g. 95% or 99%
133
 v. Estimate the population mean as under
Population mean µ = χ ± (appropriate number) ×S x
‗Appropriate number‘ means confidence level e.g. at 95%
confidence level is 1.96 this number is usually denoted by Z and is
obtained from the normal tables.

Example
The quality department of a wire manufacturing company periodically
selects a sample of wire specimens in order to test for breaking strength.
Past experience has shown that the breaking strengths of a certain type of
wire are normally distributed with standard deviation of 200 kg. A random
sample of 64 specimens gave a mean of 6200 kgs. Find out the population
mean at 95% level of confidence

Solution

ke
Population mean = χ ± 1.96 S x
Note that sample size is already n > 30 whereas s and x are given thus step

o.
i), ii) and iv) are provided.
Here: X = 6200 kgs .c
es
s 200
Sx = = = 25
ot

n 64
cn

Population mean = 6200 ± 1.96(25)

ne

= 6200 ± 49
= 6151 to 6249
.k

At 95% level of confidence, population mean will be in between 6151 and

w

6249
w
w

Finite Population Correction Factor (FPCF)

If a given population is relatively of small size and sample size is more than
5% of the population then the standard error should be adjusted by
multiplying it by the finite population correction factor
N n
FPCF is given by =
n 1
Where N = population size
n = sample size

Example
A manager wants an estimate of sales of salesmen in his company. A
random sample 100 out of 500 salesmen is selected and average sales are
found to be Shs. 75,000. If a sample standard deviation is Shs. 15000 then
find out the population mean at 99% level of confidence

134
Solution
Here N = 500, n = 100, X = 75000 and S = 15000
Now
Standard error of mean
s N n
= Sx = x
n n 1

=
15000
x
500  100
100 500  1

15000 400
= x
10 499
15000
= (0.895)
10

ke
Sx = 1342.50 at 99% level of confidence

o.
Population mean = X ± 2.58 S x
.c
es
=shs 75000 ± 2.58(1342.50)
=shs 75000 ± 3464
ot

= Shs 71536 to 78464

cn

(b) Estimation of difference between two means

ne

We know that the standard error of a sample is given by the value of the
standard deviation (σ) divided by the square root of the number of items in
.k

the sample ( n ).
w

But, when given two samples, the standard errors is given by

w
w

S A2 S B2
S X
AX B = 
n A nB
Also note that we do estimate the interval not from the mean but from the
difference between the two sample means i.e. X A  X B  .
The appropriate number of confidence level does not change
Thus the confidence interval is given by;
X A  X B  ± Confidence level SX  X  A B
= X  XB  ± Z S AX B 
A
X

Example
Given two samples A and B of 100 and 400 items respectively, they have the
means X 1 = 7 ad X 2 = 10 and standard deviations of 2 and 3 respectively.
Construct confidence interval at 70% confidence level?

135
Solution
Sample A B
X1 = 7 X 2 = 10
n1 = 100 n2 = 400
S1 = 2 S2 = 3
The standard error of the samples A and B is given by
4 9
S X  X  = 
A B 100 400
25 5
= =
400 20

=¼ = 0.25

At 70% confidence level, then appropriate number is equal to 1.04 (as read

ke
from the normal tables)

o.
X 1  X 2 = 7 – 10 = - 3 = 3
.c
We take the absolute value of the difference between the means e.g. the
es
value of X = absolute value of X i.e. a positive value of X.
Confidence interval is therefore given by
ot

= 3± 1.04 (0.25 ) From the normal tables a z value of 1.04 gives a

cn

value of 0.7.
ne

= 3± 0.26
.k
w

= 3.26 and 2.974

w

Thus 2.974 ≤ X ≤ 3.26

w

Example 2
A comparison of the wearing out quality of two types of tyres was obtained
by road testing. Samples of 100 tyres were collected. The miles traveled
until wear out were recorded and the results given were as follows
Tyres T1 T2
Mean X1 = 26400 miles X 2 = 25000 miles
Variance S21= 1440000 miles S22= 1960000 miles
Find a confidence interval at the confidence level of 70%

Solution
X1 = 26400
X 2 = 25000

136
Difference between the two means
 
X 1  X 2 = (26400 – 25000)
= 1,400

Again we take the absolute value of the difference between the two means
We calculate the standard error as follows
S12 S22
S X = 
AX B  n1 n2

1, 440, 000 1,960, 000

= 
100 100

ke
= 184.4

o.
Confidence level at 70% is read from the normal tables as 1.04 (Z = 1.04).
Thus the confidence interval is calculated as follows
= 1400 ± (1.04) (184.4)
.c
es
ot

= 1400 ± 191.77
cn

or (1400 – 191.77) to (1400 + 191.77)

ne

1,208.23 ≤ X ≤ 1591.77
.k
w

c) Estimation of population proportions

w

This type of estimation applies at the times when information cannot be

w

given as a mean or as a measure but only as a fraction or percentage

The sampling theory stipulates that if repeated large random samples are
taken from a population, the sample proportion ―p‘ will be normally
distributed with mean equal to the population proportion and standard
error equal to
Pq
Sp = = Standard error for sampling of population
n
proportions
Where n is the sample size and q = 1 – p.
The procedure for estimating a proportion is similar to that for estimating a
mean, we only have a different formula for calculating standard.

Example 1
In a sample of 800 candidates, 560 were male. Estimate the population
proportion at 95% confidence level.

137
Solution
Here
560
Sample proportion (P) = = 0.70
800
q = 1 – p = 1 – 0.70 = 0.30
n = 800
pq
=
 0.70  0.30 
n 800

Sp = 0.016

Population proportion
= P ± 1.96 Sp where 1.96 = Z.
= 0.70 ± 1.96 (0.016)

ke
= 0.70 ± 0.03

o.
= 0.67 to 0.73 .c
es
= between 67% to 73%
ot
cn

Example 2
A sample of 600 accounts was taken to test the accuracy of posting and
ne

balancing of accounts where in 45 mistakes were found. Find out the

.k

population proportion. Use 99% level of confidence

w

Solution
w

Here
w

45
n = 600; p = = 0.075
600
q = 1 – 0.075 = 0.925

Sp =
pq
=
 0.075 0.925
n 600

= 0.011

Population proportion
= P ± 2.58 (Sp)
= 0.075 ± 2.58 (0.011)

= 0.075 ± 0.028

138
 = 0.047 to 0.10

= between 4.7% to 10%

d) Estimation of difference between population proportions

Let the two proportions be given by P1 and P2, respectively
Then the difference (absolute) between the two proportions is given by (P1
– P2)
The standard error is given by
pq pq p n  p2 n2
S P  P  =  where p = 1 1 and q = 1 - p
1 2 n1 n2 n1  n2

Then given the confidence level, the confidence interval between the two
population proportions is given by
(P1 – P2) ± Confidence level S P1  P2 

ke
o.
pq pq
= (P1 – P2) ± Z 
n1 n2 .c
es
p1n1  p2 n2
Where P = always remember to convert P1 & P2 to P.
n1  n2
ot
cn

2. SMALL SAMPLES
ne

(a) Estimation of population mean

If the sample size is small (n<30) the arithmetic mean of small samples are
.k

not normally distributed. In such circumstances, students‘

w

t-distribution must be used to estimate the population mean.

w

In this case
w

Population mean µ = X ± tsx

X = Sample mean
s
Sx =
n

 x  x
2

S = standard deviation of samples = for small samples.

n 1
n = sample size
v = n – 1 degrees of freedom.

The value of t is obtained from students t-distribution tables for the

required confidence level

Example

139
A random sample of 12 items is taken and is found to have a mean weight
of 50 grams and a standard deviation of 9 grams
What is the mean weight of population?
a) with 95% confidence
b) with 99% confidence

Solution
s 9
X  50; S = 9; v = n – 1 = 12 – 1 = 11; Sx  
n 12
µ = x‘ ± tsx

At 95% confidence level

 9 
µ = 50 ± 2.262  
 12 

= 50 ± 5.72 grams

ke
o.
.c
Therefore we can state with 95% confidence that the population mean is
es
between 44.28 and 55.72 grams
ot

At 99% confidence level

 9 
cn

µ = 50 ± 3.25  
 12 
ne
.k

= 50 ± 8.07 grams
w

Therefore we can state with 99% confidence that the population mean is
w

between 41.93 and 58.07 grams

w

Note: To use the t distribution tables it is important to find the degrees of

freedom (v = n – 1). In the example above v = 12 – 1 = 11
From the tables we find that at 95% confidence level against 11 and under
0.05, the value of t = 2.201

HYPOTHESIS TESTING

Definition
- A hypothesis is a claim or an opinion about an item or issue. Therefore it
has to be tested statistically in order to establish whether it is correct or
not correct
- Whenever testing a hypothesis, one must fully understand the 2 basic
hypotheses to be tested namely
i. The null hypothesis (H0)

140
 ii. The alternative hypothesis(H1)

The null hypothesis

This is the hypothesis being tested, the belief of a certain characteristic
e.g. Kenya Bureau of Standards (KBS) may walk to a sugar making company
with an intention of confirming that the 2kgs bags of sugar produced are
actually 2kgs and not less, they conduct hypothesis testing with the null
hypothesis being: H0 = each bag weighs 2kgs. The testing will set out to
confirm this or to refute it.

The alternative hypothesis

While formulating a null hypothesis we also consider the fact that the belief
might be found to be untrue hence we will reject it. We therefore
formulate an alternative hypothesis which is a contradiction to the null
hypothesis, thus when we reject the null hypothesis we accept the
alternative hypothesis.
In our example the alternative hypothesis would be

ke
H1 = each bag does not weigh 2kg

o.
Acceptance and rejection regions
.c
All possible values which a test statistic may either assume consistency with
es
the null hypothesis (acceptance region) or lead to the rejection of the null
hypothesis (rejection region or critical region)
ot
cn

The values which separate the rejection region from the acceptance region
ne

are called critical values

.k

Type I and type II errors

w

While testing hypothesis (H0) and deciding to either accept or reject a null
w

hypothesis, there are four possible occurrences.

w

a) Acceptance of a true hypothesis (correct decision) – accepting the null

hypothesis and it happens to be the correct decision. Note that
statistics does not give absolute information, thus its conclusion could
be wrong only that the probability of it being right are high.
b) Rejection of a false hypothesis (correct decision).
c) Rejection of a true hypothesis – (incorrect decision) – this is called type I
error, with probability = α.
d) Acceptance of a false hypothesis – (incorrect decision) – this is called
type II error, with probability = β.

Levels of significance
A level of significance is a probability value which is used when conducting
tests of hypothesis. A level of significance is basically the probability of one
making an incorrect decision after the statistical testing has been done.
Usually such probability used are very small e.g. 1% or 5%

141
 0.5000 0.4900

1% provision for
errors

0
Critical value

ke
o.
0.45

.c
es
5% = 0.05
ot

Critical region
cn
ne

0
.k

Crititical value = -1.65

w
w
w

NB: If the standardized value of the mean is less than –1.65 we reject the
null hypothesis (H0) and accept the alternative Hypothesis (H1) but if the
standardized value of the mean is more than –1.65 we accept the null
hypothesis and reject the alternative hypothesis

The above sketch graph and level of significance are applicable when the
sample mean is < (i.e. less than the population mean)

142
The following is used when sample mean > population mean

Acceptance region

Critical region (rejection

region)

5% = 0.05

ke
0 Z = 1.65 (critical value)

o.
NB: If the sample means standardized value < 1.65, we accept the null
.c
hypothesis but reject the alternative. If the sample means value > 1.65 we
es
reject the null hypothesis and accept the alternative hypothesis
The above sketch is normally used when the sample mean given is greater
ot

than the population mean

cn

Accept null hyp( reject Alternative

ne

hyp)
.k
w
w
w

Reject null hyp (accept alt hyp) Reject null hyp

(accept alt hyp)

0.05% = 0.05 0.495 0.495 0.5% = 0.05

-2.58 +2.58
NB: if the standardized value of the sample mean is between –2.58 and
+2.58 accept the null hypothesis but otherwise reject it and therefore
accept the alternative hypothesis

143
TWO TAILED TESTS
A two tailed test is normally used in statistical work(tests of significance)
e.g. if a complaint lodged by the client is about a product not meeting
certain specifications i.e. the item will generate a complaint if its
measurements are below the lower tolerance limit or above the upper
tolerance limit

Region of acceptance
for
H0

ke
Critical region Critical region

o.
.c
es
15cm 17 ½ cm
ot

NB: Alternative hypothesis is usually rejected if the standardized value of

cn

the sample mean lies beyond the tolerance limits (15cm and 17 ½ cm).
ne

ONE TAILED TEST

.k

This is a test where the alternative hypothesis (H1:) is only concerned with
w

one of the tails of the distribution e.g. to test a business complaint if the
w

complaint is above the measurements of item being shorter than is

w

required.
E.g. a manufacturer of a given brand of bread may state that the average
weight of the bread is 500 gms but if a consumer takes a sample and weighs
each of the pieces of bread and happens to have a mean of 450 gms he will
definitely complain about the bread which is underweight. The statistical
analysis to be done will concentrate on the left tail of the normal
distribution in which one will have to establish whether 450 gms being less
than 500g is statistically significant. Such a test therefore is referred to as
one tailed test.

144
left

On the other hand the test may compuliate on the right hand tail of the

ke
normal distribution when this happens the major complaint is likely to do

o.
with oversize items bought. Therefore the test is known as one tailed as the
focus is on one end of the normal distribution.
.c
es
Number of standard errors
ot

Two tailed One tailed

test test
cn

5% level of 1.96 1.65

ne

significance
1% level of 2.58 2.33
.k

significance
w
w

HYPOTHESIS TESTING PROCEDURE

w

Whenever a business complaint comes up there is a recommended

procedure for conducting a statistical test. The purpose of such a test is to
establish whether the null hypothesis or alternative hypothesis is to be
accepted.
The following are steps normally adopted
1. Statement of the null and alternative hypothesis
2. Statement of the level of significance to be used.
3. Statement about the test statistic i.e. what is to be tested e.g. the
sample mean, sample proportion, difference between sample means
or sample proportions
4. Type of test whether two tailed or one tailed.
5. Statement on critical values using the appropriate level of
significance
6. Standardizing the test statistic
7. Conclusion showing whether to accept or reject the null hypothesis

145
STANDARD HYPOTHESIS TESTS
In principal, we can test the significance of any statistic related to any
probability distribution. However we will be interested in a few standard
cases. The sample statistics mean, proportion and variance, are related to
the normal, t, F, and chi squared distributions
Thus

1. Normal test
Test a sample mean ( X ) against a population mean (µ) (where samples
size n > 30 and population variance σ2 is known) and sample proportion,
P(where sample size np >5 and nq >5 since in this case the normal
distribution can be used to approximate the binomial distribution

ke
2. t test

o.
Tests a sample mean ( X ) against a population mean and especially
where the population variance is unknown and n < 30.
.c
es
3. Variance ratio test or f test
ot

It is used to compare population variances and it is used with samples of

any size drawn from normal populations.
cn
ne

4. Chi squared test

It can be used to test the association between attributes or the
.k

goodness of fit of an observed frequency distribution to a standard

w

distribution
w
w

Example 1
A certain NGO carried out a survey in a certain community in order to
establish the average at which the girls are married. The results of the
survey indicated that the marriage age for the girls is 19 years
In order to establish the validity of the mean marital age, a sample of 50
women was interviewed and the average age indicated that they got
married at the age of 16 years. However the different ages at which they
were married differed with the standard deviation of 2.1years
The sample data indicates that the marital age is less 19 years. Is this
conclusion true or not?

Required
Conduct a statistical test to either support the above conclusion drawn
from the sample statistics i.e. the marriage age is less than 19 years, use a
level of significance of 5%

146
Solution
1. Null hypothesis
H0: μ (mean marital age) = 19 years
Alternative hypothesis H1: μ (mean marital age) < 19 years
2. The level of significance is 5%
3. The test statistics is the sample mean age, X = 16 years
4. The critical value of the one tailed test (one tailed because the
alternative hypothesis is an inequality) at 5% level of significance is –
1.65

Acceptance region

ke
o.
Rejection region
.c
es
ot
cn

- 1.65 0
ne

5. The standardizes value of the sample mean is

.k

X -μ S
Z = where S x =
w

Sx n
w
w

Where, X = Sample mean

µ = Population mean
S = sample standard deviation
n = sample size
z = standard value (as per computation)
The standard value Z must fall within the acceptance region for us to
accept the null hypothesis. Thus it must be > - 1.65 otherwise we
accept the alternative hypothesis.
16  19
Z = 2.1
= - 10.1
50

6. Since –10.1 < -1.65, we reject the null hypothesis but accept the
alternative hypothesis at 5% level of significance i.e. the marriage
age in this community is significantly lower than 19 years

147
Example 2
A foreign company which manufactures electric bulbs has assured its
customers that the lifespan of the bulbs is 28 month with a standard
deviation of 4months
Recently the company embarked on a quality improvement research for
their product. After the research using new technology, a sample of 70
bulbs was tested and they gave a mean lifespan of 30.2 months
Does this justify the research undertaken? Use 1% level of significance to
conduct a statistical test in order to establish the truth about the above
question.
Testing procedure
1. Null hypothesis H0: µ = 28
Alternative hypothesis H1: µ > 28
2. The level of significance is 1% (one tailed test)
3. The test statistics is the sample mean age, x‘ = 30.2

ke
4. The critical value of the one tailed test at 5% level of significance is
+ 2.33

o.
.c
es
0.4900
ot
cn
ne

1% = 0.01
.k
w
w
w

2.33

5. The standardized value of the sample mean is

X  30.2  28
Z = = 4
= 4.6
Sx 70

6. Since 4.6 > 2.33, we reject the null hypothesis but accept the
alternative hypothesis at 1% level of significance i.e. the new sample
mean life span is statistically significant higher than the population
mean
Therefore the research undertaken was worth while or justified

Example 3

148
A construction firm has placed an order that they require a consignment of
wires which have a mean length of 10.5 meters with a standard deviation of
1.7 m
The company which produces the wires delivered 90 wires, which had a
mean length of 9.2 m., The construction company rejected the consignment
on the grounds that they were different from the order placed.

Required
Conduct a statistical test to indicate whether you support or not support
the action taken by the construction company at 5% level of significance.

Solution
Null hypothesis µ = 10.5 m
Alternative hypothesis µ ≠ 10.5 m
Level of significance be 5%
The test statistics is the sample mean X = 9.2m
The critical value of the two tailed test at 5% level of significance is ± 1.96

ke
(two tailed test).

o.
.c
es
ot
cn
ne
.k
w
w
w

149
 - 1.96 +1.96

The standardized value of the test Z =

ke
X -μ 9.2  10.5
Z = = = - 7.25
SX 1.7

o.
90

.c
Since 7.25 < 1.96, reject the null hypothesis but accept the alternative
es
hypothesis at 5% level of significance i.e. the sample mean is
ot

statistically different from the consignment ordered by the construction

company. Therefore support the action taken by the construction
cn

company
ne

TESTING THE DIFFERENCE BETWEEN TWO SAMPLE MEANS (LARGE

.k

SAMPLES)
w

A large sample is defined as one which contains 30 or more items (n≥30)

w

Where n is the sample size

w

In a business those involved are constantly observant about the standards or

specifications of the item which they sell e.g. a trader may receive a batch
of items at one time and another batch at a later time at the end he may
have concluded that the two samples are different in certain specifications
e.g. mean weight mean lifespan, mean length e.t.c. further it may become
necessary to establish whether the observed differences are statistically
significant or not. If the differences are statistically significant then it
means that such differences must be explained i.e. there are known causes
but if they are not statistically significant then it means that the difference
observed have no known causes and are mainly due to chance
If the differences are established to be statistically significant then it
implies that the complaints, which necessitated that kind of test, are
justified

150
Let X1 and X2 be any two samples whose sizes are n1 and n2 and mean X 1
and X 2. Standard deviation S1 and S2 respectively. In order to test the
difference between the two sample means, we apply the following formulas
X1  X 2
Z = 
where S X 1  X 2 =  S12 S22


S X1  X 2  n1 n2
Example 1
An agronomist was interested in the particular fertilizer yield output. He
planted maize on 50 equal pieces of land and the mean harvest obtained
later was 60 bags per plot with a standard deviation of 1.5 bags. The crops
grew under natural circumstances and conditions without the soil being
treated with any fertilizer. The same agronomist carried out an alternative
experiment where he picked 60 plots in the same area and planted the
same plant of maize but a fertilizer was applied on these plots. After the
harvest it was established that the mean harvest was 63 bags per plot with
a standard deviation of 1.3 bags

ke
o.
Required
.c
es
Conduct a statistical test in order to establish whether there was a
ot

significant difference between the mean harvest under the two types of
field conditions. Use 5% level of significance.
cn
ne

Solution
.k

H0 : µ1 = µ2
w
w

H1 : µ1 ≠ µ2
w

Critical values of the two tailed test at 5% level of significance are 1.96
The standardized value of the difference between sample means is given by
Z where

X1  X 2
Z =  
where S X 1  X 2 =
1.52 1.32


S X1  X 2  50 60

 60  63
Z =
0.045  0.028
= 11.11

151
 - 1.96 0 +1.96

Since 11.11 < -1.96, we reject the null hypothesis but accept the
alternative hypothesis at 5% level of significance i.e. the difference

ke
between the sample mean harvest is statistically significant. This implies

o.
that the fertilizer had a positive effect on the harvest of maize
Note: You don‘t have to illustrate your solution with a diagram.
.c
es
ot
cn
ne
.k
w
w
w

152
Example 2
An observation was made about reading abilities of males and females. The
observation lead to a conclusion that females are faster readers than
males. The observation was based on the times taken by both females and
males when reading out a list of names during graduation ceremonies.
In order to investigate into the observation and the consequent conclusion a
sample of 200 men were given lists to read. On average each man took 63
seconds with a standard deviation of 4 seconds
A sample of 250 women were also taken and asked to read the same list of
names. It was found that they on average took 62 seconds with a standard
deviation of 1 second.

Required
By conducting a statistical hypothesis testing at 1% level of significance
establish whether the sample data obtained does support earlier
observation or not

ke
Solution
H0: µ1 = µ2

o.
H1: µ1 ≠ µ2
.c
Critical values of the two tailed test is at 1% level of significance is 2.58.
es
X1  X 2
Z =
ot


S X1  X 2 
cn

63  62
ne

Z = = 3.45
42
 250
2
1
.k

200
w
w
w

Acceptance region

Rejection
region

- 2.58 0 +2.58 +3.45

153
 Since 3.45 > 2.33 reject the null hypothesis but accept the alternative
hypothesis at 1% level of significance i.e. there is a significant
difference between the reading speed of Males and females, thus
females are actually faster readers.

TEST OF HYPOTHESIS ON PROPORTIONS

This follows a similar method to the one for means except that the standard
error used in this case:
Pq
Sp =
n
P
Z score is calculated as, Z = Where P = Proportion found
Sp
in the sample.
Π – the hypothetical
proportion.

ke
Example
A member of parliament (MP) claims that in his constituency only 50% of the

o.
total youth population lacks university education. A local media company
.c
wanted to acertain that claim thus they conducted a survey taking a sample
es
of 400 youths, of these 54% lacked university education.
ot

Required:
cn

At 5% level of significance confirm if the MP‘s claim is wrong.

ne

Solution.
.k

Note: This is a two tailed tests since we wish to test the hypothesis that
the hypothesis is different (≠) and not against a specific alternative
w

hypothesis e.g. < less than or > more than.

w
w

H0 : π = 50% of all youth in the constituency lack university

education.
H1 : π ≠ 50% of all youth in the constituency lack university
education.
pq 0.5 x0.5
Sp = = = 0.025
n 400
0.54  0.50
Z= = 1.6
0.025

at 5% level of significance for a two-tailored test the critical value is 1.96

since calculated Z value < tabulated value (1.96).
i.e. 1.6 < 1.96 we accept the null hypothesis.
Thus the MP‘s claim is accurate.

154
HYPOTHESIS TESTING OF THE DIFFERENCE BETWEEN PROPORTIONS

Example
Ken industrial manufacturers have produced a perfume known as
―fianchetto.‖ In order to test its popularity in the market, the
manufacturer carried a random survey in Back rank city where 10,000
consumers were interviewed after which 7,200 showed preference. The
manufacturer also moved to area Rook town where he interviewed 12,000
consumers out of which 1,0000 showed preference for the product.

Required
Design a statistical test and hence use it to advise the manufacturer
regarding the differences in the proportion, at 5% level of significance.

Solution
H0 : π1 = π2
H1 : π1 ≠ π2

ke
The critical value for this two tailed test at 5% level of significance = 1.96.

o.
 P1  P2    1   2  .c
es
Now Z =
S  P1  P2 
ot
cn

But since the null hypothesis is π1 = π2, the second part of the numerator
disappear i.e.
ne

π1 - π2 = 0 which will always be the case at this level.

.k

 P1  P2 
w

Then Z =
w

S  P1  P2 
w

Where;

Sample 1 Sample 2
Sample size n1 = n2 =
10,000 12,000
Sample proportion of P1 =0.72 P2 = 0.83
success
Population proportion of Π1 Π2
success.

pq pq
Now S  p1  p2  = 
n1 n2

155
 p1n1  p2 n2
Where P =
n1  n2
And q = 1 – p
in our case
10, 000(0.72)  12, 000(0.83)
P=
10, 000  12, 000
84, 000
=
22, 000
= 0.78
 q = 0.22
0.78  0.22  0.78  0.22 
S  P1  P2   
10, 000 12, 000
= 0.00894
0.72  0.83
Z= = 12.3
0.00894

ke
o.
Since 12.3 > 1.96, we reject the null hypothesis but accept the
alternative. The differences between the proportions are statistically
.c
significant. This implies that the perfume is much more popular in Rook
es
town than in back rank city.
ot

HYPOTHESIS TESTING ABOUT THE DIFFERENCE BETWEEN TWO

cn

PROPORTIONS
ne

Is used to test the difference between the proportions of a given attribute

found in two random samples.
.k

The null hypothesis is that there is no difference between the population

w

proportions. It means two samples are from the same population.

w

Hence
w

H0 : π1 = π2
The best estimate of the standard error of the difference of P1 and P2 is
given by pooling the samples and finding the pooled sample proportions (P)
thus
p n  p2 n2
P= 1 1
n1  n2

Standard error of difference between proportions

pq pq
S  p1  p2   
n1 n2

P1  P2
And Z =
S  p1  p2 

156
Example
In a random sample of 100 persons taken from village A, 60 are found to be
consuming tea. In another sample of 200 persons taken from a village B,
100 persons are found to be consuming tea. Do the data reveal significant
difference between the two villages so far as the habit of taking tea is
concerned?

Solution
Let us take the hypothesis that there is no significant difference between
the two villages as far as the habit of taking tea is concerned i.e. π1 = π2
We are given
P1 = 0.6; n1 = 100
P2 = 0.5; n2 = 200

Appropriate statistic to be used here is given by

p n  p2 n2
P = 1 1
n1  n2

ke
=
 0.6 100    0.5 200   60  100

o.
100  200 300
= 0.53 .c
es
q = 1 – 0.53
= 0.47
ot

pq pq
S  P1  P2  = 
cn

n1 n2
ne

=
 0.53 0.47    0.53 0.47 
.k

100 200
w

= 0.0608
w

0.6  0.5
Z=
w

0.0608
= 1.64

Since the computed value of Z is less than the critical value of Z = 1.96 at
5% level of significance therefore we accept the hypothesis and conclude
that there is no significant difference in the habit of taking tea in the two
villages A and B

t distribution (student’s t distribution) tests of hypothesis (test for small

samples n < 30)
For small samples n < 30, the method used in hypothesis testing is exactly
similar to the one for large samples except that t values are used from t
distribution at a given degree of freedom v, instead of z score, the standard
error Se statistic used is also different.
Note that v = n – 1 for a single sample and n1 + n2 – 2 where two sample are
involved.
157
 a) Test of hypothesis about the population mean
When the population standard deviation (S) is known then the t statistic is
defined as
X  S
t = where S X 
SX n
Follows the students t distribution with (n-1) d.f. where
X = Sample mean
μ = Hypothesis population mean
n = sample size
and S is the standard deviation of the sample calculated by the formula

 X  X 
2

S= for n < 30
n 1
If the calculated value of t exceeds the table value of t at a specified level
of significance, the null hypothesis is rejected.

ke
Example

o.
Ten oil tins are taken at random from an automatic filling machine. The
.c
mean weight of the tins is 15.8 kg and the standard deviation is 0.5kg. Does
es
the sample mean differ significantly from the intended weight of 16kgs?
Use 5% level of significance.
ot
cn

Solution
ne

Given that n = 10; x = 15.8; S = 0.50; μ = 16; v = 9

H0 : μ = 16
.k

H1 : μ ≠ 16
w

0.5
= SX 
w

10
w

15.8  16
t = 0.5
10

0.2
=
0.16
= -1.25
The table value for t for 9 d.f. at 5% level of significance is 2.26. the
computed value of t is smaller than the table value of t. therefore,
difference is insignificant and the null hypothesis is accepted.

b) Test of hypothesis about the difference between two means

The t test can be used under two assumptions when testing hypothesis
concerning the difference between the two means; that the two are
normally distributed (or near normally distributed) populations and that the
158
standard deviation of the two is the same or at any rate not significantly
different.

Appropriate test statistic to be used is

X1  X 2
t = at (n1 + n2 – 2) d.f.
S X X 2
 1 

The standard deviation is obtained by pooling the two sample standard

deviation as shown below.

Sp =
 n1  1 S12   n2  1 S22
n1  n2  2
Where S1 and S2 are standard deviation for sample 1 & 2 respectively.
Sp Sp
Now S X 1 = and S X 2 =
n1 n2

ke
S X1X 2 = S X2  S X2 2
 

o.
1

Alternatively S X 1  X 2 = Sp
 
n1  n2 .c
es
n1n2
ot

Example
cn

Two different types of drugs A and B were tried on certain patients for
increasing weights, 5 persons were given drug A and 7 persons were given
ne

drug B. the increase in weight (in pounds) is given below

.k
w

Drug A 8 12 16 9 3
Drug B 10 8 12 15 6 8 11
w
w

Do the two drugs differ significantly with regard to their effect in increasing
weight? (Given that v= 10; t0.05 = 2.23)

Solution
H0 : μ1 = μ2
H1 : μ1 ≠ μ2

X1  X 2
t=
S X1X 2
 

Calculate for X 1 , X 2 and S

159
X1 X1 – X 1 (X1 – X 1 )2 X2 (X2 – X 2 ) (X2 – X )2
8 -1 1 10 0 0
12 +3 9 8 -2 4
13 +4 16 12 +2 4
9 0 0 15 +5 25
3 -6 36 6 -4 16
8 -2 4
11 +1 1
ΣX1 = 45 Σ(X1– X 1 ) = 0 Σ (X1 – X 1 )2= 62 ΣX2= 70 Σ (X2 – X 2 ) = 0 Σ (X2– X 2 )2= 54

X1 =
X 1
=
45
=9 X2 =
X 2

70
 10
n1 5 n2 7

62 54
S1 = = 3.94 S2 = 3
4 6

ke
 4 15.4   6  9

o.
Sp =
10 .c
es
= 3.406
ot
cn

11.6 11.6 75

S X1X 2   Or 3.406
  57
ne

5 7
.k

= 1.99
w

X1  X 2 9  10
w

t = =
w

S X1X 2 1.99
 
= 0.50

Now t0.05 (at v = 10) = 2.23 > 0.5

Thus we accept the null hypothesis.

Hence there is no significant difference in the efficacy of the two drugs in

the matter of increasing weight

Example
Two salesmen A and B are working in a certain district. From a survey
conducted by the head office, the following results were obtained. State
whether there is any significant difference in the average sales between
the two salesmen at 5% level of significance.

160
 A B
No. of sales 20 18
Average sales in shs 170 205
Standard deviation in shs 20 25

Solution
H0 : μ1 = μ2
H1 : μ1 ≠ μ2
Where

Sp =
 n1  1 S12   n2  1 S22
n1  n2  2
n1  n2
S X 1  X 2 = Sp
  n1n2

Where: X 1 =170, X 2 = 205, n1 = 20, n2 = 18, S1 = 20, S2 = 25, V = 36

ke
19   202   17   252 

o.
Sp =
20  18  2 .c
es
= 22.5
ot
cn

38
S X 1  X 2  22.5
 
ne

360
= 7.31
.k
w

170  205
t=
w

7.31
w

= 4.79
t0.05(36) = 1.9 (Since d.f > 30 we use the normal tables)

The table value of t at 5% level of significance for 36 d.f. when d.f. >30,
that t distribution is the same as normal distribution is 1.9. Since the value
computed value of t is more than the table value, we reject the null
hypothesis. Thus, we conclude that there is significant difference in the
average sales between the two salesmen

Testing the hypothesis equality of two variances

The test for equality of two population variances is based on the variances
in two independently selected random samples drawn from two normal
populations

161
Under the null hypothesis σ12  σ 22
s12
σ 12
F= Now under the H0 : σ12  σ 22 it follows that
s 22
σ 22
S12
F= which is the test statistic.
S22
Which follows F – distribution with V1 and V2 degrees of freedom. The larger
sample variance is placed in the numerator and the smaller one in the
denominator
If the computed value of F exceeds the table value of F, we reject the null
hypothesis i.e. the alternate hypothesis is accepted

Example
In one sample of observations the sum of the squares of the deviations of

ke
the sample values from sample mean was 120 and in the other sample of 12
observations it was 314. test whether the difference is significant at 5%

o.
level of significance
.c
es
Solution
Given that n1 = 10, n2 = 12, Σ(x1 – X 1 )2 = 120
ot

Σ(x2 – X 2 )2 = 314
cn
ne

Let us take the null hypothesis that the two samples are drawn from the
same normal population of equal variance
.k

H0 : σ12  σ 22
w

H1: σ12  σ 22
w
w

Applying F test i.e.

S12
F=
S22
  X1  X 1 
2

n1 1
=
 
2
X2 X 2
 n2 1
120
= 9
314
11
13.33
=
28.55
since the numerator should be greater than denominator

162
 28.55
F=  2.1
13.33

The table value of F at 5% level of significance for V 1 = 9 and V2 = 11. Since

the calculated value of F is less than the table value, we accept the
hypothesis. The samples may have been drawn from the two populations
having the same variances.

Chi square hypothesis tests (Non-parametric test)(X2)

They include amongst others
i. Test for goodness of fit
ii. Test for independence of attributes
iii. Test of homogeneity
iv. Test for population variance

The Chi square test (χ2) is used when comparing an actual (observed)
distribution with a hypothesized or explained distribution.

ke
O  E 
2

It is given by; χ =2
 Where O = Observed frequency

o.
E
.c
E = Expected frequency
The computed value of χ2 is compared with that of tabulated χ2 for a given
es
significance level and degrees of freedom.
ot
cn

i. Test for goodness of fit

These tests are used when we want to determine whether an actual sample
ne

distribution matches a known theoretical distribution

The null hypothesis usually states that the sample is drawn from the
.k

theoretical population distribution and the alternate hypothesis usually

w

states that it is not.

w
w

Example
Mr. Nguku carried out a survey of 320 families in Ateka district, each family
had 5 children and they revealed the following distribution
No. of boys 5 4 3 2 1 0
No. of girls 0 1 2 3 4 5
No. of families 14 56 110 88 40 12

Is the result consistent with the hypothesis that male and female births are
equally probable at 5% level of significance?

Solution
If the distribution of gender is equally probable then the distribution
conforms to a binomial distribution with probability P(X) = ½.
Therefore

163
 H0 = the observed number of boys conforms to a binomial distribution
with P = ½
H1 = the observations do not conform to a binomial distribution.
On the assumption that male and female births are equally probable the
probability of a male birth is P = ½. The expected number of families can
be calculated by the use of binomial distribution. The probability of male
births in a family of 5 is given by
P(x) = 5cX Px q5-x (for x = 0, 1, 2, 3, 4, 5,)

= 5cX (½)5 (Since P = q = ½)

To get the expected frequencies, multiply P(x) by the total number N =
320. The calculations are shown below in the tables

x P(x) Expected frequency =

NP(x)
0 5
c 0 ( ½ )5 = 1 320 × 1 = 10
32 32

ke
1 5
c1 ( ½ ) 5
= 5 320 × 5 = 50
32 32

o.
2 5
c2 ( ½ ) 5
= 10 320 × 10 = 100
3 5
c3 ( ½ ) 5
= 10
32
.c
320 × 10
32
= 100
es
32 32
4 5
c 4 ( ½ )5
ot

=5 320 × 5 = 50
32 32
cn

5 5
c5 ( ½ ) 5
=1 320 × 1 = 10
32 32
ne

Arranging observed and expected frequencies in the following table and

.k

calculating x2
w
w
w

O E (O – E) 2 (O – E) 2 /E
14 10 16 1.60
56 50 16 0.72
110 100 100 1.00
88 100 144 1.44
40 50 100 2.00
12 10 4 0.40
Σ(0 – E) 2 /E = 7.16

O  E 
2

χ 2
=  E

= 7.16

164
The table of χ2 for V = 6 – 1 = 5 at 5% level of significance is 11.07. The
computed value of χ2 = 7.16 is less than the table value. Therefore the
hypothesis is accepted. Thus it can be concluded that male and female
births are equally probable.

ii) Test of independence of attributes

This test disclosed whether there is any association or relationship between
two or more attributes or not. The following steps are required to perform
the test of hypothesis.
1. The null and alternative hypothesis are set as follows
H0: No association exists between the attributes
H1: an association exists between the attributes
2. Under H0 an expected frequency E corresponding to each cell
in the contingency table is found by using the formula
RC
E=
n
Where R = a row total, C = a column total and n = sample size

ke
3. Based upon the observed values and corresponding expected
frequencies the χ2 statistic is obtained using the formula

o.
O  E  .c
2
2
χ = 
es
E
4. The characteristic of this distribution are defined by the
ot

number of degrees of freedom (d.f.) which is given by

cn

d.f. = (r-1) (c-1),

Where r is the number of rows and c is number of columns
ne

corresponding to a chosen level of significance, the critical

value is found from the chi squared table
.k

5. The calculated value of χ2 is compared with the tabulated

w

value χ2 for (r-1) (c-1) degrees of freedom at a certain level of

w

significance. If the computed value of χ2 is greater than the

w

tabulated value, the null hypothesis of independence is

rejected. Otherwise we accept it.

Example
In a sample of 200 people where a particular devise was selected, 100 were
given a drug and the others were not given any drug. The results are as
follows
Drug No drug Total
Cured 65 55 120
Not cured 35 45 80
Total 100 100 200
Test whether the drug will be effective or not, at 5% level of significance.

Solution

165
Let us take the null hypothesis that the drug is not effective in curing the
disease.
Applying the χ2 test
The expected cell frequencies are computed as follows
RC 120  100
E11 = 1 1 = = 60
n 200

R1C2 120  100

E12 = = = 60
n 200

R2C1 80  100
E21 = = = 40
n 200

R2 C 2 80  100
E22 = = = 40
n 200

ke
o.
.c
es
ot
cn

The table of expected frequencies is as follows

ne

60 60 120
.k

40 40 80
100 100 200
w
w

O E (O – E) 2 (O – E) 2 /E
w

65 60 25 0.417
55 60 25 0.625
35 40 25 0.417
45 40 25 0.625
Σ(O – E) 2 /E = 2.084

Arranging the observed frequencies with their corresponding frequencies in

the following table we get

O  E 
2
2
χ =  E

= 2.084

166
V= (r –1) (c-1) = (2 – 1) (2 –1) = 1;  tabulated
2
( 0.05) = 3.841

The calculated value of χ2 is less than the table value. The hypothesis is
accepted. Hence the drug is not effective in curing the disease.

Test of homogeneity
It is concerned with the proposition that several populations are
homogenous with respect to some characteristic of interest e.g. one may be
interested in knowing if raw material available from several retailers are
homogenous. A random sample is drawn from each of the population and
the number in each of sample falling into each category is determined. The
sample data is displayed in a contingency table
The analytical procedure is the same as that discussed for the test of
independence

Example
A random sample of 400 persons was selected from each of three age

ke
groups and each person was asked to specify which types of TV programs be

o.
preferred. The results are shown in the following table
.c
es
ot

Type of program
Age group A B C Total
cn

Under 30 120 30 50 200

ne

30 – 44 10 75 15 100
45 and above 10 30 60 100
.k

Total 140 135 125 400

w

Test the hypothesis that the populations are homogenous with respect to
w

the types of television program they prefer, at 5% level of significance.

w

Solution
Let us take hypothesis that the populations are homogenous with respect to
different types of television programs they prefer
Applying χ2 test

O E (O – E) 2 (O – E) 2 /E
120 70.00 2500.00 35.7143
10 35.00 625.00 17.8571
10 35.00 625.00 17.8571
30 67.50 1406.25 20.8333
75 33.75 1701.56 50.4166
30 33.75 14.06 0.4166

167
50 62.50 156.25 2.500
15 31.25 264.06 8.4499
60 31.25 826.56 26.449
Σ(O – E) 2 /E = 180.4948

O  E 
2
2
χ =  E

The table value of χ2 for 4d.f. at 5% level of significance is 9.488

The calculated value of χ2 is greater than the table value. We reject the
hypothesis and conclude that the populations are not homogenous with
respect to the type of TV programs preferred, thus the different age groups
vary in choice of TV programs.

SUMMARY OF FORMULAE IN HYPOTHESIS

ke
Testing

o.
(a) Hypothesis testing of mean
For n>30 .c
es
X  S
Z= Where S X  at  level of significance.
ot

SX n
cn

For n < 30
X 
ne

S
t= where S X 
SX n
.k

at n – 1 d.f
w

 level of significance
w
w

(b) Difference between means (Independent samples)

For n > 30
X1  X 2
Z=
S X1X 2
 
S12 S22
Where S X 1  X 2  
  n1 n2
At  = level of significance
For n < 30
X1  X 2
t= at n1 + n2 – 2 d.f
S X1X 2
 
n1  n2
where S X 1  X 2   S p
n1n2

168
 and S p 
 n1  1 S12   n2  1 S22
n1  n2  2

(c) Hypothesis testing of proportions

p 
Z=
Sp
pq
Where: Sp =
n
p = Proportion found in sample
q=1–p
 = hypothetical proportion

(d) Difference between proportions

P P
Z= 1 2
S P1  P2 

ke
Where:

o.
pq pq
S P1  P2   
n1 n2 .c
es
p1n1  p2 n2
p=
ot

n1  n2
cn

q=1–P
(e) Chi-square test
ne

O  E 
2
2
X = 
.k

E
w

Where O = observed frequency

w

Column total × Row total

E= = expected frequency
w

Sample Size
(f) F – test (variance test)
S12
F= 2
S2
Here the bigger value between the standard deviations makes the
numerator.

169
LESSON 6 REINFORCING QUESTIONS

QUESTION ONE
A firm purchases a very large quantity of metal off-cuts and wishes to know
the average weight of an off-cut. A random sample of 625 off-cuts is
weighed and it is found that the mean sample weight is 150 grams with a
sample standard deviation of 30 grams. What is the estimate of the
population mean and what is the standard error of the mean? What would
be the standard error if the sample size was 1225?

QUESTION TWO
A sample of 80 is drawn at random from a population of 800. The sample
standard deviation was found to be 6 grams.
- What is the finite population correction factor?
- What is the approximation of the correction factor?

ke
- What is the standard error of the mean?

o.
QUESTION THREE
.c
es
State the Central Limit Theorem
ot

QUESTION FOUR
cn

a) What is statistical inference?

b) What is the purpose of estimation?
ne

c) What are the properties of good estimators?

.k

d) What is the standard error of the mean?

w

e) What are confident limits?

w

f) When is the Finite Population Correction Factor used? What is the

formula?
w

g) How are population proportions estimated?

h) What are the characteristics of the t distribution?

QUESTION FIVE
A market research agency takes a sample of 1000 people and finds that 200
of them know of Brand X. After an advertising campaign a further sample of
1091 people is taken and it‘s found that 240 know of Brand X.
It is required to know if there has been an increase in the number of people
having an awareness of Brand X at the 5% level.

QUESTION SIX

170
The monthly bonuses of two groups of salesmen are being investigated to
see if there is a difference in the average bonus received. Random samples
of 12 and 9 are taken from the two groups and it can be assumed that the
bonuses in both groups are approximately normally distributed and that the
standard deviations are about the same. The same level of significance is to
be used.

n1=12 n2=9
x1=£1060 x2=£970
s1=£63 s2=£76

The sample results were

ke
o.
QUESTION SEVEN
.c
es
Torch bulbs are packed in boxes of 5 and 100 boxes are selected randomly
to test for the number of defectives
ot

Number of Number Total

cn

Defectives of boxes defectives

0 40 0
ne

1 37 37
2
.k

17 34
3 5 15
w

4 1 4
w

5 0 0
w

100 90
The number of any individual bulb being a reject is
90
 5  0.18
100
and it is required to test at the 5% level whether the frequency of rejects
conforms to a binomial distribution.

QUESTION EIGHT
a) Define type I and type II errors.
b) What is a two-tail test?
c) What is the best estimate of the population standard deviation when the
two samples are taken

QUESTION NINE

171
Express Packets guarantee 95%of their deliveries are on time. In a recent
week 80 deliveries were made and 6 were late and the management says
that, at the 95%level there has been a significant improvement in
deliveries.

Can the MD‘s statement be supported?

If not, at what level of confidence can it be supported?
A batch of weighing machines has been purchased and one machine is
selected at random for testing. Ten weighing tests have been conducted
and the errors found are noted as follows:

ke
o.
Test Errors (gms)
1 4.6 .c
es
2 8.2
3 2.1
ot

4 6.3
cn

5 5.0
6 3.6
ne

7 1.4
.k

8 4.1
9 7.0
w

10 4.5
w
w

The purchasing manager has previously accepted machines with a mean

error of 3.8 gms and asserts that these tests are below standard.
Test the assertions at 5% level.
Compare your answers with those given in lesson 9 of the study pack

172
 CHAPTER NINE

TIME SERIES ANALYSIS

Specific Objectives
At the end of the this topic the trainee should be able to:
 Describe the components of time series;
 Describe the time series models;
 Evaluate trend and seasonal variation in the series;
 Explain the concept of deseasonalization;
 Apply time series analysis to business problems

INTRODUCTION

Businesses and governments use statistical analysis of information collected

at regular intervals over extensive periods of time to plan future policies.
For example, sales values or unemployment levels recorded at yearly,

ke
quarterly or monthly intervals are examined in an attempt to predict their
future behavior. Such sets of values observed at regular intervals over a

o.
period of time are called time series. The analysis of this data is a complex
.c
problem as many variable factors may influence the
es
changes. The first step is to plot the observations on a scattergraph, which
differs from those scattergraphs we have considered previously as the
ot

points are evenly spaced on the time axis in the order in which they are
cn

observed, and the time variable is always the independent Variable. This
scattergraph gives us a good visual guide to the actual changes, but is of
ne

very little help in showing the component factors causing these changes or
.k

in predicting future movements of the dependent variable.

Statisticians have constructed a number of mathematical models to
w

describe the behavior of time series, and several of these will be discussed
w

in this study unit.

w

COMPONENTS OF A TIME SERIES

These mathematical models assume that the changes are caused by the
variation of four main factors; they differ in the relationship between these
factors. It will be easier to understand the theory in detail if we relate it to
a simple time series so that we can see the calculations necessary at each
stage. Consider a factory employing a number of people in producing a
particular commodity, say thermometers. Naturally, at such a factory
during the course of a year some employees will be absent for various
reasons. The following table shows the number of days lost through sickness
over a five-year period. Each year has been broken down into four quarters
of three months. We have assumed that the number of employees at the
factory remained constant over the five years.

173
Table 9.1: Days lost through sickness at a thermometer factory
Year Quarter Days Lost
2003 1 30
2 20
3 15
4 35
2004 1 40
2 25
3 18
4 45
2005 1 45
2 30
3 22
4 55
2006 1 50
2 32

ke
3 28

o.
4 60
2007 1
2
.c 60
35
es
3 30
ot

4 70
cn

We will begin by plotting a time-series graph for the data, as shown in

ne

Figure 9.1.
.k

Note the following characteristics of a time-series graph:

w

 It is usual to join the points by straight lines. The only function of

w

these lines is to help your eyes to see the pattern formed by the
w

points.
 Intermediate values of the variables cannot be read from the graph.
 Every time-series graph will look similar to this, but a careful study
of the change of
 Pattern over time will suggest which model should be used for
analysis.

There are four factors that influence the changes in a time series – trend,
seasonal variations, cyclical fluctuations, and irregular or random
fluctuations. Now we will consider each in turn.

Trend
This is the change in general level over the whole time period and is often
referred to as the secular trend. You can see in Figure 9.1 that the trend is
definitely upwards, in spite of the obvious fluctuations from one quarter to

174
the next. A trend can thus be defined as a clear tendency for the time
series data to travel in a particular direction in spite of other large and
small fluctuations.

Seasonal Variations
These are variations which are repeated over relatively short periods of
time. Those most frequently observed are associated with the seasons of
the year, e.g. ice cream sales tend to rise during the summer months and
fall during the winter months. You can see in our example of employees'
sickness that more people are sick during the winter than in the summer. If
you can establish the variation throughout the year then this seasonal
variation is likely to be similar from one year to the next, so that it would
be possible to allow for it when estimating values of the variable in other
parts of the time series. The usefulness of being able to calculate seasonal
variation is obvious as, for example, it allows ice cream manufacturers to
alter their production schedules to meet these seasonal changes.

ke
As this type of fluctuation is difficult to determine, it is often considered

o.
with the final (fourth) element, and the two together are called the
residual variation. .c
es
Irregular or Random Fluctuations
ot

Careful examination of Figure 9.1 shows that there are other relatively
cn

small irregularities which we have not accounted for and that do not seem
to have any easily seen pattern. We call these irregular or random
ne

fluctuations; they may be due to errors of observation or to some one-off

.k

external influence which is difficult to isolate or predict. In our example

(Table1) there may have been a measles epidemic in 2006, but it would be
w

extremely difficult to predict when and if such an epidemic would occur

w

again.
w

Summary
To sum up, a time series (Y) can be considered as a combination of the
following four factors:
 Trend (T)
 Seasonal variation (S)
 Cyclical fluctuation (C)
 Irregular fluctuations (I).

Time series models:

175
Additive Model
Time series value = T +S +C +R
Where S, C and R are expressed in absolute value.
This model is best suited where the component factors are independent
e.g. where the seasonal variation is unaffected by trend.

Multiplicative Model:
Time series value = T × S× C × R
Where S, C and are expressed as percentage or proportions.

This model is best applied where characteristics interact e.g. where high
trends increase seasonal variations. Multiplicative model is more commonly
used in practice.
Of the four elements of time series the most important are trend and
seasonal variation. The following illustration shows how the trend (T) and
seasonal variation (S) are separated out from a time series and how the
calculated T and S values are used to prepare forecast. The process of

ke
separating out the trend and seasonal variation is known as deseasonalising
the data.

o.
There are two approaches to this process: one is based on regression
.c
through the actual data points and the other calculates the regression line
es
through moving average trend points. The method using the actual data is
demonstrated first followed by the moving average method.
ot
cn
ne
.k
w
w
w

176
 CHAPTER TEN

INDEX NUMBERS

SPECIIFIC OBJECTIVES
At the end of this topic the trainee should be able to:
 Define index numbers;
 Differentiate between the types of index numbers
 Explain the uses of index numbers
 List the limitations of index numbers.

INTRODUCTION
An index number is a device which shows by its variation the changes in a
magnitude which is not capable of accurate measurement in itself or of
direct valuation in practice. An index number may be described as a
specialized average designed to measure the level of a phenomenon with
respect to time geographic location or other characteristics such as income.

ke
Uses of index numbers
 They help in framing suitable policies

o.
 They reveal trends and tendencies
 Index numbers are very useful in deflating.c
es
Classification of index numbers.
Index numbers may be classified in terms of what they measure. In
ot

economics and business the classification is;

cn

 Price
ne

 Quantity
 Value
.k

 Special purpose
w

Problems in the constructing of index numbers

w

Before constructing index numbers a careful thought must be given to the

w

following problems.
i) The purpose of the index; as the very outset the purpose of
constructing the must be very clearly decided.
ii) Availability and comparability of data; it is needles to say that it
is impossible to make appropriate comparison unless the
necessary statistical data can be obtained
iii) Selection of base period; whenever index number are constructs,
a reference is made to some base period ,safe period should be
normal , it should not be too distant.
iv) Selection of number of items; every item cannot be included
while constructing an index number and hence one has to select a
sample.
v) Price quotation; after the commodities have been selected the
next problem is to obtain price quotation for these commodities
pricing vary from place to place and even from shop to shop.

177
 vi) Choice of an average; since index numbers are specific average a
decision has to be made as to which particular average (i.e.
arithmetic mean, median , mode , geometric mean)
vii) Selection of appropriate weight; the problem of selecting suitable
weights are quite important and at the same time quite difficult
to decide.
viii) Selection of appropriate formula; a large number of formula have
been devised for often constructing the index numbers. The
problem very often is that in selecting the most appropriate
formula.

An index number is an attempt to summarize a whole mass of data into one

figure. The single figure shows how one year differs from another year. It is
a statistical devise used to measure the change in the level of prices, wages
output and other variables at given times, relative to their level at an
earlier time which is taken as the base for comparison purposes
 Pn
A simple price index = × 100 (an unweighted price index)

ke
 Po

o.
A simple quantity index =
 Qn
.c
× 100 (an unweighted quantity index)
es
 Qo
ot

Where pn is the price of a commodity in the current year (the year for
cn

which the price index to be calculated)

ne

Where po is the price of the same commodity in the base year (the year for
.k

comparison purposes)
w

Similarly Qn and Qo are defined in the same way

w
w

Methods of constructing index numbers

a) unweighed indices
b) weighed indices

In the unweighted indices weight are not expressively assigned whereas in

the weighed indices weights are assigned to the various items. Each of
these types may further be divided under two heads.
i) simple aggregate
ii) simple average of price relations

Unweighted index numbers

i) simple aggregate method
This is the simplest method of constructing index numbers. When this
method is used to construct a price index number, the total of current year

178
prices for the various commodities is divided by total of base year and
quotient is multiplied by 100.
P01 = ∑p1 x 100
∑p0
Where ∑p1 = total current price for various commodities
∑p0 = total of base year price for various commodities
Limitation of the method
 The units in which prices of commodities are given affect the
price index.
 No consideration is given to the relative importance of the
commodities.
ii) simple average of relative method
When this method is used to construct a price index, price relative are
obtained for the various items include intercontinental he index and then
an average of these relative
P01 = ∑p1 x 100 /N
∑p0

ke
Where N= refers to the number of items
Merits

o.
 Extreme items do not influence the index
.c
 The index is not influenced by the units in which prices are quoted.
es
Limitations
 Difficulty is faced with regard to the selection of an appropriate
ot

average.
cn

 The relatives are assumed to have equal importance.

ne

Weighed index numbers

Weighed index numbers are of two types
.k

i) weighed aggregate index numbers

w

ii) weighed average of relative index numbers

w
w

Weighed aggregate method

An index is called a weighed aggregate index when it is constructed fro an
aggregate of items (price) that have been weighed.
Some important formulae of construction a weighed aggregate index are
i) Laspeyres Method
ii) Paasche Method
iii) Fishers Ideal Method
iv) Marshall – Edgeworth

Laspeyres method
In this method the base year‘s quantities are taken as weights. The formula
for constructing index is
P01 = ∑ p1q0 x 100
∑ p0q0

Paasche method
179
In this method the current year quantities are taken as the formula for
constructing index is
P01 = ∑ p1q1 x 100
∑ p0q1

Fischer‘s ideal method

Prof Fischer has given a number of formulas for constructing index numbers

P01 = √ ∑ p1q0 x ∑ p1q1 x 100

∑ p0q0 ∑ p0q1 or

ke
o.
po1 = √ (L*P) .c
es
ot
cn

Marshall Edge Worth Method

In this method also the current year as well as base year and quantities are
ne

considered
.k

P01= ∑ (q0 + q1) p1 x 100

∑ (q0 + q1) p0
w
w
w

INDEX NUMBERS APPLICATION

The index of industrial production

It is a quantity index compiled by the government. It measures changes in
the volume of production in major industries. The index is a good indication
of the state of national economy.
It covers the following major industries in the UK
i. Mining and quarrying
ii. Manufacturing such as food, drinks and tobacco, chemicals, metal
manufacture, engineering e.t.c
iii. Textile
iv. Construction
v. Gas electricity, water e.t.c

180
It excludes agriculture, fishing, trade, transport, finance and other such
industries.
Each industries order is given a weighting. The weighting is based on
average monthly production in each industry in a fixed base year. It gives
each item its relative importance amongst all other items and thus gives a
better estimate of the index for comparison purposes.

The Geometric Index (Industrial Share index)

This index is an index of 30 selected top industrial companies. It is
calculated by taking an unweighted geometric mean of the price relatives
of the selected shares.

Example
The share prices of ordinary shares of four companies on 1 st January 1990
and 1st January 1991 were as follows.

Share Price on Price on

ke
1.1.1990 1.1.1991
Company A Shs 10 Shs 12

o.
Company B Shs 12 Shs 15
Company C Shs 20 Shs 25 .c
es
Company D Shs 5 Shs 6
ot

Using an unweighted geometric index, calculate the index of share prices at

cn

1.1.1991 if 1.1.1990 is the base date, index 100

ne

Solution
.k

1 1
 12 15 25 6  4  27000  4
w

   2.25
1

     
4

 10 12 20 5   12000 
w
w

 1.225
Percentage increase = 22.5% index = 122.5

Inflation
The inflation rate for a given period can be calculated using the following
formula;
Current retail price index
Inflation = × 100
Retail price index in the base year

Marshal Hedge Worth Index

Marshal Hedge worth index =

p p
n o  qn 
× 100
 p q
o o  qn 

Tests for an Ideal Index Number

181
1. Factor Reversal Test
This test indicates that when the price index is multiplied with a quantity
index i.e. factors are reversed), it should result in the value index.

2. The time reversal test

If we reverse the time subscripts of a price or quantity index, the result
should be reciprocal of the original index.

PRACTICE QUESTION

ke
QUESTION FIVE

o.
Prodco PLC manufactures an item of domestic equipment which requires a

model have been used.

.c
number of components which have varied as various modifications of the
The following table shows the number of
es
components required together with the price over the last three years of
ot

production.
cn

COMPONENT 1981 1982 1983

ne

Prices Quantity Prices Quantity Prices Quantity

A 3.63 3 4.00 2 4.49 2
.k

B 2.11 4 3.10 5 3.26 6

w

C 10.03 1 10.36 1 12.05 1

w

D 4.01 7 5.23 6 5.21 5

w

Required:
a) Establish the base weighted price indices for 1982 and 1983 based on
1981 for the item of equipment. (8 marks)
b) Establish the current weighted price indices for 1982 and 1983 based on
1981 for the item of equipment. (8 marks)
c) Using the results of (a) and (b) as illustrations, compare and contrast
Laspeyre‘s and Paasche price index numbers.(4 marks)
(Total: 20 marks)

182
 QUESTION SIX
a) A company manufacturing a product known as 257 uses five components
in its assembly.

The quantities and prices of the components used to produce a unit of

K257 in 1982, 1983 and 1984 are tabulated as follows:

COMPONENT 1982 1983 1984

Quantity Prices Quantity Prices Quantity Prices
A 10 3.12 12 3.17 14 3.20
B 6 11.49 7 11.58 5 11.67
C 5 1.40 8 1.35 9 1.31
D 9 2.15 9 2.14 10 2.63
E 50 0.32 53 0.32 57 0.32

Required:
i) Calculate Laspyere‘s type price index number for the cost of one unit of

ke
K257 for 1983 and 1984 based on 1982. (6 marks)

o.
ii) Calculate Paasche type price index numbers for the cost of one unit of
K257 for 1983 and 1984 based on 1982. .c
(6 marks)
es
ot

iii) Compare and contrast the Laspeyre and Paasche price-index numbers
you have obtained in (i) and (ii) (3 marks)
cn
ne

A number of employers manufacturing plastic components used in plumbing

have formed themselves into an association for the purpose of negotiating
.k

with the trade union for this industrial sector.

w
w

The negotiations cover pay and contributions in this sector.

w

Required:
Explain the usefulness of an index of Industrial Production and an index of
retail prices to both sides in a series of pay negotiations.
(5 marks)
(Total: 20marks)

183
QUESTION SEVEN
The data given below indicates the prices and production of some
horticultural products in Central Territory:

Produce Production Price per box (Shs)

(1000 boxes)
1980 1990 1980 1990
Cabbages 48,600 62,000 100 150
Tomatoes 22,000 37,440 220 310
Onions 47,040 61,430 180 200
Spinach 43,110 55,720 130 170
Required:
Calculate the increase or decrease in prices from 1980 on the basis of the
following indices:

a) Mean relatives
b) Laspeyres index

ke
c) Paasche index

o.
d) Marshall – Hedgeworth index
e) Fishers index.
.c
es
ot
cn
ne
.k
w
w
w

184
 CHAPTER ELEVEN
MATRIX ALGBRA

SPECIFIC OBJECTIVES
At the end of this topic the trainee should be sable to:
 Define matrices;
 Describe the types of matrices,
 Apply matrix operations
 Form matrix models from practical problems;
 Apply matrix to decision problems.

INTRODUCTION
A matrix is a rectangular array of items or numbers. These items or
numbers are arranged in rows and columns to represent some information.
The position of an element in one matrix is very important as well be seen
later; therefore an element is located by the number of the row and
column which it occupies. The size of a matrix is defined by the number of

ke
its rows (m) and column (n).
a b c

o.
a b  
For example = 
c d
 and B =  d e f 
g h i 
.c
es
 
ot

are (2 x 2) and (3 x 3) matrices since A has 2 rows and 2 columns and B has
3 rows and 3 columns.
cn

A matrix A with three rows and four columns is given by one of:
ne

 a11 a12 a13 a14 

 
.k

A=  a 21 a 22 a 23 a 24 
a 
w

 31 a 32 a 33 a 34 
w
w

or

A =  a ij  i = 1, 2, 3
j = 1, 2, 3, 4 where i represents the row number whereas j represents the column number

185
Types of matrices
Equal Matrices
Two matrices A and B are said to be equal, that is

A=B or a  = b 
ij ij

If and only if they are identical if they both have the same number of rows
and columns and the elements in the corresponding locations in the two
matrices should be the same, that is, aij = bij for all i. And j.

Example
 3 4 0  3 4 0
The following matrices are equal  2 2 3  =  2 2 3 
5 1 1 5 1 1
   

ke
Column Matrix or column vector

o.
A column matrix, also referred to as column vector is a matrix consisting of
a single column. .c
es
 x1 
 
ot

 x2 
cn

 . 
For example x =  
ne

 . 
 . 
.k

 
w

 xn 
w
w

Row matrix or row vector

It is a matrix with a single row
For example y =  y1 , y2 , y3 .........yn 

Transpose of a Matrix
The transpose of an mxn matrix A is the nxm matrix A T obtained by
interchanging the rows and columns of A.
A = aij

The transpose of A i.e. AT is given by

AT = aij = aji

186
 mxn nxm

Example
Find the transposes of the following matrices

1 5 7
 
A=  2 1 4 
0 9 3
 

B=  b1 , b 2 , b3 , b 4 

 x1 
 
C=  x 2 
x 

ke
 3

o.
Solution
.c
es
T
1 5 7 1 2 0
   
ot

i. A T = 2 1 4 = 5 1 9
 0 9 3  7 4 3
cn

   
ne

 b1 
.k

 
b
w

ii. BT =  b1 , b 2 , b3 , b 4  =  2
T

 b3 
w

 
w

 b4 
T
 x1 
 
iii. C =  x 2  
T
 x1 x 2 x 3 
x 
 3

Square Matrix
A matrix A is said to be square when it has the same number of rows as
columns
e.g. 2 5
A= 3 7 is a square matrix of order 2

187
 B = n × n is a square matrix of the order n

Diagonal matrices
It is a square matrix with zeros everywhere in the matrix except on the
principal diagonal
e.g.

3 0 0 9 0 0
   
A = 0 1 0 , B = 0 0 0
0 0 7  0 0 0
   
An identity of unity matrix
It is a diagonal matrix in which each of the diagonal elements is a positive

ke
one (1)

o.
e.g.
1 0 0 
.c
es
1 0
I2    and I 3  0 1 0 
 
ot

0 1 
0 0 1
cn

2  2 unit matrix 3  3 unit matrix

ne

A null or zero matrix

.k
w

A null or zero matrix is a matrix whose elements are all equal to zero.
w

Sub matrix
w

The sub matrix of the matrix A is another matrix obtained from A by

deleting selected row(s) and/or column(s) of the matrix A.
7 9 8
 
e.g, if A =  2 3 6 
1 5 0
 

 2 3 6 7 9
then A1 =   and A 2 =  
1 5 0   1 5
are both sub matrices of A

OPERATION ON MATRICES
188
Matrix addition and subtraction
We can add any number of matrices (or subtract one matrix from another)
if they have the same sizes. Addition is carried out by adding together
corresponding elements in the matrices. Similarly subtraction is carried out
by subtracting the corresponding elements of two matrices as shown in the
following example
Example: Given A and B, calculate A + B and A – B

 6 1 10 5  12 4 7 3 
   
A=  3 4 2 5  B =  0 4 10 4 
 9 13 6 0   7 3 7 9 
   

ke
o.
 6 1 10 5  12 4 7 3   18 3 3 8
     
A+B=  3 4 .c
2 5  +  0 4 10 4  =  3 0 12 9 
 9 13 6 0   7 3 7 9   2 16 1 9 
es
     
ot
cn

 6 1 10 5  12 4 7 3   6 5 17 2
     
2 5  -  0 4 10 4  =  3 8 1 
ne

A-B=  3 4 8
 9 13 6 0   7 3 7 9   16 10 13 9 
     
.k
w
w

If it is assumed that A, B, C are of the same order, the following properties

w

are fulfilled:
a) Commutative law: A+B =B+A
b) Associative law: (A + B) + C = A + (B + C) = A + B + C

Multiplying a matrix by a number

In this case each element of the matrix is multiplied by that number

Example

189
  6 1 10 5 
 
If A =  3 4 2 5 
 9 13 6 0 
 

 60 10 100 50 
 
then (10)A =  30 40 20 50 
 90 130 60 0 
 

Matrix Multiplication

a) Multiplication of two vectors

Let row vector A represent the selling price in shillings of one unit of
commodity P, Q, R respectively and let column vector B represent the

ke
number of units of commodities P, Q, R sold respectively. Then the vector
product A  B will be equal to the total sales value

o.
i. e. A  B = Total sales value
.c
es
 100 
ot

 
Let A =  4 5 6  and B =  200 
cn

 300 
 
ne
.k

 100 
w

 
then  4 5 6   200  = 400 + 1,000 + 1,800 = Shs 3,200
w

 300 
 
w

Rules of multiplication
i. The row vector must have the same number of elements as the
column vector
ii. The first vector is a row vector and the second is a column vector
iii. The corresponding elements in each vector are multiplied
together and the results obtained are added. This addition is
always a single number
Going back to the example given before
 100 
 
A × B =  4 5 6   200  = 4 × 100 + 5 × 200 + 6 × 300=Shs3,200, a single number
 300 
 

190
 b) Multiplication of two matrices
Rules
i. Multiplication is only possible if the first matrix has the same
number of columns as the second matrix has rows. That is if A is
the order a×b, then B has to be of the order b×c. If the A×B = D,
then D must be of the order a×c.
ii. The general method of multiplication is that the elements in row
m of the first matrix are multiplied by the corresponding
elements in columns n of the second matrix and the products
obtained are then added giving a single number.
We can express this rule as follows

 a11 a12  b b12 b13 

Let A =   and b =  11 
 a 21 a 22   b 21 b 22 b 23 

ke
 d11 d12 d13 

o.
Then A  B = D =  
 d 21 d 22 d 23 
.c
es
A = 2 x 2 matrix B = 2 x 3 matrix D = 2 x 3 matrix
ot

Where
cn

d11 = a11  b11 + a12  b21

ne

d12 = a11  b12 + a12  b22

.k
w

Example I
w

 6 1  3 0 2  6  3  1 4 6  0  1 5 6  2  1 8 
w

     = 
 2 3 4 5 8  2  3  3 4 2  0  3 5 2  2  38

 22 5 20 
=  
 18 15 28 

Example II
Matrix X gives the details of component parts used in the make up of two
products P1 and P2 matrix Y gives details of products made on each day of
the week as follows:

191
 Matrix Y
Products
Matrix X
P1 P2
Parts
A B C Mon 1 2
Tues 2 3
P 3 4 2  
Products 1  Wed 3 2
P2 2 5 3
Thur 2 2
 
Fri 1 1

Use matrix multiplication to find the number of component parts used on

each day of the week.
Solution:
After careful consideration, it will be easy to decide that the correct order
of multiplication is YXX (Note the order of multiplication). This
multiplication is compatible and also it gives the desired answer.

ke
o.
1 2  1×3+2×2 1×4+2×5 1×2+2×5 
   
2 3
 3 4 2  2×3+3×2 .c2×4+3×5 2×2+3×3 
es
Y × X = 3 2 ×   =  3×3+2×2 3×4+2×5 3×2+2×3 
   2 5 3  
ot

2 2  2×3+2×2 2×4+2×5 2×2+2×3 

1 1   1×3+1×2 1×2+1×3 
cn

  1×4+1×5
ne

5 x 2 matrix 2 x 3 matrix = 5 x 3 matrix

.k

A B C
w

Mon 7 14 8 
w

 
Tues 12 23 13 
w

Wed  13 22 12 
 
Thur 10 18 10 
5 9 5 
Fri 
Interpretation
On Monday, number of component parts A used is 7, B is 14 and C is 8. in
the same way, the number of component parts used for other days can be
interpreted.

The determinant of a square matrix

The determinant of a square matrix det (A) or A is a number associated to
that matrix. If the determinant of a matrix is equal to zero, the matrix is
called singular matrix otherwise it is called non-singular matrix. The
determinant of a non square matrix is not defined.
192
Determination of a 2 x 2 matrix
a b
 A =  = ad - cb
c d
ii. Determinant of a 3 x 3 matrix

a b c
  e f  d f  d e 
A  d e f  = a  b  +c  
g h i  h i  g i  g h
 

a  ei - fh  - b  di - gf  +c  dh - eg 
simplify

iii. Determinant of a 4 x 4 matrix

ke
o.
 a b c d
 
A=  e f g h .c
es
 i j k l
 
ot

m n o p
cn
ne

f g h e g h  e f h  e f g
.k

       
A = a j k l   b i k l  +c  i j l  d i j k
w

n o p n o p m o p m n o
       
w
w

Simplify 3 x 3 determinants as in ii and then evaluate the 4 x 4

determinants.

Inverse of a matrix
If for an n ( n square matrix A, there is another n ( n square matrix B such
that there product is the identity of the order n X n, In, that is A X B = B X A
= I, then B is said to be inverse of A. Inverse if generally written as A-1
Hence AA-1 = I

Note: Only non singular matrices have an inverse and therefore the inverse
of a singular matrix is non defined.

General method for finding inverse of a matrix

193
In order to introduce the rule to calculate the determinant as well as the
inverse of a matrix, we should introduce the concept of minor and cofactor.

The minor of an element

Given a matrix A = (aij), the minor of an element aij in row i and column j
(call it mij), is the value of the determinant formed by deleting row i and
column j in matrix A.

Example
Let A = EMBED Equation.DSMT4

The minors are,

6 1
m11 = = 6×0  3×1 =  3
3 0

ke
5 1
= 5×0  1×2 =  2

o.
m12 =
2 0
.c
es
Similarly
ot

5 6 2 3 4 3 4 2
cn

m13 = m 21 = m 22 = m 23 =
2 3 3 0 2 0 2 3
ne

=15  12 = 3 =0  9 =  9 = 06 = 6 = 12  4 = 8
.k
w

2 3 4 3 4 2
m 32  m 33 
w

m31 =
6 1 5 1 5 6
w

 2 -18  -16  4 -15  -11  24 -10  14

The cofactor of an element

The cofactor of any element aij (known as cij) is the signed minor associated
with that element.
The sign is not changed if (i+j) is even and it is changed if (i+j)is odd. Thus
the sign alternated whether vertically or horizontally, beginning with a plus
in the upper left hand corner.
  

i.e. 3 x 3 signed matrix will have signs     

  
 

194
Hence the cofactor of element a11 is m11 = -3, cofactor of a12 is –m12 = +2
the cofactor of element a13 is +m13 = 3 and so on.
 3 2 3 
 
Matrix of cofactors of A =  9 6 8 
 16 11 14 
 
a b c
in general for a matrix M =  
d e f 
g h i 
 
Cofactor of a is written as A, cofactor of b is written as B and so on.
Hence matrix of cofactors of M is written as
A B C
=  D E F 
G H I 
 

ke
The determinant of a n×n matrix

o.
The determinant of a n×n matrix can be calculated by adding the products
of the element in any row (or column) multiplied by their cofactors. If we
use the symbol ∆ for determinant.
.c
es
ot

Then ∆ = aA + bB + cC
cn

or
ne

= dD + eE + fF e.t.c
.k
w

Note: Usually for calculation purposes we take ∆ = aA + bB + cC

w
w

Hence in the example under discussion

∆ = (4  –3) + (2  2) + (3  3) = 1

The ad joint or transposition of a matrix

A B C
 
The ad joint of matrix  D E F  is written as
G H I 
 

195
 A D G
 
 B E H  i.e. change rows into columns and columns into rows
C F I 
 
(transpose)
a b c
The inverse of the matrix  d e f 
g h i 
 

1
is written as  (adjoint of the matrix of cofactors)
determinant

A D G
1  
i.e. A-1 =  B E H
  

ke
C F I 

o.
Where ∆ = aA + bB + cC

 4 2 3 .c
es
Hence inverse of  5 6 1 
 2 3 0
ot

 
cn

is found as follows
ne

∆ = (4  –3) + (2  2) + (3 ( 3) = 1
.k

A = -3 B=2 C=3
w

D=9 E = -6 F = -8
w
w

G = -16 H = 11 I = 14

EMBED Equation.DSMT4
(note: Check if A ( A-1 = A-1  A = 1)

APPLICATION OF MATRICES

Solution of simultaneous equations

In order to determine the solutions of simultaneous equations, we may use
either of the following 2 methods
i. The cofactor method

196
 ii. Cramers rule

The cofactor method

This method requires that we obtain
a) The minors and cofactors
b) The adjoint of the matrix
c) The inverse of the matrix
d) Multiply the original by the inverse on both sides of the matrix
equation

Example
Solve the following
a) 4x1 + x2 – 5x3 = 8
-2x1 + 3x2 + x3 = 12
3x1 – x2 + 4x 3 = 5

ke
o.
b) 4x1 + 3x3 + 5x3 = 27
.c
es
x1 + 6x2 + 2x3 = 19
ot

3x1 + x2 + 3x3 = 15
cn
ne

c) 4x1 + 2x2 + 6x3 = 28

3x1 + x2 + 2x3 = 20
.k
w

10x1 + 5x2 + 15x3 = 70

w
w

d) 2x1 + 4x2 – 3x3 = 12

3x1 – 5x2 + 2x3 = 13
-x1 + 3x2 + 2x3 = 17

Solution
a) From a, we have

 4 1 -5   x1  8
     
 -2 3 1   x 2  = 12 
 3 -1 4   x  5
   3  
A X b

197
 We need to determine the minors and the cofactors for the above
matrix

Definition
A minor is a determinant of a sub matrix obtained when other elements are
detected as shown below.
A cofactor is the product of (-1) i + j and a minor where
i = Ith row i = 1, 2, 3 …….
j = Jth row j = 1, 2, 3 …….

1+1 3 1
Cofactor of 4 (a11) = (-1) = 13
1 4

2+1 1 5
Cofactor of -2 (a21) = (-1) = 1
1 4

ke
3+1 1 5
Cofactor of 3 (a31) = (-1) = 16

o.
3 1

Cofactor of 1 (a12) = (-1) 1+2 2 1

= 11 .c
es
3 4
ot

2+2 4 5
Cofactor of 3 (a22) = (-1)
cn

= 31
3 4
ne

2+3 4 5
Cofactor of -1 (a23) = (-1) = 6
.k

2 1
w

1+3 2 3
= 7
w

Cofactor of -5 (a13) = (-1)

3 1
w

2+3 4 1
Cofactor of +1 (a23) = (-1) = 7
3 1

3+3 4 1
Cofactor of 4 (a33) = (-1) = 14
2 3

The matrix of C of cofactors is

 13 11 7 
 
 1 31 7 
16 6 14 
 

198
  13 1 16 
CT =  11 31 6  = Adjoint of the original matrix of coefficients
 7 7 14 
 
The original matrix of coefficients

 4 1 5 
=  2 3 1 
 3 1 4 
 
Therefore

ke
= (48 + 3 – 10) – (-45 – 4 – 8)

o.
= 41 + 57
.c
es
= 98
ot

The inverse of the matrix of coefficients, see (*) will be

cn

13 1 16 
ne

1  
=  11 31 6 
.k

98  
 -7 7 14 
w
w

by multiplying the inverse on both sides of * we have,

w

199
 13 1 16   4 1 5   x1 
1      

98 
11 31 6   2 3 1   x2 
  3 1 4  x 
 -7 7 14     3

13 1 16  8
1    
  11 31 6  12 
98   5
 -7 7 14   

 98 0 0   x1   196 
1     1  
  0 98 0   x2  =  490 
98   x  98  
 0 0 98   3  98 

1 0 0  x1   2

ke
     
= 0 1 0  x 2  = 5

o.
0 0 1 x  1
   3  
.c
es
 x1   2
ot

   
  x2  =  5 
cn

x  1
 3  
ne

 X1 = 2, X2 = 5, X3 = 1
.k
w

c) 4x1 + 2x2 + 6x3 = 28

w

3x1 + x2 + 2x3 = 20
w

10x1 + 5x2 + 15x3 = 70

4 2 6  x1   28 
     
= 3 1 2  x 2  =  20 
10 5 15  x   70 
   3  

4 2 6 4 2 6 4 2
 
3 1 2 = 3 1 2 3 1
10 5 15 
  10 5 15 10 5

= (60 + 40 + 90) – (60 + 40 + 90)

200
 =0
Hence the solutions of x1, x2, and x3 do no exist. The equations are
independent
Now work out part (b) on your own.
Cramers Rule in Solving Simultaneous Equations
Consider the following system of two linear simultaneous equations in two
variables.
a11 x1 + a12 x2 = b1 ……………(i)
a21 x1 + a22 x2 = b2 ……………(ii)
after solving the equations you obtain
b1 a12
b1a 22  b 2 a12 b 2 a 22

ke
x1 = 
a11a 22  a12 a 21 a11 a12

o.
a 21 a 22
.c
es
and
ot
cn

a11 b1
ne

a11b 2 - a 21b1 a 21 b 2
x2 = 
.k

a11a 22 - a12 a 21 a11 a12

w

a21 a22
w

Solutions of x1 and x2 obtained this way are said to have been derived using
w

Cramers rule, practice this method over and over to internalize it. It is
advisable for exam situation since it is shorter.
Example
Solve the following systems of linear simultaneous equations by Cramers‘
rule:
i) 2x1 – 5x2 = 7
x1 + 6x2 = 9
ii) x1 + 2x2 + 4x3 = 4
2x1 + x3 = 3
3x2 + x3 = 2
Solutions

201
 i. 2x1 – 5x2 = 7
x1 + 6x2 = 9
can be expressed in matrix form as

 2 5   x1  7
  x  =  
1 6   2  9

A X b
and applying cramers‘ rule
7 -5
9 6 87 2
x1 = = = 5
2 -5 17 17
1 6

ke
2 7

o.
1 9 11
x2 =
2 -5
=
17 .c
es
1 6
ot

(ii) can be expressed in matrix form as

cn

1 2 4  x1   4
ne

     
2 0 1  x2  =  3
.k

0 3 1 x   2
   3  
w

and by cramers‘ rule

w
w

4 2 4
3 0 1
2 3 1 22
x1 = =
1 2 4 17
2 0 1
0 3 1

202
 1 2 4
2 0 3
0 3 2 7
x3  =
1 2 4 17
2 0 1
0 3 1

1 4 4
2 3 1
0 2 1 9
x2  =
1 2 4 17
2 0 1
0 3 1

ke
Solving simultaneous Equations using matrix algebra

o.
i. Solve the equations
2x + 3y = 13 .c
es
3x + 2y = 12
ot

in matrix format these equations can be written as

cn

 2 3  x   13 
ne

    =  
 3 2  y 12 
.k

Pre multiply both sides by the inverse of the matrix

w
w

2 3
 = 5
w

3 2

and inverse of the matrix is

2 3 
1  2 3  5 5 
   =  
5  3 2  3 2
  
5 5
Pre multiplication by inverse gives

 2 3   2 3 
 5 5   2 3   5 5   13   2
    =     =  
 3 2  3 2   3  2  12   3
   
 5 5  5 5
203
Therefore x = 2 y=3

ii. Solve the equations

4x + 2y + 3z = 4
5x + 6y + 1z = 2
2x + 3y = -1
Solution:
Writing these equations in matrix format, we get

A  BX = b
 4 2 3 x 4
     
5 6 1  y =  2 
 2 3 0 z  -1
     

ke
Pre-multiply both sides by the inverse

o.
 3 9 16 
the inverse of A as found before is A-1 =  2 6 11 
.c
es
 3 8 14 
 
ot

 3 9 16   4 3 2 x  3 9 16  4  22 

cn

           
 2 6 11  5 6 1  y  =  2 6 11   2  =  -15 
ne

 3 8 14  2 3 0 z  3 8 14   -1  -18 

           
.k
w

Hence x = 22 y = -15 z = -18

w

(Note: under examination conditions it may be advisable to check the

w

solution by substituting the value of x, y, z into any of the three original

equations)

SIMPLE INPUT – OUTPUT ANALYSIS

The input output analysis is a topic which requires application of matrices
The technique analyses the flow of inputs from one sector of the economy
to the other sectors thus the technique is quite useful in studying the
interdependence of sectors within a single economy.
The input – output analysis was first developed by Prof Leontief hence the
Leontief matrix has been developed. See the following example

Example (B)
204
INPUT OUTPUT TABLE
TO Final Total
FROM Agric Industry Service Demand Demand
(output)
Agric 300 360 320 1080 2060
Industry 450 470 410 800 2130
Service 610 500 520 270 1900
Primary inputs 700 800 650 - -
Total inputs 2060 2130 1900 - -

NB: In the above table, one should be able to interpret the table e.g. of the
total demand of 2060 metric tones from the agricultural sector; 300 is
produced for the agricultural sector, 360 for industrial sector, 320 for the
service sector and 1080 metric tones makes up the final demand.

ke
The final demand is the additional demand besides the sectoral demand
which is normally made by other users e.g. government, foreign countries,

o.
other manufacturers not included in the other sectors.
.c
For production if items besides the inputs from other sectors namely labour
es
capital e.t.c
ot

Technical coefficients: (‘to’ sectors)

cn

300
Agriculture 300 = = 0.14
ne

2060
450
.k

450 = = 0.22
2060
w

610
w

610 = = 0.30
2060
w

360
Industry 360 = = 0.7
2130
470
470 = = 0.22
2130
500
500 = = 0.23
2130

320
Service 320 = = 0.17
1900
410
410 = = 0.22
1900

205
 520
520 = = 0.27
1900

The matrix of technical coefficients is:

TO Final Total
FROM Agric Industry Service Demand Demand
(output)
Agric 0.14 0.7 0.17 1080 2060
(y1)
Industry 0.22 0.22 0.22 800 (y2) 2130
Service 0.30 0.23 0.27 270(y3) 1900
Primary x x x - -
inputs
2060(x1) 2130(x2) 1900(x3) - -

From the above table, we may develop the following equations

ke
0.14x1 + 0.7x2 + 0.17x3 + y1 = x1

o.
0.22x1 + 0.22x2 + 0.22x3 + y2 = x2
.c
es
0.30x1 + 0.23x2 + 0.27x3 + y3 = x3
ot
cn

 0.14 0.17 0.17   x1   y1   x1 

       
ne

 0.22 0.22 0.22   x 2  +  y 2  =  x 2  ...................(*)

 0.30 0.23 0.27   x  y  x 
   3  3  3
.k

A X Y X
w
w

Let the coefficient matrix be represented by

w

 a11 a12 a13   y1   x1 

     
A   a 21 a 22 a 23  y =  y2  x =  x2 
a a 33  y  x 
 31 a 32  3  3
Equation (*) may be written as
AX + Y = X
Y = X – AX
Y = X (I-A)
⇒ (I – A)–1 Y = X
The matrix I – A is known as Leontief Matrix

206
Technical Coefficients
These show the units required from each sector to make up one complete
product in a given sector e.g. in the above matrix of coefficients it may be
said that one complete product from the agricultural sector requires 0.14
units from the agricultural sector itself, 0.22 from the industrial sector and
0.30 from the service sector
NB: The primary inputs are sometimes known as ―value added‖

Example 1
Determine the total demand (x) for the industry 1, 2, 3 given the matrix of
technical coefficients (A), Capital and the final demand vector B.
1  0.3 0.4 0.1   20 
   
A = 2  0.5 0.2 0.6  B =  10 
3  0.1 0.3 0.1   30 
 

ke
From the input – output analysis

o.
 X1 
–1
X = (I – A) B, Where X = X2 
.c
  is the demand vector
es
x 
 3
ot
cn

1 0 0  0.3 0.4 0.1 

ne

   
I - A =  0 1 0  -  0.5 0.2 0.6 
0 0 1  0.1 0.3 0.1 
.k

   
w
w

 0.7 -0.4 -0.1 

w

 
  -0.5 0.8 -0.6 
 -0.1 -0.3 0.9 
 

1
 0.7 -0.4 -0.1 
 
 (I - A)-1 =  -0.5 0.8 -0.6 
 -0.1 -0.3 0.9 
 

The matrix of cofactors of  I - A  is

 0.54 0.51 0.23 
=
 
 0.39 0.62 0.25 
 0.32 0.47 0.36 
 

207
The transpose (adjoint) of the above matrix is

 0.54 0.39 0.32 

=  
 0.51 0.62 0.47 
 0.23 0.25 0.36 
 

Δ of (I - A) = 0.495 – (0.008 + 0.126 + 0.18

= 0.809

 0.54 0.39 0.32   0.66 0.48 0.40 

1    
 (I - A) =
-1
 0.51 0.62 0.47  =  0.63 0.77 0.58 
0.809    0.23 0.31 0.44 
 0.23 0.25 0.36   

ke
 0.66 0.48 0.40   20   X1 
     

o.
 0.63 0.77 0.58   10  =  X 2 
 .c
 0.23 0.31 0.44   30 
  
X 
 3
es
Therefore X =
ot

 30 
 
cn

X =  37.7 
 21.9 
ne

 
.k

The total demand from the three industries 1, 2 and 3 is 30 from 1, 37.7
w

from 2 and 21.9 from 3.

w
w

Example 2
Three clients of Disrupt, Ltd P, Q and Rare direct competitors in the retail
business. In the first week of the year P had 300 customers Q had 250
customers and R had 200 customers. During the second week, 60 of the
original customers of P transferred to Q and 30 of the original customers of
P transferred to R. similarly in the second week 50 of the original customers
of Q transferred to P with no transfers to R and 20 of the original customers
of R transferred to P with no transfers to Q.

Required
a) Display in a matrix the pattern of retention and transfers of customers
from the first to the second week
(4 marks)

208
b) Re-express the matrix that you have obtained in part (a) showing the
elements as decimal fractions of the original numbers of customers of P,
Q and R (2 marks) Refer to this re expressed matrix as B
c) Multiply matrix B by itself to determine the proportions of the original
customers that have been retained or transferred to P, Q and R from the
second to the third week. (4 marks)
d) Solve the matrix equation (xyz)B = (xyz) given that x + y + z = 1
(8 marks)
e) Interpret the result that you obtain in part (d) in relation to the
movement of customers between P, Q and R
(2marks (Total 20 marks)

Solution
a). Think of each row element as being the point from which the customer
originated and each column element as being the destination e.g. 210
customers move from P to P, 60 move from P to Q and 30 move from P
to R. The sum of the elements of the first row totaling 300 that is the

ke
number of customers originally with P.

o.
Hence required matrix is
P Q R To .c
es
P  210 60 30   row total 300 
   
ot

From Q  50 200 0   row total 250 

cn

R  20 0 180   row total 200 

 
ne

b). the requirement of this part is to express each element as a decimal

fraction of its corresponding row total. The second row, first element is
.k

therefore 50/250, that is 0.2 and the second element is therefore

w

200/250 that is 0.8.

w
w

 0.7 0.2 0.1 

Hence B =
 
 0.2 0.8 0 
 0.1 0 0.9 
 

 0.7 0.2 0.1   0.7 0.2 0.1   0.54 0.30 0.16 

c).
     
 0.2 0.8 0   0.2 0.8 0  =  0.30 0.68 0.02 
 0.1 0 0.9   0.1 0 0.9   0.16 0.02 0.82 
     

The result can be checked by the normal rules of matrix multiplication.

209
  0.7 0.2 0.1 
d).  x y x  X  0.2 0.8 0  =  x y z 
 0.1 0 0.9 
 

This produces from the first row

0.7x + 0.2y + 0.1z = x
Which reduces to 0.2y + 0.1z = 0.3x
Or 2y + z = 3x ………………………...(i)

Or
The second row produces, 0.2x + 0.8y = y
Reducing to 0.2x = 0.2 y

ke
X = y ………………………..(ii)

o.
Or
The third row produces 0.1x + 0.9z = z .c
es
Reducing to 0.1x = 0.1z
ot

X = z ………. (iii)
cn
ne

At this point you will notice that condition h (ii) and condition (iii) produce
2x + x = 3x when substituted into condition (i), we therefore need extra
.k

condition x + y + z = 1 to solve the problem.

w

Thus x+x+x=1
w

Or 3x = 1
w

That is x= 1
3

Leading to x=1 , y=1 , z= 1

3 3 3
e). In proportion terms this solution means that P, Q, and R will in the long
term each have one third of the total customers

Example 4
There are three types of breakfast meal available in supermarkets known as
brand BM1, brand BM2 and Brand BM3. In order to assess the market, a
survey was carried out by one of the manufacturers. After the first month,
the survey revealed that 20% of the customers purchasing brand BM1
switched to BM2 and 10% of the customers purchasing brand BM1 switched
to BM3. similarly, after the first month of the customers purchasing brand
210
BM2, 25% switched to BM1 and 10% switched to BM3 and of the customers
purchasing brand BM3 0.05% switched to BM1 and 15% switched to BM2

Required
i. Display in a matrix S, the patterns of retention and transfers of
customers from the first to the second month, expressing percentage
in decimal form. (2marks)
ii. Multiply matrix S by itself (that is form S2) (5
Marks)
iii. Interpret the results you obtain in part ii with regard to customer
brand loyalty (3 marks)

Solution
The objective of the first part of the question was to test the candidate‘s
ability to formulate and manipulate a matrix, then interpret the result of
such manipulation.

ke
a. i. The matrix showing the pattern of retention and transfer from the
first to the second month is

o.
BM1 BM2 BM3
 0.70 0.20 0.10  BM1 .c
es
 
S =  0.25 0.65 0.10  BM2
ot

 0.05 0.80  BM3

 0.15
cn

(The second element in the first row shows the 20% movement from BM1 to
ne

BM2 and so on)

.k

i. The product of matrix S with itself is demonstrated as follows

w
w
w

 0.70 0.20 0.10   0.70 0.20 0.10   0.5450 0.2850 0.1700 

     
 0.25 0.65 0.10   0.25 0.65 0.10  =  0.3425 0.4875 0.1700 
 0.05 0.80   0.05 0.80   0.1125 0.6600 
 0.15   0.15   0.2275 

Where for example second element in the first row, that is 0.2850 is the
result of multiplying the corresponding elements of the first row of S by the
second column of S and summing the product.

0.2850= 0.70 × 0.20 + 0.20 × 0.65 + 0.10 × 0.15

= 0.14 + 0.13 + 0.015 e.t.c.

ii. The resulting matrix may be interpreted in the following way

Of the original customers who buy BM1, 54.5% will remain loyal to
the brand in month three, 28.5% will have switched to BM2 and
17% will have switched to BM3.
211
 Of the original customers who buy BM2, 48.7% will remain loyal to
the brand in month three 34.25% will have switched to BM1 and
17% will have switched to BM3
Of the original customers who buy BM3, 66% will remain loyal to
the brand in month three 11.25% will have switched to BM1 and
22.75% will have switched to BM2
Alternatively
In month three 54.5% of the customers buying BM1 are original
customers. 34.5% came from BM2 originally and remaining 11.25%
have switched from BM3 and so on.

SIMPLE MARKOV PROCESS

MARKOV CHAINS/PROCESSES
The Markov processes are defined as a set of trials which follow a certain

ke
sequence which depend on a given set of probabilities known as transition
probabilities. These probabilities indicate how a particular activity or

o.
product moves from one state to another.
.c
es
Applications of Markov Chains in Business
The Markov processes or chains are frequently applied as follows:-
ot
cn

1. Brand Switching
ne

By using the transitional probabilities we can be able to express the manner

in which consumers switch their tastes from one product to another.
.k
w

2. Insurance industry
w

Markov analysis may be used to study the claims made by the insured
w

persons and also decide the level of premiums to be paid in future.

3. Movement of urban population

By formulating a transition matrix for the current population in the urban
areas, one can be able to determine what the population will be in say 5
years.

4. Movement of customers from one bank to another.

It is a fact that customers tend to look for efficient banks. Therefore at a
certain time when a given bank installs such machinery as computers it will
tend to attract a number of customers who will move from certain banks to
efficient ones.

PROPERTIES OF MARKOV CHAINS

212
1. Each outcome in a markov process belongs to a state space or transition
matrix. E.g.
S1 S2 S3
S1  P11 P12 P13 
  Where S1, S2, S3 are states and P11 P12 e.t.c are
S2  P21 P22 P23 
S3  P31 P32 P33 
probabilities

2. The outcome of each trial depends on the immediate preceding activities

but not on the previous activities

BASIC TERMS IN MARKOV CHAINS

a) Probability Vector
This is a row matrix whose elements are non-negative and also they add

ke
up to 1 e.g. u = 0.2, 0.1, 0.2, 0.5)

o.
Example
.c
es
Consider u = (¾, 0, - ¼, 1/2) Not because – ¼ is negative
v = (¾, ½, 0, ¼) Not because the sum of the element <1
ot
cn

w= (¼, ¼, 0, ½) Adds up to 1, each element is non

ne

negative.
Therefore it‘s a prob, vector
.k
w
w

State the ones which are probability vectors

w

b) Stochastic matrix
A matrix whose row elements are all non negative and also add up to 1.
 0.1 0.2 0.3 0.4 
 
0.0 0.7 0.1 0.2 
Example (i) M =
 0.5 0.1 0.1 0.3 
 
 0.3 0.4 0.2 0.1 
Example ii) = Consider the following matrices
 13 0 2
 0 1 0
1 
3
3   
3
A= 4
1
2   1
4 B =  14 4
1
C =  12 16 13 
1 1 1  3 3  1 2 0
3 3 3  3 3 

213
A is not stochastic matrix because the element in the 2 nd row and 3rd
column is negative.
B is not stochastic matrix because the elements in the second row do not
add up to 1
C is stochastic matrix because each element is non negative and they add
up to 1 in each row.

c) Regular stochastic matrix

A matrix P is said to be regular stochastic matrix if all the elements in
Pm are all positive, where m is a power, m = 1, 2, 3 e.t.c
0 1
Let A = 1 1 Where A is a Stochastic Matrix
2 2
0 1  0 1   12 12 
A2 = 1 1  1 1 = 1 3

ke
2 2 2 2 4 4

o.
3
A =
0 1  12
1 1  1
1
2  .c
 14
 = 3
3
4 

es
3 5
2 2 4 4  8 8 
ot

Since the elements in A2 and A3 are all positive then A is regular

cn

stochastic matrix.
ne

ABSORBING STATES
.k

A state Si (I = 1, 2, 3 …) of a markov chain is called absorbing if the system

w

remains in the state, Si once it enters there. Thus a state, Si is absorbing if

w

and only if the ith row of the transition matrix p has a 1 on the main
w

diagonal and zeroes every where else. See the following example.
The following matrix, P is a transition matrix of the markov chain.
S1 S2 S3 S4 S5

S1  4
1 0 1 1 1 
4 4 4
S2  0 1 0 0 0 
 
P= S3  1 0 1 1 0 
2 4 4
S4  
 0 1 0 0 0 
S5 
 0 0 0 0 1 

The States S2 and S5 are absorbing states since the 2 nd and 4th rows have 1
on the main diagonal.

214
Probability Transition Matrices (Brand switching)
These are matrices in which the individual elements are in the form of
probabilities. The probabilities represent the probability of one event
following another event i.e. the probability of transition from one event to
the next. The probabilities of the various changes applied to the initial
state by matrix multiplication, give a forecast of the succeeding state.
Normally a transition matrix is defined with its columns adding up to one
and state vectors as column vectors.
In this case the succeeding state is found by pre-multiplying the transition
matrix by the proceeding state (column) vector. If the transition matrices
are given with their rows adding up to one, then the succeeding state is
found by post multiplying the transition matrix by the preceding state (row)
vector.

Example 1
The probability transition matrix of the switching probabilities, consider

ke
that two brands G and X share the market in the ratio of 60% to 40%

o.
respectively of customers. If in every week 70% of G‘s customers retain the
.c
brand but 30% switch to product x where as 80% of X‘s customers retain
brand but 20% percent switch to brand G. Analyze the exchange in share
es
market per week.
ot

G X
cn

G 0.7 0.2 
State the system next week
X 0.3 
0.8 
ne
.k

G  60 
w

Column vector for initial market share

X  40 
w
w

 0.7 0.2   60   50 
Share next week     =  
 0.3 0.8   40   50 
 0.7 0.2   50   45 
Share week after     =  
 0.3 0.8   50   55 
and so on
This process can continue till equilibrium is reached.

G
Let the market share be  
X

215
  0.7 0.2  G G
∴    =  
 0.3 0.8   X  X

0.7G + 0.2 X = G or 0.3G + 0.8X = X

0.2X = 0.3G or 0.3X = 0.2X

G 0.3 3
i.e. = =
X 0.2 2

Hence G‘s share is 60% and X‘s share is 40%

Example 2
A marketing division toothpaste manufacturing company has worked out the

ke
following transition probability matrices concerning the behaviors of
customers before and after an advertising campaign.

o.
.c
es
ot
cn

Transition probability matrix

ne

(before advertising campaign)

.k

TO
w

Our Brand Another Brand

w

FROM (State I) (Sate II)

w

Our brand (State I) 0.8 0.2

Another Brand (sate II) 0.4 0.6

Transition probability matrix

(After advertisement)
TO
Our Brand Another Brand
FROM (State I) (Sate II)
Our brand (State I) 0.9 0.1
Another Brand (sate II) 0.5 0.5

If the advertising campaign costs Shs 20,000 per year, would it be

worthwhile for the company to undertake the campaign?

216
You may suppose there are 60,000 buyers of toothpaste in the market and
for each customer average annual profit of the company is Shs 2.50

Solution
Let P1 be the fraction share of our brand and P 2 be the fraction share of
another brand

0.8 0.2
 P1 P2      P1 P2 
0.4 0.6

 0.8P1  0.4P2 0.2P1  0.6P2    P1 P2 

0.8P1  0.4 P2  P1 and 0.2 P1  0.6 P  P2

ke
o.
0.4 P2  0.2 P1 and 0.2 P1  0.4 P2

Thus:
.c
es
P1  2 3 and P2  1 3
ot

After Advertising
cn
ne

0.9 0.1
 P1 P2      P1 P2 
0.5 0.5
.k
w

 0.9P1  0.5P2 0.1P1  0.5P2    P1 P2 

w
w

0.5P2  0.1P1 and 0.5P2  0.1P1

Thus:
Before Advertising P1  5 6 and P2  1 6

If there are 60,000 buyers

Before Advertising 2  60, 000  40, 000 Customers

3
P1  2 3 will buy our
brand
this implies that, contribution  40, 000  2.5
= Sh.100, 000
217
After Advertising
P1  5 6

this implies that,

5  60, 000  50, 000 customers will buy

6

our brand
contribution  (50, 000  2.5)  20, 000
= Sh.105, 000

ke
o.
.c
es
ot
cn
ne
.k
w
w
w

218
the difference between advertising and not advertising is

105,000 – 100,000 = Sh.5, 000 in favour of advertising,

Thus the advertising campaign is worthwhile.

PRACTICE QUESTIONS

QUESTION ONE
Because of inreasing cost increasing cost energy, the population within Maueni
district seem to be shifting from the north to the south the transition matrix S
describes the migration behaviour observed between the regions.
to north to south

 0.90 0.10 
from north
S 

ke

 0.05 0.95 
from south

o.
.c
determine whether the populations will attain an equillibrium condition and if
so, the population of the two regions.
es
ot

QUESTION TWO
cn

A simple hypothetical economy of three industries A, B and C is represented in

ne

the following table (data in millions of shillings).

.k
w

A B C Final Total
w

User Demand Output

w

Producer
A 80 100 100 40 320
B 80 200 60 60 400
C 80 100 100 20 300

Determine the output vector for the economy if the final demand changes to 60
for A, 60 for B and 60 for C

QUESTION THREE
A tea blender uses two types of tea, T1, and T2, to produce two blends, B1 and
B2 for sale. B1 uses 40% of available T1 and 60% of the available T2 whilst B2
uses 50% of the available T1 and 25% of the available T2.

Required:
a) Given that t1 kilos of T1 and t2 of T2 are made available to produce b1 kilos
of B1 and b2 kilos of B2. Express the blending operation in the matrix
format.

b) If 400 kilos of T1 and 700 kilos of T2 were made available for blending, what
quantities of B1 and B2 would be produced?

c) If 600 kilos of B1 and 450 kilos of B2 were produced, use a matrix method to
determine what quantities of T1 and T2 would be used to produce the
blends.

QUESTION FOUR
2 2 
Let A =  
 3 -3 
a) Find A2 and A3

b) If F(x) = x3 – 3x2 – 2x + 4I

ke
Find F (A)

o.
c) Find the inverse of matrix A
.c
es
QUESTION FIVE
ot

A child‘s toy is marketed in three sizes standard size contains 10 squares (S), 15
triangles (T) and 6 hexagons (H). The deluxe set contains 15 S, 20 T, 6 H and 4
cn

octagons (O). The super set contains 24 T, 8 H, 16 H, 16 S and 6 (O). Squares

ne

cost 12 pence to produce, triangles cost 8p, hexagons cost 18p and octagons
22p.
.k
w

The standard set is sold at £6, the deluxe set for £10 and super for £15. The
w

manufacturer produces 100 standard sets, 80 deluxe sets and 50 super sets per
w

week.

Use matrix form and matrix multiplications to find:

The cost of producing each set.

The number of each shape required each week
Total expenditure on shapes each week

QUESTION SIX
Matrix N below shows the number of items of type A, B, and C in warehouses Y
and W. Matrix p shows the cost in pence per day of storing (S) and maintaining
(M) one item each of A, B and C

220
 A B C S M
A  2 0.5 
Y  10 12 50   
N   P = B  3 1.5 
W  60 0 20 
C  2 0.5 
a) Evaluate the matrix (N×P) and say what it represents.

b) Stock movement occurs as follows:

At the start of the day 1:

Withdrawal of 2 type B from warehouse Y, 20 of type A from warehouse
W.

At the start of day 2:

Delivery of 7 type B and 10 of type C to warehouse Y and 15 of type B to
warehouse W.

Evaluate the total cost of storage and maintenance for days 1 and 2.

ke
c) Write down without evaluating a matrix expression which could be used

o.
to evaluate the storage and maintenance cost of items A, B and C for the
.c
period from day 1 to 4. Allow for the stock movements on days 1 and 2,
es
as described in part (b). There were no stock movements on days 3 and
4.
ot
cn
ne
.k
w
w
w

221
 CHAPTER TWELVE

INTRODUCTION TO CALCULUS

SPECIFIC OBJECTIVES
At the end of this topic the trainee should be able to:
 Differentiate algebraic functions;
 Integrate algebraic functions

INTRODUCTION

Calculus is a branch of mathematics which explains how one variable changes

in relationship to another variable. It enables us to find the rate of change of
one variable with respect to another variable.

Example
i. The rate at which business revenue is increasing at a particular stage
when volume of sales is increasing.
ii. The rate at which costs are changing at a particular stage when

ke
volume of sales is given

o.
iii. The evaluation of ‗rate of change‘ can help us to identify when the

iv.
.c
change in one variable reaches a maximum or minimum.
Calculus may be used in production management when the
es
production manager wants to know
ot

a) How much is to be manufactured in order to maximize the profits,

revenues e.t.c
cn

b) How much is to be produced in order to minimize the production

ne

costs
.k

Calculus is divided into two sections namely:

w
w

Differentiation and integration

w

Differentiation deals with the determination of the rates of change of business

activities or simply the process of finding the derivative of a function.
Integration deals with the summation or totality of items produced over a given
period of time or simply the reverse of differentiation

The derivative and differentiation

The process of obtaining the derivative of a function or slope or gradient is
referred to as derivation or differentiation.
dy
The derivative is denoted by or f΄(x) and is given by dividing the change in
dx
y variable by the change in x variable.
The derivative or slope or gradient of a line AB connecting points (x,y) and
(x+dx, y + dy) is given by
y

Change in y

 y  dy   y  dy
x Change in x x  dx  x dx
222
Where dy is a small change in y and dx is a small change in x variables.

Illustration

Line AB

(y + dy) B = (x + dx, y +dy)

dy
(x,y) = A
y dx
x (x + dx)

Rules of Differentiation

ke
1. The constant function rule
dy

o.
If given a function y = k where k is a constant then = 0
dx
.c
es
Example
Find the derivative of (i) y = 5
ot
cn
ne

Solution
i. y = 5 dy = 0
.k

dx
w
w
w

Illustration
y

5 y=5
dy 5  0
slope   0
dx 0
dy

Derivative of a constant function x

2. Power function rule

223
Given a function y  x r
dy
Then  rx r 1
dx
Example

Find dy for;
dx
(i). y = x7
(ii). y = x2ˠ
(iii). y = x-3
(iv). y=x

Solution
i. y = x7
dy = 7x 7-1 = 7x6

ke
dx

o.
ii. y = x2ˠ

dy = 2ˠ x (2ˠ - 1)
.c
es
dx
ot

iii.
y = x-3
cn

dy = -3x –3-1 = -3x-4

ne

dx
.k

iv. y=x
w

dy = 1x 1-1 = 1.x0= 1 (since x0=1)

w

dx
w

3. Power function multiplied by a constant

If given y = Axr, then dy = rAxr-1
dx

4. The sum rule

The derivative of the sum of two or more functions equals the sum of the
derivatives of the functions.
For instance
If H(x) = h(x) + g(x)
Then dy or H´(x) = h´(x) + g´(x)
dx

5. The difference rule

224
The derivative of the difference of two or more functions equals the difference
of the derivatives of the functions
If H (£) = h(x) – g(x)
Then H´ (£) = h´(x) – g´(x)

Examples.
Find the derivatives of
i. y = 3x2 + 5x + 7
ii. y = 4x2 – 2xb

Solution
i. y = 3x2 + 5x + 7

dy d  3x  d  5 x  d  7 
2

  
dx dx dx dx

ke
o.
 6x  5  0
 6x  5 .c
es
ii. y = 4x2 – 2xb
ot

dy d  4 x  d  2 x 
2 b
cn

 
dx dx dx
ne
.k

 8 x  2bxb 1
w
w

6. The product rule – both factors are functions

w

The derivative of the product of two functions equals the derivative of the first
function multiplied by the second function PLUS the derivative of the second
function multiplied by the first function.
given that H  x   h  x  .g  x 

Then H   x   h  x  .g  x   h  x  .g   x 

Example
Find dy for
dx
i. y = x2(x)
ii. y = (x2+ 3) (2x3+ x2- 3)

Solution
225
 i. y = x2(x)

dy d x  2
d  x
 x.  x2 .
dx dx dx

 x.2 x  x 2 .1
 2x2  x2
 3x 2

Note that y = x2(x) = x3. Directly differentiating this we get 3x2.

ii. y = (x2+ 3) (2x3+ x2- 3)

dy d  x  3 d  2 x 3  x 2  3
2

 .  2 x  x  3   x  3 .
3 2 2

dx dx dx

2 x.  2 x3  x 2  3   x 2  3 .  6 x 2  2 x 

ke
10 x 4  4 x3  18 x 2

o.
.c
es
7. Quotient Rule
ot

The derivative of the quotient of two functions equals the derivative of the
cn

numerator times the denominator MINUS the derivative of the denominator

times the numerator, all which are divided by the square of the denominator
ne
.k

h x 
If given H (x) =
w

g x 
w
w

h  x  .g  x   h  x  .g   x 
then H   x  
 g  x  
2

For example
Find dy for
dx
x
i.
3  x2

x
ii.
3x  7

Solutions
226
 x
i.
3  x2
d  x d 3  x2 
.3  x  
2
.x
dy
 dx dx
3  x2 
2
dx

=
3  x   2x.x
2

3  x  2

3  x 2  2x 2 3  x2
= 

3  x2  3  x2 
x3
ii. y
3x  7 
dy

3x2 3x  7  3x 3   6x3  21x2
dx 3x  72 3x  72

ke
o.
Example
.c
A farmer of a large farm of poultry announced that egg production per month
follows the equation;
es
w = 3m3 – m2
ot

m2 + 10
cn

Where w – Total no of eggs produced per month

m – amount in kilograms of layers mash feed.
ne
.k

Required
w

Determine the rate of change of w with respect to m (i.e. the rate at which the
number of eggs per month increase or decrease depending on the rate at which
w

the kilos of layers marsh are increased).

w

Solution
Let u = 3m3 – m2
∴ du = 9m2 – 2m
dm

Let v = m2 + 10
∴ dv = 2m
dm

dw  m  10  9m  2m    3m  m  2m
2 2 3 2

∴ 
 m2  10
2
dm

227
 9m4  90m2  2m3  20m   6m4  2m3 

m  10 
2 2

3m4  90m2  20m


m  10 
2 2

8. Chain Rule
This rule is generally applied in the determination of the derivatives of
composite functions, which can be defined as a function in which another
function can be considered to have taken the place of the independent
variable. The composite function is also referred to as a function of a function.
It is normally of the form y = (2x2 + 3)3. If we let u = (2x2 + 3), then y = u3.
In order to differentiate such an equation we use the formula
dy dy du
 
dx du dx
Solution

ke
y = (2x2 + 3)3

o.
Let u = 2x2 + 3
.c
es
∴ du = 2x
dx
ot
cn

Let y = u3
ne

dy
∴ = 3u2
.k

du
w
w

dy = dy . du = 3u2 x 4x = 12xu2
w

dx du dx
= 12x(2x2 + 3)2

Example
Consider the function
y = (x2 + 16x + 5)2
which can be decomposed into
y = u2 and u = x2 + 16x + 5. in this case y is a function of (x2 + 16x + 5)
Hence y = f(u) and u = g(x)
dy = dy . du
dx du dx
= (2u) (2x +16)

228
 = 2 (x2 + 16x + 5) (2x + 16)

The derivative of a function raised to power r; the composite function

rule.
The derivative of a function raised to power r equals to the power r times the
function which is raised by power (r-1), all of which is multiplied by the
derivative of the function

If y = [g(x)]r

Then dy = r[g(x)]r-1 . g´(x)

dx

For example
given y   3 x 2  4 x 
dy 5
Find
dx

ke
Solution
 5  3x 2  4 x  .  6 x  4 
dy 4

o.
dx
.c
es
Differentiation of an implicit function
An Implicit function is one of the y = x2 y + 3x2 + 50. it is a function in which
ot

the dependent variable (y) appears also on the right hand side.
cn

To differentiate the above equation we use the differentiation method for a

product, quotient or function of a function.
ne
.k

Solution
w

y = x2 y + 3x2 + 50
w

dy d  x y  d  3x  d  50 
w

2 2

  
dx dx dx dx

dy  dy 
  y  2x   x2   6x  0
dx  dx 

dy dy
0  2 xy  x 2   6x
dx dx

229
 0  2 xy   x 2  1
dy
 6x
dx

  x 2  1
dy
 2 xy  6 x
dx

dy   2 xy  6 x 

dx  x 2  1
Partial derivatives
These derivatives are used when we want to investigate the effect of one
independent variable on the dependent variable.
For example, the revenues of a farmer may depend on two variables namely;
the amount of fertilizer applied and also the type of the natural soil.
Let ㄫ = 30x2y + y2 + 50x + 60y
Where ㄫ = annual revenue in £ ‗000‘

ke
x = type of soil

o.
y = amount of fertilizer applied

Required
.c
es
Determine the rate of change of the ㄫ with respect to x and y
ot
cn

Solution
ne

ㄫ = 30x2y + y2 + 50x + 60y

.k

Differentiating ㄫ with respect to x keeping y constant we have

w

dㄫ = 60xy + 50
w

dx
w

Differentiating with respect to y keeping x constant we have

dㄫ = 30x2 + 2y + 60
dy

Maxima, minima and points of inflexion

a) Test for relative maximum
Consider the following function of x whose graph is represented by the figure
below
y = f(x)
dy = f´(x)
dx

230
y

dy
0
C dx

dy
0
dy dx
0 B D
dx

y  f  x
A E

ke
x1 x2 x3 x

o.
Relative maximum point
.c
The graph of the function slopes upwards to the right between points A and C
es
and hence has a positive slope between these two points. The function has a
ot

negative slope between points C and E. At point C, the slope of the function is
Zero.
cn

dy
Between points X1 and X2  0 Where X1 ≤ X < X2
ne

dx
.k

dy
and between X2 and X3  0 Where X2 < X ≤ X3.
w

dx
Thus the first test of the maximum points require that the first derivative of a
w

function equals zero or

w

dy
 f  x  0
dx
The second text of a maximum point requires that the second derivative of a
function is negative or
d2y
 f   x   0
dx 2

Example
Determine the critical value for the following functions and find out the critical
value that constitutes a maximum
y = x3 – 12x2 + 36x + 8

Solution
y = x3 – 12x2 + 36x + 8
then dy = 3x2 – 24x + 36 +0
231
 dx
iii. The critical values for the function are obtained by equating the first
derivative of the function to zero, that is:
dy = 0 or 3x2 – 24x + 36 = 0
dx
Hence (x-2) (x-6) = 0
And x = 2 or 6
The critical values for x are x = 2 or 6 and critical values for the function are y
= 40 or 8
ii. To ascertain whether these critical values of x will give rise to a maximum,
we apply the second text, that is
d2y <0
d2x
dy = 3x2 – 24x + 36 and
dx
d2y = 6x - 24
d2x
a) When x = 2
Then d2y = -12 <0

ke
d2x

o.
b) When x = 6 .c
es
Then d2y = +2 > 0
ot

d2x
Hence a maximum occurs when x = 2, since this value of x satisfies the second
cn

condition. X = 6 does not give rise to a local maximum

ne

b) Tests for relative minimum

.k

There are two tests for a relative minimum point

w

i. The first derivative, that is

w

dy = f´(x) = 0
w

dx
ii. The second derivative, that is
d2y = f´(x) > 0
dx2

Example:
For the function
h(x) = 1/3 x3 + x2 – 35x + 10
Determine the critical values and find out whether these critical values are
maxima or minima. Determine the extreme values of the function

Solution
i. Critical values
h(x) = 1/3 x3 + x2 – 35x + 10 and
h´(x) = x2 + 2x – 35
by first text,
then h´(x) = x2 + 2x – 35 = 0
232
 or (x-5) (x+7) = 0
Hence x = 5 or x = -7

ii. The determinant of the maximum and the minimum points requires
that we test the value x = 5 and –7 by the second text
H´´(x) = 2x + 2
a) When x = -7 h´´(x) = -12 <0
b) When x = 5 h´´(x) = 12>0
There x = -7 gives a maximum point and x = 5 gives a minimum point.
iii. Extreme values of the function
h(x) = 1/3 x3 + x2 – 35x + 10
when x = -7, h(x) = 189 2/3
when x = 5, h(x) = -98 1/3
The extreme values of the function are h(x) = 189 2/3 which is a
relative maximum and h(x) = -98 1/3 , a relative minimum

c) Points of inflexion
Given the following two graphs, points of inflexion can be determined at points

ke
P and Q as follows:

o.
y y=g(x)
.c
es
P
ot
cn
ne
.k
w
w
w

k1 x

233
 Diagram (i)
y
y =f(x)
Q

k2 x

The points of inflexion will occur at point P when

g´´(x) = 0 at x = k1

ke
´´
g (x) < 0 at x < k1

o.
g´´ (x) > 0 at x > k1
and at point Q when
f´´(x) = 0 at x = k1
.c
es
´´
f (x) > 0 at x < k1
ot

f´´(x) < 0 at x > k1

cn

Example
ne

Find the points of inflexion on the curve of the function

y = x3
.k
w

Solution
w

The only possible inflexion points will occur where

w

d2y
0
dx 2
From the function given
dy d2y
 3 x 2 and  6x
dx dx 2
Equating the second derivative to zero, we have
6x = 0 or x = 0
We test whether the point at which x = 0 is an inflexion point as follows
d2y
When x is slightly less than 0,  0 which means a downward concavity
dx 2
d2y
When x is slightly larger than 0,  0 which means an upward concavity
dx 2
Therefore we have a point of inflexion at point x = 0 because the concavity of
the curve changes as we pass from the left to the right of x = 0
Illustration

234
 y

y=x3
point of
inflexion

0 x

ke
o.
Example
.c
1. The weekly revenue Sh. R of a small company is given by
es
x3
R  14  81x  Where x is the number of units produced.
12
ot
cn

Required
i. Determine the number of units that maximize the revenue
ne

ii. Determine the maximum revenue

.k

iii. Determine the price per unit that will maximize revenue
w

Solution
w

i. To find maximum or minimum value we use differential calculus as

w

follows
x3
R  14  18 x 
12

dR 1
 81  .3 x 2
dx 12

d 2R 1 x
2
 0  .3.2 x  
dx 12 2

235
 dR 1
put 0 i.e. 81  x 2  0
dx 4

which gives x  18 or x  18

d 2R x
2

dx 2
d 2R
thus when x  18;  9 which is negative  indicating a maximum value
dx 2

Therefore at x = 18, the value of R is a maximum. Similarly at x = -18, the

value of R is a minimum. Therefore, the number of units that maximize the
revenue = 18 units
ii. The maximum revenue is given by
R = 14 + 81 + 18 – (18)3

ke
12
= Shs. 986

o.
ii. .c
The price per unit to maximize the revenue is
es
986 = 54.78 or Shs.54.78
18
ot
cn

INTEGRATION
ne

It is the reversal of differentiation. An integral can either be indefinite (when it

has no numerical value) or definite (have specific numerical values)
.k

It is represented by the sign ʃf(x)dx.

w
w

Rules of integration
w

i. The integral of a constant

ʃadx = ax +c where a = constant
Example
Find the following
a) ʃ23dx
b) ʃɤ2dx. (where ɤ is a variable independent of x, thus it is treated
as a constant).

Solution
i. ʃ23dx = 23x + c
ii. ʃɤ2dx. = ɤ2 x + c

ii. The integral of x raised to the power n

236
 1
 x dx  n  1 x
n 1
n
c
Example
Find the following integrals
a) ʃx2dx
b) ʃx-5/2 dx
Solution
 x dx  x  c
2 1 3
i) 3

 x dx   x
 52  32
ii ) 2
3
c

iii). Integral of a constant times a function

 af  x dx  a  f  x  dx
Example
Determine the following integrals

ke
i. ʃax3dx
ʃ20x5dx

o.
ii.

Solution .c
es
 ax dx  a  x dx
3 3
a)
ot

 a4 x 4  c
cn

 20 x dx  20 20 x dx
5 5
b)
ne

 103 x 6  c
.k

iv). Integral of sum of two or more functions

w

ʃ{f(x) + g(x)} dx = ʃf(x)dx + ʃg(x) dx

w

ʃ{f(x) + g(x) + h(x)}dx = ʃf(x)dx + ʃg(x)dx + ʃh(x)dx

w

Example
Find the following
i. ʃ(4x2 + ½ x-3) dx
ii. ʃ(x3/4 + 3/7 x- ½ + x5)

Solution
  4x 
 12 x 3 dx   4 x 2dx   12 x 3dx
2
i)
= 43 x3  14 x 2  c

x 
 x5 dx   x 4 dx   73 x 2 dx   x 5dx
3
 12 3
1
ii ) 4
 73 x
7 1
 74 x 4  76 x 2  16 x 6  c

237
5. Integral of a difference
ʃ{f(x) - g(x)} dx = ʃf(x)dx - ʃg(x) dx

Definite integration
Definite integrals involve integration between specified limits, say a and b
b
The integral  f  x  dx Is a definite integral in which the limits of integration are
a
a and b
The integrals is evaluated as follows
1. Compute the indefinite integral ʃf(x)dx. Supposing it is F(x) + c
2. Attach the limits of integration
3. Substitute b(the upper limit) and then substitute a (the lower
limit) for x.
4. Take the difference and the result is the numerical value for the
definite integral.

Applying these steps to the definite integral

ke
b

 f  x  dx   F  x   c 
b

o.
a


  F  b   c    F  a   c  .c
es
 F b  F  a 
ot

Example
cn
ne

Evaluate
 (3x 2 + 3)dx
3
i.
.k

1
w


5
ii. (x + 15)dx
w

0
w

Solution

a.  (3x 2 + 3)dx = [(x 3 + 3x + c)]

3

= (27 + 9 + c) – (1 + 3 + c)
= 32

 (x + 15)dx = [( ½ x2 + 15x + c)] 50

5
b.
0

= (12 ½ + 75 + c) – (0 + 0 + c)
= 87 ½

238
 
b
The numerical value of the definite integral f(x)dx can be interpreted as the
a
area bounded by the function f(x), the horizontal axis, and x=a and x=b see
figure below

y = f(x)

0
f(x)
20

25

ke
30

o.
0 a
35
.c
b x
es
40
area 45
under curve
ot
cn


b
Therefore f(x)dx = A or area under 50
the curve
ne

a
55
.k

Example 60
w

1. You are given the following marginal revenue function

w

MR  a  a1q 65
w

Find the corresponding total revenue function

70
Solution A
Total revenue   MR.dq    a  a1q dq
 aq  12 a1q2  c

Example 2
A firm has the following marginal cost function
MC  a  a1q  a2q 2
Find its total cost function.

Solution
The total cost C is given by
C = ʃMC.dq
= ʃ(a + a1q + a2q2).dq

 aq  21 q 2  q3  c
a a2
3

239
Note: Exams focus: Note the difference between marginal function and total
function. You differentiate total function to attain marginal function, this is
common in exams,
total profit = total revenue – total cost.

Example 3.
Your company manufacturers large scale units. It has been shown that the
marginal (or variable) cost, which is the gradient of the total cost curve, is (92
– 2x) Shs. thousands, where x is the number of units of output per annum. The
fixed costs are Shs. 800,000 per annum. It has also been shown that the
marginal revenue which is the gradient of the total revenue is (112 – 2x) Shs.
thousands.

Required
i. Establish by integration the equation of the total cost curve
ii. Establish by integration the equation of the total revenue curve
iii. Establish the break even situation for your company
iv. Determine the number of units of output that would

ke
a) Maximize the total revenue and

o.
b) Maximize the total costs, together with the maximum total
revenue and total costs
.c
es
Solution
ot

i. First find the indefinite integral limit points of the marginal cost as
the first step to obtaining the total cost curve
cn

Thus ʃ(92 – 2x) dx = 92x – x2 + c

ne

Where c is constant
.k

Since the total costs are the sum of variable costs and fixed costs, the constant
w

term in the integral represents the fixed costs, thus if Tc are the total costs
w

then,
w

Tc = 92x – x2 + 800
or Tc = 800 + 92x - x2

ii. As in the above case, the first step in determining the total revenue
is to form the indefinite integral of the marginal revenue
Thus ʃ(112 - 2x) dx = 112x – x2 + c
Where c is a constant

The total revenue is zero if no items are sold, thus the constant is zero and if
Tr represents the total revenue, then
Tr = 112x – x2
iii. At break even the total revenue is equal to the total costs
Thus 112x – x2 = 800 + 92x - x2
20x = 800
x = 40 units per annum
240
iv.
a) Tr = 112x – x2
d Tr 
 112 x  2 x
dx
d 2 Tr 
 2
dx 2
at the maximum point
d 2 Tr 
 0 that is 112  2 x  0
dx 2
x = 56 units per annum
d 2 Tr 
Since  2 this confirms the maximum
dx 2

ke
The maximum total revenue is Shs. (112 x 56 – 56 x 56) x 1000
= Shs. 3,136,000

o.
.c
es
ii. Tc = 800 + 92 x – x2
ot

d Tc 
 92  2 x
cn

dx
ne

d 2 Tc 
 2 x
.k

dx 2
w

At this maximum point

w

d Tc 
w

0
dx
92 – 2x = 0
92 = 2x
x = 46 units per annum
since
d 2 Tc 
 2 x this confirms the maximum
dx 2
the maximum costs are Shs. (800 + 92 x 46 - 46 x 46) x 1000
= Shs. 2,916,000

241
PRACTICE QUESTIONS

QUESTION ONE
Find the derivative of

a) y = 6x – x
1
b) y
x2

c) y  1  2x

1
d) y
x

QUESTION TWO
A cost function is

ke
Ksh.(c) = Q2 – 30Q + 200

o.
Where Q = quantity of units produced

Find the point of minimum cost. .c

es
ot

QUESTION THREE
cn

250 members of a certain society have voted to elect a new chairman. Each
member may vote for either one or two candidates. The candidate elected is
ne

the one who polls most votes.

.k

Three candidates x, y, z stood for election and when the votes were counted,
w

it was found that,

w

59 voted for y only, 37 voted for z only

w

12 voted for x and y, 14 voted for x and z

147 voted for either x or y or both x and y but not for z
102 voted for y or z or both but not for x.

Required:
i) How many voters did not vote?
ii) How many voters voted for x only?
iii) Who won the election?

QUESTION FOUR
The weekly revenue Ksh.R of a small company is given by:
R = 14 + 81x – x3 where x is the number of units produced
12

Required:
a) Determine the number of units that maximize the revenue.

242
b) Determine the maximum revenue.
c) Determine the price per unit that will maximize the revenue

QUESTION FIVE
A furniture firm has two operating departments; Production and sales. The
firms‘s operating costs are split between these two departments with the
resultant period of fixed costs of Shs.20,000 and Shs.6,000 respectively. The
production department has a basic variable cost per unit of Shs.6 plus
additional variable cost per unit of Shs.0.0002 which relates to all the
manufactured items during the period. The sales department has a variable
cost per unit of Shs.2. The sales department receives the finished goods from
the production department and pay the basic variable cost per unit plus 80% of
the same.

NB: Demand Q is given by the following function:

Q = 40,000 – 2,000P, where P is the selling price of the sales

department.

ke
Required:

o.
a) Calculate the quantity that maximizes the profits of the production
department.
.c
b) Calculate the selling price that maximizes the profits of the sales
es
department.
ot

c) Determine the firm‘s profit as a result of adopting the quantity and selling
prices
cn

in i and ii.
ne

d) Determine the quantity and selling price that maximize the ship‘s profit.
What is the amount of this profit?
.k
w

QUESTION SIX
w

a) Describe how quadratic equations can be used in decision making.

w

b) The demand for a commodity is given by p = 400 – q. The average total

cost of producing the commodity is given by

1000
ATC   100  5q  q 2
q Where p is the price in shillings and q is the
quantity in kilograms.

Required
i) What does 1000
q
in the ATC equation represent economically?
(1 mark)
ii) Determine the output that leads to maximum profit and the profit at the
level of output. (9
marks)

243
c) Alpha industries sells two products, X and Y, in related markets, with
demand functions given by:

Px – 13 + 2X + Y = 0
Py– 13 + X + 2Y = 0

The total cost, in shillings, is given by:

TC = X + Y

Required:
Determine the price and the output for each good which will maximize
profits. (7 marks)
(Total: 20 marks)

QUESTION SEVEN
a) The following table shows the Fixed Cost (F) and the variable cost (V) of
producing 1 unit of X and 1 unit of Y:

ke
Product
X Y

o.
Cost F 5 8 (Shs ‗000‘)
Cost V 4 12 .c
es
ot

When x units of X and y units of Y are produced, the total fixed cost is
cn

Shs.640,000 and total variable cost is Shs.820,000. Express this information as

a matrix equation and hence find the quantities of x and y produced using
ne

matrix algebra. (10 marks)

.k

The marginal productivity of an industrial operation (the production of electric

w

furnaces) is given by:

w

f  x   60  10
w

x2

Where x is capitalization in millions of shillings. Given that, when the

capitalization is Shs. Million they can produce 62 of the furnaces per week.

Required:
a) How many furnaces they will be able to produce if their capitalization
increased to Shs 10 million.
b) What does the term marginal of productivity mean?
(10 marks)
(Total: 20 marks)

244
 CHAPTER THIRTEEN

NETWORK ANALYSIS

Specific objectives
At the end of this topic the trainee should be able to:
 Describe networks and need for the analysis;
 Construct project networks;
 Determine the critical path;
 Apply network analysis to decision problems.

INTRODUCTION
This is a system of interrelationship between jobs and tasks for planning and
control of resources of a project by identifying critical path of the project.

Terminology
Activity. Task or job of work, which takes time and resources e.g building a
bridge. It is represented by an arrow which indicates where the task
begins and ends

ke
o.
Event (node). This is a point in time and it indicates the start or finish of an
.c
activity e.g in building a bridge, rails installed. It is represented by a
circle.
es
ot

Dummy activity. An activity that doesn‘t consume time or resources, it is

cn

merely to show logical dependencies between activities so as abide by

ne

rules of drawing a network, it is represented by a dotted arrow

.k
w

Network. This is a combination of activities and events (including dummy

w

activities)
w

Rules for Drawing a Network

a) A network should only have one start point and one finish point (start
event and finish event )
b) All activities must have at least one preceding event (tail event) and at
least one succeeding event (head event), but an activity may not share
the same tail event and head event.
c) An activity can only start after its tail event has been reached
d) An event is only complete after all activities leading to it are complete.
e) Activities are identified by alphabetical or numeric codes i.e. A,B,C;
1,2,3 or identification by head or tail events 1-2, 2-4, 3-4,1-4…

245
 f) Loops (a series of activities leading back to the same event) and danglers
(activities which do not link to the overall project) are not allowed

Dangling activity
Loop

Dummy Events
This is an event that does not consume time or resources, it is represented by
dotted arrow. Dummies are applied when two or more events occur
concurrently and they share the same head and tail events e.g. when a car
goes to a garage tires are changed and break pads as well, instead of
representing this as;

A- Tires Changed

Car Arrives (CA) Car ready (CR)

ke
B- Break pads Changed

o.
These events are represented as;
.c
es
ot

B
cn

CA CR
A
ne

Example of a network.
.k
w

Activities
w

1-2 - where 1 is the preceding event where as 2 is the succeeding event of

w

the activity
1-3
2-4
2-5
3-5
4-5
4-6
5-6
6-6

246
 4
2

6 7
1

3 5

Network Analysis-Time Analysis

Assessing the time
a) After drawing the outline of the network time durations of the activities
are then inserted.
a) Time estimates. The analysis of the projects time can be achieved by
using :
i. Single time estimates for each activity. These estimates would
be based on the judgment of the individual responsible or by
technical calculations using data from similar projects
ii. Multiple time estimates for each activity. the most usual

ke
multiple time estimates are three estimates for each activity ,

o.
i.e. optimistic (O), Most Likely (ML), and Pessimistic (P). These
.c
three estimates are combined to give an expected time and the
accepted time formula is:
es
ot

O  P  4ML
Expected time =
cn

6
For example assume that the three estimates for an activity
ne

are
Optimistic 11 days
.k

Most likely 15 days

w

Pessimistic 18 days
w

11  18  4 15 
w

Expected time =
6
= 14.8 days
b) Use of time estimates. as three time estimates are converted to a
single time estimate. There is no fundamental difference between
the two methods as regards the basic time analysis of a network.
However, on completion of the basic time analysis, projects with
multiple time estimates can be further analyzed to give an
estimate of the probability of completing the project by a
scheduled date.
c) Time units. Time estimates may be given in any unit, i.e. minutes
, hours, days depending on the project. All times estimates within
a project must be in the same units otherwise confusion is bound
to occur.

Basic time analysis – critical path

b) The critical path of a network gives the shortest time in which the
whole project can be completed. It is the chain of activities with the
247
 longest duration times. There may be more than one critical path which
may run through a dummy.

Earliest start times (EST) – Forward pass, Once the activities have been
timed we can assess the total project time by calculating the ESTs for
each activity. The EST is the earliest possible time at which a
succeeding activity can start.
Assume the following network has been drawn and the activity times
estimated in days.

2
B D
2 4

0 A 1 C 3 E 4 F 5
1 3 1 2

ke
o.
The ESTs can be inserted as follows.

EST
.c
es
ot

2
3
cn

B D
ne

2 4
.k

0 A 1 C 3 E 4 F 5
w

0 1 1 3 4 1 7 2 9
w
w

The method used to insert the ESTs is also known as the forward pass,
this is obtained by;
EST = The greater of [EST (tail event) + Activity duration]

a) Start from the start event giving it 0 values,

b) For the rest of the events EST is obtained by summing the EST of the tail
event and the activity duration
c) Where two or more routes converge into an activity, calculate individual
EST per route and then select the longest route (time)
d) The EST of the finish event is the shortest time the whole project can be
completed.

Latest Start Times (LST) – Backward pass. this is the latest possible time
with which a preceding activity can finish without increasing the project
duration. After this operation the critical path will be clearly defined.

From our example this is done as follows;

248
 2 LST
3 3
B D
2 4

0 A 1 C 3 E 4 F 5
0 0 1 1 1 3 4 6 1 7 7 2 9 9

LST = Lowest of [LST (head event) – activity duration]

a) Starting at the finish event, insert the LST (i.e. 9 for our example) ,and
work backwards through the network.
b) deduct each activity duration from the previously calculated LST (i.e.
head LST).
c) Where the tails of activities join an event, the lowest number is taken as
the LST for that event

ke
o.
Critical path determination
.c
es
Critical Path. . This is the chain of activities in a network with the longest
duration Assessment of the resultant network shows that one path
ot

through the network (A, B, D, F) has EST's and LST's which are
cn

identical this is the critical path.

ne

The critical path can be indicated on the network either by a

different colour or by two small transverse lines across the arrows
.k

along the path, thus in our example we have;

w
w
w

2
3 3
B D
2 4

0 A 1 C 3 E 4 F 5
0 0 1 1 1 3 4 6 1 7 7 2 9 9

Activities along the critical path are vital activities which must be
completed by their EST's/LST's otherwise the project will be delayed.

Non critical activities (in the example above, C and E) have spare
time or float available. C and/ or E could take up to an additional 2
days in total without delaying the project duration. If it is required to
reduce the overall project duration then the time of one or more of
the activities on the critical path must be reduced perhaps by using
more labour, or better equipment to reducing job times.

249
Float
Float or spare time can only be associated with activities which are non-
critical. By definition, activities on the critical path cannot have float. There
are three types of float, Total Float, Free Float and Independent Float. To
illustrate these types of float we use the following example.
6
A 5 B 40 50 C
10 20 10

Section of the network

a) Total float. Amount of time by which a path of activities could be

delayed without affecting the overall project duration. The path in this
example consists of one activity only i.e. B

ke
Total Float = Latest Finish time (LFT) - Earliest Start time(EST) time –

o.
Activity Duration
Total Float = 50 - 10 - 10
= 30 days
.c
es
ot

b) Free float Amount of time an activity can be delayed without affecting

the commencement of a subsequent activity at its earliest start time,
cn

but may affect float of a previous activity.

ne

Free Float = Earliest Finish Time(EFT) - EST - Activity Duration

.k

Free Float = 40-10-10

w

= 20 days
w

c) Independent float. Amount of time an activity can be delayed when all

w

preceding activities are completed as late as possible and all succeeding

activities commenced as early a possible. Independent float therefore
does not affect the float of either preceding or subsequent activities.

Independent float = EFT- Latest Start time (EST) - Activity Duration

Independent float = 40 - 20 - 10
= 10 days
Note:
 for examination purposes, float always refers to total float
 The total float can be calculated separately for each activity but it is often
useful to find the total float over chains of non-critical activities between
critical events
Example.
The following represents activities of a network.

250
 Activity Preceding Activity Duration Days
A - 4
B A 7
C A 5
D A 6
E B 2
F C 3
G E 5
H B,F 11
I G,H 7
J C 4
K D 3
L I,J,K 4
Required:
a) Draw the network diagram and find the critical path
b) Calculate the floats of the network in question

Solution. (a)
4 8

ke
E 13 G
2 5 23

o.
3
11 .c H
es
11
5 I 7
ot

12
cn

B 7
ne

F 10
1 9
.k

3 L 34
0 A 30
4
w

2 J
4 C
w

4 6 4
5
w

K
D 3
6

7
10

 First we draw the network structure ensuring it fits the data above
 We then label all activities from 1 to 12 and indicate activity duration
 Conduct a forward pass operation (to obtain the diagram above)
 Operate backward pass to establish the critical path, thus we have…

251
 E 13 18 G
2 5 23 23

11 15 H
11
I 7
12 12
B 7

F 3
L 34 34
0 0 A 30 30
4
4 J
4 4 C
4
5 9 9

ke
D 3
6

o.
.c
es
10 27
ot
cn

Therefore we get the critical path to be, A- C- F- H- I- L

ne

b) The floats of the network,

Activit Total Float Free Independ
.k

y Float ent Float

w

Durati
w

on
w

Activity EST LST EFT LFT D LFT -EST- EFT-EST-D EFT-LST-D

D
*A 0 0 4 4 4 - - -
B 4 4 11 15 7 4 - -
*C 4 4 9 9 5 - - -
D 4 4 10 22 6 12 - -
E 11 15 13 21 2 8 - -
*F 9 9 15 15 3 - - -
G 13 21 23 23 5 5 5 -
*H 12 12 23 23 11 - - -
*I 23 23 30 30 7 - - -
J 9 9 30 30 4 17 17 17
K 10 22 30 30 3 17 17 5
*L 30 30 34 34 4 - - -
The total float on the non-critical chains are;

Non-critical Time required Time available Total Float over

chain (sum of (LFT of last activity-EST of 1st chain
252
 duration) activity)
B,E,G 14 19 5
B,Dummy 7 8 1
D,K 9 26 17
J 4 21 17

Slack
This is the difference between the EST and LST for each event. Strictly it does
not apply to activities but on occasions the terms are confused in examination
questions and unless the context makes it abundantly clear that event slack is
required, it is likely that some form of activity float is required. Events on the
critical path have zero slack.

APPLICATIONS

Cost Scheduling
This is done by calculating the cost of various project durations, cost analysis
seeks to find the cheapest way of reducing the overall cost duration of a

ke
project by increasing labour hours, equipment e.t.c.

o.
Terminologies
.c
Normal cost. The costs associated with a normal time estimate for an activity.
es
Often the normal time estimate is set at the point where resources
(labour, equipment, etc.) are used in the most efficient manner.
ot
cn

Crash cost. The costs associated with the minimum possible time for an
activity. Crash costs, because of extra wages, overtime premiums,
ne

extra facility costs are always higher than normal costs.

.k

Crash time. The minimum possible time that an activity is planned to take. .
w

The minimum time is invariably brought about by the application of

w

extra resources, e.g. more labour or machinery.

w

Cost slope. This is the average cost of shortening an activity by one time unit
(day, week, month as appropriate). The cost slope is generally
assumed to be linear and is calculated as follows:

Cost slope = Crash cost – Normal cost

Normal time – Crash time

Example
A project has the following activities and costs. You are required to prepare
the least cost schedules for all possible durations from normal time – normal
cost to crash time – crash cost.

253
Activity Preceding Duration Crash Cost Crash Cost
Activity days time (Shs). cost slope
A - 4 3 360 420 60
B - 8 5 300 510 70
C A 5 3 170 270 50
D A 9 7 220 300 40
E B,C 5 3 200 360 80

1
D 3
4 4 9 14 14
A C
4 5 E
0 5
0 0 B
2
8
9 9

ke
Project duration and costs

o.
(a) Normal duration = 14 days
Critical path = A,C,E
.c
es
Project cost (cost of all activities in normal time) = Shs. 1,250.
ot
cn

(b) Reduce by 1 day the activity on the critical path with the lowest cost
slope. Thus we reduce C at extra cost of Shs. 50.
ne

Now
.k

Project duration = 13 days

w

Project cost = Shs. 1,300

w
w

Note: that all activities are now critical.

(c) Further reducing the critical path by 1 day will require that more than
one activity is affected because there exist several critical paths.

Reduce by 1 day Extra cost Activities

critical
A and B 60 + 70 = 130 All
D and E 40 + 80 = 120 All
B, C and D 70 + 50 + 40 = All
160
A and E 60 + 80 = 140 A, D, B, E

From this we realize that reducing D and E is the cheapest.

254
 However closer examination of the fourth alternative reveals that C is
now non-critical and has 1 day float. Since we earlier reduced C for Shs.
50, if we reduce A and E and increase C by a day which will save Shs. 50.

Then the net cost for 12 day duration = 1,300 + (140 – 50) = 1,390.

The network becomes………

1 D
3
3 3 9
12 12

A
3 (crash) 5
C E
4
0 B 2

ke
0 0 8 7 7

o.
.c
es
ot

(d) Next we reduce D & E

Project duration = 11 days
cn

Project cost = 1,510

ne

Critical activities = All

.k

(e) Final reduction possible is by reducing B, C & D for Shs. 160 the network
w

then becomes.
w
w

1 D
3
3 3
7 (Crash) 10 10
A
3 (crash)
4
C E
3 (crash)
0 B 2
0 0 7 7 7

Duration = 10 days
255
 Cost = Shs. 1,670
Critical activities = All.

Note: only critical activities affect project duration.

: Always look for a possibility of increasing the duration of a
previously
Crashed activity.

Scheduling Resources and Gantt Chart

Apart from time, cost network analysis also help in controlling and planning of
resources.

Example
A project has the following activity durations and resource requirements.

Activity Preceding Duration Resource requirement (man

activity (days) power)

ke
A - 6 3

o.
B - 3 2
C - 2 .c
2
es
D C 2 1
E B 1 2
ot

F D 1 1
cn

Required
ne

i) What is the networks critical path

ii) Draw a gantt chart diagram indicating activity times, using their
.k

estimate.
w

iii) Show resource requirement on a day to day basis assuming all events
w

commence at their estimates.

w

iv) Assuming that only six employees are available, how will the activities
be planned for?

Solution
i)
Activities Duration EST LST Man power
A 6 0 0 3
B 3 0 0 2
C 2 0 0 2
D 2 2 3 1
E 1 3 5 2
F 1 4 5 1

256
 w
w
w
.k
ne
cn

257
ot
es
.c
o.
ke
ii) A gantt chart or a bar chart. This is a diagram indicating a resource
scaled network.

ke
o.
iii) Resource requirements on a day to day basis.
.c
es
ot
cn
ne
.k
w
w
w

iv) When on 6 manpower resources are available then we adjust the

activities to accommodate this and still end at the given critical time
duration i.e.

258
 ke
o.
Node Networks
.c
This network also known as a procedure diagram is represented with the same
information as a network diagram.
es
Its characteristics are;
ot

i) Activities are shown in boxes instead of arrows

ii) Events are not represented.
cn

iii) The arrows linking boxes indicate the sequence precedence of

ne

activities.
iv) Dummies aren‘t necessary.
.k

E.g.
w
w
w

Would appear as

259
A full activity node network is represented as;

This is represented as;

ke
o.
.c
es
ot
cn
ne
.k
w
w
w

Note:
i) EST and LST are calculated by the same process we learnt earlier.
ii) EFT and LFT are calculated by adding the activity time duration to EST
and LST respectively.
iii) Critical path is similarly identified by identifying equal EST and LST
throughout the path.

260
 LESSON 8 REINFORCING QUESTIONS

QUESTION ONE
Regal Investments has just received instructions from a client to invest in two
shares; one an airline share, the other an insurance share. The total maximum
appreciation in share value over the next year is to be maximized subject to
the following restrictions:
- the total investment shall not exceed Sh.100,000
- at most Sh.40,000 is to be invested in the insurance shares
- quarterly dividends must total at least Sh.2,600

The airline share is currently selling for Sh.40 per share and its quarterly
dividend is Sh.1per share. The insurance share is currently selling for Sh.50 per
share and the quarterly dividend is Sh.1.50 per share. Regal‘s analysts predict
that over the next year, the value of the airline share will increase by Sh.2 per
share and the value of the insurance share will increase by Sh.3 per share.
A computer software provided the following part solution output:

ke
Objective Function Value = 5,400

o.
Variable Number Reduced cost
Airline shares 1,500 0.000 .c
es
Insurance shares 800 0.000
ot

Constraint Slack/Surplus Dual prices

cn

Total investment 0.000 0.050

ne

Investment in insurance 0.000 0.010

Dividends 100.000 0.000
.k
w
w
w

261
Objective Coefficient Ranges

Variable Lower limit Current value Upper limit

Airline share 2.500 3.000 No upper limit
Insurance share 0.000 2.000 2.400

Right-hand Side Ranges

Constraint Lower limit Current value Upper limit

Total investment 96,000.00 100,000 No upper limit
Investment in insurance20,000.00 40,000 100,000.00
Dividends No lower limit 2,600 2,700.00

Required:
a) Formulate the above problem.
b) Explain what reduced cost and dual prices columns above mean.
c) How should the client‘s money be invested to satisfy the restrictions?
d) Suppose Regal‘s estimate of the airline shares appreciation is an error,

ke
within what limits must the actual appreciation lie for the answer in (c)
above to remain optimal?

o.
(Q 6 Dec
2001) .c
es
QUESTION TWO
ot

a) A baker makes two products; large loaves and small round loaves. He can
cn

sell up to 280 of the large loaves and up to 400 small round loaves per day.
ne

Each large loaf occupies 0.01m3 of shelf space, each small loaf occupies
0.008m3 of space, and there is 4m3 of shelf space available. There are 8
.k

hours available each night for baking, and he can produce large loaves at
w

the rate of 40 per hour, and small loaves at the rate of 80 per hour. The
w

profit on each large loaf is Sh.5.00 and Sh.3.00 profit on the small round
w

loaf.

Required:
In order to maximize profits, how many large and small round loaves should
he produce?

b) Summarize the procedure for solving the kind of quantitative technique you
have used to solve part (a) above. (Q 6 June
2001)

QUESTION THREE
a) A small company will be introducing a new line of lightweight bicycle
frames to be made from special aluminium alloy and steel alloy. The
frames will be produced in two models, deluxe and professional. The
anticipated unit profits are currently Sh.1,000 for a deluxe frame and
Sh.1,500 for a professional frame. The number of kilogrammes of each alloy

262
 needed per frame is summarized in the table below. A supplier delivers 100
kilogrammes of the aluminium alloy and 80 kilogrammes of the steel alloy
weekly.

Aluminium alloy Steel alloy

Deluxe 2 3
Professional 4 2

Required:
i) Determine the optimal weekly production schedule.
ii) Within what limits must the unit profits lie for each of the frames for
this solution to remain optimal?

b) Explain the limitations of the technique you have used to solve part (a)
above.
(Q 6 Dec 2000)

QUESTION FOUR

ke
a) Define the following terms as used in linear programming:

o.
i) Feasible solution
ii) Transportation problem
iii) Assignment problem .c
es
ot

b) The TamuTamu products company ltd is considering an expansion into five

new sales districts. The company has been able to hire four new
cn

experienced salespersons. Upo analysing the new salesperson‘s past

ne

experience in combination with a personality test which was given to them,

the company assigned a rating to each of the salespersons for each of the
.k

districts .These ratings are as follows:

w
w

c)
w

Districts
1 2 3 4 5
A 92 90 94 91 83
Salespersons B 84 88 96 82 81
C 90 90 93 86 93
D 78 94 89 84 88

The company knows that with four salespersons, only four of the five
potential districts can be covered.

Required:
i) The four districts that the salespersons should be assigned to in order
to maximize the total of the ratings
ii) Maximum total rating. (Q 6 June
2002)

263
QUESTION FIVE
a) Explain the value of sensitivity analysis in linear programming problems and
show how dual values are useful in identifying the price worth paying to
relax constraints.

b) J.A Computers is a small manufacturer of personal computers. It

concentrates on production of three models- a Desktop 386, a Desktop 286,
and a Laptop 486, each containing one CPU Chip. Due to its limited
assembly facilities JA Computers are unable to produce more than 500
desktop models or more than 250 Laptop models per month. It has one
hundred and twenty 80386 chips (these are used in Desktop-386) and four
hundred 80286 chips (used in desktop 286 and Laptop 486) for the month.
The Desktop 386 model requires five hours of production time, the Desktop
286 model requires four hours of production time, and the Laptop 486
requires three hours of production time. J.A Computers have 2000 hours of
production time available for the coming month. The company estimates
that the profit on Desktop 386 is Sh. 5,000. for a desktop 286 the profit is
Sh.3,400 and Sh.3,000 profit for a laptop 486.

ke
Required:
Formulate this problem as a profit maximization problem and mention the

o.
basic assumptions that are inherent in such models.
.c
es
c) An extract of the output from a computer package for this problem is given
below:
ot
cn

Output solution
X1=120, X2 = 200, X3 = 200
ne

Dual values Constraints 3 150

.k

Constraints 4 90
Constraints 5 20
w

Sensitivity analysis of objective function coefficients:

w
w

Variable Lower Original Upper

limit value limit
X1 100 250 No limit
X2 150 170 200
X3 127.5 150 170

Sensitivity analysis on R.H.S ranges.

Constrain Lower Original Upper

ts limit value limit
1 320 500 No limit
2 200 250 No limit
3 80 120 130
4 350 400 412.5
5 1950 2000 2180

264
 X1=Monthly production level for Desktop 386.
X2 =Monthly production level for Desktop 286.
X3=Monthly production level for Laptop 486.

Required:
i) Interpret the output clearly, including optimum product mix, monthly
profit, unused resources and dual values
ii) Explain the purpose of upper limits and lower limits for the variables
X1,X2,X3 and constraints 1 to 5.
iii) Calculate the increase in profit if the company is able to produce a
further 10 CPU 80386 chips. (Q7
July 2000 Pilot paper)

QUESTION SIX
Preface Retailers is a high-technology retailer and mail order business. In order
to improve its process the company decides to install a new microcomputer
system to manage its entire operation (i.e. payroll, accounts, inventory).

ke
Terminals for each of its many stores will be networked for fast, dependable
service. The specific activities that Preface will need to accomplish before the

o.
system is up and running are listed below. The table also includes the
.c
necessary increased staffing to undertake the project.
es
Activity Preceding Duration Increase
ot

Activities (Days) d
cn

Staff
A. Build insulated enclosure - 4 1
ne

B. Decide on computer - 1 3
.k

system A 3 2
C. Electrical wiring of room B 2 1
w

D. Order and collect A 4 2

w

computer D, E 2 2
w

E. Install air conditioning B 5 1

F. Install computer C, F 2 1
G. Staff testing G, H 3 1
H. Install software
I. Staff training

Required:
a) Draw a network diagram for the project and determine the critical
path and its duration.
b) Assuming that all activities start as soon as possible, draw a progress
chart for the project, showing the times at which each activity takes
place and the manpower requirements.
c) The union has decided that any staff employed on the project must
be paid for the duration of the project whether they work or not, at
a rate of £500 per day.

265
Assuming that the same staff is employed on the different activities,
determine the work schedule that will minimise labour costs though
not necessarily the project time. What is the cost associated with
this schedule?

Comment on the validity of the assumption.

ke
o.
.c
es
ot
cn
ne
.k
w
w
w

266
 CHAPTER FOURTEEN

INVENTORY CONTROL MODELS

Specific objectives
At the end of this topic the trainee should be able to:
 Define inventory;
 Describe the control systems;
 Calculate the EOQ;
 Determine safety stock (ss) and re-order level.

INTRODUCTION
The activities of a business during a financial year combine investment projects
in progress with new projects commencing and others terminate within the year.
It would appear reasonable to presume, therefore, that business financial
reports are presented in the cash-flow mode used to appraise investments, to
facilitate comparison of actual with planned cash flows.

ke
o.
Some businesses do make such comparisons as part of their retrospective
.c
monitoring of investment decisions, but there is no obligation to do so. Cash-
flow accounting, as it is called, has its supporters, but its introduction is
es
frustrated by statutory and non-statutory regulations.
ot

The Companies Act requires limited companies to produce profit and loss
cn

accounts and balance sheets in prescribed form. The Inland Revenue assumes
ne

that taxable profit has been computed by applying recognized accounting

principles. The Accounting Standards Committee recommends the application of
.k

standard practices in the measurement of profit and portrayal of a company's

w

financial position in its balance sheet. More compellingly, profit and loss
w

reporting is compatible with the investors' objectives of stable and growing

w

earnings.

Profit is measured conventionally by setting against the sales revenue for a

period the costs expired in earning that revenue. That is, sales are matched
against their relevant costs. Profit is therefore more evenly reported than it
would be if all cash receipts and payments, capital and revenue, were fully
reflected in the accounts of the period in which they are received and paid.

The management accountant also adopts the matching principle when preparing
control information in both actual and budgeted form, and also ascertains full
product cost as a starting point for setting selling prices.

This outlines the systems and methods used to control the flow of resources
through production and service cost centres, for their eventual inclusion in
product and period costs.

267
MATERIAL CONTROL
It is said that "any fool can sell"—it is buying at the right price that is more
critical to the achievement of a satisfactory return on capital employed. Buying
price is important of course, but buying the right materials, is equally important
if production targets are to be achieved and investment in inventories to be
minimized.

WHAT TO ORDER
This is governed by product specifications, but an efficient buyer will always
have his ear to the ground to discover new and substitute materials and
components of advantageous quality and price. Other economies can be realized
by reducing the variety of materials purchased by standardization, e.g. reducing
the variety of colors of paint stocked, or by introducing value analysis into the
decision process.

VALUE ANALYSIS

ke
Is a formalized technique involving a rigorous analysis of products at the design
stage or at any time during the saleable lives, to determine their value

o.
characteristics. These are the attributes that a customer looks for in a product
.c
and include its use value (functional qualities), appeal value (color, style etc.)
es
and second-hand value (e.g. trade-in-price). The object of value analysis is to
build into the product the optimum of desired value at minimum cost, by
ot

introducing the most up-to-date designs, materials and methods of manufacture.

cn

No more value need be built into the product than is desired by the customer.
For example, moulded plastic bumper bars are now fitted to many cars, because
ne

they are cheaper than and equally as functional as chromium-plated steel ones.
.k
w
w
w

268
How much to order
Supposing the estimated annual usage of a component by Harambee Agricultural
Machinery Ltd is 20,000 units. Usage is even throughout the year and only one
order per annum is placed with the supplier. Because only one delivery is made,
average stock will be high, i.e. 20,000  2 = 10,000 and consequently
stockholding costs will be very high. On the other hand, the costs of ordering
will be negligible. If two orders are placed there will be less in stock (i.e.
average 5,000), which will reduce holding costs, but ordering costs will increase.
Thus, the higher the number of orders placed, the lower are stockholding costs,
but the higher are ordering costs.

Stockholding costs include interest on the capital invested in stocks, storage,

insurance, rates, security, building maintenance, heating, etc. Ordering costs
include buying-department staff costs, receiving and handling.

Assuming that the cost of each Harambee component is £10, that holding cost is
10% of stock value and the cost of placing an order is £1, the total annual cost of
stockholding and ordering when different numbers of orders are placed, is as

ke
follows:

o.
Number of orders 4 20 50 100 200 400
Size of order 5,000 1,000 .c
400 200 100 50
es
Average Stock (50% 2,500 500 200 100 50 25
order)
ot

Holding cost £2,500 £500 £200 £100 £50 £25

cn

Ordering cost (£1 per £4 £20 £50 £100 £200 £400

ne

order)
Total Annual Cost £2,504 £520 £250 £200 £250 £425
.k
w
w
w

269
 ke
o.
Figure 1: Economic Order Quantity
.c
es
Placing 100 orders a year results in the lowest of ordering and holding cost of
£200, therefore the economic order quantity is 200 units.
ot

The same information is graphed in Figure 1 above, showing that the economic
cn

order quantity (EOQ) is the point where ordering and holding costs are equal,
and total £200.
ne

As costs of ordering and holding stock are equal at the EOQ point, we can build a
.k

simple mathematical model to solve the problem, as follows:

w
w

Q X H = A X P
w

2 Q

Where Q = EOQ
H = holding cost per unit
A = annual demand
P = cost of placing an order

2 AP
finally Q =
H
Using the data in the previous example:
2 x 20,000 x 1
EOQ =
1

= 40,000
= 200

270
Although the model assumes that holding and ordering costs are fixed, this
simplification is acceptable given a relatively unchanging level of production
activity. In addition, because the total cost curve in the Figure 1 is relatively
flat either side of the EOQ, minor errors and approximation in the variables used
in the calculation may not affect the end result significantly.

Practical constraints on the use of the model include restrictions on the available
storage space, the availability of quantity discounts (though the model can be
modified in this respect), the seasonal nature of supplies, the shelf-life of
products and delivery schedules imposed by suppliers.

WHEN TO ORDER
If deliveries from suppliers normally take two weeks to arrive, then
replenishment orders should be placed with them when the level of stocks
represents two weeks' supply. For example, if usage is 200 units a week, an
order (the EOQ) will be placed when the stock level falls to 400 units. Figure
6(a) illustrates that, with certain knowledge of usage and lead time, delivery

ke
takes place just as stock is exhausted.

o.
.c
es
ot
cn
ne
.k
w
w
w

Figure 2(a): Stock levels when usage and replenishment times are known

271
Figure 2(b): Stock levels when usage and replenishment are uncertain

Lead times and usage may not be stable and provision against running out of
stock the becomes necessary (Figure 2 (b)). Safety stocks have a cost, however,
and this has to be balanced against the cost of running out of stock. `Stock outs'

ke
may cause loss of customers and the probability of this happening at various

o.
levels of safety stock must be estimated. The point, at which the cost of
carrying safety stocks plus the cost of `stock outs' is lowest, indicates the safety
.c
stock level. Notice that uncertainty causes the reorder level to be at a higher
es
level to include the required safety stock.
ot
cn

CONTROLLING MATERIAL FLOW

ne

Figure 3 outlines the progressive stages in purchasing, issuing and recording

materials in a manufacturing concern. An efficient system of documenting and
.k

recording is vitally necessary, not only for accounting purposes, but to ensure
w

that the right materials arrive at the right place at the right time.
w
w

The purchase requisition submitted to the buyer may be triggered automatically

if the system is computerised, by a message from the stores ledger that the
reorder level has been reached. Other requests to purchase may be raised by
the production planning department for new product materials not yet carried in
stock, and also by any departmental head for supplies and equipment of any
kind.

272
The buyer, ideally after making enquiries of several suppliers, sends a purchase
order, and eventually the material is received, checked by the good-inwards
department as to quality and quantity, and is detailed on a goods-received note
(GRN). One copy of the GRN goes to the buyer to write off the outstanding order
record; one to the accounts department for checking against the order and
invoice—the latter authorizing payment to the supplier; and one to the stores
department with the materials.

ke
o.
.c
es
ot
cn
ne
.k
w
w
w

Figure 3: Procedure of material acquisition, recording and issue

A stores record is maintained into which the quantity and value of materials
received is entered. Issues of materials to production are made by authorized
materials requisitions which are also entered into the stores ledger to keep that
record up to date continuously, and also into the appropriate job or process
const record.

As already indicated, all the above procedures may be integrated into a

computerised stock record which can provide information at the press of a
button to the storekeeper, buyer, production planner, financial manager or any

273
other person authorized to key into it. For example, information on slow-moving
stock items can be obtained automatically and without delay.

MATERIAL STORAGE
Sophisticated mathematical models to control economic buying, and systems
control the flow of material may all be for nought if the obvious—efficient
storekeeping—ignored. Good practice in this respect implies:

• The employment of a well-trained stores staff

• Use of the most efficient equipment—for storage and handling
• Easy access to items—stored in logical order
• Siting of stores convenient to users
• Security against theft and fire
• Protection against deterioration
• A system of continuously checking physical with recorded stocks.

ke
INVENTORY PLANNING AND CONTROL

o.
The major goal of "inventory control" is to discover and maintain the optimum
.c
level of investment in all types of inventories, from raw materials and supplies
to finished goods that helps to maximize long-run profits.
es
ot

Two limits must be imposed in controlling inventory levels, because there are
cn

two danger points that management usually wants to avoid. They are:
ne

i. That inadequate inventories, disrupts production and may lose sales.

ii. That excessive inventories, introduces unnecessary carrying costs and
.k

obsolescence risks.
w
w
w

MOTIVES FOR HOLDING INVENTORIES

If production and delivery of goods were instantaneous, there would be no need
for inventories. However in reality, the manufacturing and purchasing processes
do not function quickly enough to avoid the need for having inventories. Hence
inventories must be maintained so that he or she does not turn to another source
of supply. In turn, production operations cannot flow smoothly without having
inventories of direct materials, work-in-progress, supplies, etc. In other words
inventories must be viewed as cushions:

i. To absorb planning errors and unforeseen fluctuations in supply and

demand. (i.e. precautionary motive).

ii. To facilitate smooth production and marketing operations.

Note: The fundamental questions to which answers are required are:

a. How much should the company order (when stocks are re-ordered)?
b. When should the company reorder?
274
 c. How much should the company produce, and when (if internal
production is involved).

Accordingly regardless of their complexity, all inventory planning models

focus on the twin problem of size and timing.

1. CHOOSING ORDER QUANTITY (SIZE—PROBLEM)

The objective of inventory decisions is usually to minimize total inventory
costs to the company. Costs are ascribed to all elements which are of
interest in reaching its inventory decisions (e.g. purchasing costs, stock
out costs etc.), and solutions, are derived based on these costs.

Several inventory planning models exist. These models can be classified

into two basic classified into two basic classifications:

i. Deterministic Models:—whereby all parameters are known with

certainty, e.g. lead-time, annual demand, etc.

ke
o.
ii. Stochastic Models:—in which parameters (particularly demand and
lead time) are not known with certainty, but follow known
.c
probability distributions (i.e. risks)
es
ot

A. THE DETERMINISTIC MODELS

cn

1. THE BASIC EOQ MODEL

ne

This is the most simple of all the models discussed. In addition to

.k

the general assumptions which relate to all deterministic models

w

(i.e. certainty of all parameters) it is further assumed that:

w
w

a. Demand is continuous and constant over time.

b. That suppliers lead time is zero i.e. stocks are delivered
immediately on the day the order is made.
c. That stock-outs are not allowed.
d. There are no bulk quantity discounts.
e. Holding costs per unit, ordering costs per order and costs per
unit are constant.

Relevant costs of basic EOQ model

The relevant costs that should be considered when
determining optimum inventory levels can be classified into
two categories:
i. Ordering costs.
ii. Holding (Carrying) Costs.
I.e. TC = Ordering Costs + Holding Costs

i. Ordering Costs
275
 These are incurred in getting purchased items into
the company‘s inventory or stores, and usually consist
of clerical costs of:

1. Making the purchase requisition.

2. Issuing of a purchase order
3. Follow-up action
4. Receiving the goods
5. Inspection for quality control
6. Placing goods in stores
7. Paying vendors (Suppliers)
Note:
The basic EOQ model assumes that these costs are fixed constant for each order
made.

ii. Carrying costs of inventory

These are costs incurred because the firm has decided to maintain

ke
inventories. They usually consist of:

o.
1. Stock-out costs
2. Insurance costs .c
es
3. Warehouse and storage costs
4. Material handling costs
ot

5. Costs of obsolescence
cn

Total Ordering Cost = Total demand for period x Ordering

ne

Costs per period

.k

Quantity Ordering
w

= DO
w

Q
w

Total Holding Costs = Quantity Ordered x Holding Costs per

unit
2

= QH
2

Therefore total relevant costs (TC) for any order quantity can be
expressed as:

TC = DO + QH
Q 2

We can determine a minimum of this total cost function by:

276
i. Differentiating the above formula with respect to Q and
setting the derivative (1st) equal to zero.

dTC = -DO + H = 0
dQ Q2 2

H = DO
2 Q2

Q2 = 2DO

H
2DO
Q=
H

ke
o.
.c
es
ot
cn
ne
.k
w
w
w

277
 ii. Equating ordering costs to holding costs.

DO = QH
Q 2

Q2 = 2DO
H

2DO
Q=
H
ILLUSTRATION
1. Assume a wholesaler has to supply his customer with 40,000 units of a
given product every year.
2. Assume that demand is fixed and known
3. Assume the cost of placing each order is Shs 2.00 while the holding cost
per unit is Shs 1.00.

ke
o.
Required:
a. .c
Determine the optimum order size using the basic EOQ model.
es
b. Determine total costs incurred at optimum order size.
ot
cn

SOLUTION
ne

a. 2 2 x 40,000 x 2
Q =
.k

1
w
w
w

2
Q = 160,000 = 16 x 104

Q = 400 units

b. TC = 40,000 (2) + 400 (1)

400 2

= 200 + 200

TC = Shs 400

1. EOQ Model with quantity discounts

Circumstances frequently occur where firms are able to obtain quantity
discounts for large purchase orders. Buying in bulk has some advantages
and disadvantages.
278
 Advantages

i. (A saving in purchase rule) Decreases in unit cost, which consists of

the total amount of discount for the period.

ii. A reduction in the total ordering costs because fewer orders are
placed to take advantages of the discounts.
Disadvantages
Increased holding cost arising from higher stock levels when large quantities are
purchased.

Such as:
i. Stock out cost
ii. Insurance.
iii. Deterioration
iv. Security etc.

ke
o.
Broadly there are two types of discount structures:
i. .c
Fixed discount on "all units" when the order placed is for a minimum
quantity.
es
ii. Variable discounts for given ranges.
ot
cn

1. The case of a fixed discount

When evaluating inventory decisions when a fixed discount rate exists,
ne

the appropriate procedure is to compare the total costs of the EOQ with
.k

the total costs when discounts are taken. The option giving lower costs is
w

then chosen.
w

Note:
w

The Unit (variable) cost (i.e. Purchase Price) behaves in the following manner.

C = Co if 0  Q  Qb
Co (1 - P) if Q  Qb

Where Co = basic unit cost without a discount

P = Discount rate allowed.
Qb = Break-point (Quantity)—where discounts
become operational.

In order to determine the optimal ordering quantity, it is necessary to include

the costs of the inventory with the carrying ordering costs.

Total costs of Inventory = Total Purchase cost + Total order

cost + Total carrying cost

279
 TC = DCo + Q* H + Do If 0  Q  Qb (i)
2 Q2

TC = DC. (1 - P) + Q H + Do If Q  Qb (ii)
2 Q

Note:
Equation (ii) i.e. with discounts will give a lower TC than equation (i) for the
same. The decision whether to go for the discount lies on a trade-off between
extra carrying costs vs. a reduction in acquisition costs.

ILLUSTRATION
1. Assume X Ltd purchases a raw material from an outside supplier at cost of
Shs 70 per unit.
2. Assume total annual demand for the product is 9,000 units.
3. Assume the holding cost is Shs 40 per unit and the ordering cost is Shs 50
per order.

ke
4. Assume a quantity discount of 3% of the purchase price is available for
orders in excess of 1,000 units.

o.
Required: .c
es
a. Calculate the EOQ and the associated costs.
ot

b. Calculate the total costs if the company purchased in batches of 1,000

units (N.B. It is not wise to buy in batches with more than 1,000 units
cn

because of the increase rate in carrying exceed the rate at which ordering
ne

costs decline).
c. Advise the management on the appropriate inventory policy.
.k
w
w

SOLUTION
w

a. i. EOQ = 2 x 9,000 x 50
40

= 22,500 = 225 x 102

= 150 units

ii. TC = 9,000 (50) + 150 (40) + 9,000 (70)

150 2

= 3,000 + 3,000 + 630,000

= Shs 636,000

280
c. Q = 1,000

TC = 9,000 (50) + 1,000 (40) + 9,000 (70)(1 - 0.03)

1,000 2

= 450 + 20,000 + 611,100

= Shs 631,500

Note: Discount price = 70 (1 - 0.03) = Shs 67.90

d. Decline in inventory costs = 631,550 - 636,000

= Shs 4,450

DECISION
The firm saves Shs 4,450 by taking the quantity discount.

ke
o.
.c
es
ot
cn
ne
.k
w
w
w

281
THE CASE OF VARIABLE QUANTITY DISCOUNTS
In practice, suppliers may offer different discounts for different quantities
purchased. For example:

SEGMENT QUANTITY PURCHASED UNIT PRICE

1 0 — 500 Shs 100
2 501 — 1,000 Shs 90
3 1001 — 1,500 Shs 80
4 over — 1,500 Shs 70

The best approach to the solution in this case is to apply the price-breaks
theorem. This works as follows:

1. For each segment an EOQ is calculated. There are two possible requests:

i. The EOQ is within the quantity segment (i.e. valid)

In this case, the EOQ is used as the minimum cost quantity for that

ke
segment.

o.
ii. The EOQ is outside the quantity segment (i.e. invalid)
.c
In this case the minimum cost quantity will be the quantity within
es
the segment closest to the EOQ as calculated.
ot

2. Select the quantity that leads to the lowest total inventory costs (i.e.
cn

Purchase, Ordering & Carrying).

ne

Illustration:
.k

1. Assume a manufacturer uses 3,300 drums of a certain chemical per year.

w

2. Assume delivery costs incurred per order are Shs 40 and inventory carrying
w

costs are estimated to be 30% of stock value.

w

3. Assume the normal cost per drum is Shs 22 but the supplier offers discount
of 1.5% on orders for 500 drums or more, and 3% on orders for 1,000
drums or more.

Required:
Determine the order quantity the manufacturer should adopt to minimize total
costs.

Solution:
Note:
There are 3 discount levels (0, 1.5% & 3%) and hence 3 segments.

STEPS
1. Calculating the EOQ for each segment.

282
a. For Segment 1 (0 - 499 drums)
C = Shs 22

EOQ = 200 drums

7
2 x 3,300 x 40
EOQ =
0.3 (22)

264,000
= = 40,000
6.6

= 200

ke
b. For Segment 2 (500 - 999 drums)

o.
C
C
=
=
22(1 - 0.015)
22(0.985)
.c
es
C = Shs 21.67
ot
cn
ne

2 x 3,000 x 40
EOQ =
0.3 (21.67)
.k
w
w

264,000
= = 40,615
w

6.50

= 201.5 drums (can round to 202 drums)

c. For Segment 3 (1,000 or more)

C = 22(1 - 0.03)
C = Shs 21.34

264,000
EOQ =
0.3 (21.34)

264,000
= = 41,237
6.40 283
 = 203 drums

Price Unit Valid

Segment Break Quantity Price EOQ Quantity
1 0 – 499 Shs 22 200 200 (Valid)
2 500 - 999 Shs 21.67 202 500 (Invalid)
3 over 1000 Shs 21.34 203 1,000 (Invalid)

2. Calculate total costs

SEGMENT UNITS ORDERED TOTAL COSTS

1 200 (3,300 x 22) + (2002 x 6.6) + (3,300200 x 40) =

ke
Shs 73.920
2 500 (3,300 x 21.67) + (5002 x 6.5) + (3,300500 x 40)

o.
= Shs 73.400
3 .c
1,000 (3,300 x 21.34) + (1,0002 x 6.4) + (3,3001000 x 40
es
= Shs 73.754
ot

Decision:
cn

The firm should order 500 drums per order per annum and incur Shs 73.400 total
ne

costs.
.k
w

NON-ZERO LEAD TIME (DETERMINING REORDER POINT)

w
w

This basic EOQ model assumes that the suppliers lead time is zero (i.e. goods are
delivered immediately on the day the order was made). In reality, however,
supplies are rarely ordered and received on the same day. Accordingly, orders
must be placed some time before stocks reach zero. In world of certainty (when
demand is continuous and constant) the reorder point will be the number of
days/weeks lead time multiplied by the daily/weekly usage during the period.

i.e. Reorder point = Average daily usage x Lead time in days.

284
Note:

ke
The reorder point has no cost implications, since it does not affect the EOQ.

o.
Illustration:
1.
.c
Assume X Ltd uses 50,000 kg of a raw material annually.
es
2. Assume ordering costs are Shs 160 per order and stock holding costs are
ot

Shs 0.25 per kg per annum.

cn

3. Assume the purchase price is Shs 20 per kg and no quantity discounts are
offered.
ne

4. Assume lead time for delivery of orders is 4 weeks.

5. Assume working time is 50 weeks a year.
.k
w

Required:
w

a. Calculate the EOQ

w

b. Calculate the inventory reorder level.

c. Calculate the total costs per annum.

SOLUTION

a. 2 x 50,000 x 160
EOQ =
0.25

16,000,000
= = 64,000,000
0.25

EOQ = 8000 kgs

285
 Make 6.25 orders (50,0008,000) per annum.

b. Reorder level = Demand per week x Lead time.

 50,000 
= x4
 50 

= 1,000 x 4
= 4,000 units.

c. Total costs = (50,000 x 20) + (8,0002 x 0.25) + 50,0008,000 (160)

= 1,000,000 + 1,000 + 1,000
= Shs 1,002,000

ke
INVENTORY PLANNING & CONTROL UNDER UNCERTAINTY

o.
The basic EOQ model assumes that all the parameters (elements) in the model
.c
are certain (i.e. can be predicted accurately in advance). These parameters
es
are:
ot

i. Demand or usage of stocks

cn

ii. Lead times.

iii. Holding costs per unit, ordering costs per order and costs per unit.
ne

In reality however, stock demand, supplies lead times and cost date are not
.k

known with certainty. Accordingly to make the models applicable to real

w

situations we must consider uncertainty when planning for inventory levels.

w
w

To protect itself from conditions of uncertainty, a firm will maintain a level of

safety stocks for raw materials, work-in-progress and finished goods stocks. Thus
safety stocks are the amount of stocks that are carried in excess of the expected
use during the lead time to provide a cushion against running out of stocks. Thus
the reorder point is computed as safety stock plus the average usage during the
lead time
i.e. reorder point = Average usage during lead time + safety (buffer)
stock.

DETERMINING THE SAFETY STOCKS LEVEL

1. Uncertainty of demand

Demand is the most troublesome variable to predict accurately. Actually,

demand may fluctuate from day to day, from week to week or from
286
 month to month. Thus, the firm takes the risk of running out of stocks if
there are sudden increases in demand. Hence safety stock is the extra
inventory held as a buffer of protection against the possibility of stock
due to higher demand.

However, a larger inventory of safety stock will involve a higher inventory

carrying costs, and on the other hand, the higher safety stock will
decrease stock-out costs. Therefore one has to make a balance between
these two costs in order to find out an optimal safety costs.

Note:
The optimum safety-stock level exists where the costs of carrying an extra unit
are exactly counter balanced by the expected stock-out costs. This would be
the level that minimizes the annual total stock-out and carrying costs.

Stock-out costs

These are the opportunity costs of running out of stock. They include:

ke
i. The costs of lost customer sales, and therefore lost contribution to fixed

o.
costs.
ii. Potential loss of goodwill with customers whose demand cannot be net.
iii. .c
Acquiring emergency supplies at higher prices to meet demand.
es
iv. Cost production of finished goods, where raw material stock-outs occur.
ot

The computation of safety stocks lingers on demand forecasts. The manager will
cn

have some notion (usually based on past experience) of the range of daily
ne

demand. That is the probability that exists for usage of various quantities.
.k

Hence total inventory costs will be as follows:

w
w

Total inventory costs = Purchase price cost + carrying costs +

w

stock-out cost + order costs.

= Purchase price costs + "normal~ carrying costs

(Q2 H) + Buffer Stock holding costs (B x H) + Stock-out costs + order costs.

Total inventory costs = DC + Q2H + (B x H) + stock-out costs + DQ 

Co

Where: D = Total annual demand

H = Holding costs per unit
B = Buffer stock
Q = EOQ

Note:

1. The normal EOQ formula is used to compute order quantity Q. Hence

purchase costs, carrying costs, and ordering costs remain unchanged.
287
 Only buffer stock holding costs and stock-out costs change. Accordingly,
the minimization of these two costs will also mean that total/overall costs
will be minimized.

2. Stock-out costs = Number of units short x Probability of being

short.

= {(Number of units short x Stock-out costs per unit

Probability of being short)—Stock-out costs for every usage duration} x
Number of orders per year.

3. Buffer Stock-holding costs = Bx H

ke
o.
.c
es
ot
cn
ne
.k
w
w
w

288
ILLUSTRATION
A company has an annual demand for material X of 250 tonnes per annum.
Order lead time is 4 days and usage during lead time as shown by past record is

USAGE PROBABILITY
0 0
1 0.01
2 0.05
3 0.15
4 0.25
5 0.30
6 0.10
7 0.09
8 0.05

The cost per tonne is £20 and stock holding cost is 25% per annum of the stock
value. Delivery cost per batch is £4. The costs of stock out also estimated to be

ke
£4.

o.
You are required to:
.c
es
a. Calculate the economic batch quantity and the expected number of
orders per annum
ot

b. Ascertain the re-order level taking the information given above into
cn

consideration.
ne

Solution:
.k
w

a. EBQ or EOQ where stock-outs are permitted

w

2DK h + Cs
w

EBQ or EOQ =
h CS

Where D is annual demand

K is order cost

h is holding cost
and Cs is stock out cost
2 x 250 x 4 5+4
EBQ =
5 4

9
= 400
4

289
 = 30

No. of orders = 250

30

(Note: Stock holding cost b = 25% of £20 = £5)

We are required to find out expected demand during lead time.

Usage Prob x p(x)

x p(x)
0 0 0.00
1 0.01 0.01
2 0.05 0.10
3 0.15 0.45
4 0.25 1.00

ke
5 0.30 1.50
6 0.10 0.60

o.
7 0.09 0.63
8 0.05 .c
0.40
es
xp(x) 4.69
ot
cn

Hence normal usage during lead time is 4.69 tonnes but maximum usage
ne

during lead time can be as large as 8 tonnes with a probability of 0.05.

.k

Average usage in four days = 4.69 tonnes and at this level there is no
w

buffer stock. The problem is how much buffer stock we should have, so that the
w

cost of holding stock together with the cost of expected stock-out cost is
w

minimum.

Let 1 = 4.69 tonnes.

B = Buffer stock
S = Re-order level (for lead time) = 1 + B

Hence B = S - 1
Here B = S - 4.69

290
1 2 3 4 5 6 7
Re- Buffer Expecte Expecte Expecte Holding Total
order Stock d Stock- d Annual d stock- Cost (£) cost
Levels B out per Shortage out cost (£)
S order (£)
4.69 0 0.5974 4.978 19.912 0 19.91
2
5 0.31 0.43 3.58 14.32 1.55 15.87
6 1.31 0.19 1.583 6.332 6.55 12.88 
2 minimum
cost
7 2.31 0.05 0.4166 1.664 11.55 13.21
4
8 3.31 0 0 0 16.55 16.55
Working Column 4 Column
column 3 x4 2x5
x i.e.

ke
250

o.
30 Bxh
.c
es
Hence for minimum cost S=6
ot

Buffer stock S = 6 - 4.69

= 1.31 tones
cn

and minimum cost = £12.88

ne
.k

WORKINGS AND EXPLANATIONS

w

1. When Re-order Level S = 4.69

w

B = 0
w

Hence possible shortages will occur when demand is 5, 6, 7 or 8 units.

Demand Shortage Prob p(x) Expected Value xp(x)

x
5 5 - 4.69 0.30 0.31 x 0.30 = 0.93
6 6 - 4.69 0.10 1.31 x 0.10 = 0.131
7 7 - 4.69 0.09 2.31 x 0.09 = 0.2079
8 8 - 4.69 0.05 3.31 x 0.05 = 0.1655
xp(x) 0.5974

This is expected stock out per order when demand is 5, 6, 7 or 8

2. Similarly if Re-order level is 5

Shortages will occur if demand is 6, 7, 8

Working in the way as above

291
 Demand Shortage Prob p(x) Expected Stock-
x out cost
6 1 0.10 0.10
7 2 0.09 0.18
8 3 0.05 0.15
xp(x) 0.43

3. If re-order level is 6, shortage will occur if demand is 7 or 8

Demand Shortage Prob p(x) Expected Stock-

x out cost
7 1 0.09 0.09
8 2 0.05 0.10
xp(x) 0.19

ke
4. If re-order level is 7, shortage if demand is 8

o.
Expected shortage cost = (8 - 7) x 0.05 = 0.05
.c
es
5 If re-order level is 8, there is no probability of shortage
ot

a. Expected Annual Shortage = (Expected Stock-out per order) x

cn

No. of orders
ne

i.e (Expected stock-out per

.k

order) x 250
w

3
w

3
w

b. Expected stock-out cost = (Expected annual shortage) x

Shortage cost per item

c. Holding cost = Buffer stock x cost of holding per item

Important Note:
If shortage Cs is not taken into account
2 x 250 x 4
EOQ =
5

= 400 = 20

292
and number of orders = 250 = 12.5
20

Using this model, we can still develop a table similar to the table worked
out in this problem.

Graph of total cost against re-order level can be drawn using the table.
The minimum cost and the re-order level can then read from the graph. Once
re-order level is known, safety stock can be worked out.

In the next example, the initial EOQ value has been taken using the
formula

2DK
EOQ =
h

ke
ILLUSTRATION

o.
1. Assume a manufacturer has experienced trouble from stock shortages
.c
(stock-outs) of raw materials X which is required in a manufacturing
process. Usage of X averages 6,000 units per year where a year consists
es
of 50 weeks.
ot
cn

2. Assume the costs of ordering each batch of X is Shs 30 and the lead time is
2 weeks (known for certain). The annual holding costs amount to Shs 1
ne

per unit of X held. The cost of a stock-out has been estimated to be Shs 5
per unit short.
.k
w

3. Assume the demand (usage) is unknown. However, the total usage of raw
w

material X over the 2 week lead time is expected to be as follows:

w

Usage in units Probability

60 0.07
120 0.08
180 0.20
240 0.30
300 0.20
360 0.08
420 0.07

Required:
a. Calculate the EOQ, and the expected number of orders per annum
b. Calculate the average usage per 2 weeks (Lead-time)

SOLUTION
Annual demand = 6,000 units
293
Order cost = Shs 30
Holding cost = Shs 1
Shortage cost = Shs 5 per unit short

2 x 6,000 x 30
EOQ = = 600 units
1

Number orders = 6,000 = 10

600

Average usage for two weeks = 60 x 0.07 + 120 x 0.08 + 180 x 0.20 + 240
x 0.30 + 300 x 0.20 + 360 x 0.08 + 420 x 0.07

= 240 units

ke
Re-order Buffer Expected Expected Expected Holding Total cost
Levels Stock Stock- Annual stock-out Cost at Shs

o.
out per Shortage cost 1
order .c
es
240 0 34.2 342 1,710 0 1,710
300 60 13.2 132 660 60 720
ot

360 120 4.2 42 210 120 330

cn

420 180 0 0 0 180 180

ne

i. Expected Stock out per order where re-order level is 240

.k
w

(300 - 240) x 0.2 + (360 - 240) x 0.08 + (420 - 240) x 0.07 = 34.2
w
w

ii. similarly for re-order level = 300

(360 - 300) x 0.08 + (420 - 300) x 0.07 = 13.2

iii. and for re-order level = 360

(420 = 360) x 0.07 = 4.2

iv. No shortage when re-order level is 420

Hence total cost is minimum when re-order level is 420 units.

Best policy is to have safety stock of 180 units.

SENSITIVITY ANALYSIS OF EOQ MODEL

Sensitivity Analysis is concerned with the way in which those results of solutions
change in response to change in model parameters.
294
 2 DO
EOQ =
H

Note:
It is important to appreciate that in formulating our inventory models, we have
really been performing a planning exercise. Thus we have made certain
assumptions and estimates (e.g. annual demand D, holding costs H, & ordering
costs O) and out solutions have obviously been affected by these. When for
example calculating a deterministic EOQ value with maximization of total
inventory costs as the objective, expected annual demand (D) is taken into
account. If we subsequently find that annual demand has differed from that
expected then we will find that the EOQ we selected was not the optimum and,
as a result, the total inventory cost was not actually minimum.

Illustration:

1. Assume X Ltd expected annual demand for 1991 for 62,500 units of raw

ke
material X per annum.

o.
2. Assume holding costs are Shs 15 per unit per annum and each order costs
Shs 10. .c
es
3. Assume that the end of the year actual demand has been found to have
ot

been for 90,000 units, not the 62,500 expected.

cn

Required:
ne

Calculate the additional costs borne by X Ltd through basing the size and
.k

frequency of orders on expected figures.

w
w

Solution:
w

a. Calculating EOQ based on expected figures, and the associated inventory

costs.
2 x 62,500 x 1O
EOQ =
5

1,250,000
EOQ = = 250,000
5

= 500 units

Total inventory costs = 62,500 (10) + 500 (5)

500 2

295
 = 1,250 + 1,250

= Shs 2,500

b. Calculating the total inventory costs based on actual demand.

TC = 90,000 (10) + 500 (5)

500 2

= 1,800 + 1,250

= Shs 3,050

c. Calculating the total inventory costs had the company forecasted the
actual demand accurately.

2 x 90,000 x 1O
EOQ =
5

ke
o.
EOQ =
1,800,000
= 360,000 .c
es
5
ot

= 600 units
cn
ne

TC = 90,000 (10) + 600 (5)

500 2
.k
w

= 1,500 + 1,500
w
w

= Shs 3,000

Therefore, X Ltd total inventory costs are actually Shs 50 (3,050 - 3,000) higher
than they would have been if the EOQ had been set in accordance with perfect
information.

Observation:
Hence the change in demand by 44% (99,000 - 62,500) has result in a 20%
(600 - 500)
62,000 500

Increase in EOQ but only 91.64% (3,050 - 3,000) increase in total costs.
3,000

Hence we can conclude that total cost is relatively insensitive to changes in

demand.
296
SELECTIVE INVENTORY MANAGEMENT
The inventory of an industrial firm generally comprises thousands of items with
diverse prices, usage and lead time, as well as procurement and/or technical
problems. It is neither desirable nor possible to exercise the same degree of
control over all those items. The organisation should pay more attention and
care to those items whose usage value is high and less attention to those whose
usage and consumption value is low. The organisation has, therefore, to be
selective in its approach to control its investment in various types of stocks and
inventories. Such a system is known as `selective inventory control' system.

ABC ANALYSIS (PARETO ANALYSIS)

In ordinary parlance, ABC analysis can be best compared with our class society
where the population is categorised into Top, Middle and Lower classes. In the
case of inventories also, it has been noticed that out of a large number of items
(in a million-tonne capacity steel plant there would be usually about 50,000

ke
items of inventory of various types) that are generally held in stock, some of the

o.
items are quite significant whereas the others are not that important. Through
.c
ABC plan which is in fact an analytical approach based on common statistical
techniques, the relative importance of the various items is established for the
es
purpose of individual scrutiny and subsequent control. Through this technique
ot

`VIP' or the `privileged few' and the `trivial many' are distinguished and treated
cn

as such.
ne

ABC analysis contemplates to classify all the inventory items in a number of

categories, generally in three categories based on their values. Items of high
.k

value but small in number are classified as `A' items which would be under a
w

strict control. `C' items represent relatively small value items and would be
w

under simple control. Items of moderate value and size are classified as `B'
w

items and would attract reasonable attention of the management. Since this
plan concentrates attention on the basis of the relative importance of the
various items of inventory, it is also known as `control by importance and
exception'. As items are classified in order of their relative importance in terms
of value, it is also known as the `proportional value analysis'.

It has been found that normal inventory items in most organisations show the
following distribution pattern:

A— 5 to 10% of total number of items account for about 70% of the

total consumption value. These items may be called "A" items.

B— 10 to 20% account for 20% of total consumption value.

C— the remaining large number of items account for the balance of

15% of the consumption value.
297
Remark When a detailed scrutiny was conducted in respect of inventories
held by the Ford Motor Company, the following results were obtained.

1. 9% of the total items (in number) were accounting for 57% of the
total value of the inventory. These were classed as `A' items.

2. 10% of the items (in number) were found to be accounting for 18%
of the total value. These were categorized as `B' items.

3. 81% of the items (in number) were found to be accounting for only
25% of the total value. These were classified as `C' items.

Similarly, when ABC ANALYSIS was done in the case of G.E.C., the results
obtained were as under:

`A' Category—Items accounting for 8% of the total number but 75% of total
value.

ke
`B' Category—Items accounting for 25% of the total number and 20% of

o.
total value.
.c
`C' Category—Items accounting for 67% of the total number and only 5% of
es
the total value.
ot

It would thus be observed that substantial and effective controls are made
cn

possible if greater attention is focused on `A' category items since these would
ne

be covering quite a substantial part of the inventory in terms of shilling value.

.k

Some Remarks:
w

1. There is no hard and fast rule that all the inventory items should be
w

classified only in these three categories. There can be a large number

w

of classifications based on the requirements of the company and the

nature of the items. For example `A' items may be further sub-
classified as A1, A2, A3, etc. The same principle may be extended to `B'
items also or alternatively all the inventory items may be classified into
A, B, C, D, E, F, etc.

2. All items that the company consumes must be considered together while
classifying into ABC classes. Separate classification of raw materials,
spares and consumable is not really meaningful.

While classifying as ABC items, what counts is the consumption shillings and not
the unit price of an item or its consumptions in terms of units. Thus of the three
items given below, the last one is most important since its annual consumption
in terms of value is more than the other two.

Item Price/U Annual Annual

nit Consumption Consumption
298
 Shs in units in Shs
X 20,000 2 40,000
Y 0.02 100,000,000 200,000
Z 1,000 500 500,000

4. Even though, so far we have referred to annual consumption, it is not at

all necessary that the consumption figures should be taken only for one
year. It can be for 6 months or even 3 months. But the period should be
so selected that the consumption figures would be representative.
However, annual figures are far more convenient and are universally
followed.

5. If a firm follows ABC analysis, it will devote much time and effort on the
control of `A' items. For example, extra care will be taken in the
determination of minimum, maximum, reorder level, etc. of the `A'
items, whereas so much control may not be exercised on `C' items. `A'
items may be purchased only once in a year. For `A' items perpetual
inventory system may be applied whereas in the case of `C' items, only a

ke
bin card may be maintained. In the same way an appropriate accounting
method for `B' items may be devised. However, in the classification of

o.
items into ABC categories if there are some critical items which are of
.c
small value whose non-availability may hamper the production, may in
es
the normal situation, be classified as `C' items but, due to the critical
nature of these items extra care may be taken so that these may not go
ot

out of stock.
cn

6. The objective of classifying inventory items into `A', `B' and `C' categories
ne

is to develop policy guidelines for selective control. Such a policy can be

designed in a variety of ways. In general `A' items merit a tightly
.k

controlled inventory system with periodic attention, and `C' items to

w

subject to loose control with casual attention.

w
w

7. TABLE SHOWING FEATURES OF ABC ANALYSIS

Nature A items B items C items

(High value) (Moderate value) (Low value)
1. Extent of Rigid control Moderate control Loose control
control (close day to day (Regular review) (infrequent
control) review)
2. Safety stock Low safety stocks Medium safety Large safety
coverage stock stocks
3. Frequency of Frequently Less frequently Bulk ordering
order
4. Degree of Individual posting Small group Group postings
posting postings
5. Period of Every fortnight Quarterly Yearly

299
 Nature A items B items C items
(High value) (Moderate value) (Low value)
review
6. Sources of Good number of Few reliable One or two
supplies sources sources sources
7. Follow up Vigorous Periodic Occasional
8. Control Weekly control Monthly control Quarterly control
statements statements reports reports
9. Forecasting Emphasis on Focus on past Rough estimate
accurate forecast trend
10. Level of Senior Middle Stores supervisor
management management management
11. Lead time Maximum efforts Moderate efforts Minimum clerical
to reduce lead efforts
time
12. Value % and 80% of the value 15% of the value 5% of the value in
item percentage in 20% of the in 30% of the 50% of the items
(Approximation) items items

ke
o.
ADVANTAGES OF "ABC ANALYSIS"
.c
The benefits derived from this analysis and its subsequent follow up are
es
summarized below:
ot

• Facilities selective control and thereby save valuable time of busy

cn

executives.
ne

• Eliminates lot of unnecessary paper work involved in various other control

.k

procedures. Tangible savings can be affected in this behalf by following

w

Two-Bin System which is very closely related with this technique.

w
w

• Facilitates Inventory Control and control over usage of stores materials

which ultimately results in cost control.

LIMITATIONS OF ABC ANALYSIS

Although ABC analysis is a fundamental tool for exercising selective control over
numerous inventory items, it does not, in its present form, permit precise
consideration of all relevant problems of inventory management. For instance, a
never-ending problem in inventory management is that of adequately handling
thousands of low-value `C' items. Low-value purchases frequently require more
items and thereby reduce the time allowance available to purchasing personnel
for value analysis, vendor investigation, and other `B' items.

Besides, if ABC analysis is not periodically reviewed and updated, the very
approach of control may be defeated. For example, `C' items like diesel oil in a
firm, will become most high-value items during power crisis should, therefore,
300
deserve more attention, but this point may be overlooked if classification of
items is not reviewed and updated.

The following steps are involved in implementing the ABC analysis:

1. Classify the items of inventories, determining the expected use in units

and price per unit for each item.
2. Determine the total value of each item by multiplying the expected units
by its unit price.
3. Rank the item in accordance with the total value, giving first rank to the
item with highest total value and so on.
4. Compute the ratios (percentage) of number of units of each item to total
units of all items and the ratio of total value of each item total value of
all items.

Combine items on the basis of their relative value to form three categories—A, B
and C.

ke
o.
.c
The data in the table below illustrates the ABC analysis.
es
TABLE: ABC ANALYSIS
ot

Item Units % of Cumulati Unit Total % of Cumulativ

cn

total ve price cost total e

ne

1 10,00 10 30.4 304,000 38.0

0 0
.k

1 15 7 70
w

5 0
w

2 5,000 5 51.20 256,000 32.0

w

0
3 16,00 16 5.50 88,000 11.0
0 0
3 45 2 90
0 0
4 14,00 14 5.14 72,000 9.00
0
5 30,00 30 1.70 51,000 6.38
0
5 100 1 100
5 0
6 15,00 15 1.50 22,500 2.81
0
7 10,00 10 0.65 6,500 0.81
0
TOTAL 100,0 800,000
00
301
The tabular and graphic representation indicates that "Item A" forms a minimum
proportion, 15 per cent of total units of all items, but represents the higher
value, 70 per cent. On the other hand, "Item C" represents 55 per cent of the
total units and only 10 per cent of the total value. "Item B" occupies the middle
place. Items A and B jointly represent 45 per cent of the total units and 90 per
cent of the investment. More than half of the total units are item C,
representing merely 10 per cent of the investment. Thus, a tighter control
should be exercised on "Item A" in order to maximize profitability on its
investment. In case of "Item C" simple controls will be sufficient.

ke
o.
.c
es
ot
cn
ne

JUST-IN TIME (JIT) INVENTORY MANAGEMENT

.k
w

JIT is a system whose objective is to produce or to purchase products or

w

components as they are required by customers or for use rather than for stock.
A JIT system is a pull system which responds to demand as opposed to a push
w

system in which stocks acts as buyers* between the different element of the
system such as purchasing, production and sales

JIT PRODUCTION
Is a production system, which is driven by demand for the finished products
whereby each component on the production line is produced only when needed
for the next stage.

JIT PURCHASING
On the other hand is a purchasing system in which material purchased are
contracted so as that the receipt and usage of materials to the maximum
extent possible, coincide.

302
JIT concept can be traced back to the Japanese company whose success in the
international market generates interest among many western companies as t6o
how this success was achieved.
The implementation of JIT production methods was considered to be pursuit of
excellence in all phase of manufacturing systems design and operations.
The JIT are to produce the required items at the required quality and in the
required quantities, at the precise time that they are required.

JIT seeks to achieve the following goals-

(1). Elimination of non-value adding activities.
(2). Zero inventory
(3). Zero defects
(4). Batch size of one.
(5). Zero break-downs
(6). 100% on time delivery services.

The above goals represent perfection and are most unlikely to be archived in
practice. They do however offer targets and create a climate for continuous

ke
improvement and excellence.

o.
Major features of JIT.
.c
es
(1). Elimination of non-value added activity
ot

JIT manufacturing can be described as a philosophy of management, dedicate

cn

to the elimination of waste. Waste is defined as anything that does not add
value to a product.
ne

The cycle time involved in manufacturing and selling a product consist of –

.k

 Process time-add values

w

 Inspection time
w

 More time
w

 Queue time
 Storage time

Of these 5 steps only process time actually adds value to the products. All the
other activities add cost and No value to the production and therefore are
deemed as non-value within the JIT philosophy.
Usually in many companies, process time is less than 10% of total
manufacturing, lead and cycle time. Therefore 90% of the manufacturing lead
time disassociated with the product, adds cost but no-value to the product by
adapting a JIT philosophy and forecasting, on reducing lead time, it is claimed
that total cost can be significantly reduced.
The ultimate products with lead-time = processing time, and eliminating all
non-value adding activities.

303
(2). Factory Layout
The first stage of implementing the JIT manufacturing techniques is to
rearrange the factory floor away from the batch production functional layout
towards a production layout using low lines with a functional plant layout
production through a No. of special departments that normally contain a group
of similar machines.
Products are processed in stage batches so as to minimize the set times when
machine settings are changed between processing batches of different
products. Batches move via different and complex routes through the various
departments, traveling over much of the factory floor before they are
complete.
Each process normally involves a considerable amount of waiting time and
which much time is taken transporting items from one process to another.

A further problem is that it is not easy at any point in time to determine what
progress has been made on individual batches and therefore detailed cost
accumulation records are necessary to track work in progress. This results in
long manufacturing cycles and high work in progress levels.

ke
The JIT solution is to reorganize the production process by dividing the many

o.
different products that an organization makes into families of similar products
.c
or component. All the products is a in a particular group will have similar
es
production requirement and routing. Production is necessary used* so that each
production family is manufactured in a well-defined production cell based on
ot

flow line principles. In a production how lines specialist department containing

cn

similar markets no longer exist. Instead groups of dissimilar markets are

organised into products or component family flow lines that function like an
ne

assembly line.
.k

For each production lines the market are placed close together in the order in
w

which they are required by the group of products to be processed. Items in

w

each product family can now move one at time from process to process more
w

easily, thereby reducing wip, and lead time.

The aim is to produce products or component from start to finish without
returning to the stock room or stores

(3). Batch size of one.

Set up time is the amount of time required to adjust equipment and to retool
for a different product. Long set ups a change over time make the production
of batches with a small no. of units uneconomical.
However, the creation of large batches lead to substantial lead time delays and
the creation of high inventory levels. The JIT philosophy is to reduce and
eventually eliminate set-up times.E.g. by investing in advanced manufacturing
technology some machines setting can be adjusted automatically instead of
manually.
Alternatively some set up times can be eliminated entirely by reducing
products, so that markets do not have to be reset each time a different
product has to be made.

304
If the set up times are approaching zero, then there‘s no advantage production
in batches and therefore the optimal batch size can be one. With a batch size
of one, the work can flow smoothly to the next stage without the need for and
to schedule the next machine to accept this item.

JIT PURCHASING ARRANGEMENTS

JIT philosophy also extends to adapting JIT purchasing techniques whereby
delivery of material immediately precedes their use. By arranging with
suppliers for more frequent deliveries stocks can be out to a minimum
Considerable savings in material handling expenses can be obtained by
requiring suppliers to inspect materials before their delivery and guarantee
their quality.
This improved service is obtained by giving more business to fewer suppliers
and placing long tern purchasing orders, therefore the suppliers has an
assurance of long term sales and can plan to meet this demand.
Companies that have implemented JIT purchasing techniques have claimed to
substantially reduce their investment in raw materials and work in progress
stocks.

ke
o.
Other advantages include

1. Substantial savings in factory space.

.c
es
2. Large quantity discount.
ot

3. Savings in time from negotiating with fewer suppliers

cn

4. Reduction in paper work arising from issuing (long term orders) to a few
suppliers rather than individual purchase order to many suppliers.
ne

JIT and Management Accounting.

.k

Management accountants in many organisations have been criticised because of

w

their failure to change their managing accounting system to reflect the mode
w

from a traditional manufacturing to a JIT manufacturing system.

w

Conventional management accounting systems can encourage behaviour that is

inconsistent with JIT phylosopy, management, accounting must support JIT
manufacturing by monitoring, identifying, and communicating to decision
makers any delays errors and waste in the system.

Modern management accounting systems are now placing greater emphasis on

providing information on suppliers relialibility set up times cycle times,
percentage of deliveries that are on time and defect rates
All these measures are critical in supporting JIT manufacturing philosophy.

FINANCIAL MANAGER'S ROLE IN INVENTORY MANAGEMENT

The techniques of inventory management, discussed above are very useful in
determining the optimum level of inventory and finding answers to the problems
of the economic order quantity, the re-order point and the safety stock. The
techniques are very essential to economize the use of resources by minimizing
305
the total inventory cost. Although out treatment of inventory management has
been simple, it indicates the broad framework of managing inventories. Many
sophisticated techniques have been evolved to handle inventory management
problems more efficiently and effectively and the improvements are still
continuing. For the majority of the companies, inventory represents a
substantial investment. Thus, the goal of the wealth maximization is related to
the efficiency with which inventory is managed. Consequently, the financial
manager has an important role to play in the management of inventory, although
it is not his operating responsibility to control inventory. The financial manager
should see that only an optimum amount is invested in inventory. He should be
familiar with the inventory control techniques and ensure that inventory is
managed well. He should introduce the policies which reduce the lead time,
regulate usage and thus, minimize safety stock. The net effect would be to
reduce inventory investment and increase the firm's prospects of making more
profits.

ke
o.
PRACTICE QUESTIONS
QUESTION .c
es
Boots Ltd. manufactures a range of five similar products, A, B, C, D and E. The
table below shows the quantity of each of the required inputs necessary to
ot

produce one unit of each product, together with the weekly inputs available
cn

and selling prices of each product.

ne

Inputs A B C D E Weekly inputs

.k

available
w

Raw materials 6.0 6.5 6.1 6.1 6.4 35,000 Kgs

(Kg)
w

Forming (hours) 1.00 0.75 1.25 1.00 1.00 6,000 hours

w

Firing (hours) 3.00 4.50 6.00 6.00 4.50 30,000 hours

Packing (hours) 0.50 0.50 0.50 0.75 1.00 4,000 hours
Selling price 40 42 44 48 52
(Sh.)

The costs of each input are as follows:

Material Sh.2.10 per Kg
Forming Sh.3.00 per hour
Firing Sh.1.30 per hour
Packing Sh.8.00 per hour

Required:
a) Formulate this problem as a Linear Programming problem.
(7 marks)

306
 b) The problem has been solved using a computer package and the
following final tableau of a simplex solution has been produced:

Basis A B C D E X S T U Value
1 1.18 1.04 0.46 0 0.36 0 0 -2.29 3,357
A
B 0 -0.34 0.23 0.02 0 -0.18 1 0 0.14 321
T 0 1.37 2.97 2.28 0 -0.27 0 1 -2.79 9,482
E 0 -0.09 -0.02 0.52 0 -0.18 0 0 2.14 2,321
Zj 0 1.26 1.06 0.51 0 2.02 0 0 105,791
8.81

Where A, B, C, D and E are the weekly production levels for the five products;
X is the amount of raw material that falls short of the maximum available; S, T
an U are the respective number of hours short of maximum weekly input of
forming, firing and packing time.

ke
i Use this tableau to find the optimum weekly production plan.
(4 marks)

o.
ii Describe the implications of using this plan in terms of unused
.c
resources and overall contribution to profit. (3 marks)
es
iii In the context of this problem explain the meaning of ―The dual or
shadow price of a resource‖ (3 marks)
ot

iv There is a proposition that the company manufactures an additional

cn

product which would sell at Sh.50 per unit. Each unit will need 6 kg of raw
material, one hour of forming time, five hours of firing time and one hour of
ne

packing time. Is it s worthwhile proposition?

(3 marks)
.k

(Total: 20 marks)
w
w
w

307
 CHAPTER SIXTEEN

ITRODUCTION TO LINEAR PROGRAMMING

Specific Objectives
At the end of this to the trainee should be able to:
 Explain the model concepts;
 Describe the various types of models;
 State the requirements of LP models;
 Model decision problems as linear models;
 Solve LP models.

INTRODUCTION
These are mathematical technique that deals with optimism of linear function
variables known as objective functions. It is a method of determine an
optimum program of independent activities in view of available resources.
Linear programming is a technique of decision making used by managers to
allocate limited resources e.g. machinery, raw materials and labor in order to
minimize costs or maximize production. Decision variable are the amounts of

ke
each product to be made in a given time period. Linear programming assumes

o.
that the variable has a linear relationship.
.c
N.B; it is a technique concerned with scarce resources.
es
ot

Advantages
1. It helps in retaining the optimum use of production factor e.g.
cn

efficient use of man power.

ne

2. it improves quality of decision

3. it gives possible practice positively
.k

4. It improves the knowledge and skills of executive.

w
w

Limitation
w

1. it assumes that the activity is infinitely divisible

2. it treats all relationships as linear
3. it takes into account single object only.
4. It can be adopted only under the condition of certainty i.e. when the
quantity of resources is known.

Assumption of linear programming

1. linearity
2. Divisibility- assumes that quantities, costs and revenues are infinitely
divisible.
3. certainty- the technique assumes that there is certainty in all
4. positive solutions- non negative constraints are introduced to ensure the
values are considered
5. Interdependent between demands for products is ignored.
6. Time factor is ignored. It assumes that production is instantaneous.

308
Application of linear programming
Production department to decide the quantity of pots to be produced subject
to limited resources (constraints) e.g. labor, power, machine hours, raw
materials etc.
Marketing department: Allocation of salesmen to different sales regions subject
to their expected performance.
Human Resource: Scheduling personnel‘s work hours and job description to
either maximize production or minimize cost.

Steps in solving linear programming problems (problem formulation)

1. Identify variables (e.g. product x and product y)
2. Identify the objective (To maximize contribution or to minimize cost), and
write down its mathematical presentation in terms of variables.
3. Identify the constraints (i.e. the limited resources shared among the
variables), and write down its mathematical representation in terms of
variables.
4. Write down the objectives and the constraints in terms of the variables.
These steps apply regardless of the number of the variables.

ke
NOTE: If only two variables are involved, a graphical solution can be used

o.
otherwise for multivariable problems, an algebraic method is applied to find
the solution.
.c
es
Example 1:
ot

Long Castling Breweries manufactures two brands of beer, Benko lager and
Benoni lager. Benko has a contribution of Sh.4 per unit and Benoni has a
cn

contribution of Sh.3 per unit. Benko requires 30 machine minutes and 30 labor
ne

minutes to manufacture a unit. Total available machine hours per day are 12hrs
whereas total available labor hours per day are 14hrs.
.k
w

Required:
w

1. Formulate linear programming model.

w

2. How much of each brand should Long Castling produce if it wishes to

maximize its daily contribution assuming that all the lager produced is sold.

309
Solution:
1. Formulating a linear programming model

Step 1: Identifying variables:

The variables here are the number of units of Benko and Benoni lager produced
by Long castling breweries per day; we can represent them as:
X1= a unit of Benko lager.
X2= a unit of Benoni lager.

Step 2: Identify the objective:

Definition: An objective is the desired result i.e. optimization of a function

dependent on decision variable and subject to some constraints.

The objective of Long Castling Breweries is to maximize daily contribution.

Objective function is the formula that will give us the total contribution in a
day for both Benko lager and Benoni lager.
The information above can thus be represented in a tabular form as:

ke
PRODUCT Maximum

o.
available
(PER DAY)
Machine hours
X1
0.5
.c X2
0.33
hours/day
12
es
Labor hours 0.5 0.5 14
ot

Contribution 4 3
cn

Objective function = 4X1 + 3X2

ne

The objective is to maximize 4X1 + 3X2

.k

Step 3: Identifying constraints (constraints formulation)

w
w

Definition: Constraints are circumstances that govern achievement of an

w

objective.
Limitations must be quantified mathematically and they must be linear.

For Long castling breweries we have limited machine hours (12hrs/day), which
must be shared among production of Benko and Benoni lagers.
Therefore production must be such that the numbers of machine hours required
is less than or equal to 12 hours per day.
0.5X1+0.33X2 ≤ 12hrs

Similarly for labor hours we have:

0.5X1+0.5X2 ≤ 14hrs

Non-Negativity: It is logical assumption to assume that the company cannot

manufacture negative amounts of a product, thus it can only manufacture
either zero product or more. Therefore we have: X1≥0
X2≥0 or X1, X2≥0

310
Thus the complete linear programming model is;
Maximize 4X1 + 3X2

Subject to the constraints;

0.5X1 + 0.33X2 ≤ 12
0.5X1 + 0.5X2 ≤ 14
X1 X2 ≥ 0

Solving linear programming problems

The question requires us to optimize (in our case, maximize) the objective (the
contribution function), or in simple terms we are required to solve the linear
programming model.
Solving linear programming model entails finding the values of variables that
satisfy all inequalities simultaneously and optimize the objective.

Graphical solution
This method is used to solve LP models in case where only two variables are
involved. For more than two variables (multivariable) then the simplex
technique (algebraic method may be used).

ke
Now in solving the problem above we first draw the axis, taking X1 to be the y

o.
axis and X2 to be the X axis.
.c
es
ot

X1
cn
ne
.k
w
w
w

0
X2

Next we plot the scales on each axis to approximate the scales to use them, we
consider each constraint equation. We get the value of one of variables putting
the other variable to be zero and by substituting the inequality ≤or ≥ with
equality sign (=).

For: 0.5X1 + 0.33X2 ≤ 12

When X1 = 0
0.5(0) + 0.33X2 = 12
0.33X2 = 12
X2 = 12/0.33 = 36
Therefore point to plot is ( 36, 0) … Implying that when
X2 = 36 then X1 = 0
311
When X2 = 0
0.5X1 + 0.33(0) = 12
0.5X1 = 12
X1 = 12/0.5 = 24
Therefore the point is (0, 24)

For 0.5X1 + 0.5X2 ≤ 14

When X1 = 0
0.5(0) + 0.5X2 = 14
0.5X2 = 14
X2 = 14/0.5 = 28
Therefore the point is (28, 0)

When X2 = 0
0.5X1 + 0.5(0) = 14
0.5X1 = 14
X1 = 14/0.5 = 28
Therefore the point is (0, 28)

ke
o.
Comparing these values we see that X2 ranges between 0 – 28, therefore we
can have the graph plotted as:
.c
es
ot
cn
ne
.k
w
w
w

Next draw each limitation (constraint) as separate line on the graph.

For 0.5X1 + 0.33X2 ≤ 12
The two points that represent this line are (36, 0) and (0, 24). This is plotted as
a straight line from 36 on X2 axis to 24 on X1 axis.

312
Now including the Non-Negativity constraints since no negative product can be
produced;

ke
X1 ≥ 0; x2 ≥ 0

o.
.c
es
ot
cn
ne
.k
w
w
w

We must now consider how to choose the production which will maximize
contribution. This we do by plotting a line representing the objective function
(4x1 + 3x2).
313
First choose a convenient point inside the feasible region

E.g. X2(10) +3X1 (20) = 40 +60

= Sh 100

All of the other product mixes that give a contribution of Sh.100 lies on the
line:

100 = 4X1 + 3X2 ........................................... (i)

<<This line is called a contribution line>>

Picking another point, say X2 =10 ad X1 = 20

Its contribution value is SH 110, thus give a contribution line of

110 = 4X1 + 3X2 ............................................. (ii)

Plotting these two contribution lines to our graph we get two parallel lines.

ke
o.
.c
es
ot
cn
ne
.k
w
w
w

Until we reach the last feasible solution(s) before the line moves entirely out of
the feasible region.

314
 ke
o.
.c
es
Point X is the last feasible solution. Coordinates of this point give a
ot

combination of the two lager‘s production volumes that fetches the highest
cn

contribution.
Coordinates of point X can be read from the graph, but for precision they are
ne

calculated by solving simultaneously the equations of the two lines that

.k

intersect at point X.
The two constraints are called binding or limiting constraints. They are the
w

resources being fully used thus preventing daily contribution from increasing
w

further.
w

Therefore to get point x we solve:

0.5X1 + 0.33X2 = 12…. (i)

0.5X1 + 0.5X2 = 14 …. (ii)
Since X is the intersection of these two constraints, solving by deducting (i)
from (ii) we get
0.17X2=2
X2 =11.76

And substituting X2 = 11.76 to equation (i) we get

X1 = 16.24

Therefore 11.76 units of Benko lager and 16.24 units of Benoni lager need to be
produced for maximum contribution.

Contribution = 4(16.24) + 3(11.76) = 100.24

315
Requirements in linear programming
Assumptions that are made to solve these types of problems are that:.
 Proportionality: all activities in linear programming problems are
proportional to the level of decision variables.
 Divisibility: the solution to a linear programming problem does not have
to be an integer but for strictly whole number solutions, use integer
programming.
 Non-negativity: no decision variable can be negative.
 Additivity: the total of all activities in linear programming problems are
assumed to equal to the sum of individual activities.

Special cases in linear programming

 Infeasibility: This is when all constraints don‘t satisfy a particular point
thus there is no feasible solution.
 Redundancy: A constraint is considered redundant if it does not affect
the feasible region. This happens in cases of excess resources since it
does not limit attainment of the objective.
 Multiple optimal solutions: This occurs when the objective function has

ke
the same slope as a binding constraint.

o.
Minimizing problem
.c
es
Example
A manufacturing company has acquired new machine for producing product P
ot

at a rate of 25 units per hour with a 98% rate of efficiency. The company
cn

requires producing at least 1800 units of P per day. The 10 old machines that
the company has produce 15 units of P with 95% efficiency.
ne

The cost of operating the new machine is Sh. 4 per hour and Sh. 3 per hour for
.k

the old ones. The cost incurred due to inefficiency is Sh. 2 per unit; It is
government policy that at least 2 of the new machines must be indulged into
w

production.
w

The company wishes to optimally allocate the machines in order to minimize

w

the total manufacturing cost if the total available hours for production in a day
are 8 hours.

Solution
In minimizing problems we use ≥ (greater or equal to) type inequality.

Step 1: Identifying variables.

Since the problem requires us to appropriately allocate the machines in order
to minimize the costs thus our variables are the new machines and the old
machines, we can let;
X1 =new machines
X2=old machines

Step 2: Identify objectives:

The objective is to minimize manufacturing costs.
316
Total manufacturing cost per machine=Operating cost + (inefficiency rate ×
number of units × cost of loss)
Therefore cost for new machines
Cost = 4+ (0.02 × 25 × 2) = Sh. 5 per hour
=5 × 8= Sh. 40 per day]
Similarly for old machines
Cost = 3+ (0.05×15×2) = Sh. 4.5 per hour
=4.5×8=Sh. 36 per day
Therefore the objective function is to minimize 40X1 + 36X2
Step 3: Identifying constraint functions.
X1 ≤ 8
X2 ≤ 10
(25 × 8) X1 +(8 × 15) X2 ≥1800 (This can be simplified further as)
200 X1 +120 X2 ≥1800 (dividing through by 40)
5X1 +3 X2 ≥45
X1 ≥2
X1, X2 ≥0

Thus the LP model is:

ke
Minimize 40X1 + 36X2

o.
Subject to: X1 ≤ 8
X2 ≤ 10
5X1 +3 X2 ≥45
.c
es
X1 ≥2
ot

X1, X2 ≥0
Plotting this on a graph we get:
cn
ne

X2
.k
w
w
w

8
X1

317
The line X1 ≥2 does not affect the feasible region (doesn‘t cause reduction of
the feasible region), this constraint doesn‘t limit attainment of the objective,
thus its known as a redundant constraint.
Now picking a convenient point inside the feasible region, say (6, 10)

We get a total cost of 600 = (6(40)+10(36))

Thus the objective function line of 40X1 + 36X2 = 600

Moving this line parallel toward the origin to locate the last apex before the
line completely fall off the feasible region, we get:

ke
o.
.c
es
ot
cn
ne
.k
w
w

Point X is the point of optimal solution

w

The binding constraints here are:

X1 ≤ 8
And 5X1 +3 X2 ≥45
Solving this to get coordinates of point X
5X1 +3 X2 = 45 (when X1 = 8)
40 + 3 X2 = 45
X2 = 5/3
Thus the solution is X1 = 8
X2 = 5/3

Shadow or dual prices

Definition: A shadow price or a dual price is the amount increase (or decrease)
of the objective function when one more (or one less) of the binding
constraints is made available.
Consider example 1.

Maximize 4X1 + 3 X2
318
Subject to: 0.5X1 + 0.33X2 ≤ 12 (Machine hours)
0.5X1 + 0.5X2 ≤ 14 (labor hours)
Starting with machine hours; let‘s assume that one more machine hour is
available (with labor hours remaining constant)
We get:
0.5X1 + 0.33X2 = 13
0.5X1 + 0.5X2 = 14
Solving this simultaneously we get the values of X1 and X2 as
0.17 X2 = 1
X2 = 5.88
X1 = 22.12
Thus the contribution is
4(22.12) + 3(5.88) = Sh.106.12
Comparing this with its original contribution of Sh.100.24 (see example 1) we
see increasing machine hours by one unit has increased contribution by
Sh.5.88, which is the shadow price per machine hour.
Note: This figure is also arrived at if we assume that machine hours are
reduced by 1 unit i.e. 12-1.
Similarly assuming that one more labor hour is made available, and then

ke
contribution change is:

o.
0.5X1 + 0.33X2 = 12
0.5X1 + 0.5X2 = 15
Solving this simultaneously gives:
.c
es
0.17 X2 =3
ot

X2 = 17.65
X1 = 12.35
cn

Which give a contribution of.

ne

4(12.35) + 3(17.65) = Sh.102.35

The contribution change is Sh.2.11 which is the shadow price per labor hour.
.k
w

Note:
w

The shadow prices apply in so far as the constraint is binding for example if
w

more and more labor hours are available it will reach a point where labor hours
are no longer scarce thus labor hours cease to be a binding constraint and its
shadow price becomes a zero.(All non-binding constraints have zero shadow
price). Logically its senseless to pay more to increase a resource, which is
already abundant.

Interpretation of shadow prices

A shadow price of a binding constraint indicates to management how much
extra contribution will be gained by increasing a unit of the scarce resource.
In the example above Sh.2.11 is the shadow price for labor hours. This implies
that management is ready to pay up to Sh.2.11 extra per hour for the extra
hours i.e. say an employee is paid sh.5 per hour and one day he works for two
hours extra (overtime), the management is prepared to pay up to sh.7.11 per
hour for the two hours overtime worked.

Sensitivity Analysis

319
Definition: Sensitivity analysis is the test of how certain changes in resources
affect the optimal solution.

In sensitivity analysis we consider the effect of additional limiting or non-

limiting constraints.
We already know that adding more non-limiting constraints does not change
the optimal solution.
We also know that adding more binding constraints affects the objective
function.
It is very important for the management to know how much of a limit resource
can be made available until it has no effect on the objective function (ie
ceases to be a binding resource)

SIMPLEX METHOD
When analyzing linear programming problems with three or more variables the
graphical method becomes enadequate, in such cases we employ simplex
method . Simplex method is an algebraic procedure for solving systems of
equations requiring optimization of the objective function..

ke
This method can be applied to any number of variables, the more they are the

o.
more complex it becomes to workout a solution on paper. Computer programs
e.g. Tora are used to solve the most intricate problems.
.c
The first step in simplex method is conversion of inequalities to linear
es
equations
ot

Example
cn

Consider the linear problem.

ne

Maximize 45 x1  80 x2
Subject to 5 x1  20 x2  400
.k
w

10 x1  15 x2  450
w

x1  0, x2  0
w

Solution.
1. To convert this problem to a system of linear equation, we introduce
slack variables to each constraint.
Z  45 x1  80 x2
Subject to 5 x1  20 x2  x3  400
10 x1  15 x2  x4  450
x1 , x2 , x3 , x4  0

Where the structural variables x3 and x4 are slack variables

2. we then place this information in a tabular form known as a tableau

Initial tableau
Solutio Products Slack Solutio
n Variables n
Variabl x1 x2 x3 x4 Quantit

320
 e y
x3 5 20 1 0 400
x4 10 15 0 1 450
Z 45 80 0 0 0

The tableau shows that

x3  400
x4  450
Z 0 (Total contribution)
It also show that unused capacity is at maximum i.e. the value for the slack
x3 and x4 is 0.
After several operations and when an optimal solution has been attained, these
values will change to give an optimal feasible solution.

3. Select the column with the highest value of Z (i.e. 80), then devide the
positive numbers in that column (i.e. the x 2 column) into the quantity
column.

ke
i.e. 400  20  20

o.
450 15  30
.c
Chose the row giving the lowest result (in our case the row with x 3 gives 20)
es
and mark the element falling on the intersection of the selected row and
ot

selected column (i.e. 20: selected element)

cn

4. Divide all the elements in the selected row by the value of the selected
ne

and change the solution variable to the heading of the identified column
(from x3 to x2)
.k

Thus the tableau will appear as follows.

w

Solutio Products Slack Solutio

w

n Variables n
w

Variabl x1 x2 x3 x4 Quantit
e y
x2 1
4
1 1
20
0 20
x4 10 15 0 1 450
Z 45 80 0 0 0

5. Next we conduct row operations that aim to reduce elements falling in

the same column as the previously marked element to zero. These row
operations may sometime necessitate multiplying or dividing the
selected row with an arbitrary number.

Therefore:
Row 2 – 15×Row 1

321
 x4 10 15 0 1  450

15  x2 154 15 15
20 0  300
new x4 6 14 0* 43 1  150
Note that the aim was to attain the zero.

Z 45 80 0 0  0

80  x2 20 80 4 0  1600
new Z 25 0* 4 0  1600

On replacing the new rows to the tableau we get

2nd tableau.
Solutio Products Slack Solutio
n Variables n
Variabl x1 x2 x3 x4 Quantit
e y
x2 1
4
1 1
20
0 20

ke
x4 6 14 0 3
4
1 150

o.
Z 25 0 -4 0 -1600
.c
Since in the Z row under products column we still have values greater than
es
zero, we conduct another operation.
ot

Taking the column with a Z value of 25, we repeat the process in the same
cn

manner.
ne

20  14  80
150  6 14  24
.k
w

Thus we pick the x4 row and mark the element 6 14 , the row solution
w
w

variable is changed to x1 and we divide the row by 6 14 to convert the

marked element to 1.
Therefore; x1 1 0 3
25
4
25 = 24

Next we do the row operations

x2 14 1 201 0  20

 4  x1 4 0 100 100  6
1 3 4

new x2 0* 1 1
50  251  24

Z 25 0 4 0  1600

25  x1 25 0 3 4  600
new Z 0* 0 7  4  2200

the tableau becomes, 3rd Tableau

Solutio Products Slack Solutio
322
 n Variables n
Variabl x1 x2 x3 x4 Quantit
e y
x2 0 1 1
50  251 24
x1 1 0 3
25
4
25
24
Z 0 0 -7 -4 -2200

This is the final tableau since the Z row has no values greater than zero thus
we have the optimal solution.

Interpretation
 to maximize Z we need to produce 24 units x2 and 24 units of x1, we
obtain these values from the solutions quantity column
Thus, Z = 24(45) + 24(80) = 3000

 We have zero slack (unused quantities of constraints).

Assume tableau 2 is the final tableau and let the constraint with variable
x3 be labor hours and x4 are raw materials, the slack would‘ve been

ke
interpreted as:

o.
o 150 units of raw materials were unused
o To maximize Z we produce 20units of x2 and none of x1.
.c
 The value represents in Z row under slack variable column represents
es
shadow prizes. Thus the shadow prize for the first constraint with x 3
ot

is 7 and the shadow prize for the second constraint with the variable
x4 is 4.
cn
ne

INTERPRETATION OF COMPUTER GENERATED SOLUTION

.k

Example
w

Maximize 25 x1  20 x2  24 x2
w

where : x1  Xtragrow, x2  Youngrow, x3  Zupergrow

w

Subject to 0.3x1  0.2 x3  500

0.5 x2  0.4 x3  1000
0.2 x1  0.1x2  0.1x3  800
0.4 x2  0.3 x3  600
x1  1500
x1  0, x2  0, x3  0
The computer generated solution for this problem is as follows;

Objective value = 71666.7

Variable Value Obj. Coeff Obj Value
Contribution
X1: Xtragrow 1666.7 25 4166.7
X2: Youngrow 1500 20 30000
X3: Zupergrow 0 24 0
323
Constraint RHS Slack-
/Surplus+
1(<) 500 0
2(<) 1000 250-
3(<) 800 316.67-
4(<) 600 0
5(>) 1500 166.7+

Sensitivity Analysis
Variable Current obj Min obj coeff Max Obj Reduced
coeff Coeff cost
X1: Xtragrow 25 13.50 Infinity 0
X2: Youngrow 20 9.78 Infinity 0
X3: 24 -Infinity 31.67 7.67
Zupergrow
Constraint Current RHS Min RHS Max RHS Dual price
1(<) 500 450 975 83.33

ke
2(<) 1000 750 Infinity 0
3(<) 800 483.3 Infinity 0

o.
4(<) 600 0 800 50
5(>) 1500 .c
-Infinity 1666.7 0
es
Required
ot

Interpret the data generated by the computer.

cn

Solution.
ne

Table 1:
Objective value, is the solution to objective function (e.g. the solution to this
.k

example is 71,436)
w

The four columns of table 1 are to be interpreted as follows;

w

 Variable: these are the variables of the model. In our example we have
w

x1 = Xtragrow, x2 = Young row and x3 = Zupergrow

 Value: this is value that the variables assume at optimal solution (to
optimize the objective function one needs to produce this amounts of
the variables). In our example we are required to produce 1,666.67 of x1
and 1,750 of x2 and none of x3
 Objective coefficient: these are the coefficients of the objective
function
 Objective value contribution: this is the value contributed by each
variable to the objective function (for x1=25×1,666.67), the total of this
is equal to our objective value (i.e. 41,666.67+35,000=76,666.67).

The 3 columns of the second part of table1 can be interpreted as follows;

 Constraints: this is constraints of the model representing the limited
resources.
 RHS: the Right hand side value is the limiting value of the constraint.
E.g. for the first constraint the maximum amount of material A is 500
tons.
324
  Slack/surplus: at optimal production not all the materials for some of
the constraints will be fully utilized, slack is the amount of material that
is left over after production. For constraint 1 and 4 no material
remained, this also implies that these are the binding constraints i.e.
their adjustment directly affects the objective solution

Sensitivity Analysis
This is the analysis of the effect of adjusting variables or constraint, whether
the objective solution will be affected. How much of the objective coefficient
(or the maximum available amount of a constraint) can be reduced or
increased without affecting the objective solution.
The columns of this table can be interpreted as follows;
Variable: as explained above
Current objective coefficient: this is the value of coefficients of the
objective function
Minimum objective coefficient: this is how low the coefficient can be

ke
reduced without affecting the optimal basis. The coefficient for x1 can

o.
be reduced from 25 to 13.50 (the prize for Xtragrow fall to as low as
$13.50 from $25) but the optimal solution will remain the same
.c
es
Maximum objective coefficient: this is how high the coefficient can be
ot

increased without affecting the optimal basis

cn

Reduced cost: this is amount by which the coefficient of the variable has to
ne

be adjusted with for it to become a basic variable (included to the

objective optimal solution). X1 and x2 have 0 reduced costs implying
.k

that they already make part of the optimal solution, x3 will require to
w

be increased by 7.67 for it to make part of the basic variable.

w
w

The second part of the table is interpreted as follows;

 Constraints: as described above
 Current Right Hand Side: the limiting value of the constraint. E.g. for
the first constraint the maximum amount of material A is 500 tons.
 Minimum RHS: the lowest the available amount of the constraint can be
reduced without affecting the optimal basis
 Maximum RHS: the highest the available amount of the constraint can
be increased without affecting the optimal solution
 Dual price: this is amount increase to the objective contribution due to a
unit increase of the available constraint. Since there is only 500 tons of
material A, if management decides to increase it by a unit to 501tons
then the objective optimal solution will be 76,666.67+83.33.

325
 PRACTICE QUESTIONS

QUESTION ONE
Regal Investments has just received instructions from a client to invest in two
shares; one an airline share, the other an insurance share. The total maximum
appreciation in share value over the next year is to be maximized subject to
the following restrictions:
- the total investment shall not exceed Sh.100,000
- at most Sh.40,000 is to be invested in the insurance shares
- quarterly dividends must total at least Sh.2,600

The airline share is currently selling for Sh.40 per share and its quarterly
dividend is Sh.1per share. The insurance share is currently selling for Sh.50 per
share and the quarterly dividend is Sh.1.50 per share. Regal‘s analysts predict
that over the next year, the value of the airline share will increase by Sh.2 per
share and the value of the insurance share will increase by Sh.3 per share.
Computer software provided the following part solution output:

ke
Objective Function Value = 5,400

o.
Variable Number Reduced cost
Airline shares 1,500 0.000 .c
es
Insurance shares 800 0.000
ot

Constraint Slack/Surplus Dual prices

cn

Total investment 0.000 0.050

ne

Investment in insurance 0.000 0.010

Dividends 100.000 0.000
.k
w
w
w

326
Objective Coefficient Ranges

Variable Lower limit Current value Upper limit

Airline share 2.500 3.000 No upper limit
Insurance share 0.000 2.000 2.400

Right-hand Side Ranges

Constraint Lower limit Current value Upper limit

Total investment 96,000.00 100,000 No upper limit
Investment in insurance20,000.00 40,000 100,000.00
Dividends No lower limit 2,600 2,700.00

Required:
e) Formulate the above problem.
f) Explain what reduced cost and dual prices columns above mean.
g) How should the client‘s money be invested to satisfy the restrictions?
h) Suppose Regal‘s estimate of the airline shares appreciation is an error,

ke
within what limits must the actual appreciation lie for the answer in (c)
above to remain optimal?

o.
(Q 6 Dec
2001) .c
es
QUESTION TWO
ot

c) A baker makes two products; large loaves and small round loaves. He can
cn

sell up to 280 of the large loaves and up to 400 small round loaves per day.
ne

Each large loaf occupies 0.01m3 of shelf space, each small loaf occupies
0.008m3 of space, and there is 4m3 of shelf space available. There are 8
.k

hours available each night for baking, and he can produce large loaves at
w

the rate of 40 per hour, and small loaves at the rate of 80 per hour. The
w

profit on each large loaf is Sh.5.00 and Sh.3.00 profit on the small round
w

loaf.

Required:
In order to maximize profits, how many large and small round loaves should
he produce?

d) Summarize the procedure for solving the kind of quantitative technique you
have used to solve part (a) above. (Q 6 June
2001)

QUESTION THREE
c) A small company will be introducing a new line of lightweight bicycle
frames to be made from special aluminum alloy and steel alloy. The frames
will be produced in two models, deluxe and professional. The anticipated
unit profits are currently Sh.1, 000 for a deluxe frame and Sh.1,500 for a
professional frame. The number of kilograms of each alloy needed per

327
 frame is summarized in the table below. A supplier delivers 100 kilogram‘s
of the aluminum alloy and 80 kilogram‘s of the steel alloy weekly.

Aluminum alloy Steel alloy

Deluxe 2 3
Professional 4 2

Required:
iii) Determine the optimal weekly production schedule.
iv) Within what limits must the unit profits lie for each of the frames for
this solution to remain optimal?

d) Explain the limitations of the technique you have used to solve part (a)
above.
(Q 6 Dec 2000)

QUESTION FOUR
a) Define the following terms as used in linear programming:

ke
i) Feasible solution

o.
ii) Transportation problem
iii) Assignment problem
.c
es
b) The TamuTamu products company ltd is considering an expansion into five
ot

new sales districts. The company has been able to hire four new
experienced salespersons. Upon analysing the new salesperson‘s past
cn

experience in combination with a personality test which was given to them,

ne

the company assigned a rating to each of the salespersons for each of the
districts .These ratings are as follows:
.k
w

c)
w

Districts
w

1 2 3 4 5
A 92 90 94 91 83
Salespersons B 84 88 96 82 81
C 90 90 93 86 93
D 78 94 89 84 88

The company knows that with four salespersons, only four of the five
potential districts can be covered.

Required:
iii) The four districts that the salespersons should be assigned to in order
to maximize the total of the ratings
iv) Maximum total rating. (Q 6 June
2002)

QUESTION FIVE
328
d) Explain the value of sensitivity analysis in linear programming problems and
show how dual values are useful in identifying the price worth paying to
relax constraints.

e) J.A Computers is a small manufacturer of personal computers. It

concentrates on production of three models- a Desktop 386, a Desktop 286,
and a Laptop 486, each containing one CPU Chip. Due to its limited
assembly facilities JA Computers are unable to produce more than 500
desktop models or more than 250 Laptop models per month. It has one
hundred and twenty 80386 chips (these are used in Desktop-386) and four
hundred 80286 chips (used in desktop 286 and Laptop 486) for the month.
The Desktop 386 model requires five hours of production time, the Desktop
286 model requires four hours of production time, and the Laptop 486
requires three hours of production time. J.A Computers have 2000 hours of
production time available for the coming month. The company estimates
that the profit on Desktop 386 is Sh. 5,000. For a desktop 286 the profit is
Sh.3, 400 and Sh.3,000 profit for a laptop 486.
Required:
Formulate this problem as a profit maximization problem and mention the

ke
basic assumptions that are inherent in such models.

o.
f)
below:
.c
An extract of the output from a computer package for this problem is given
es
ot

Output solution
X1=120, X2 = 200, X3 = 200
cn

Dual values Constraints 3 150

ne

Constraints 4 90
Constraints 5 20
.k

Sensitivity analysis of objective function coefficients:

w
w

Variable Lower Original Upper

w

limit value limit

X1 100 250 No limit
X2 150 170 200
X3 127.5 150 170

Sensitivity analysis on R.H.S ranges.

Constrain Lower Original Upper

ts limit value limit
1 320 500 No limit
2 200 250 No limit
3 80 120 130
4 350 400 412.5
5 1950 2000 2180

X1=Monthly production level for Desktop 386.

X2 =Monthly production level for Desktop 286.
329
X3=Monthly production level for Laptop 486.

Required:
iv) Interpret the output clearly, including optimum product mix, monthly
profit, unused resources and dual values
v) Explain the purpose of upper limits and lower limits for the variables
X1, X2, X3 and constraints 1 to 5.
vi) Calculate the increase in profit if the company is able to produce a
further 10 CPU 80386 chips. (Q7
July 2000 Pilot paper)

ke
o.
.c
es
ot
cn
ne
.k
w
w
w

330