STATISTICS LECTURE NOTES Page |1

STATISTICS DIPLOMA M3

STATISTICS

GENERAL OBJECTIVE

This paper is intended to equip the candidate with knowledge, skills and attitudes that will enable
him/her to use statistical tools in business operations and decision making.

LEARNING OUTCOMES

A candidate who passes this paper should be able to:

 Use mathematical techniques in solving business problems

 Apply set theory in business decision making

 Apply operation research techniques in decision making

CONTENT

Fundametals of Mathematics and Statistics

Equations and Functions

 Functions, equations and graphs: Linear, quadratic, cubic, exponential and logarithmic

 Counting Techniques

 Set theory

 Binomial Theorem and progression

 Application of mathematical functions in solving business problems

Mathematics of finance

 Simple and compound interest

 Cash flow generation

 present and future values

 project appraisal techiniques

Inventory control models

 Definition of inventory control

 Types of inventory control systems

 EOQ model safety stock and reorder level

Index numbers

 Definition and types of index numbers

STATISTICS LECTURE NOTES Page |2

 Construction of index numbers

 Fixed based and chain based index numbers

 Applications of index numbers – Consumer price indices, inflarion rates, NSE price indices

Matrix algebra

 Definition and Types of Matrices and operations (addition, subtraction, multiplication,

transposition and inversion)

 Application of matrices: statistical modelling and general applications

Introduction to statistics

 Introduction: definitions and branches of statistics

 Methods of data collection: primary and secondary data

 Sampling techniques

Collection and presentation of data

 Tables

 Diagrams: bar charts and pie charts

 Graphs: time series graphs, Z-charts, Lorenz curves and semi-log graphs

 Frequency distribution tables

 Histogram and frequency polygons

 Cumulative frequency curve (ogive) and its application

Descriptive statistics

– Measures of central tendency: mean: arithmetic mean, weighted arithmetic mean; median,
mode, geometric mean and harmonic mean

 Measures of dispersion: range, quartile, deciles, percentiles, mean deviation, standard deviation
and coefficient of variation

 Measures of skewness; pearsons coefficient of skewness, product coefficient of skewness

 Measures of kurtosis; pearsons coefficient of kurtosis, product coefficient of kurtosis.

Probability theory and distribution Probability theory

 Definitions: event, outcome, experiment, sample space

 Types of events: elementary, compound, dependent, independent, mutually exclusive,

exhaustive, mutually inclusive

 Laws of probability: additive and multiplicative rules

STATISTICS LECTURE NOTES Page |3

 Baye’s Theorem

 Probability trees

 Expected value, variance, standard deviation and coefficient of variation using frequency and
probability

Probability distributions

 Discrete and continuous probability distributions (uniform, normal, binomial, poisson

and exponential)

 Application of probability to business problems



Correlation and regression analysis Correlation analysis

 Scatter diagrams

 Measures of correlation –product moment and rank correlation coefficients (Pearson and
Spearman)

 Regression analysis

 Simple linear regression analysis

 Assumptions of linear regression analysis

 Coefficient of determination, standard error of the estimate

 Computer output of linear regression

Hypothesis testing and estimation

 Hypothesis tests on the mean (when population standard deviation is unknown)

 Hypothesis tests on proportions

 Hypothesis tests on the difference between means (independent samples)

 Hypothesis tests on the difference between means (matched pairs)

 Hypothesis tests on the difference between two proportions

Time series

 Definition of time series

 Components of time series (circular, seasonal, cyclical, irregular/ random, trend)

 Application of time series

 Methods of fitting trend: free hand, semi-averages, moving averages, least squares methods

 Models – additive and multiplicative models

STATISTICS LECTURE NOTES Page |4

 Measurement of seasonal variation using additive and multiplicative models

 Forecasting time series value using moving averages, ordinary least squares method and
exponential smoothing

 Comparison and application of forecasts for different techniques

 Trend projection methods: linear, quadratic and exponential

Network planning and analysis

 Basic concepts – network, activity, event

 Activity sequencing and network diagram

 Critical path analysis (CPA)

 Float and its importance

 Crashing of activity/project completion time

 Project evaluation and review technique (PERT)

 Resource scheduling (leveling) and Gantt charts

 Advantages and limitations of CPA and PERT

Emerging issues and trends

 biostatistics, causal inference, cloud computing, data visualization, distributed inference and
learning, differential privacy, federated learning, integrative analysis, interpretable machine
learning, replicability and reproducibility, scalable statistical inference, study design

Linear Algebra and Matrices

Contents
- Functions and graphs
- Linear equations, higher order equations, inequalities and simultaneous equations
- Matrix algebra
- Application of matrix algebra to input-output analysis and elementary Markovian process.
Functions and graphs
A function is a mathematical relationship in which the value of a single dependent variable are determined
from the values of one or more independent variables. The following is an example of a function in which y
is said to be a function of x.
y = a + bx
In the above example, both x and y are variables this is because they may assume different values throughout
the analysis of the function. On the other hand, a and b are referred to as constants because they assume
fixed values.
STATISTICS LECTURE NOTES Page |5

The variable y is a dependant variable in the sense that its values are generated from an independent variable
x.
The collection of all the values of the independent variable for which the function is defined is referred to as
the domain of the function corresponding to this we have the range of the function, which is the collection of
all the values of the dependent variable defined by the function
The fact that it is a function of x can also be denoted by the following general form
y = f(x)
Functions of a single independent variable may either be linear or non linear.
Linear functions can be represented by:
y = a + bx
Whereas non – linear functions can be represented by functions such as:

3
i. y = α0 + α 1 x + α2x3
ii. y2 = 3x + 18
iii. y = 2x2 + 5x + 7
iv. ax2 + bx + cy + d = 0 Where α, a, b, c, d, k = constants
v. xy = k
vi. y = ax

Graph of a function
A graph is a visual method of illustrating the behaviour of a particular function. It is easy to see from a graph
how as x changes, the value of the f(x) is changing.
The graph is thus much easier to understand and interpret than a table of values. For example by looking at a
graph we can tell whether f(x) is increasing or decreasing as x increases or decreases.
We can also tell whether the rate of change is slow or fast. Maximum and minimum values of the function
can be seen at a glance. For particular values of x, it is easy to read the values of f(x) and vice versa i.e. graphs
can be used for estimation purposes
Different functions create different shaped graphs and it is useful knowing the shapes of some of the most
commonly encountered functions. Various types of equations such as linear, quadratic, trigonometric,
exponential equations can be solved using graphical methods.

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)
2x3 + 4x2 + 3x + 8 = 0 (cubic equation)
The solution of equations or the values of the variables for which the equations hold is called the roots of the
equation or the solution set.

Solution of Linear Equation

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 the solution of linear equations
Rule 1: Additional rule
STATISTICS LECTURE NOTES Page |6

If M = N then M + P = N + P
Rule 2: Subtraction rule
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 3 = - 4
Solutions
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
3  4  3
ii. 3
y = –12 (simplifying)
Solution of quadratic equations
Suppose that we have an equation given as follows
ax2 + bx + c = 0
Where a, b and c are constants, and a≠ 0. such an equation is referred to as the general quadratic equation in
x. if b = 0, then we have
ax2 + c = 0
Which is a pure quadratic equation
There are 3 general methods for solving quadratic equations; solution by factorization, solution by
completing the square and solution by the quadratic formula.

Solution by Factorization
The following are the general steps commonly used in solving quadratic equations by factorization
(i) Set the given quadratic equation to zero
(ii) Transform it into the product of two linear factors
(iii) Set each of the two linear factors equal to zero
STATISTICS LECTURE NOTES Page |7

(iv) Find the roots of the resulting two linear equations

Example
Solve the following equation by factorization
i. 6x2 = 18x
ii. 15x2 + 16x = 15
Solutions
i. 6x2 = 18x
6x2 – 18x = 0 (step 1)
6x(x – 3) = 0 (step 2)
6x = 0 (step 3)
and x – 3 = 0
∴ x = 0 or x = 3 (by step 4)
ii. 15x2 + 16x = 15
15x2 + 16x – 15 = 0 (step 1)
(5x – 3) (3x +5) = 0 (step 2)
(5x – 3) = 0} Step 3
{3x + 5 = 0}
5 3
∴ x = - 3 or + 5 (step 4)
Solution by Completing the Square
The process of completing the square involves the construction of a perfect square from the members of the
equation which contains the variable of the equation.
Consider the equation – 9x2 – bx = 0
The method of completing the square will involve the following steps
i. Make the coefficient of x2 unity
ii. Add the square of ½ the coefficient of x to both sides of the equal sign. The
left hand side is now a perfect square
iii. Factorize the perfect square on the left hand side.
iv. Find the square root of both sides
v. Solve for x

Example
Solve by completing the square.
i. 3x2 = 9x
ii. 2x2 + 3x + 1 = 0
STATISTICS LECTURE NOTES Page |8

Solutions
i. 3x2 = 9x or
(3x2 - 9x = 0)
x2 - 3x = 0 (Step 1)
2 2
 3  3
x 2  3x        
 2  2 (Step 2)

2
 3 9
x  
 2 4 (Step 3)

9
x 3  
4 (Step 4)

∴ 3 3
x 
2 2
33 3 3
 or  
2 2 2
(= 3 or 0)

ii. 2x2 + 3x + 1 = 0 or (2x2 + 3x = -1)

3x 1
x2 + =-
2 2 …………………...….. (Step 1)
2 2
3x  3   3  1
X  2
     
2 4 4 2 ……… (Step 2)
2
 3 1
x +  =
 4 16 …………………….. (Step 3)
x+ 3
4 =± 1
16

x =  34 = ± 1
4

 34 + 1
4 or - 34 - 1
4

x   12 or x  1

Solution by Quadratic Formula

Consider the general quadratic equation
ax 2 + bx + c = 0 where a  0
STATISTICS LECTURE NOTES Page |9

The roots of the equation are obtained by the following formula:

 b  b 2  4ac
x
2a
Example
Solve for x by formula
5x2 + 2x – 3 = 0
Solution
a = 5, b = 2, c = - 3

 b  b 2  4ac
x
2a

 2  2 2  4(5)(3)
x
2(5)

3
x  or  1
5

Inequalities
An inequality or inequation is an expression involving an inequality sign (i.e. >, <, ≤, ≥, i.e. greater than, less
than, less or equal to, greater or equal to) The following are some examples of inequations in variable x.
3x + 3 > 5
x2 – 2x – 12 < 0
The first is an example of linear inequation and the second is an example of a quadratic in equation.

Solutions of inequations
The solutions sets of inequations frequently contain many elements. In a number of cases they contain
infinite elements.

Example
Solve and graph the following inequation
x – 2 > 2 ; x  w (where x is a subset of w)

Solution
x – 2 > 2 so x–2+2>2+2
Thus, x>4
The solution set is infinite, being all the elements in w greater than 4
STATISTICS LECTURE NOTES P a g e | 10

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

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

 3x - 7 + 7 < -13 + 7

 3x < -6

3x -6
<
3 3

x < -2
….. R Line

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

Rules for solving linear inequations

Suppose M, M1, N, N1 and P are expressions that may or may not involve variables, then the corresponding
rules for solving inequations will be:
Rule 1: Addition rule
If M > N and M1> N1
Then M + P > N + P and
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)
STATISTICS LECTURE NOTES P a g e | 11

Rule 4: Division
If M > N and M1< N1 and P≠ 0
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:
i. 7 – 2x > - 11 ;
ii. –5x + 4 ≤ 2x – 10 ;
iii. –3 ≤ 2x + 1 < 7 ;
Solutions
i. 7 - 2x > -11

-2x > -18 (subtraction rule)

-2x -18
< (bydivision rule)
-2 -2

x<9

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

ii. -5x + 4  2x - 10

-7x + 4  -10 (by subtraction rule)

-7x  -14 (by subtraction rule)

x2 (by division rule)

STATISTICS LECTURE NOTES P a g e | 12

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

iii. -3  2x + 1 < 7

-4  2x < 6 (by substraction rule)

-2  x < 3 ( by division rule)

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

Linear inequation in two variables: relations

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.

1.2 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.
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.
STATISTICS LECTURE NOTES P a g e | 13

10

9
.

.
7
(2,4)
.
6

.
5 x + 2y = 10
2x + y = 8

Example -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

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
6x + 8y = - 10 …… (iv).
Subtract iii from iv.
17y = -34 …….. (v).
 y = -2

Step 2
STATISTICS LECTURE NOTES P a g e | 14

Multiply (i) by 4
8x – 12y = 32 ……. (vi)
Multiply (ii) by 3
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).
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
Substitute this value of x into equation (ii). And obtain an equation in y only
3x + 4y = -5
3 (4 + 3/2 y) + 4y = -5
8 ½ y = - 17 ……. (iv)
Step 3
Solve the equation (iv). For y
8½y = -17
y = -2
Step 4
Substitute this value of y into equation (i) or (iii) and obtain the value of x
2x – 3y = 8
STATISTICS LECTURE NOTES P a g e | 15

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
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
application of equations in business
1 analysis of cost
a. Fixed Costs (FC): These are costs which do not vary with the level of production i.e. they are fixed at
all levels of production. They are associated with fixed factors of production in the Short Run. Examples are
rent or premises, interest on loans and insurance.

Costs

TFC

Output

b. Variable Costs (VC): These are costs, which vary with the level of production. The
higher the level of production, the higher will be the variable costs. They are associated
with variable factors of production in the Short Run. Examples are costs of materials,
cost of fuels, labour costs and selling costs.
STATISTICS LECTURE NOTES P a g e | 16

Costs TVC

Output

c. Total Cost (TC): This is the sum of fixed costs and variable costs i.e. TC = FC + VC.

Costs TC = TVC + TFC

TVC

TFC

Output

d. Average Fixed Cost (AFC): This is fixed cost per unit of output, obtained
by dividing fixed costs by total output i.e.

AFC = Fixed Costs

Total output

As the output increases, the Average Fixed Cost falls.

STATISTICS LECTURE NOTES P a g e | 17

Costs

AFC
Output

e. Average Variable Cost (AVC): This is the average cost per unit of output, obtained
by dividing variable costs by total output i.e.

AVC = Variable Cost

Total Output

Costs AVC

Output

f. Average Total Costs (ATC): This is total cost per unit of output, obtained by dividing total cost by
total output i.e.

ATC = Total Cost

Total Output

Costs ATC
STATISTICS LECTURE NOTES P a g e | 18

g. Marginal Cost: This is the increase in total cost resulting from the production of an extra unit of
output. Thus, if TCn is the total cost of producing n units of output and TCn-1 is the total cost of
producing n-1 units of output, then the marginal cost of producing the ‘nth’ of unit of output is
calculated as:
Marginal Cost = TCn - TCn-1

It will be observed that since fixed costs are fixed, it follows that:

Marginal Cost = VCn - VCn-1

(GRADIENT OF THE TOTAL COST FUNCTION)

2. REVENUE
TOTAL REVENUE (TR)- total sales of a firm based on a given quantity of goods. It is the total income of
a company TR = price
Price is also known as average revenue
MARGINAL REVENUE(MR)- revenue gained by producing one additional unit of a product or service.

PROFIT ()is the difference between TR and TC. The most optimal point of operation of a firm is the proint
where profits are at maximum where MR.=MC. It is obtained by differentiating the profit function with
respect to Q and equating it to 0 as follows
0

therefore MR.=MC
Illustration
QUESTION TWENTY-FOUR
A monopoly firm is faced with the following demand function
STATISTICS LECTURE NOTES P a g e | 19

P = 13 – 0.5Q

The Marginal Cost function for the firm is given by 3 + 4Q and the total fixed cost is 4.

Determine:
a) The profit maximizing output. (6 marks)
b) The level of supernormal profit if any. (3 marks)
c) The output level at the break-even point. (2 marks)

A firm operating in a perfectly competitive market has to sell all its output at the price of Sh.10 per unit. Its
marginal cost function is given by Q + 4 and the total fixed cost is 1.

Determine:
d) The profit maximizing output level. (6 marks)
e) The level of supernormal profit if any. (3 marks)
(Total: 20 marks)
The total cost equation in the production of bacon at some hypothetical factory is
C = 1000 + 100Q – 15Q2 + Q3

Where C = Cost measured in shillings, while Q = quantity measured in kilogrammes.

a) Compute the total and average costs at output level of 10 and 11 kilogrammes.(6 marks)
b) What is the Marginal cost of the 12th Kilogramme? (4 marks)

MATRICES
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 its rows (m) and column (n).
a b c
a b  
For example =   and B = d e f 
c d g h i 
 
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.
A matrix A with three rows and four columns is given by one of:
STATISTICS LECTURE NOTES P a g e | 20

 a11 a12 a13 a14 

 
A=  a 21 a 22 a 23 a 24 
a a 34 
 31 a 32 a 33

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

Properties 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
   
 2 2 3 =  2 2 3
   
The following matrices are equal  5 1 1   5 1 1 

Column Matrix or column vector

A column matrix, also referred to as column vector is a matrix consisting of a single column.

 x1 
 
 x2 
 . 
 
 . 
 . 
 
For example x =  xn 

Row matrix or row vector

It is a matrix with a single row

For example y =  y1 , y 2 , y3 .........y n 

Transpose of a Matrix
The transpose of an mxn matrix A is the nxm matrix AT obtained by interchanging the rows and columns of
A.
STATISTICS LECTURE NOTES P a g e | 21

A = aij

The transpose of A i.e. AT is given by

AT = aij = aji
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 
 3

Solution
T
1 5 7 1 2 0
   
i. A T =  2 1 4 =  5 1 9
 0 9 3  7 4 3
   

 b1 
 
b
ii. BT =  b1 , b 2 , b3 , b 4  =  2
T

 b3 
 
 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.
STATISTICS LECTURE NOTES P a g e | 22

2 5
A= 3 7 is a square matrix of order 2

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 one (1)
e.g.

1 0 0 
1 0 
I2    and I 3  0 1 0 
0 1   
0 0 1
2  2 unit matrix 3  3 unit matrix

A null or zero matrix

A null or zero matrix is a matrix whose elements are all equal to zero.
Sub matrix
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
Matrix addition and subtraction
STATISTICS LECTURE NOTES P a g e | 23

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 
   

 6 1 10 5  12 4 7 3   18 3 3 8
     
A+B=  3 4 2 5  +  0 4 10 4  =  3 0 12 9 
 9 13 6 0   7 3 7 9   2 16 1 9 
     

 6 1 10 5  12 4 7 3   6 5 17 2
     
A-B=  3 4 2 5  -  0 4 10 4  =  3 8 8 1 
 9 13 6 0   7 3 7 9   16 10 13 9 
     

If it is assumed that A, B, C are of the same order, the following properties 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

 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
STATISTICS LECTURE NOTES P a g e | 24

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 number of units of commodities P, Q, R sold respectively. Then the vector
product A  B will be equal to the total sales value
i. e. A  B =Total sales value

 100 
 
Let A =  4 5 6  and B =  200 
 300 
 

 100 
 
then  4 5 6   200  = 400 + 1,000 + 1,800 = Shs 3,200
 300 
 

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 
 
STATISTICS LECTURE NOTES P a g e | 25

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 

  and b =  11 
Let A =  a 21 a 22   b 21 b 22 b 23 

 d11 d12 d13 

 
Then A  B = D =  d 21 d 22 d 23 
A = 2 x 2 matrix B = 2 x 3 matrix D = 2 x 3 matrix
Where
d11 = a11  b11 + a12  b21
d12 = a11  b12 + a12  b22

Example I

 6 1  3 0 2  6  3  1 4 6  0  1 5 6  2  1 8 
     = 
 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:

Matrix X Matrix Y
Parts Products
A B C P1 P2
P 3 4 2 Mon 1 2
Products 1 
P2 2 5 3  Tues 2 3
 
Wed 3 2
Thur 2 2
 
Fri 1 1
STATISTICS LECTURE NOTES P a g e | 26

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.

1 2  1×3+2×2 1×4+2×5 1×2+2×5 

   
2 3
 3 4 2  2×3+3×2 2×4+3×5 2×2+3×3 
Y × X = 3 2 ×   =  3×3+2×2 3×4+2×5 3×2+2×3 
   2 5 3  
2 2  2×3+2×2 2×4+2×5 2×2+2×3 
1 1   1×3+1×2 1×4+1×5 1×2+1×3 
 
5 x 2 matrix 2 x 3 matrix = 5 x 3 matrix

A B C
Mon 7 14 8 
 
Tues 12 23 13 
Wed  13 22 12 
 
Thur 10 18 10 
Fri 5 9 5 

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.

Determination of a 2 x 2 matrix

a b
A =  = ad - cb
c d

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.
STATISTICS LECTURE NOTES P a g e | 27

2x2 Matrix
Consider a 2x2 matrix:
whose determinant is D=ad-cb
Then its inverse is

PERMUTATION AND COMBINATION

Fundamental Counting Principle
In a sequence of events, the total possible number of ways all events can performed is the product of the
possible number of ways each individual event can be performed.
The Bluman text calls this multiplication principle
Factorials
If n is a positive integer, then
n! = n (n-1) (n-2) ... (3)(2)(1)
n! = n (n-1)!
A special case is 0!
0! = 1
Permutations
The permutation key (nPr) is located under the math probability menu. Enter the number of objects, n,
first; then the permutation key; then the number of objects to take at one time, r.
Consider the different ways in which a pair of objects can be selected from five distinguishable
objects—such as the letters A, B, C, D, and E. If both the letters selected and the order of selection are
considered, then the following 20 outcomes are possible:

Each of these 20 different possible selections is called a permutation. In particular, they are called the
permutations of five objects taken two at a time, and the number of such permutations possible is denoted
by the symbol 5P2, read “5 permute 2.” In general, if there are n objects available from which to select,
and permutations (P) are to be formed using k of the objects at a time, the number of different
P.
permutations possible is denoted by the symbol n k A formula for its evaluation is

nPk.=
STATISTICS LECTURE NOTES P a g e | 28

The expression n!—read “n factorial”—indicates that all the consecutive positive integers from 1 up to
and including n are to be multiplied together, and 0! is defined to equal 1. For example, using this
formula, the number of permutations of five objects taken two at a time is 20

5P2 =
Example: The final night of the Folklore Festival will feature 3 different bands. There are 7 bands to

choose from. How many different programs are possible? 7 3 P

Combinations
k objects are selected from a set of n objects to produce subsets without ordering. Contrasting the
previous permutation example with the corresponding combination, the AB and BA subsets are no longer
distinct selections; by eliminating such cases there remain only 10 different possible subsets—AB, AC,
AD, AE, BC, BD, BE, CD, CE, and DE.

C
The number of such subsets is denoted by n k, read “n choose k.” For combinations, since k objects
have k! arrangements, there are k! indistinguishable permutations for each choice of k objects; hence
dividing the permutation formula by k! yields the following combination formula:

nCk = nPk/ k! =
10 C4 = =210
The Classical Studies Department has 7 faculty members. Three must attend the graduation
ceremonies. How many different groups of 3 can be chosen?
• If order mattered, the answer would be 7·6·5 = 210
• Let’s look at one set of three professors: A, B, C:
A B C A C B B C A B AC CAB C B A
• Why are there 6 listings for the same set of 3

Correlation
is a statistical method that determines the degree of relationship between two different
variables. It is also known as a “bivariate” statistic, with bi- meaning two and variate indicating variable
or variance. The two variables are usually a pair of scores for a person or object. The relationship
STATISTICS LECTURE NOTES P a g e | 29

between any two variables are can vary from strong to weak or none. When a relationship is strong, this
means that knowing a person's or object’s score on one variable helps to predict their score on the
second variable. In other words, if a person has a high score of variable A (compared to all the other
peoples’ scores on A, then they are likely to have a high score on variable B (compared to the other
peoples’ scores on B). The latter would be considered a strong positive correlation. If the correlation or
relationship between variable A and B is a weak one, then knowing a person's score on variable A does
not help to predict their score on variable B. One very nice feature of the correlation coefficient is that it
can only range from –1.00 to +1.00. Any values outside this range are invalid. Here is a graphicrepresentation
of correlation’s range. Note that the correlation coefficient is represented in a sample by
the value “r.”

When the correlation coefficient approaches r = +1.00 (or greater than r = +.50) it means there is a
strong positive relationship or high degree of relationship between the two variables. This also means
that the higher the score of a participant on one variable, the higher the score will be on the other
variable. Also, if a participant scores very low on one variable then their score will also be low on the
other variable. For example, there is a positive correlation between years of education and wealth.
Overall, the greater the number of years of education a person has, the greater their wealth. A strong
correlation between these two variables also means the lower the number of years of education, the
lower the wealth of that person. If the correlation was perfect one (r = +1.00), then there would be not a
single exception in the entire sample to increasing years of education and increasing wealth. It would
mean that there would be a perfect linear relationship between the two variables. However, perfect
relationships do not exist between two variables in the real world of statistical sampling. Thus, a strong
but not perfect relationship between education and wealth in the real world would mean that the
relationship holds for most people in the sample but there are some exceptions. In other words, some
highly educated people are not wealthy, and some uneducated people are wealthy.
When the correlation coefficient approaches r = -1.00 (or less than r = -.50), it means that there is a strong
negative relationship. This means that the higher the score of a person on one variable, the
lower the score will be on the other variable. For example, there might be a strong negative relationship
between the value of gold and the Dow Jones Industrial Average. In other words, when the value of
gold is high, the stock market will be lower and when the stock market is doing well, the value of gold
will be lower.
A correlation coefficient that is close to r = 0.00 (note that the typical correlation coefficient is
STATISTICS LECTURE NOTES P a g e | 30

reported to two decimal places) means knowing a person's score on one variable tells you nothing about
their score on the other variable. For example, there might be a zero correlation between the number of
letters in a person's last name and the number of miles they drive per day. If you know the number of
letters in a last name, it tells you nothing about how many miles they drive per day. There is no
relationship between the two variables; therefore, there is a zero correlation
The Four Common Types of Correlation
1. Pearson's r: A measure of the strength of a relationship between two continuous variables.
2. Spearman’s r: A measure of the similarity between two ordinal rankings of a single set of data.
3. Point-Biserial r: A measure of the strength of a relationship between one continuous variable
and one dichotomous variable (a two-level-only variable like gender).
4. Phi (φ) Correlation: A measure of the strength of a relationship between two dichotomous
variables.
The Karl Pearson Product-Moment Correlation Coefficient
Pearson’s coefficient r is obtained for a sample drawn from a population. The population value of
Pearson’s coefficient is called rho (ρ), and thus, r is an estimate of ρ.
The formula for r is as follows:

Illustration
A tobacco company statistician wishes to know whether heavy smoking is related to longevity.
From a sample of recently deceased smokers, the number of cigarettes (estimated on a per day for
their
last five years after visits with their surviving relatives) is paired with the number of years that they
lived.
STATISTICS LECTURE NOTES P a g e | 31

We will arbitrarily name one variable x and the other variable y. The results of Pearson’s r will be
exactly same no matter which variable is labeled x or y.
Step 1. First, obtain Σx, Σx2 (Σx)2, and (Σy)2
Σx = 25 + 35 + 10 + 40 + 85 + 75 + 60 + 45 + 50 = 425
Σx2 = 252 + 352 + 102 + 402 + 852 + 752 + 602 + 452 + 502 = 24,525
Σy = 63 + 68 + 72 + 62 + 65 + 46 + 51 + 60 + 55 = 518
Σy2 = 632 + 682 + 722 + 622 + 652 + 462 + 512 + 602 + 552 = 33,188
(Σx)2 = HINT: square Σx or (25 + 35 + 10 + 40 + 85 + 75 + 60 + 45 + 50)2
(Σx)2 = 4252 = 180,625
(Σy)2 = HINT: square Σy or (63 + 68 + 72 + 62 + 65 + 46 + 51 + 60 + 55)2
(Σy)2 = 5422 = 293,764
223
Step 2. Obtain the sum of the cross products (Σxy) by multiplying each x score by its paired y score.
Σxy = (25 x 63) + (35 x 68) + (10 x 72) + (40 x 62) + (85 x 65) + (75 x 46) + (60 x 51) + (45 x 60) +
(50 x 55)
Σxy = 1575 + 2380 + . . . + 2750
Σxy = 24,640
Step 3. Obtain the value of the numerator in the r formula. Remember, N is equal to the number of
pairs of scores. In this example, there are 9 pairs of scores.
NΣxy - (Σx) (Σy)
(9) (24,640) - (425) (518)
221,760 - 230,350 = -8,590
Step 4. Obtain the value of the denominator in the r formula. Remember that the square root is
obtained
after all other denominator values have been computed, simplified, and reduced to a single number.
STATISTICS LECTURE NOTES P a g e | 32

Step 5. Divide the result of Step 3 by the result of Step 4.

Note that the Pearson r is usually rounded off to two decimal places. Thus, r = -.61 means that there
is a strong negative correlation between smoking and longevity. This indicates that the higher the
number of cigarettes smoked in the past five years, the lower the number of years lived. And the
lower
the number of cigarettes, the higher the number of years lived. Remember, this relationship between
these two variables DOES NOT mean that heavy smoking causes one to live less. It may, however,
give
STATISTICS LECTURE NOTES P a g e | 33

clues as to further research ideas for experiments. In this case, an experiment might be set up
(perhaps
with animals) with an experimental group and a control group to determine whether cigarette
smoking
actually has a causal relationship with early morbidity.
Testing for the Significance of a Correlation Coefficient
A correlation coefficient may be tested to determine whether the coefficient significantly differs
from zero. The value r is obtained on a sample. The value rho (ρ) is the population's correlation
coefficient. It is hoped that r closely approximates rho. The null and alternative hypotheses are as
follows:
Ho: ρ = 0
Ha: ρ ≠ 0
The value of r and the number of pairs of scores are converted through a formula into a distribution
(similar to the z distribution) called the t distribution (in Appendix B, page 276 of the required text,
Statistics: A Gentle Introduction). The t formula can only be used to test whether r is equal to zero. It
cannot be used to test to see whether r might be equal to some number other than zero. It is also
important to note that the t distribution may be used to test other types of inferential statistics.
Therefore, if someone says that a t test is being used, it would be a legitimate question to ask
“why?”
The t distribution is most commonly used to test whether two means are significantly different,
however,
it may also be used to test the significance of the correlation coefficient. The t distribution also has
other
uses. Interestingly, the t distribution becomes z distribution when the data is infinite but they are
also
strikingly visually similar when there are only several hundred numbers in the set of data.
The t test formula in order to test the null hypothesis for a correlation coefficient is:
STATISTICS LECTURE NOTES P a g e | 34

Spearman's Correlation
Spearman's correlation coefficient (sometimes referred to as Spearman's rho or rs where the sub s
is in honor of Spearman) determines the degree of relationship for ranked data. Spearman's
correlation
244
is also called the rank-order correlation coefficient. Although Spearman's correlation is far less
common
than Pearson's r, occasionally variables are ordered according to rank (like 1st through 10th), or
variables may be subsequently ranked on the basis of a continuous variable. The formula for
Spearman's r:

where N = the number of pairs of scores.

Spearman's might be most typically used in situations where there are a number of variables and
they
are all ranked by two independent judges. For example, a grocery store owner wishes to know
whether
STATISTICS LECTURE NOTES P a g e | 35

married couples have similar tastes in vegetables. The members of the couple were independently
asked
to rate their preference for seven vegetables from most preferred (#1 rank) to least preferred (#7).
Their
data is as follows:

Note that the D score is the difference between the pairs of ranks on the first variable ranked, etc.
The number "6" in the formula is a constant and remains "6" regardless of the numbers of ranked
variables. N is the number of pairs of ranks (or the number of variables that are ranked).
In this example, N = 7.

Note that Pearson's r and Spearman's r are most typically reported to two decimal places.
STATISTICS LECTURE NOTES P a g e | 36

Spearman's r may be interpreted as a measure of the linear correlation between ranks. Pearson's r
will produce the same value as Spearman's r on the same set of ranked data. In the case where the
variables are expressed in their original form as continuous measures, the Pearson's r will not equal
the
Spearman's r after they have been converted to ranks, but they will have similar values.
Correlation can be graphically illustrated using scatter diagrams the numbers represent the values
of correlation coefficients

4.2 REGRESSION
- 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 regression techniques one can be able to determine the estimated
expenditure of a given family if the annual income is known and vice versa
- The general equation used in simple regression analysis is as follows
y = a + bx
Where y = Dependant variable
a= Interception y axis (constant)
b = Slope on the y axis
STATISTICS LECTURE NOTES P a g e | 37

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
STATISTICS LECTURE NOTES P a g e | 38

y x x Line of best fit

x x
x x
x x
x x
x x

x
The following sets of equations normally known as normal equation are used to determine the
equation of the above regression line when given a set of data.
Σy = an + bΣx
Σxy = aΣx + bΣx2
Where Σy = Sum of y values
Σxy = sum of the product of x and y
Σx = sum of x values
Σx2= sum of the squares of the x values
a = The intercept on the y axis
b = Slope gradient line of y on x
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.
Regression line of x on y i.e. x = a + by
- The fact that regression lines can only be used in one way leads to 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
STATISTICS LECTURE NOTES P a g e | 39

Piece of land (x)Acreage (y) Cost £ 000 xy x2

Hectares
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
Σx =44.0 Σy = 4400 Σxy= 25607 Σx2 = 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 hectares
ii. Estimate the expected average if the piece of land costs £ 900,000
Σy = an + bΣxy
Σxy = a∑x + bΣx2

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

4400 = 9a + 44b …….. (i)
25607 = 44a + 254.96b ……..(ii)
By multiplying equation …. (i) by 44 and equation …… (ii) by 9 we have
193600 = 396a + 1936b …….. (iii)
230463 = 396a + 2294.64b ……..(iv)
By subtraction of equation …. (iii) from equation …… (iv) we have
36863 = 358.64b
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
STATISTICS LECTURE NOTES P a g e | 40

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
b is the slope of the line or regression coefficient
n is the sample size
then,

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

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
3.5 16 56.0 12.25 256
2.4 13 31.0 5.76 169
4.9 19 93.1 24.01 361
4.2 18 75.6 17.64 324
3.0 12 36.0 9.0 144
1.3 11 14.3 1.69 121
1.0 8 8.0 1.0 64
3.0 14 42.0 9.0 196
1.5 9 13.5 2.25 81
4.1 16 65.6 16.81 256
STATISTICS LECTURE NOTES P a g e | 41

Σx = 28.9 Σy = 136 Σxy = 435.3 Σx2 = 99.41 Σy2= 1972

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)
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, that 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

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  ax b
Both of these can be reduced to linear model. Simple or multiple linear regression methods are then
used to determine the values of the coefficients

i. Exponential model
STATISTICS LECTURE NOTES P a g e | 42

y  ab x
Take log of both sides
log y = log a + log bx
log y = log a + xlog b
Let log y = Y and log a = A and log b = B

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

ii. Geometric model

y  ax b
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

TIME SERIES AND ANALYSIS

This is the mathematical or statistical analysis on past data arranged in a periodic sequence.
Decision making and planning in an organization involves forecasting which is one of the time
series analysis.

Impediments in time series analysis

Accuracy of data in reflecting
a) Drastic changes e.g. in the advent of a major competitor, period of war or sudden change of
taste.
b) For long term forecasting internal and external pressures makes historical data less effective.
STATISTICS LECTURE NOTES P a g e | 43

1. Moving Average
Periodical data e.g. monthly sales may have random fluctuation every month despite a general trend
being evident. Moving average helps in smoothing away these random changes.

A moving average is the forecast for a period that takes the average of the previous periods.

Example:
The table below represents company sales, calculate 3 and 6 monthly moving averages, for the data

Months Sales
January 1200
February 1280
March 1310
April 1270
May 1190
June 1290
July 1410
August 1360
September 1430
October 1280
November 1410
December 1390
Solution.
These are calculated as follows
Jan + Feb + Mar 1200 +1280 +1310
April’s forecast = 3 = 3
Feb + Mar + Apr 1280 +1310 +1270
May’s forecast = 3 = 3

And so on…

Similarly for 6 monthly moving average

Jan + Feb + Mar + Apr + May + Jun 1200 +1280 +1310 +1270 +1190 +1290
July forecast = 6 = 6
 STATISTICS LECTURE NOTES P a g e | 44

And so on…
3 months moving average 6 months moving average
April 1263
May 1287
June 1257
July 1250 1257
August 1297 1292
September 1353 1305
October 1400 1325
November 1357 1327
December 1373 1363

Note:
When plotting moving average on graphs the points are plotted as the midpoint of the period of the
average, e.g. in our example the forecast for April (1263) is plotted on mid Feb.
Characteristics of moving average
1) The more the number of periods in the moving average, the greater the smoothing effect.
2) Different moving averages produce different forecasts.
3) The more the randomness of data with underlying trend being constant then the more the periods
should be involved in the moving averages.

Limitations of moving averages.

1) Equal weighing with disregard to how more recent data is more relevant.
2) Moving average ignores data outside the period of the average thus it doesn’t fully
utilise available data.
3) Where there is an underlying seasonal variation, forecasting with unadjusted moving
average can be misleading.

2. Exponential smoothing
This is a weighted moving average technique, it is given by:
New forecast = Old forecast +  (Latest Observation – Old forecast)
Where  = Smoothing constant
This method involves automatic weighing of past data with weights that decrease exponentially with
time.

Example
Using the previous example and smoothing constant 0.3 generate monthly forecasts
STATISTICS LECTURE NOTES P a g e | 45

Months Sales Forecasts:  = 0.3

January 1200
February 1280 1200
March 1310 1224
April 1270 1250
May 1190 1256
June 1290 1233
July 1410 1250
August 1360 1283
September 1430 1327
October 1280 1358
November 1410 1335
December 1390 1357

Solution
Since there were no forecasts before January we take Jan to be the forecast for February.
 Feb – 1200
For March;
March forecast = Feb forecast + 0.3 ( Feb sales – Feb forecast)
= 1200 + 0.3 (1280 – 1200)
=1224

Note:
 The value  lies between 0 and 1.
 The higher the  value, the more the forecast is sensitive to the current status.

Characteristics of exponential smoothing

 More weight is given to the most recent data.
 All past data are incorporated unlike in moving averages.
 Less data is needed to be stored unlike in periodic moving averages.

Decomposition of time series

Time series has the following characteristics.
a) A long term trend (T) –tendency of the whole series to rise and fall.
STATISTICS LECTURE NOTES P a g e | 46

b) Seasonal variation (S) – short term periodic fluctuations in values. e.g. in Kenya
maize yield is high in November and low in March or matatus have better business
on Friday and very low on Sundays.
c) Cyclical variation (C) – These are medium term changes caused by factors which
apply for a while then disappear, and come back again in a repetitive cycle. e.g.
drought hits Kenya every 7 years.
Note that cyclic variation has a longer term than seasonal variation e.g. seasonal variation may occur
once every year while cyclic variation occurs once every several years.
d) Random residual variation (R) – These are non-recurring random variations e.g. war,
fire, coup e.t.c.

For accurate forecasts these aspects are qualified separately (i.e. T,C,S and R) from data. This is
known as time decomposition or time series analysis
The separate elements are then combined to produce a forecast.

Time series models:

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
separating out the trend and seasonal variation is known as deseasonalising the data.

There are two approaches to this process: one is based on regression through the actual data points
and the other calculates the regression line through moving average trend points. The method using
the actual data is demonstrated first followed by the moving average method.
STATISTICS LECTURE NOTES P a g e | 47

1. Time series analysis: trend and seasonal variation using regression on the data

The following data will be used to illustrate how the trend and seasonal variation are calculated.
Example 1

Sales of widgets in ‘000s

Quarter 1 Quarter 2 Quarter 3 Quarter 4
Year 1 20 32 62 29
2 21 42 75 31
3 23 39 77 48
4 27 39 92 53

It will be apparent that there is a strong seasonal element in the above data (low in Quarter 1 and
high in Quarter 3) and there is a generally upward trend.

The steps in analyzing the data and preparing a forecast are:

Step 1: Calculate the trend in the data using the least squares method.

Step 2: Estimate the sales for each quarter using the regression formula
established in step 1.

Step 3: Calculate the percentage variation of each quarter’s actual sales from the
estimates, obtained in step 2.

Step 4: Average the percentage variations from step 3. This establishes the
average seasonal variations.

Step 5: Prepare forecast based on trend percentage seasonal variations.

Solution

Step 1
Calculate the trend in the data by calculating the linear regression line y = a + bx.
STATISTICS LECTURE NOTES P a g e | 48

x (quarters) x (sales) xy x2
 1 20 20 1
2 32 64 4
Year 1 3 62 186 9
4 29 116 16

 5 21 105 25
6 42 252 36
Year 2 7 75 525 49
8 31 248 64

 9 23 207 81
10 39 390 100
Year 3 11 77 847 121
12 38 576 144

 13 27 351 169
14 39 546 196
Year 4 15 92 1380 225
16 53 848 256
x=136 y= 710 xy= 6661  x2 =1496

Least square equations

y = an + bx

xy = ax + bx2

710 = 16a + 136b

6661 = 136a + 1496b

626 = 340b

b = 1.84 and substituting we obtain

STATISTICS LECTURE NOTES P a g e | 49

a = 28.74

Trend line = 28.74 + 1.84x

Steps 2 and 3
Use the trend line to calculate the estimated sales for each quarter.

For example, the estimate for the first quarter in year 1 is

estimate = 28.74 + 1.84 (1) = 30.58

The actual value of sales is then expressed as a percentage of this estimate. For example, actual sales
in the first quarter were 20 so the seasonal variation is
Actual sales 20
%  65%
Estimate 30.58

x (quarters) y (sales) Trend Actual

%
Trend
 1 20 30.58 65
2 32 32.42 99
Year 1 3 62 34.26 181
4 29 36.10 80

 5 21 37.94 55
6 42 39.78 106
Year 2 7 75 41.62 180
8 31 43.46 71

 9 23 45.30 51
10 39 47.14 83
Year 3 11 77 48.98 157
12 48 50.82 94

 13 27 52.66 51
STATISTICS LECTURE NOTES P a g e | 50

14 39 54.50 72
Year 4 15 92 56.34 163
16 53 58.18 91

Trend estimates and percentage variations table.

Step 4

Average the percentage variations to find the average seasonal variations.

Q1 Q2 Q3 Q4
% % % %
65 99 181 80
55 106 180 71
51 83 157 94
51 72 163 91
222 360 681 336
4= 56% 90% 170% 84%

These then are the average variations expected from the trend for each of the quarters; for example,
on average the first quarter of each year will be 56% of the value of the trend. Because the variations
have been averaged, the amounts over 100% (Q3 in this example). This can be checked by adding
the average and verifying that they total 400% thus:

56% + 90% + 170% + 84% = 400%.

On occasions, roundings in the calculations will make slight adjustments necessary to the average
variations.

Step 5

Prepare final forecasts based on the trend line estimates from “trend estimates and percentages
variation table” (i.e. 30.58, 32.42, etc) and the averaged seasonal variations from the table above. (i.e.
56%, 90%, 170% and 84%)

The seasonally adjusted forecast is calculated thus:

Seasonally adjusted forecast = Trend estimate × Seasonal variation%

STATISTICS LECTURE NOTES P a g e | 51

X (quarters) Y (sales) Seasonally

adjusted
forecast
 1 20 17.12
Year 1 2 32 29.18
3 62 58.24
4 29 30.32

 5 21 21.24
Year 2 6 42 35.80
7 75 70.75
8 31 36.51

 9 23 25.37
Year 3 10 39 42.43
11 77 83.27
12 48 42.69

 13 27 29.49
Year 4 14 39 49.05
15 92 95.78
16 53 48.87

Seasonally adjusted forecasts

The forecasts are compared with the actual data to get some idea of how good extrapolated forecasts
might be. With further analysis they enable us to quantify the residual variations.

Extrapolation using the trend and seasonal factors

Once the formulae above have been calculated, they can be used to forecast (extrapolate) future
sales. If it is required to estimate the sales for the next year (i.e. Quarters 17, 18, 19 and 20 in our
series) this is done as follows:

Quarter 17 Basic trend = 28.74 + 1.84 (17)

= 60.02
 STATISTICS LECTURE NOTES P a g e | 52

Seasonal adjustment for a first quarter = 56%

Adjusted forecast = 60.02 × 56%
= 33.61

A similar process produces the following figures:

Adjusted forecasts Quarter 18 = 55.67

19 = 108.29
20 = 55.05

Notes:
a) Time series decomposition is not an adaptive forecasting system like moving averages and
exponential smoothing.
b) Forecasts produced by such an analysis should always be treated with caution. Changing conditions
and changing seasonal factors make long term forecasting a difficult task.
c) The above illustration has been an example of a multiplicative model. This is the seasonal variations
were expressed in percentage or proportionate terms. Similar steps would have been necessary if the
additive model had been used except that the variations from the trend would have been the
absolute values. For example, the first two variations would have been

Q1: 20 – 30.58 = absolute variation = -10.58

Q2: 32 – 32.42 = absolute variation = - 0.42
And so on.

The absolute variations would have been averaged in the normal way to find the average absolute
variation, whether + or -, and these values would have been used to make the final seasonally
adjusted forecasts.

2. Trend and seasonal variation using moving averages

When the correlation coefficient is low the method of calculating the regression line through the
actual data points should not be used. This is because the regression line is too sensitive to changes
in the data values.

In such circumstances, calculating a regression line through the moving average trend points is
more robust and stable.

Example 1 is reworked below using this method and, because there are many similarities to the
earlier method, only the key stages are shown.
STATISTICS LECTURE NOTES P a g e | 53

x y 3 point moving Trend line (2) Actual

average (1) %
Trend
1 20 34.38 58
2 32 38 35.70 90
3 62 41 37.02 167
4 29 37.3 38.34 76
5 21 30.7 39.66 53
6 42 46 40.98 102
7 75 49.3 42.30 177
8 31 43 43.62 71
9 23 31 44.94 51
10 39 46.3 46.26 84
11 77 54.7 47.58 162
12 48 50.7 48.90 98
13 27 38 50.22 54
14 39 52.7 51.54 76
15 92 61.3 52.86 174
16 53 54.18 98

Trend estimates and percentage variations utilizing moving averages

The first three moving average is calculated as follows

20  32  62
= 38 which is entered opposite period 2
3

The next calculated:

32  62  29
= 41, and so on
3
The regression line y = a + bx of the moving average values is calculated in the normal manner and
results in the following:
y = 33.06 + 1.32x

This is used to calculate the trend line:

STATISTICS LECTURE NOTES P a g e | 54

e.g. For Period 1:y = 33.06 + 1.32(1) = 34.38

For Period 2:y = 33.06 + 1.32 (2) = 35.70

The percentage variations are averaged as previously shown, resulting in the following values:

Q1 Q2 Q3 Q4
Average seasonal variation % 54 89 170 86

The trend line and the average seasonal variations are then used in a similar manner to that
previously described.

For example, to extrapolate future sales for the next year (i.e. quarters 17, 18, 19 and 20) is as follows:
Quarter 17
Forecast sales = (33.06 + 1.32(17)) × 0.54 = 29.97

A similar process produces the following figures:

Quarter 18 = 50.57
19 = 98.84
20 = 51.13

Forecast errors
Differences between actual results and predictions may arise from many reasons. They may arise
from random influences, normal sampling errors, choice of the wrong forecasting system or alpha
value or simply that the future conditions turn out to be radically different from the past. Whatever
the cause(s) management wish to know the extent of the forecast errors and various methods exist to
calculate these errors.

A commonly used technique, appropriate to time series, is to calculate the mean squared error of the
deviations between forecast and actual values then choose the forecasting system and/or parameters
which gives the lowest value of mean squared errors, i.e. akin to the ‘least squares’ method of
establishing a regression line.

Longer- term forecasting

Moving averages, exponential smoothing and decomposition methods tend to be used for short to
medium term forecasting. Longer term forecasting is usually less detailed and is normally concerned
with forecasting the main trends on a year to year basis. Any of the techniques of regression analysis
described in the preceding chapters could be used depending on the assumptions about linearity or
STATISTICS LECTURE NOTES P a g e | 55

non- linearity, the number of independent variables and so on. The least squares regression
approach is often used for trend forecasting.

Forecasting using least squares

Example 2
Data have been kept of sales over the last seven years

Year 1 2 3 4 5 6 7
Sales (in ‘000 units 14 17 15 23 18 22 27

It is required to forecast the sales for the 8th year

STATISTICS LECTURE NOTES P a g e | 56

Solution
Years (x) Sales (y) xy x2
1 14 14 1
2 17 34 4
3 15 45 9
4 23 92 16
5 18 90 25
6 22 132 36
7 27 189 49
x=28 y = 136 xy=596  x2= 140

136 = 7a + 28b

596 = 28a + 140b

 b = 1.86
And substituting in one of the equations we obtain
a = 12
 Regression line = y = 12 + 1.86x

Or, Sales in (‘000s of units) = 12.00 + 1.86 (no of years)

We use this expression for forecasting, for 8th year sales = 12 + 1.86 (8)
=26.88 i.e. 26,888 units
Applications: The usage of time series models is twofold:
 Obtain an understanding of the underlying forces and structure that produced the observed
data
 Fit a model and proceed to forecasting, monitoring or even feedback and feedforward
control.
Time Series Analysis is used for many applications such as:
 Economic Forecasting
 Sales Forecasting
 Budgetary Analysis
 Stock Market Analysis
 Yield Projections
 Process and Quality Control
 Inventory Studies
STATISTICS LECTURE NOTES P a g e | 57

 Workload Projections
 Utility Studies
 Census Analysis

PROJECT/INVESTMENT APPRAISAL TECHNIQUES

Interest is defined as the cost of borrowing money as in the case of interest charged on a loan
balance. Conversely, interest can also be the rate paid for money on deposit as in the case of a
certificate of deposit. Interest can be calculated in two ways, simple interest or compound interest.
 Simple interest is calculated on the principal, or original, amount of a loan.
 Compound interest is calculated on the principal amount and also on the accumulated
interest of previous periods, and can thus be regarded as "interest on interest."
There can be a big difference in the amount of interest payable on a loan if interest is calculated on a
compound rather than simple basis. On the positive side, the magic of compounding can work to
your advantage when it comes to your investments and can be a potent factor in wealth creation.
While simple interest and compound interest are basic financial concepts, becoming thoroughly
familiar with them may help you make more informed decisions when taking out a loan or investing.
Simple Interest Formula
The formula for calculating simple interest is:
Simplle Interest
Ammount =Principal +Simple interest
Thus, if simple interest is charged at 5% on a $10,000 loan that is taken out for three years, the total
amount of interest payable by the borrower is calculated as $10,000 x 0.05 x 3 = $1,500.
Interest on this loan is payable at $500 annually, or $1,500 over the three-year loan term.
Compound Interest = Total amount of Principal and Interest in future
Compound Interest =Ammount -Principal
Ammount =
Compound Interest is the effect of ‘interest on interest’ and happens when interest is added to the
capital or principal sum so that from that moment on the interest which has been added also itself
earns interest. If you borrow £10,000 on a credit card at 1% per month after one month you owe
£10,100 but at the end of 2 months £10210. After 10 years with no repayments you would owe over
£33000 (not the £22,000 you may expect on simple annual interest of 10 years times 12% =£12000
interest plus original principal) This addition of interest to the principal is called compounding (i.e.
interest is compounded).
The Positives about Compound Interest
If you have money on deposit earning regular interest your capital grows reasonably quickly.
To demonstrate if you deposit £14,000 in a retirement fund for 45 years at 10% you will retire with
over a million pounds. Do you remember the story of a grain of rice on each chess board square
being double the last square until there isn’t enough rice grains in the world to cover the 64th square
– well compound interest is similar.
STATISTICS LECTURE NOTES P a g e | 58

If you understand compound interest you can make better comparisons and judgements.
Some mortgages compound every day some every year – which would you choose? Once a year in
arrears for preference but in advance at 12% is better than compounding daily that would compound
to 12.75% (excluding repayments).
Compound interest is usually the only game in town so take care out there
The Negative of Compound Interest
 The cost is disguised and can run away with your money.
 Missing a payment by a day may mean interest falls due to be calculated before the payment
is recorded. Time your monthly payments and try to stop them slipping.
 Compound interest is designed to help lenders. Credit card monthly repayments are usually
set so you are encouraged to keep borrowing and thus keep paying interest. Try eat into the
capital owed by repaying interest plus some capital every time.

Therefore Compounding method is used to know the future value of present money.
Conversely, discounting is a way to compute the present value of future money.
the formulas used are

FV = PV (1 + r)^n for compounding PV = FV / (1 + r)^n for discounting

Usually calculations involving compounding and discounting can be worked out using tables of
present value(or future value) interest factor.
Example 1-1:
Assume you put 20,000 dollars (principal) in a bank for the interest rate of 4%. How much money
will the bank give you after 10 years?
FV=
So the bank will pay you 29604.8 after 10 years (the future value of 20,000 invested at 4% interest rate
for 10 years) note that from the table of future values interest factor we multiply 20000by
1.4802=29604
STATISTICS LECTURE NOTES P a g e | 59

Example 1-2:discounting
Assuming the discount rate of 10 %, present value of 10000 dollars which will be received in 5 years
from now can be calculated as:
Future Value = $10000
Present Value =
From the table of present value interest factor at 10% for 5 years is 0.6209 therefore the PV= FV*
present Value interest factor =10000*0. 6209= dollars

The investment appraisal techniques can be categorised into two groups:

(a) Discounted Cashflow methods

i. Net present value method

ii. Internal rate of return
iii. Profitability index

(b) Non-discounted cashflow method

i. Accounting rate of return

ii. Payback period

DISCOUNTED CASHFLOW METHODS

1. Net Present Value (NPV)

STATISTICS LECTURE NOTES P a g e | 60

NPV analysis is a form of intrinsic valuation and is used extensively across finance and accounting
for determining the value of a business, investment security, capital project, new venture, cost
reduction program, and anything that involves cash flow
This is defined mathematically as the present value of cashflow less the initial
outflow.
n
Ct
NPV =  -
t Io
t=1 (1 + K )

Where Ct is the cashflow

K is the opportunity cost of capital
Io is the initial cash outflow
n is the useful life of the project

Decision Rule using NPV

The decision rule under NPV is to:

- Accept the project if the NPV is positive

- Reject the project if NPV is negative

Note: if the NPV = 0, use other methods to make the decision.

Example of Net Present Value (NPV)
Consider a project whose initial cost is $60000 that is expected to have an annual return of $10000 per
year for 10 years. Calculate its NPV at the discount rate required of 10% and determine whether to
accept or reject the project
Solution
irst calculate the net present value of a series of cash inflows(in the screenshot below), the
assumption is that an investment will return $10,000 per year over a period of 10 years

NPV = present value of all cash inflow - the initial cost = 61446-60000 =1446
STATISTICS LECTURE NOTES P a g e | 61

Since NPV >0 (NPV =1446) we Accept the project as the NPV is positive
2. Internal Rate of Return (IRR)
The internal rate of return of a project is that rate of return at which the projects NPV = 0

Therefore IRR occurs where:

n
Ct
NPV =  - =0
t Io
t=1 (1 + r )

Where r = internal rate of return

Note that IRR is that ratio of return that causes the present value of cashflows to be
equal to the initial cash outflow.

Decision Rule under IRR

If IRR > opportunity cost of capital - accept the project

- IRR < opportunity cost of capital - reject the project

- IRR = opportunity cost of capital - be indifferent

Usuallly irr is calculated using iteration(trial and error) method
Compute the NPV of the project using an arbitrary selected discount rate
If the NPV so computed is positive then try a higher rate and if negative try a lower rate.
Continue this process until the NPV of the project is equal to zero
Use linear interpolation to determine the exact rate

where
rn and rn-1​ are randomly selected discount rates
NPVn=higher net present valueNPVn-1=lower net present value​

Illustration:A project has the following cash flows

Year 1 2 3 4
STATISTICS LECTURE NOTES P a g e | 62

Cash inflow 300 400 700 900

The cost of the project is 1500 Br.Determine whether project is acceptable if the cost of capital is 18%
using the IRR method
PV at
year cf PVIF at 15% PV at 15% PVIF at 20% 20%
1 300 0.8696 260.88 0.8333 249.99
2 400 0.7561 302.44 0.6944 277.76
3 700 0.6575 460.25 0.5787 405.09
4 900 0.5718 514.62 0.4823 434.07
total 1538.19 total 1366.91
NPV 38.19 NPV -133.09
Take an arbitrary interest rate (say 15% or lower)and calculate npv
Since, NPV at 15% is positive but not large; we select a slightly higher rate, say, 20% and calculate
npv until npv becomes -ve
Since NPV at 20% is negative, IRR therefore lies between 15% and 20%, and since zero NPV willthe
between -133.09and 38.19, to get the correct (exact) IRR we have to interpolate between 15%and 18%
using interpolation formula

Therefore IRR< acceptable cost of capital

Decision: Reject the project since IRR is less than the required rate of return (cost of capital
3. Profitability Index

This is a relative measure of projects profitability. It is given by the following

formula.

 (1+CK )
t=1
t
t
PI =
Io

Decision Rule

If PI > 1 - Accept the project

PI < 1 - Reject the project
PI = 1 - Be indifferent
Example of Profitability Index
Company A is considering two projects:
STATISTICS LECTURE NOTES P a g e | 63

Project A requires an initial investment of $1,500,000 to yield estimated annual cash flows of:
 $150,000 in Year 1
 $300,000 in Year 2
 $500,000 in Year 3
 $200,000 in Year 4
 $600,000 in Year 5
 $500,000 in Year 6
 $100,000 in Year 7
The appropriate discount rate for this project is 10%.

Project B requires an initial investment of $3,000,000 to yield estimated annual cash flows of:
 $100,000 in Year 1
 $500,000 in Year 2
 $1,000,000 in Year 3
 $1,500,000 in Year 4
 $200,000 in Year 5
 $500,000 in Year 6
 $1,000,000 in Year 7

The appropriate discount rate for this project is 13%.

Company A is only able to undertake one project. Using the profitability index method, which
project should the company undertake?
STATISTICS LECTURE NOTES P a g e | 64

Using the PI formula, Company A should do Project A. Project A creates value – Every $1 invested in
the project generates $.0684 in additional value.

Discounting the Cash Flows of Project A:

 $150,000 / (1.10) = $136,363.64
 $300,000 / (1.10)^2 = $247,933.88
 $500,000 / (1.10)^3 = $375,657.40
 $200,000 / (1.10)^4 = $136,602.69
 $600,000 / (1.10)^5 = $372,552.79
 $500,000 / (1.10)^6 = $282,236.97
 $100,000 / (1.10)^7 = $51,315.81
Present value of future cash flows:
$136,363.64 + $247,933.88 + $375,657.40 + $136,602.69 + $372,552.79 + $282,236.97 + $51,315.81 =
$1,602,663.18
Profitability index of Project A: $1,602,663.18 / $1,500,000 = $1.0684. Project A creates value.

Discounting the Cash Flows of Project B:

 $100,000 / (1.13) = $88,495.58
 $500,000 / (1.13)^2 = $391,573.34
 $1,000,000 / (1.13)^3 = $693,050.16
 $1,500,000 / (1.13)^4 = $919,978.09
 $200,000 / (1.13)^5 = $108,551.99
 $500,000 / (1.13)^6 = $240,159.26
 $1,000,000 / (1.13)^7 = $425,060.64
Present value of future cash flows:
$88,495.58 + $391,573.34 + $693,050.16 + $919,978.09 + $108,551.99 + $240,159.26 + $425,060.64 =
$2,866,869.07
Profitability index of Project B: $2,866,869.07 / $3,000,000 = $0.96. Project B destroys value.

NON-DISCOUNTED CASHFLOW METHODS

1. Accounting rate of return (ARR)

ARR = Average annual income

Average investment
STATISTICS LECTURE NOTES P a g e | 65

Where Average annual income = Average cashflows - Average Depreciation

Average investment = 1/2 (Cost of investment - Salvage value)
(assuming straight line depreciation method).
Projects with higher ARR are preferable.

2. Payback Period

This is defined as the time taken by the project to recoup the initial cash outlay.
The decision rule depends on the firms target payback period (i.e. the maximum
period beyond which the project should not be accepted.
Consider a project whos initial cost is $ 5 million. The expected annual cash inflows are as illustrated
below determine the payback period

From the data above, we can see that project investment is being recovered in the
4th year. So the formula for the payback period would be:

NETWORK ANALYSIS
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.
STATISTICS LECTURE NOTES P a g e | 66

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

Event (node). This is a point in time and it indicates the start or finish of an activity e.g in building a
bridge, rails installed. It is represented by a circle.

Dummy activity. An activity that doesn’t consume time or resources, it is merely to show logical
dependencies between activities so as abide by rules of drawing a network, it is represented by a
dotted arrow

Network. This is a combination of activities and events (including dummy activities)

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…
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

Loop
Dangling activity
 STATISTICS LECTURE NOTES P a g e | 67

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)

B- Break pads Changed

These events are represented as;

CA CR
A

Example of a network.

Activities
1-2 - where 1 is the preceding event where as 2 is the succeeding event of the activity
1-3
2-4
2-5
3-5
4-5
4-6
5-6
6-7

4
2

6 7
1

3 5
STATISTICS LECTURE NOTES P a g e | 68

8.4 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 multiple time
estimates are three estimates for each activity , i.e. optimistic (O), Most
Likely (ML), and Pessimistic (P). These three estimates are combined to
give an expected time and the accepted time formula is:

O  P  4ML
Expected time = 6
For example assume that the three estimates for an activity are
Optimistic 11 days
Most likely 15 days
Pessimistic 18 days
11  18  4 15 
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

STATISTICS LECTURE NOTES P a g e | 69

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 longest duration times. There may be more than one critical path
which may run through a dummy
Example
Consider a construction project with the following activity timings drawn its network and determine
its critical path
Activity Preceeding Duration
activity (WEEKS)
A - 4
B - 6
C B 4
D A 12
E A,C 7
F B 9
G E,F 5
END D,G 0

A scheduling method
The network analysis is a method used to analyze, control and monitoring of business processes and
workflows. Contrary to the work breakdown structure, a network diagram also considers the
STATISTICS LECTURE NOTES P a g e | 70

chronological order of activities, milestones and tasks, their durations and dependencies and
visualizes them graphically or as a table, e.g. in a Gantt chart.
The network analysis enables project managers to take various factors into account when creating
a project plan:
 Dependencies between activities
 Buffer times between activities
 Earliest and latest start and end dates
 Duration of activities
 Critical Path
The network analysis method is often used in procurement and production in order to control
project processes more efficiently and to complete projects on schedule and on budget.
Managerial applications of network analysis are as follows:
1. Assembly line scheduling,
2. Research and development,
3. Inventory planning and control,
4. Shifting of manufacturing plant from one site to another,
5. Launching of new products and advertising campaigns,
6. Control of traffic flow in cities,
7. Budget and audit procedures,
8. Launching space programmes,
9. Installation of new equipments
10. Long-range planning and developing staffing plans, etc.
Network techniques:
A number of network techniques, given below have been developed in recent times:
1. PERT- Programme Evaluation and Review Technique
2. CPM- Critical Path Method
:
3. RAMS- Resource Allocation and Multi-project Scheduling
4. PEP- Programme Evolution Procedure
5. COPAC- Critical Operating Production Allocation Control
6. MAP- Manpower Allocation Procedure
:
7. RPSM- Resource Planning and Scheduling Method
8. LCS- Least Cost Scheduling
9. MOSS- Multi-Operation Scheduling System
10. PCS- Project Control System
STATISTICS LECTURE NOTES P a g e | 71

:
11. GERT- Graphical Evaluation Review Technique.

INVENTORY CONTROL
Inventory control, also known as stock control, is regulating and maximizing your company’s
warehouse inventory.
The objectives of inventory management are
 To ensure adequate stocks to allow for continuous operations/production, and
 To minimize the cost of having inventory.

Inventory management is important since in most organizations it represents the largest single
investment. The major types of inventory are:
 Raw materials
 Work in progress
 Finished goods

To achieve the above objectives of material cost costing , the manger has to make decisions
regarding the following:
a) What commodities to stock?
 Use Material Requirement Planning
From the Master Production Schedule, the manager has determined the products to be produced. A
Bill of Materials can then be prepared. This lists in descending order the components required to
make the final product. The information required includes part name or description, part number,
next higher level assembly, required quality per end item, quantity per end item and quantity
required for the next higher level assembly.

Stores Ledger Account is also used to obtain information on what is currently available. The file
shows balance on hand as well as past data on how much is usually ordered, lead-time, and safety
stock.
From the above the manager can determine what need to be purchased.
b) How much to stock?
 Use The Economic Order Quantity Model
This is a simple model that helps the manager to determine the optimum quantity of stock to order
so as to keep total costs at a minimum. The main costs of inventory are:
 Holding or carrying costs
 Ordering or set up costs
 Shortage costs
STATISTICS LECTURE NOTES P a g e | 72

To determine the economic order quantity the following formula may be used:

EOQ  2DCO
Ch

Where D is the annual demand (knits)

Co is the cost of making one order
Ch is the holding cost per unit per annum

 Use Pareto Analysis

Items are classified into three classes as follows:
Class A: These are high cost, fast moving and high usage items. They are few accounting for only
20 percent of the total number of items yet account for 80 percent of the total inventory budget.
Class B: These are medium moving goods. They account for 15 percent of the total number of the
budget
Class C: These are slow moving low value items. They are very many accounting for 65 percent of
the total number of items and only 5 percent of the total inventory budget.
c) When to stock?
This will be influence by the inventory system in place as follows:
 Periodic order system.
The firm receives a new order of the amount specified by the order quantity at equal intervals of
time. The firm determines the maximum and minimum inventory, the safety stock and the reorder
level.
 Continuous Review System
The firm places orders at regular intervals but the order quantity varies according to how much a
firm requires to bring the level to some predetermined size or value.
 Just In Time Inventory System
This is a concept developed by the Japanese and advocates zero inventory and stockless production.
In addition, it calls for 100 per cent quality. Some of the major features of JIT include:
a) Frequent and reliable deliveries to avoid inventory build up. Companies are also setting
delivery dates with penalties for not meeting them.
b) Closer location to suppliers and customers
c) Improved communication between companies and suppliers through the use of
computerized purchasing systems that allows for online ordering.
d) Single sourcing and building long-term relations with a few trusted suppliers.
STATISTICS LECTURE NOTES P a g e | 73

e) Increased supplier involvement in the design aspects of a product to ensure that they
meet the company’s quality requirements.
f) Maintenance of strict quality control by all parties.

Inventory control systems

Inventory control systems are technology solutions that integrate all aspects of an organization’s
inventory tasks, including shipping, purchasing, receiving, warehouse storage, turnover, tracking,
and reordering. While there is some debate about the differences between inventory management
and inventory control, the truth is that a good inventory control system does it all by taking a holistic
approach to inventory and empowering organizations to utilize lean practices to optimize
productivity and efficiency along the supply chain while having the right inventory at the right
locations to meet customer expectations.
That being said, there are two different types of inventory control systems available today:
perpetual inventory systems and periodic inventory systems. Within those systems, two main types
of inventory management systems – barcode systems and radio frequency identification (RFID)
systems – used to support the overall inventory control process:
 Main Inventory Control System Types:
 Perpetual Inventory System
 Periodic Inventory System
 Types of Inventory Management Systems within Inventory Control Systems:
 Barcode System
 Radio Frequency Identification (RFID) System
Inventory control systems help you track inventory and provide you with the data you need to
control and manage it. No matter which type of inventory control system you choose, make sure that
it includes a system for identifying inventory items and their information including barcode labels or
asset tags; hardware tools for scanning barcode labels or RFID tags; a central database for all
inventory in addition to the ability to analyze data, generate reports, and forecast demand; and
processes for labeling, documenting, and reporting inventory along with a proven inventory
methodology like just-in-time, ABC analysis, first-in, or first out (FIFO), or last-in-first-out (LIFO).
Read on to learn more about the types of inventory and the types of inventory control systems and
inventory management apps that can help companies more efficiently manage their inventory.
What Are the 4 Types of Inventory?
Before getting into details about the types of inventory control systems, it’s important to understand
the different types of inventory.
Generally, inventory can be grouped into four primary classifications:
 Raw materials – Raw materials are inventory items used in the manufacturing process to
create finished goods. What is considered a raw material to one company may be considered
finished goods to another. For example, a company that creates parts or components for
machinery or equipment would consider those components finished goods. A manufacturer
that purchases those components for use in their manufacturing process would consider the
same components raw materials. Raw materials may consist of things like paper or steel,
nuts and bolts, chemicals, wheels, and other items.
STATISTICS LECTURE NOTES P a g e | 74

 Work-in-progress – Work-in-progress (WIP) inventory includes items that are currently

being processed. WIP inventory can include raw materials and components that are going
through the manufacturing process to produce finished goods as well as finished items that
are waiting for final inspection or quality control. After those final steps are complete, these
finished items would be considered finished goods.
 Finished goods – Finished goods are comprised of all completed items that are ready for sale
to the final customer.
 MRO goods – MRO stands for maintenance, repair, and operating supplies. MRO inventory
consists of items necessary to operate, such as equipment and machinery, and the items
needed for maintaining equipment and infrastructure. That means MRO inventory can also
include items that are sometimes considered raw materials but in this case are essentially
spare parts. Nuts and bolts are a good example. When nuts and bolts are on hand to
assemble finished products, they’d be classified as raw materials. Extra nuts and bolts a
company keeps in storage to repair equipment, on the other hand, are classified as MRO.
Other examples of MRO inventory include janitorial supplies such as cleaning solutions,
mops, and brooms, tools, packaging materials, uniforms and gloves, and office supplies such
as paper, pens, calculators, printer ink, and other items.
Inventory can be further classified in several ways depending on the industry, the company’s
operations, and the types of inventory the company manages. Companies that purchase finished
goods and sell them to customers at a markup have just one type of inventory called merchandising
inventory.
Some companies, such as manufacturers, need to manage a variety of inventory in different
classifications, making efficient inventory tracking a must. To effectively manage inventory, an
inventory tracking solution is paired with an inventory control app or inventory management app.
How Do Inventory Control Systems Work?
Inventory control systems, such as inventory control apps, offer a variety of functions that help
companies manage various types of inventory. Inventory control systems typically consist
of inventory management apps paired with barcode tagging to identify inventory assets, and
information about each item is stored in a central database. Barcode labels serve as inventory
trackers, allowing users to bring up information about the item on a computer system, such as the
item’s price, the number of items in stock, the location of an item within a warehouse, and more.
The best inventory control apps are mobile-compatible, with companion apps that allow users to
track and manage inventory while they move throughout a facility or from site to site. There are
many inventory tracking apps for smartphones, some of which are mobile-exclusive, while others
have desktop applications to allow users to track inventory from any device. There are also many
inventory tracking apps designed specifically to meet the needs of warehouse managers. When
looking for an inventory management app, look for features that accommodate your company’s
needs, such as trigger alerts when inventory levels reach pre-defined thresholds, re-ordering
capabilities, and analysis and reporting to support functions such as forecasting.
The 2 Types of Inventory Control Systems
Now that we’ve covered the basics of inventory and how inventory control systems work in general,
let’s discuss the two main types of inventory control systems.
Perpetual Inventory System
When you use a perpetual inventory system, it continually updates inventory records and accounts
for additions and subtractions when inventory items are received, sold from stock, moved from one
location to another, picked from inventory, and scrapped. Some organizations prefer perpetual
inventory systems because they deliver up-to-date inventory information and better handle minimal
STATISTICS LECTURE NOTES P a g e | 75

physical inventory counts. Perpetual inventory systems also are preferred for inventory tracking
because they deliver accurate results on a continual basis when managed properly. This type of
inventory control system works best when used in conjunction with a database of inventory
quantities and bin locations updated in real time by warehouse workers using barcode scanners.
Inventory management apps are perpetual inventory systems.
There are some challenges associated with perpetual inventory systems. First, these systems cannot
be maintained manually and require specialized equipment and software that results in a higher cost
of implementation, especially for businesses with multiple locations or warehouses. Periodic
maintenance and upgrades are necessary for perpetual inventory systems, which also can become
costly. Another challenge of using a perpetual inventory system is that recorded inventory may not
reflect actual inventory as time goes by because they do not conduct periodic physical inventory
counts, a necessary activity even when inventory trackers are used. The result is that errors, stolen
items, and improperly scanned items impact the recorded inventory records and cause them not to
match actual inventory counts.

Periodic Inventory System

Periodic inventory systems do not track inventory on a daily basis; rather, they allow organizations to
know the beginning and ending inventory levels during a certain period of time. These types of
inventory control systems track inventory using physical inventory counts. When physical inventory
is complete, the balance in the purchases account shifts into the inventory account and is adjusted to
match the cost of the ending inventory. Organizations may choose whether to calculate the cost of
ending inventory using LIFO or FIFO inventory accounting methods or another method; keep in
mind that beginning inventory is the previous period’s ending inventory.
There are a few disadvantages of using a periodic inventory system. First, when physical inventory
counts are being completed, normal business activities nearly become suspended. As a result,
workers may hurry through their physical counts because of time constraints. Periodic inventory
systems typically don’t use inventory trackers, so errors and fraud may be more prevalent because
there is no continuous control over inventory. It also becomes more difficult to identify where
discrepancies in inventory counts occur when using a periodic inventory control system because so
much time passes between counts. The amount of labor that is required for periodic inventory
control systems make them better suited to smaller businesses.

Barcode Inventory Systems

Inventory management systems using barcode technology are more accurate and efficient than
those using manual processes. When used as part of an overall inventory control system, barcode
systems update inventory levels automatically when workers scan them with a barcode scanner or
mobile device. The benefits of using barcoding in your inventory management processes
are numerous and include:
 Accurate records of all inventory transactions
 Eliminating time-consuming data errors that occur frequently with manual or paper systems
 Eliminating manual data entry mistakes
 Ease and speed of scanning
 Updates on-hand inventory automatically
 Record transaction histories and easily determine minimum levels and reorder quantities
 Streamline documentation and reporting
STATISTICS LECTURE NOTES P a g e | 76

 Rapid return on investment (ROI)

 Facilitate the movement of inventory within warehouses and between multiple locations and
from receiving to picking, packing, and shipping
Radio Frequency Identification (RFID) Inventory Systems
Radio frequency identification (RFID) inventory systems use active and passive technology to
manage inventory movements. Active RFID technology uses fixed tag readers throughout the
warehouse; RFID tags pass the reader, and the movement is recorded in the inventory management
software. For this reason, active systems work best for organizations that require real-time inventory
tracking or where inventory security has been an issue. Passive RFID technology, on the other hand,
requires the use of handheld readers to monitor inventory movement. When a tag is read, the data is
recorded by the inventory management software. RFID technology has a reading range of
approximately 40 feet with passive technology and 300 feet with active technology.
RFID inventory management systems have some associated challenges. First, RFID tags are far
more expensive than barcode labels; thus, they typically are used for higher value goods. RFID tags
also have been known to have interference issues, especially when tags are used in environments
with a lot of metal or liquids. It also costs a great deal to transition to RFID equipment, and your
suppliers, customers, and transportation companies need to have the required equipment as well.
Additionally, RFID tags carry more data than barcode labels, which means your system and servers
can become bogged down with too much information.
When choosing an inventory control system for your organization, you first should decide whether a
perpetual inventory system or periodic inventory system is best suited to your needs. Then, choose a
barcode system or RFID system to use in conjunction with your inventory control system for a
complete solution that will enable you to have visibility into your inventory for improved accuracy in
scanning, tracking, recording, and reporting inventory movement.
Stock Level and its control
Management must make decisions about the control of stock levels with a view to minimizing the
cost of the company while achieving more efficiency in the availability of material to fulfill planned
usage requirements. Consideration should be given to the following control levels:
a) Minimum stock level
b) Maximum stock level
c) Re-order level
d) Re order quantity (Note the re-order quantity is not necessary the EOQ)

a) Minimum stock level

This is the level below which stock should not fall. It is essentially a base (buffer) stock level. If
stock falls below this point, there is a danger of stock out.
Minimum stock level = Reorder level – (Normal consumption x normal reorder period)
b) Maximum stock level
This is the upper limit above which stock should not be allowed to rise. Each material to be kept in
store must have a maximum level and stock should not be allowed to go beyond this level
Maximum stock level = Re-order level +re-order Quantity - (Minimum consumption x minimum re-
order period)
c) Re-order level
STATISTICS LECTURE NOTES P a g e | 77

Is a point that lies between minimum and maximum stock levels at which purchase orders must be
placed to ensure that goods ordered are received before the minimum stock level is reached? It is the
level of stocks at which replenishment must be made to avoid a stock-out.
Re-order level = maximum consumption X maximum re-order period
STATISTICS LECTURE NOTES P a g e | 78

d) Re-Order quantity
This is the quantity of stock ordered once the re-order point is reached. The quantity is such as to
minimize stock costs taking into consideration the cost of holding stocks and making an order. This
is also regarded as the Economic Order Quantity (EOQ). It is computed as follows:
Where D is the annual demand (knits)
Co is the cost of making one order
Ch is the holding cost per unit per annum

EOQ  2DCO
Ch

Economic Order Quality (EOQ)

Define the EOQ model and the three methods of computing EOQ.
- Assumptions of the model.

Illustration

The following information was extracted from the books of Danex Holdings regarding its stocks:
i. Reorder quantity 1,800
ii. Reorder period 4 weeks
iii. Maximum consumption 450 units/week
iv. Normal consumption 300 units/week
v. Minimum consumption 150 units/week
Vi Maximum reorder period 5 weeks
Vii Minimum reorder period 3 weeks

Required
Determine the following stock levels for Danex Holdings:
i. Re-order level
ii. Maximum stock level
iii. Minimum stock level
iv. SAFETY STOCK
STATISTICS LECTURE NOTES P a g e | 79

Solution
i) Re-order level = Maximum consumption X maximum reorder period
= 450 units X 5 weeks = 2,250 units
ii) Maximum stock level = reorder level + reorder quantity-
(Minimum consumption X minimum reorder period)
= 2250 + 1800 – (150 X3) = 4050 – 450 = 3600 units
iii) Minimum stock level = Reorder level – (Normal consumption X
Normal reorder period)
= 2,250 – (300 X 4) = 2250 – 1200 = 1050 units

Safety stock= Safety stock = (Maximum daily usage * Maximum lead time in days) – (Average daily
usage * Average lead time in days). = (450 X 5) – (300 X 4) =
Economic Order Quantity (EOQ):
It constitutes the quantity purchased of either stocks or raw materials that is considered most
optimum. This is the quantity that minimizes both holding costs and ordering costs, As the quantity
of purchase increases there is a reduction in ordering costs, but an increase in holding costs as
illustrated in the graph below:

Qx represents the EOQ where the aggregate stock cost is lowest

Total cost = Total ordering costs + Total holding costs
Total ordering costs = cost per order X no of orders in a period
Total holding cost = average stock quantity X holding cost per unit

The EOQ model assumes:

STATISTICS LECTURE NOTES P a g e | 80

 Annual demand is known

 Hold costs are constant and known
 Ordering costs are known and constant
 The same quantity is ordered every time an order is made since demand as assumed
not to fluctuate significantly.
Example
ABC Ltd has an aggregate demand of 1.2 Million units. Each time they place an order there is an
ordering cost of shs 1,000, holding cost is shs 100 per unit. Determine:
i. EOQ
ii. No. of order to be made based EOQ
iii. Total cost of stocks based on the EOQ

Solution
2DCo  2 X 1200 000X1000  4899 units
EOQ = Ch 100

i) No of order = 1200000 = 244.9 ≈ 245 Orders

4899
ii) Total cost = DCo + ½ QCh = 1200000(1000) + ½ (4899)100 = 489,900
Q 4899

Descriptive Statistics and Index Numbers

Contents
- Application of statistics
- Measures of centra tendency
i. Arithmetic mean
ii. Median
iii. Mode
iv. Geometric mean
v. Harmonic mean
- Measures of dispersion
i. Simple range
ii. Quartile deviation
iii. Mean deviation
STATISTICS LECTURE NOTES P a g e | 81

iv. Standard deviation

v. Coefficient of mean deviation
vi. Coefficient of quartile deviation
- Skew ness and Kurtosis
- Indices
STATISTICS LECTURE NOTES P a g e | 82

3.1 Descriptive Statistics

a) Statistics
Definition: Statistics viewed as a subject is a process of collecting, tabulating and analyzing numerical data
upon which significant conclusions are drawn.
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 from the ministry of planning, Agricultural statistics from the
ministry of Agriculture
Statistics may also refer to the values, which have been obtained from statistical calculations e.g. the mean,
mode, range e.t.c.

b) Application of statistics
1. Quality Control
Usually there is a quality control departments in every industry which is charged with the responsibility of
ensuring that the products made do meet the customers standards e.g. the Kenya bureau of standards (KeBS)
is one of the national institutions which on behalf of the government inspects the 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 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 is due to frustration or by choice.

3.2 Measures of Central Tendency

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. These measures are as follows:
i. The arithmetic mean
ii. The mode
iii. The median
iv. The geometric mean
v. The harmonic mean

1. The arithmetic mean

STATISTICS LECTURE NOTES P a g e | 83

This is commonly known as average or mean it is obtained by first of all 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
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 cells. A sample of electronic cells was
taken and the life spans were recorded as shown in the following table.

Life span hrs No. of cells (f) Class MP(x) X–A=d fd

1600 – 1799 25 1699.5 -600 -15000
1800 – 1999 32 1899.5 -400 -12800
2000-2199 46 2099.5 -200 -9200
2200 – 2399 58 2299.5(A) 0 0
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 = 2299.5 + 
12800
f 238
= 2299.5 +-53.78

= 2245.72 hours
STATISTICS LECTURE NOTES P a g e | 84

Example 2 – (use of the coded method)

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 (f) mid points (x) x-a = d D/c = u fu
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

Actual mean = A(assumed mean) +

 fu  c
f
102
= 32 +  5
193

= 29.36 years

NB. 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 uq extended 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.

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

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

2. The mode
 STATISTICS LECTURE NOTES P a g e | 85

- 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. 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 class
- The mode is a very important statistical value in business activities quite often business firms tend to
stock specific items which are heavily on demand e.g. footwear, clothes, construction materials (beams,
wires, iron sheets e.t.c.
- The mode can easily be determined form ungrouped data by arranging the figures given and determining
the one with the highest frequency.
- When determining the values of the mode from the grouped data we may use the following methods;-
i. The graphical method which involves use of the histogram
ii. The computation method which involves use of formula

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

Required
Calculate the modal value of the IQ’s tabulated above using
i. The graphical method and
ii. Formular
STATISTICS LECTURE NOTES P a g e | 86

Graphical method

50

40

30

20

10

20 40 60 80 100 120 140

Value of the mode
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

Therefore Mode = 60.50 + 

  48 - 32   × 20

 2  48  - 32 - 27 

= 69.14
3. 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, 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 are equal.
- In order to determine the value of the median from grouped data. When data is grouped the median may
be determined by using the following methods
STATISTICS LECTURE NOTES P a g e | 87

i. Graphical method using the cumulative frequency curve (ogive)

ii. The formula

Example
Referring to the table in 105, determine the median using the methods above
The graphical method

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

xv
160
140
120
100
80
60
40
20
20 40 60 80 100 120 140 160

Value of the median

n+1 146+1
The position of the median = =
2 2
ii Computation
The formula used is

 n 1
 Cfbm 
Median = L   2
  c
 cfmc 
Where L = Lower class boundary of the class containing the median
STATISTICS LECTURE NOTES P a g e | 88

N = No of observations
Cfo= cfbm = Cumulative frequency of the class before that containing
the median
F1 fmc = Frequency of the class containing the median
n 1
- Cfbm
Therefore median  L  2

cfmc

73.5 - 56
= 60 + × 20
48
= 60 + 7.29
= 67.29

4. Geometric mean
- This is a measure of central tendency normally used to measure industrial growth rates.
- It is defined as the nth root of the product of ‘n’ observations or values

-
i.e. GM = n x1 × x 2 ×... × x n

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
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

5. Harmonic mean
This is a measure of central tendency which is used to determine the average growth rates for natural
economies. It is defined as the reciprocal of the average of the reciprocals of all the values given by HM.
STATISTICS LECTURE NOTES P a g e | 89

1
HM 
1
n ( 1 x1  1
x2  ... 1
x3 )

Example
The economic growth rates of five countries were given as 20%, 15%, 25%, 18% and 5%
Calculate the harmonic mean
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%
6. Weighted mean
- This is the mean which uses arbitrarily given weights
- It is a useful measure especially where assessment is being done yet the conditions prevailing are not the
same. This is particularly true when assessment of students is being done given that the subjects being
taken have different levels of difficulties.

Examples
The following table shows that marks scored by a student doing section 3 and 4 of CPA

Subject Scores (x) Weight (w) wx

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
Ewx
Ew

18005

285

 63.17%
STATISTICS LECTURE NOTES P a g e | 90

Merits and demerits of the measures of central tendency

The arithmetic mean (a.m)
Merits
i. It utilizes all the observations given
ii. 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.
iii. It is the best representative of a given set of data if such data was obtained from a normal
population
iv. The a.m. can be determined accurately using mathematical formulas

Demerits of the a.m.

i. If the data is not drawn from a ‘normal’ population, then the a.m. may give a wrong impression
about the population
ii. 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 mode
Merits
i. It can be determined from incomplete data provided the observations with the highest frequency
are already known
ii. The mode has several applications in business
iii. The mode can be easily defined
iv. It can be determined easily from a graph

Demerits
i. If the data is quite large and ungrouped, determination of the mode can be quite cumbersome
ii. Use of the formula to calculate the mode is unfamiliar to most business people
iii. The mode may sometimes be non existent or there may be two modes for a given set of data.
In such a case therefore a single mode may not exist

The median
Merits
i. It shows the centre of a given set of data
ii. Knowledge of the determination of the median may be extended to determine the quartiles
iii. The median can easily be defined
iv. It can be obtained easily from the cumulative frequency curve
v. It can be used in determining the degrees of skew ness (see later)

Demerits
i. In some situations where the no. of observations is even, the value of the median obtained is
usually imaginary
ii. The computation of the median using the formulas is not well understood by most businessmen
iii. In business environment the median has got very few applications

The geometric mean

STATISTICS LECTURE NOTES P a g e | 91

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

Demerits
i. The determination of the GM by using logarithms is not familiar process to all those expected to
use it e.g managers
ii. If the data contains zeros or –ve values, the GM ceases to exist

The harmonic mean and weighted mean

Merits – same as the arithmetic mean
Demerits – same as the arithmetic mean

3.3 Measures of Dispersion

- 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 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 indicating that the mean does not represent the data sufficiently
- The commonly used measures of dispersion are
a) The range
b) The absolute mean deviation
c) The standard deviation
d) The semi – interquartile and quartile deviation
e) The 10th and 90th percentile range
f) Variance

a) The range
- The range is defined as the difference between the highest and the smallest values in a frequency
distribution. This measure is not very efficient because it utilizes only 2 values in a given frequency
distribution. However the smaller the value of the range, the less dispersed the observations are from
the arithmetic mean and vice versa
- The range is not commonly used in business management because 2 sets of data may yield the same
range but end up having different interpretations regarding the degree of dispersion

b) The absolute mean deviation

- This is a useful measure of dispersion because it makes use of all the values given see the following
examples

Example 1
In a given exam the scores for 10 students were as follows
Student Mark (x)
xx
A 60 1.8
B 45 16.8
C 75 13.2
D 70 8.2
STATISTICS LECTURE NOTES P a g e | 92

E 65 3.2
F 40 21.8
G 69 7.2
H 64 2.2
I 50 11.8
J 80 18.2
Total 618 104.4

Required
Determine the absolute mean deviation

x 618
Mean, = 10 = 61.8

 X-X 104.4
= = 10.44
Therefore AMD = 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

Firms (f) Amount of loan fx

xx x x .f
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 = 459000 286736.90

Required
Calculate the mean deviation for the amount of items given

 fx 459, 000
X    24157.9
f 19

 X -X 286736.90
 AMD  
f 19

 Shs 15, 091.40

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 representative.
STATISTICS LECTURE NOTES P a g e | 93

c) The standard deviation

- This is one of the most accurate measures of dispersion. It has the following advantages;
i. It utilizes all the values given
ii. It makes use of both negative and positive values if they occur
iii. The standard deviation reflects an accurate impression of how much the sample data varies from
the mean. This is because its suitability can also be tested using other statistical methods

Example
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

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
25 625
31 961
Total 121 2651

2 2 2
x x 2651  121 
     
n  n  6  6 
 STATISTICS LECTURE NOTES P a g e | 94

= 5.93

Example 2
The following table shows the part-time rate per hour of a given no. of laborers in the month of June 1997.

Rate per hr (x) Shs No. of labourers fx fx2

(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

 fx   fx 
2
2

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

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 order to determine the standard
deviation from such data, we use any of the three following methods
i. The long method
ii. The shorter method
iii. The coded method
The above methods are used in the following examples

Example 3.1
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

i. Using long method

Bar lengths No. of bars(f) Class mid point fx fx2
(cm) (x)
STATISTICS LECTURE NOTES P a g e | 95

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

 fx   fx 
2
2

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

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

= 84.99 cm

ii. Using the shorter method

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

bars(f)
201 – 250 25 225.5 -150 -3750 562500
251 – 300 36 275.5 -100 -3600 360000
301 – 350 49 325.5 -50 -2450 122500
351 – 400 80 375.5 (A) 0 0 0
401 – 450 51 425.5 50 2550 127500
451 – 500 42 475.5 100 4200 420000
501 - 550 30 525.5 150 4500 675000
Total 313 1450 2267500

Calculate the standard deviation using the shorter method quagmire

 fd   fd 
2
2

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

2
2267500  1450 
 
= 313  313 

= 7244.40  21.50

= 7222.90

= 84.99 cm
STATISTICS LECTURE NOTES P a g e | 96

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
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 
 

2
907  29 
 50   
313  313 

= 50 × 1.6997

= 84.99

Variance
Square of the standard deviation is called variance.

d) The semi interquartile range

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)
The semi-interquartile range,
 STATISTICS LECTURE NOTES P a g e | 97

Q3 - Q1
SIR =
2

Example 1
The weights of 15 parcels recorded at the GPO were as follows:
16.2, 17, 20, 25(Q1) 29, 32.2, 35.8, 36.8(Q2) 40, 41, 42, 44(Q3) 49, 52, 55 (in kgs)
Required
Determine the semi interquartile range for the above data
Q3  Q1 44 - 25 19
SIR = = = = 8.5
2 2 2

Example 2 (Grouped Data)

The following table shows the levels of retirement benefits given to a group of workers in a given
establishment.

Retirement benefits £ ‘000 No of retirees (f) UCB cf

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

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
STATISTICS LECTURE NOTES P a g e | 98

= £36.13

The upper quartile (Q3) lies on position

 N + 1
 3 
 4 
 382 + 1 
=3  
 4 
= 287.25

 287.25 - 279 
∴ the value of Q3 = 59.5 + 52 × 10

= 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%
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 + 40 x 10
= 72.925
= £ 72925

iii. The lower 40% corresponds to position

40
= 100 (382 + 1)

= 153.20
STATISTICS LECTURE NOTES P a g e | 99

∴ retirement benefits corresponding to its position

153.2 - 119 
= 39.5 + 70 x 10
= 39.5 + 4.88
= 44.38
= £ 44380

e. 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 (see later)

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

= 38.3
∴ the value corresponding to the tenth percentile

(38.3 x 10)
= 19.5 +
50
= 19.5 + 7.66
= 27.16
The 90th percentile lies on position
90
(382 + 1) = 0.9 x 383
100
= 344.7
∴ the value corresponding to the 90th percentile

 344.7 - 331
= 69.5 + 40 x 10

= 69.5 + 3.425
= 72.925
∴ the required value of the 10th – 90th percentile = 72.925 – 27.16 = 45.765

Relative measures of dispersion

Definition:
 STATISTICS LECTURE NOTES P a g e | 100

A relative measure of dispersion is a statistical value which may be used to compare variations in 2 or more
samples.
The measures of dispersion are usually expressed as decimals or 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
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

iv. Coefficient of variation

standard deviation
×100
= mean

Example (see information above)

First group of cars: mean = 2000 kms
Standard deviation = 5 kms
∴ C.O.V = 5 x 100
2000

= 0.25%

Second group of cars: mean = 3000 kms

Standard deviation = 10kms
STATISTICS LECTURE NOTES P a g e | 101

∴ C.O.V = 10 x 100
3000

= 0.33%

Conclusion
Since the coefficient of variation is greater in the 2nd group, than in the first group we may conclude that the
distances covered in the 1st group are much closer to the mean that in the 2nd group.

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

C.O.V = 800 x 100
8500

= 9.4%
Conclusively, since 4.29% < 9.4% then the salaries offered by the firm in UK are much closer to the mean
given them in the case to the local branch in Kenya

COMBINED MEAN AND STANDARD DEVIATION

Sometimes we may need to combine 2 or more samples say A and B. It is therefore essential to know the new
mean and the new standard deviation of the combination of the samples.

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
STATISTICS LECTURE NOTES P a g e | 102

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)

40  600   50  520  50, 000

Combined mean    555.56
40  50 90

Combined standard deviation

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


40  50

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

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
STATISTICS LECTURE NOTES P a g e | 103

frequency Positively skewed frequency

frequency curve Negatively skewed
frequency curve

Normal distribution

M M M Long tail
od ed ea
e ia n M M M
ea ed od
n ia e

2. Negative Skewness
This is an asymmetrical curve in which the long tail extends to the left

NB: This frequency curve for the age distribution is characteristic of the age distribution in developed
countries
- The mode is usually bigger than the mean and median
- 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

3
 mean - median 
= Standard deviation

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.
Similarly if the coefficient of skewness is +ve it can be concluded that the amount of skew ness of deviation
from the normal distribution is quite high and also the degree of frequency distribution is positively skewed.

Example
STATISTICS LECTURE NOTES P a g e | 104

The following information was obtained from an NGO which was giving small loans to some small scale
business enterprises in 1996. the loans are in the form of thousands of Kshs.

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

(f)
46 – 50 32 48 -15 -3 -96 288 50.5 32
51 – 55 62 53 -10 -2 -124 248 55.5 94
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.
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 minusing
assumed mean (A) from X and then deviding 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

-
 f   f 
The standard deviation =c×  

2
3086  428 
- 
=5 × 610  610 

= 10.68
n +1
The Position of the median lies m = 2
610 +1
= 2 = 305.5
STATISTICS LECTURE NOTES P a g e | 105

 305.5 - 191
= 60.5 + 120 ×5

114.4 
= 60.5 + 120 ×5
Median = 65.27
Therefore the Pearsonian coefficient

3
 66.51- 65.27 
= 10.68
= 0.348
Comment
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
= 152.75
The position of Q1 lies on = 4

152.75 - 94   5  58.53
∴ actual value Q1 =55. 5 + 97

3
 610 +1 = 458.25
The position of Q3 lies on = 4
 458.25 - 403   5  73.83
∴ actual value Q3 =70.55 + 83 ×5

 610 +1
Q2 position: i.e. 2 4 = 305.5

 60.5 
 305.5 -191  5  65.27
Actual Q2 value 120

The required coefficient of skew ness

73.83  58.53  2  65.27 
 0.013
= 73.83  58.53

Conclusion
Same as above when the Pearsonian coefficient was used
STATISTICS LECTURE NOTES P a g e | 106

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 than 3 are definitely leptokurtic
b) Some frequency distributions when plotted may produce a curve similar to that of the
normal distribution. Such frequency distributions are referred to as mesokurtic. The degree
of kurtosis is usually equal to 3
c) When the frequency curve contacted produces a peak which is lower that that of a normal
distribution when such a curve is said to be platykurtic. The coefficient of such is usually less
than 3
- It is necessary to calculate the numerical measure of kurtosis. The commonly used measure of
kurtosis is the percentile coefficient of kurtosis. This coefficient is normally determined using the
following equation
1
 Q3 - Q1
2
Percentile measure of kurtosis, K (Kappa) = P90 - P10
Example
Refer to the table above for loans to small business firms/units
Required
Calculate the percentile coefficient of Kurtosis
90
 n +1 = 0.9  610 +1
P90 = 100
= 0.9 (611)
= 549.9
The actual loan for a firm in this position
 549.9 - 538 
(549.9) = 80.5 + 40 x 5 = 81.99
10
P10 = 100 (n + 1) = 0.1 (611) = 61.1
The actual loan value given to the firm on this position is
 61.1  32 
50.5 + 62 x 5 = 52.85
= 0.9 (611)
= 549.9
∴ percentile measure of kurtosis

 Q3 - Q1
K(Kappa) = ½ P90 - P10
STATISTICS LECTURE NOTES P a g e | 107

 73.83 - 58.53
= ½ 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
Kurtosis is also measured by moment statistics, which utilize the exact value of each observation.
X
i. M1 the first moment = M1 = n = Mean M1 or M1

X 2

M2 = n

X 3

M3 = n

X 4

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 = f Where m is mean

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
STATISTICS LECTURE NOTES P a g e | 108

16 6 96 -1.6 2.56 15.36 39.32

18 10 180 .4 0.16 1.60 0.256
20 7 140 2.4 5.76 40.32 232.24
22 2 44 4.4 19.36 38.72 749.62
30 528 179.20 2,676.74

528
M = 30 = 17.6
179.20
σ2 = 30 = 5.973
σ4 = 35.677

  x  m
4
f 2, 676.74

M4 = f =
30
= 89.22

89.22
Moment coefficient of Kurtosis = 35.677 = 2.5

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

3.4 Indices
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 =  Po × 100 (an unweighted price index)

 Qn
A simple quantity index =  Q o × 100 (an unweighted quantity index)

Where pn is the price of a commodity in the current year (the year for which the price index to be calculated)

Where po is the price of the same commodity in the base year (the year for comparison purposes)

Similarly Qn and Qo are defined in the same way

AGGREGATE PRICE INDEX NUMBERS AND QUANTITY INDEX NUMBERS

PRICE INDEX QUANTITY INDEX
LASPEYRE’S INDEX
p q n o q p n o

P q o o × 100 q p o o × 100
PAASCHE’S INDEX
p q n n q q n n

Pq o n × 100 q p o n × 100

STATISTICS LECTURE NOTES P a g e | 109

p q n n

Value index = P q o o × 100

MODIFIED FORM OF THE LASPEYRE’S PRICE INDEX NUMBER


  w pn
po o
100
Laspeyre’s Price index w o

Where w0 are the proportions of the total expected in the basic period. This formula is frequently used to
calculate retail price index.

CHANGING THE BASE OF THE INDEX

For comparison purposes if two series have different base years, it is difficult to compare them directly. In
such cases, it is necessary to change the base year of one of the series (or both) so that both have the same
base.
It is also necessary to keep the index relevant to current conditions hence the need to change the base from
time to time.

Example;
Year 1985 1986 1987 1988 1989 1990 1991 1992
Price index 100 104 108 109 112 120 125 140

Suppose we wish to change the base year to 1989

We recalculate each index by expressing it as a percentage of 1989

Previous index Recalculated index

1985 100 100
112 × 100 = 89.3
1986 104 104
112 × 100 = 92.9
1987 108 108
112 × 100 = 96.4
1988 109 109
112 × 100 = 97.3
1989 (new base year) 112 112
112 × 100 = 100
1990 120 120
112 × 100 = 107.1
1991 125 125
112 × 100 = 111.6
1992 140 140
112 × 100 = 125.0

When changing the base year, it is advisable to update the weights used in the base year.

CHAIN BASED INDEX NUMBERS

STATISTICS LECTURE NOTES P a g e | 110

A chain based index is one where the index is calculated every year using the previous year as the base year.
This type of index measures rate of change from year to year.
This method is suitable where weights are changing rapidly and items are constantly being brought into the
index and unwanted items taken out. It can be a price or quantity index

Previous index Recalculated chainbased fixedbased index

index
1985 100 100 100(1985 base year
1986 104 104 104
100 × 100 = 104 100 × 100 = 104
1987 108 108 108
104 × 100 = 103.8 100 × 100 = 108
1988 109 109 109
108 × 100 = 100.9 100 × 100 = 109
1989 112 112 112
109 × 100 = 102.8 100 × 100 = 112
1990 120 120 120
112 × 100 = 107.1 100 × 100 = 120
1991 125 125 125
120 × 100 = 104.2 100 × 100 = 125
1992 140 140 140
120 × 100 = 112 100 × 100 = 140

The Fisher’s index

The Fisher’s index acts as a compromise between Laspeyre’s index and Paasche’ index. It is calculated as a
geometric mean of the two indexes.

Retail price index

It is weighted average of price relatives based upon an average household in the base year. The items
consumed are divided into groups such as food, housing, transport, alcoholic drinks, footwear, fuel, light,
water, household goods, services e.t.c. each item included in the index is given a weighting and a price relative
to the base is calculated. Modified form of laspeyre’s price index formula is used as a weighted arithmetic
mean of price relatives.


   W 100
pn
po

W
0
I.e. Retail Price index 0
The index is used by the Government as a guide in determining the minimum wages, pension rates
unemployed benefits (in UK e.t.c). Trade unions use it as a basis for their wages claims.

Deflation
Indexes may be used to deflate time series so that comparisons between periods may be made in real terms
It is a process of reducing a value measured in current period prices to its equivalent in the base period prices.
The deflated value is what would have been necessary to purchase the same amount of goods as the present
value can purchase in the current period
p qn n

Deflation Factor =  p q
0 n × 100
STATISTICS LECTURE NOTES P a g e | 111

Deflation of a time series

Year Average monthly earnings (shs) Retail index Real earnings
1 5,000 100 5000 = 5000
2 5,500 120 100
5,500 × 120 = 4,583.3
3 6,000 140 100
6,000 × 140 = 4,285.7
4 6,500 170 100
6,500 × 170 = 3,823.5
5 7,200 200 100
7,200 × 200 = 3,600.0

The technique of index number construction

When preparing index numbers it is important to define
a) The exact purpose of the index
b) How the items are to be selected
c) The choice of the weights
d) The choice of the base
e) The type of average to be used
The base year should be as close to the normal trend as possible. The best methods should be used for
collection of data. The items should be selected in such a way that they are a fair representation of all the
relevant items.
Due consideration should be given to the weighting of all items selected

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
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 1st January 1990 and 1st January 1991 were as
follows.

Share Price on 1.1.1990 Price on 1.1.1991

Company A Shs 10 Shs 12
Company B Shs 12 Shs 15
STATISTICS LECTURE NOTES P a g e | 112

Company C Shs 20 Shs 25

Company D Shs 5 Shs 6

Using an unweighted geometric index, calculate the index of share prices at 1.1.1991 if 1.1.1990 is the base
date, index 100

Solution
1 1
 12 15 25 6  4  27000  4
   2.25 
1

     
4

 10 12 20 5   12000 
 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 = Retail price index in the base year × 100

Marshal Hedge Worth Index

p p n o  qn 
Marshal Hedge worth index =  p  q o o  qn  × 100

Tests For An Ideal Index Number

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.
STATISTICS LECTURE NOTES P a g e | 113

LESSON 3 REINFORCING QUESTIONS

QUESTION ONE
a) Distinguish between discrete and continuous data.
b) What is dispersion and what is the formula for the standard deviation?
c) What is the measure of relative dispersion?
d) Draw diagrams showing positive and negative skewness

QUESTION TWO
The managers of an import agency are investigating the length of time that customers take to pay their
invoices, the normal terms for which are 30 days net. They have checked the payment record of 100
customers chosen at random and have compiled the following table:

Payment in Number of customers

5 to 9 days 4
10 to 14 days 10
15 to 19 days 17
20 to 24 days 20
25 to 29 days 22
30 to 34 days 16
35 to 39 days 8
40 to 44 days 3

Required:
a) Calculate the arithmetic mean.
b) Calculate the standard deviation
c) Construct a histogram and insert the modal value.
d) Estimate the probability that an unpaid invoice chosen at random will be between 30 and 39 days old.

QUESTION THREE
The price of the ordinary 25p shares of Manco PLC quoted on the stock exchange, at the close of the
business on successive Fridays is tabulated below

126 120 122 105 129 119 131 138

125 127 113 112 130 122 134 136
128 126 117 114 120 123 127 140
124 127 114 111 116 131 128 137
127 122 106 121 116 135 142 130

Required
a) Group the above date into eight classes. (4 marks)
b) Calculate cumulative frequency, the median value, quartile values and the
semi-quartile range. (4 marks)
c) Calculate the mean and standard deviation of your frequency distribution. (7 marks)
d) Compare and contrast the values that you have obtained for:
i) The median and mean
ii) The semi-interquartile range and the standard deviation (5 marks)
(Total: 20 marks)
STATISTICS LECTURE NOTES P a g e | 114

QUESTION FOUR
Define the coefficient of variation.

The following table gives profits (in ten thousands of shillings) of two supermarkets over a duration of one
year.

Month Supermarket A Supermarket B

January 65 28
February 48 33
March 15 20
April 28 23
May 41 69
June 59 45
July 41 53
August 10 15
September 24 35
October 56 57
November 92 99
December 120 136

Required:
i) Compute the coefficient of variation for each supermarket.
ii) Indicate for which supermarket the variability of profits is relatively greater.

QUESTION FIVE
Prodco PLC manufactures an item of domestic equipment which requires a number of components which
have varied as various modifications of the model have been used. The following table shows the number of
components required together with the price over the last three years of production.

COMPONENT 1981 1982 1983

Prices Quantity Prices Quantity Prices Quantity
A 3.63 3 4.00 2 4.49 2
B 2.11 4 3.10 5 3.26 6
C 10.03 1 10.36 1 12.05 1
D 4.01 7 5.23 6 5.21 5

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)
STATISTICS LECTURE NOTES P a g e | 115

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 K257
for 1983 and 1984 based on 1982. (6 marks)

ii) Calculate Paasche type price index numbers for the cost of one unit of K257
for 1983 and 1984 based on 1982. (6 marks)

iii) Compare and contrast the Laspeyre and Paasche price-index numbers you
have obtained in (i) and (ii) (3 marks)

A number of employers manufacturing plastic components used in plumbing have formed themselves into an
association for the purpose of negotiating with the trade union for this industrial sector.

The negotiations cover pay and contributions in this sector.

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: 20 marks)

QUESTION SEVEN
The data given below indicates the prices and production of some horticulatural 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
STATISTICS LECTURE NOTES P a g e | 116

Required:
Calculate the increase or decrease in prices from 1980 on the basis of the following indices:

a) Mean relatives
b) Laspeyres index
c) Paasche index
d) Marshall – Hedgeworth index
Fishers index.

(NOTES ON SAMPLING DISTRIBUTION AND TEST OF HYPOTHES ARE ATTACHED ON

THE HANDOUT)