Paper2-QT-2022 Lecture Notes-Hassan PDF

THE INSTITUTE OF
Quantitative Techniques
CERTIFIED PUBLIC
ACCOUNTANTS UGANDA
Paper- 2
LEVEL-1 STUDY
TEXT
2022-2023
Preface
CONTENTS
1
HASSAN MBAZIIRA (0703-859 320). ICPAU- PAPER 2 (QT). 2022-2023
COMPLETE SYLLABUS
Chapter 1 Introduction to statistical data,
presentations & measures
Chapter2 Sampling and sampling methods
Chapter3 Data classification and tabulation
Chapter4 Measures of central tendency
Chapter5 Measures of dispersion
Chapter6 Introduction to probability
Chapter7 Decision making
Chapter8 Permutations and combinations
Chapter9 Discrete probability distribution
Chapter10 Continuous probability distribution
Chapter11 Estimation and hypothesis testing
Chapter12 Chi – squared tests
Chapter13 Control charts
Chapter14 Linear algebra and calculus
Chapter15 Linear Programming
Chapter16 Correlation and regression
Chapter17 Forecasting: Time series
Chapter18 Net work analysis
Chapter 19 Index numbers
FACILITATOR
MBAZIIRA HASSAN
hassanmbaziira@gmail.com
0703- 859 320
2
Chapter-1
Introduction to statistical data, presentation and measures
1.1 INTRODUCTION
Most of the quantitative methods mentioned in this course fall under the general heading of
statistics which is often used to refer simply to a set of data. For example, we can refer to a
Uganda's unemployment statistics (which might be presented in a table or chart showing the
country's unemployment rates each year for the last few years, and might be broken down by
gender, age, region and/or industrial sector, etc.).
However, we can also use the term "Statistics" (preferably with a capital letter) to refer to the
academic discipline concerned with the collection, description, analysis and
interpretation of numerical data. As such, the subject of Statistics may be divided into two
main categories; descriptive and inferential statistics.
A. DESCRIPTIVE STATISTICS This is mainly concerned with collecting and summarizing

data, and presenting the results in appropriate tables and charts. For example, companies
collect and summarize their financial data in tables (and occasionally charts) in their annual
reports, but there is no attempt to go "beyond the data".
B. STATISTICAL INFERENCE This is concerned with analyzing data and then interpreting
the results (attempting to go "beyond the data"). The main way in which this is done is by
collecting data from a sample and then using the sample results to infer conclusions about
the population. For example, prior to general elections in Uganda and many other
countries, statisticians conduct opinion polls in which samples of potential voters are asked
which political party they intend to vote for. The sample proportions are then used to
predict the voting intentions of the entire population. Of course, before any descriptive
statistics can be calculated or any statistical inferences made, appropriate data has to be
collected. We will start the course, therefore, by seeing how we collect data. This study
unit looks at the various types of data, the main sources of data and some of the
numerous methods available to collect data.
1.2 VARIABLES AND DATA

A VARIABLE OR VARIATE is any characteristic on which observations can be made. For
example, height is a variable because observations taken are of the heights of a number of
people. Variables, and therefore the data which observations of them produce, can be
categorized in various ways: - quantitative/qualitative, continuous/ discrete.
A. QUANTITATIVE VARIABLES, to which we shall restrict discussion here, are those for
which observations are numerical in nature. Qualitative variables have non-numeric
observations, such as color of hair, although of course each possible non-numeric value
may be associated with a numeric frequency.
B. A CONTINUOUS VARIABLE may take any value between two stated limits (which may
possibly be minus and plus infinity). Height, for example, is a continuous variable, because
a person's height may (with appropriately accurate equipment) be measured to any minute
fraction of a millimeter. A discrete variable however can take only certain values
occurring at intervals between stated limits. For most (but not all) discrete variables, these
intervals are the set of integers (whole numbers). For example, if the variable is the
number of children per family, then the only possible values are 0, 1, 2, ... etc., because it 3
is impossible to have other than a whole number of children. However in Uganda shoe
sizes are stated in half-units, and so here we have an example of a discrete variable which
can take the values 1, 1½, 2, 2½, etc. You may possibly see the difference between
continuous and discrete variables stated as "continuous variables are measured, whereas
discrete variables are counted". While this is possibly true in the vast majority of cases, you
should not simply state this if asked to give a definition of the two types of variables.
1.3 METHODS OF COLLECTING DATA

Before any data collection, you should know the difference between a primary source and
secondary source, primary data and secondary data. Primary data is statistical material which
the investigator originates for the purposes of the inquiry at hand. Secondary data is that
statistical material which is NOT originated by the investigator himself but obtained from
someone else’s records.
METHODS OF COLLECTING PRIMARY DATA

These include; personal interviews, mailed questionnaires and questionnaires in
charge of enumerators.
1 Personal interviews may be structured, semi- structured or unstructured

In a STRUCTURED INTERVIEW, the interviewer usually has a well-defined set of prepared
questions (i.e. a questionnaire) in which most of the questions are "closed" (i.e. each question
has a predetermined set of options for the response, such as a box to be ticked). The design
of such questionnaires is essentially the same as that discussed below under the heading Self-
Completion Questionnaires. Structured interviewing is useful if the information being sought is
part of a clearly-defined business research project (such as market research), and if the aim of
the survey is to collect numerical data suitable for statistical analysis.
In a SEMI-STRUCTURED INTERVIEW, the interviewer has a set of prepared questions, but

is happy to explore other relevant issues raised by the interviewee.
In UNSTRUCTURED INTERVIEWS, the interviewer does not have a set of prepared

questions and the emphasis is often on finding out the interviewee's point of view on the
subject of the survey. Unstructured interviews are more commonly used in qualitative (rather
than quantitative) research, though they can also be useful as pilot studies, designed to help a
researcher formulate a research problem.
ADVANTAGES OF INTERVIEWING
It is the most appropriate method when studying attitudes, beliefs, values and motives of the
respondents. It is also appropriate in studies of irregular, private behavior as compared to say,
observation, it is very flexible, allows assessment of non-verbal behavior, ideal method to get
information from incapable people, i.e. those who cannot read and write, high response rate,
possible to get deeper into the details of the subject of the interview, possibility of cross
checking whether the information provided is correct, more complex questions can easily be
administered /asked through the interview method, e.g. questions full of graphs, charts and
detailed instructions, allows to attain spontaneous answers, i.e. non specific structured
answers, it is possible to study and exploit unknown events which are only known to the
respondents and the interview ensures that all questions are answered.
4
DISADVANTAGES OF INTERVIEWING
Very costly/expensive especially if samplers are large, it is time consuming because each
respondent needs his/her own time to be interviewed, too much reliance on memory because
the interviewer is expected to do a lot of things at the same time. Also the memory of the
respondents is not reliable. No opportunity to consult records, difficulty in persuading
respondents to respond as required, problem of finding trained personnel to conduct the
interview, personal emotions of the interviewer and the respondents, Language barrier, the
“interact” between the interviewer and the respondents may sometimes bias the answers, e.g.
race, religion, sex and age differences, etc.
Obviously, the interview method is time consuming and not suitable when we have large
groups of respondents. Here one drafts a questionnaire.
THE QUESTIONNAIRE METHOD

The questionnaire used can be mailed (mailed questionnaire) to the respondents for filling in
and returning OR the questionnaire can be put in charge of enumerators who go around and
fill them after getting the desired information.
ADVANTAGES OF A MAILED QUESTIONNAIRE METHOD

Its time saving, its cost effective compared to other methods of data collection. Questionnaires
can be completed at respondent’s convenience. Greater assurance of anonymity i.e. answers
to very sensitive issues (embarrassing questions) can be acquired. No interviewer bias.
Consultation of records hence reliable information. Accessibility especially respondents who are
widely scattered. Standardized wording since the same questions are sent to all prospective
respondents
DISADVANTAGES OF A MAILED QUESTIONNAIRE

Lack of flexibility since no interviewer is present to rephrase a misunderstood question. Low
response rate. No control over question order which leads to skipping of some questions. Many
questions remain unanswered since there is no supervision. Can’t record spontaneous answers.
It’s difficult to gather spontaneous first opinions as the respondent has the chance to delete
/erase the first answer that he/she thinks is undiplomatic. A wrong person can fill the
questionnaire. Can’t use a complex questionnaire format. Questions that have high complex
and confusing questions can’t be asked. No control over the environment. That’s to say,
questions that are directed to housewives only might not be answered properly since a
husband might be around. No supplementing data through observation. No control over the
date of response.
WHEN ENUMERATORS ARE TASKED TO DO THE INTERVIEW

Can be employed on a pay basis or can offer free services. These interviewers must be trained
in the use of the questionnaire and advised how to present it so that maximum cooperation is
obtained from the respondent. This training is very important and must be carefully thought
out. The interviewers must be carefully selected so that they will be suitable for the type of
interview envisaged. The type of interviewer and the method of approach must be varied
according to the type of respondent selected, e.g. the same technique should not be used for
interviewing students and senior bank staff.
PERSONAL OBSERVATION METHOD OF DATA COLLECTION

This method is used when it is possible to observe directly the information that you wish to
collect. For example, data for traffic surveys is collected in this way: observers stand by the 5
roadside and count and classify the vehicles passing in a given time. Increasingly, computers
and automated equipment are replacing human observers in this method of data collection as
they are considerably cheaper and often more reliable. There are numerous examples of this.
For instance, most traffic information is now collected by sensors in rubber tubes laid across
the road and linked to a small computer placed alongside the road.
ADVANTAGES
Enables the researcher to study the behavior as it occurs. This guards against the difference
between reported behavior and actual behavior that are common when interviewing. Only
method to gather information about non-verbal behavior. There the researcher records the
silent features of behavior. Observation is free from non-response errors. Errors arising from
misinterpretation of question thus giving distorted answers. It is possible to adjust goals and
objectives of the study as data is collected which is not applicable in the interview or even the
questionnaire method. Research is carried out in the natural environment so the results got are
based on existing reality.
DISADVANTAGES
Time consuming. The researcher has to wait4 until the phenomenon/event occurs. Poor
methods when the researcher is interested in attitudes or opinion studies. It is a costly
technique in terms of man-hours spent in the field. Besides, it is tricky/difficult to delegate
research assistants in the observation method. Behavior that extends over a long period of
time is difficult to study under this method. Bias from the researcher or observer since he/she
may be influenced by particular moods or prevailing physiological status of the
researcher/observer. Lack of control on other natural factors that may affect our data. Only
deals with smaller sample sizes since it is not easy to observe large populations. Poor method
when studying sensitive topics issues, e.g. studies/research on sexual practices, abortion,
contraception, etc.
METHODS OF COLLECTING SECONDARY DATA

It is important to consult published sources before deciding to go out and collect your own
data, to see if all or part of the information you require is already available. Published sources
provide valuable access to secondary data for use in business and management research. We
will now describe where to look for business data. You will often find useful information from
several sources, both within an organization and outside.
Scanning Published Data. When you examine published data from whatever source, it is
helpful to adopt the following procedure: (a) Overview the whole publication Flip through the
pages so that you get a feel for the document. See if it contains tables only, or if it uses
graphs and tables to describe the various statistics. (b) Look at the Contents pages. A study of
the contents pages will show you exactly what the document contains and give you a good
idea of the amount of detail. It will also show you which variables are described in the tables
and charts. (c) Read the Introduction. This will give a general indication of the origin of the
statistics in the document. It may also describe how the survey which collected the information
was carried out. (d) Look at part of the Document in Detail. Take a small section and study
that in depth. This will give you an appreciation of just what information is contained and in
what format. It will also get you used to studying documents and make you appreciate that
most tables, graphs or diagrams include some form of notes to help explain the data.
6
1.4 INTERNAL DATA SOURCES
All types of organization will collect and keep data which is therefore internal to the
organization. More often than not it applies to the organization where you work, but you
should not think of it as meaning just that type of organization. It is important when looking
for some particular types of data to look internally because;
It will be cheaper if the data can be obtained from an internal source as it will save the
expense of some form of survey. Readily available information can be used much more quickly
especially if it has been computerized and can be easily accessed. When the information is
available from within your own organization, it can be understood much more easily as
supporting documentation is likely to be readily available. Overall there are several advantages
from using internal data, although there is a tendency when using this type of data to make do
with something that is nearly right. Companies' annual reports provide a particularly useful set
of data for financial and business research.
EXTERNAL DATA SOURCES

The sources of statistical information can be conveniently classified as: central and local
government sources together with other publications from private sources. The data produced
by these sources can be distinguished as: Data collected specifically for statistical purposes –
e.g. the population census. Data arising as a by-product of other functions – e.g.
unemployment figures. This latter distinction is well worth noting because it sometimes helps
to indicate the degree of reliability of the data. Do not forget, of course, that very often the
statistician has to be his or her own source of information; then he or she must use the
techniques of primary data collection which we have already discussed.
The main producer of statistics in Uganda is Uganda Bureau of Statistics (UBOS). UBOS exists
primarily to service the needs of central government and other interested parties. However,
much of the information it produces is eminently suitable for use by the business community
as well, and indeed central government which is increasingly becoming aware of the need to
gear its publications so that they can be used by the business sector.
Also, the United Nations publications and websites are available, which cover world-wide
statistics in subjects such as population and trade. Most Ugandan companies also provide
useful financial data in their annual reports and accounts, most of which are now available
online
7
Chapter-2
Sampling and sampling methods
2.1 INTRODUCTION
A sample is a subset of a population, a small sample is less than 30 elements and greater
than 30 elements is a large sample. By studying a sample, one can be able to draw conclusions
that are general to the entire population.
A population is the set of all the individuals or objects which have a given characteristic, e.g.
the set of all persons eligible to vote in a given country. Within an entire population, any single
member is called an element.
A study of all elements in the population is called a census where as a study of a few elements
in the sample is called sampling
Sampling is the process of selecting and inquiring from a representative fraction of the
population to make a conclusion about the whole population. Or sampling is a systematic
approach of selecting a few elements from the entire population.
Sampling unit is an item from which information is obtained. It may be a person, an
organization or an inanimate object such as a tyre. Sampling frame – a list of all the items in a
population
2.2 REASONS FOR SAMPLING

1. Reduced costs than census. Resources are not often adequate to cover entire
population
2. Greater speed in getting data. When results are required quickly
3. Greater scope covered.
4. Up-to-date information is obtained
5. Minimal errors.
6. Greater control than census
7. Better quality interviews with almost all questions answered correctly.
A. METHODS OF SAMPLING
PROBABILITY/ RANDOM SAMPLING TECHNIQUES
SIMPLE RANDOM SAMPLING
Every member of the population has an equal probability of being selected. Example; Crane
Bank manager wants to find out views about their bank services. Given the sampling
frame is the list of depositors in Crane bank. Everyone on the depositors list is given a unique
number from 1 to n, (n being the total number of people in the sampling frame). Each number
is now written on a slip of paper and put in a box. If you want a sample of a thousand people
you mix up these slips thoroughly and draw out a thousand slips. The numbers on these slips
then represent the people to be interviewed. In theory each slip would stand an equal chance
of being drawn out and so would have been chosen in a random manner.
Advantages
1. Always produces an unbiased sample.
2. Simple and easy to apply.
Disadvantages
1. Sampling units may be difficult or expensive to contact (dispersed people) 8
STRATIFIED SAMPLING
This is a probability sampling technique wherein the researcher divides the entire population
into different subgroups or strata, then randomly selects the final subjects proportionally from
the different strata.
It is important to note that the strata must be non-overlapping. Having overlapping subgroups
will grant some individuals higher chances of being selected as subject. This completely
negates the concept of stratified sampling as a type of probability sampling. Equally important
is the fact that the researcher must use simple probability sampling within the different strata.
The most common strata used in stratified random sampling are age, gender, socioeconomic
status, religion, nationality and educational attainment.
The sample size of each stratum in this technique is proportionate to the population size of the
stratum when viewed against the entire population. This means that the each stratum has the
same sampling fraction.
Example on proportionate Stratified Random Sampling

For example, you have 3 strata with 100, 200 and 300 population sizes respectively.
And the researcher chose a sampling fraction of ½. Then, the researcher must randomly
sample 50, 100 and 150 subjects from each stratum respectively.
Stratum A B C
Population Size 100 200 300
Sampling Fraction ½ ½ ½
Final Sample Size 50 100 150
The important thing to remember in this technique is to use the same sampling fraction for
each stratum regardless of the differences in population size of the strata. It is much like
assembling a smaller population that is specific to the relative proportions of the subgroups
within the population.
Advantages – the advantage of this method is that the results from such a sample will not be
distorted or biased by undue emphasis on extreme observations.
Disadvantages – the main disadvantage is the difficulty of defining the strata. This method
can also be time-consuming, expensive and complicated to analyze.
QUASI-RANDOM SAMPLING
SYSTEMATIC SAMPLING METHOD
It is a type of probability sampling method in which sample members from a larger population
are selected according to a random starting point and a fixed, periodic interval. This interval,
called the sampling interval, is calculated by dividing the population size by the desired sample
size; I= (N/n). N is the population size and n is the sample size.
Example; Assuming the bank’s population of depositors is so large, he can’t use simple
random sampling. He will use systematic sampling. On his list of depositors, he might decide to
interview every 10th person on the list. As before, each member of the population (depositors) 9
will be given a number from 1 to n, the starting number is selected from a table of random
numbers by taking the first number in the table between 1 and 9. Say a 2 was chosen, then
the 2nd, 12th, 22nd, 32nd … etc. person would be selected from the sampling frame. This
method of sampling is often used as it reduces the amount of time that the sample takes to
draw. However, it is not a purely random method of selecting a sample, since once the
starting point has been determined, then the items selected for the sample have also been set
Advantages
1. Quick to select a sample.
2. Also it is sufficiently close to simple random sampling, in most cases, to justify its
widespread use.
Disadvantages
1. A major disadvantage occurs if the sampling frame is arranged so that sampling units
with a particular characteristic occur at regular intervals, causing over-representation or
under-representation of this characteristic in the sample. For example, if you are
choosing every tenth house in a street and the first randomly chosen number is 8, the
sample consists of numbers 8, 18, 28, 38 and so on. These are all even numbers and
therefore are likely to be on the same side of the street. It is possible that the houses
on this side may be better, more expensive houses than those on the other side. This
would probably mean that the sample was biased towards those households with a
high income. A sample chosen by systematic sampling must always be examined for
this type of bias.
MULTISTAGE SAMPLING
This refers to a sampling technique where sampling is carried out in stages using smaller and
smaller sampling units at each stage. Or a more complicated form of cluster sampling in which
larger clusters are further subdivided into smaller, more targeted groupings for the purposes of
surveying.
Despite its name, multi-stage sampling can in fact be easier to implement and can create a
more representative sample of the population than a single sampling technique. Particularly in
cases where a general sampling frame requires preliminary construction, multi-stage sampling
can help reduce costs of large-scale survey research and limit the aspects of a population
which needs to be included within the frame for sampling.
Example: Ministry of Education Researchers used a multi-stage sampling design to survey

teachers in Mbarara, in order to examine whether socio-demographic characteristics determine
teachers’ attitudes towards adolescent sexuality education. First-stage sampling included a
simple random sample to select 20 secondary schools in the Mbarara region. The second stage
of sampling selected 13 teachers from each of these schools, who were then, administered
questionnaires.
Advantages – the advantages of this method are that at each stage the samples selected are
small and interviews are carried out in 50 small areas instead of in 5,000 scattered locations,
thus economizing on time and cost. There is no need to have a sampling frame to cover the
whole country. The sample is effectively a simple random sample.
Disadvantages – the main disadvantages are the danger of introducing interviewer bias and
of obtaining different levels of accuracy from different areas. The interviewers must be well
chosen and thoroughly trained if these dangers are to be avoided.
10
NON- PROBABILITY/ NON-RANDOM SAMPLING TECHNIQUES
QUOTA SAMPLING
It is a non-probability sampling technique wherein the assembled sample has the same
proportions of individuals as the entire population with respect to known characteristics, traits
or focused phenomenon. OR; a type of non-probability sample in which the researcher selects
people according to some fixed quota. That is, units are selected into a sample on the basis of
pre-specified characteristics so that the total sample has the same distribution of
characteristics assumed to exist in the population being studied.
For example, if you are a researcher conducting a national quota sample, you might need to
know what proportion of the population is male and what proportion is female as well as what
proportions of each gender fall into different age categories, race or ethnic categories,
educational categories, etc. The researcher would then collect a sample with the same
proportions as the national population. In quota sampling, the researcher aims to represent
the major characteristics of the population by sampling a proportional amount of each.
Example; let’s say, for example, that you want to obtain a proportional quota sample of 100
people based on sex.
First you would need to find out the proportion of the population that is men and the
proportion that is women. If you found out the larger population is 40% women and 60%
men, you would need a sample of 40 women and 60 men for a total of 100 respondents. You
would start sampling and continue until you got those proportions and then you would stop.
So, if you’ve already got 40 women for the sample, but not 60 men, you would continue to
sample men and discard any legitimate women respondents that came along. You don’t need
them because you have already “met your quota.” The difficulty here is that you have to
decide in advance the specific characteristics on which you will base the quota. Will it be by
gender, age, education race, religion, etc.?
CLUSTER /AREA SAMPLING

We have already considered the problems of cost and time associated with simple random
sampling, and cluster sampling is another probability sampling method which may be used to
overcome these problems. It is also a useful means of sampling when there is an
inadequate sampling frame or when it is too expensive to construct the frame.
Definition; the method consists of dividing the sampling area into a number of small
concentrations or clusters of sampling units. Some of these clusters are chosen at random, and
every unit in the cluster is sampled.
For example, suppose you decided to carry out the bank survey using the list of all the
customers as the sampling frame. If you wished to avoid the cost of simple random sampling,
you could take each branch of the bank as a cluster of customers. Then you select a
number of these clusters randomly, and interview every customer on the basis of the branches
chosen. As you interview all the customers at the randomly selected branches, the sum of all
interviews forms a sample which is representative of the sampling frame, thus fulfilling your
major objective of a random sample of the entire population.
A variation of this method is often used in Uganda, because of the vast distances involved in
that country (often referred to as AREA SAMPLING). With the use of map references, the
entire area to be sampled is broken down into smaller areas, and a number of these areas are 11
selected at random. The sample consists of all the sampling units to be found in these selected
areas.
Advantages: Reduction in cost and increase of speed in carrying out the survey. The method
is especially useful where the size or constitution of the sampling frame is unknown. Nothing
needs to be known in advance about the area selected for sampling, as all the units within it
are sampled; this is very convenient in countries where electoral registers or similar lists do not
exist.
Disadvantages – one disadvantage is that often the units within the sample are
homogeneous, i.e. clusters tend to consist of people with the same characteristics. For
example, a branch of a bank chosen in a wealthy suburb of a town is likely to consist of
customers with high incomes. If all bank branches chosen were in similar suburbs, then the
sample would consist of people from one social group and thus the survey results would be
biased. This can be overcome to some extent by taking a large number of small clusters rather
than a small number of large clusters. Another disadvantage of taking units such as a bank
branch for a cluster is that the variation in size of the cluster may be very large, i.e. a very
busy branch may distort the results of the survey.
Quota sampling is a non-probability sampling technique wherein the assembled sample has
the same proportions of individuals as the entire population with respect to known
characteristics, traits or focused phenomenon. In addition to this, the researcher must make
sure that the composition of the final sample to be- used in the study meets the research's
quota criteria.
Step-by-step Quota Sampling

The first step in non-probability quota sampling is to divide the population into exclusive
subgroups. Then, the researcher must identify the proportions of these subgroups in the
population; this same proportion will be applied in the sampling process. Finally, the researcher
selects subjects from the various subgroups while taking into consideration the proportions
noted in the previous step. The final step ensures that the sample is representative of the
entire population. It also allows the researcher to study traits and characteristics that are noted
for each subgroup.
When to Use Quota Samples

The main reason why researchers choose quota samples is that it allows the researchers to
sample a subgroup that is of great interest to the study. If a study aims to investigate a trait or
a characteristic of a certain subgroup, this type of sampling is the ideal technique.
Quota sampling also allows the researchers to observe relationships between subgroups. In
some studies, traits of a certain subgroup interact with other traits of another subgroup. In
such cases, it is also necessary for the researcher to use this type of sampling technique.
Disadvantages of Quota Samples
It may appear that this type of sampling technique is totally representative of the population.
In some cases it is not. Keep in mind that only the selected traits of the population were taken
into account in forming the subgroups. In the process of sampling these subgroups, other
traits in the sample may be overrepresented. In a study that considers gender, socioeconomic
status and religion as the basis of the subgroups, the final sample may have skewed
representation of age, race, educational attainment, marital status and a lot more.
12
Chapter-3
Data classification and tabulation
3.1 INTRODUCTION
Information collected from primary sources is often unorganized and can’t easily be
interpreted. As such, we have to organize and classify it.
3.2 DATA CLASSIFICATION is a process of arranging data into sequences and groups
according to their common characteristics (resemblance) or separating them into different but
related parts. Thus, all CPA paper5 students in my class can be classified according to their
marital status, height, religion etc. If students are classified according to marital status then all
married are put in one group and unmarried in another. If its classification of students heights,
then students who fall in a certain range (group) of heights will be put in one group etc
OBJECTIVES OF DATA CLASSIFICATION

1. To condense the raw data so that similarities and differences / dissimilarities are
brought out
2. To eliminate unnecessary details
3. To be able to utilize data for further statistical analysis
4. To clearly show significant features of the data.
A. ORGANIZING/ CLASSIFYING QUANTITATIVE DATA

The first thing to organize / arrange quantitative data is to prepare a data array. This is an
arrangement of values of a variable in ascending/ descending order. This will help us to know
the range over which the items are spread and get an idea of their general distribution.
EXAMPLE OF A DATA ARRAY (AGES OF 50 WORKERS FROM MTN)

34, 28, 22, 36, 27, 18, 52, 39, 42, 29, 35, 31, 27, 22, 37, 34, 19, 20, 57, 49, 50, 37, 46, 25,
17, 37, 42, 53, 41, 51, 35, 24, 33, 41, 53, 60, 18, 44, 38, 41, 48, 27, 39, 19, 30, 61, 54, 48,
26, 18.
QN: Make a table grouping the data into classes from 10 to 19, 20 to 29, 30 to 39, 40 to 49
etc.
The bulk of such data must be reduced so that the eye can easily understand them. Here, we
represent any repetitions with a tally instead of re-writing them. The frequency will correspond
to the number of tallies for each particular salary. Thus, a Frequency distribution is a list of
values obtained in the data and the frequency with which these values occur in the data.
For grouped data like ages above, we can generate a frequency distribution as below;
13
Age Tally Number of
workers
(Check the total:
10-19 //// / 6
6+11+14+10+7=2=50, this
20-29 //// //// / 11
is correct)
30-39 //// //// //// 14
40-49 //// //// 10
50-59 //// // 7
60-69 // 2
Total 50
It can be noted that we have two types of classes which must be taken seriously; inclusive
and exclusive.
Inclusive class intervals: Here the upper limit of the class interval is included in the
given class interval. These class intervals are used when the variable under consideration
happens to be a discrete one (i.e. takes on one figure at a time) e.g. the number of workers.
11-19 20-29 30-39 40-49 Etc

Note: here the term whose value is 19 will be put in a class interval (11-19) and the one
having a value of 39 will be put in a class interval of (30-39)
Exclusive class intervals: Here, all values equal to the upper class limit are grouped into the
next class. These class intervals are usually used when the variable under consideration is a
continuous one (capable of being measured in fractions like height, age, weight and
so on).
10-20 20-30 30-40 40-50 Etc
Note: here, the term whose value is say 40 won’t be put in (30-40) but will instead fall under
(40-50).
RELATIVE FREQUENCY
It’s the frequency of the class expressed as a percentage of the total frequency. Relative
frequency = [(frequency / Total Frequency) x 100%]
Class Limits
These are the smallest and largest values or observations that do fall in a given class. The
smallest value is the lower class limit and the largest value is the Upper class limit. e.g. (60 -
62) is the class interval, then 60 is the lower class limit and 62 is the upper class limit.
Class boundaries (Real class limits / mathematical limits)

These are values within which all possible observations that fall in a given class interval must
lie. Class boundaries are obtained by adding 0.5 to the upper class limit and subtracting 0.5 to
the lower class limit. E.g. (11 - 20), becomes (10.5 - 20.5) as the class boundary.
Class mark (class midpoint)

This is the midpoint of the class interval. E.g. for an interval of (60-62), the class mark is
(60+62)/2 = 61
14
Cumulative frequency distribution
These helps us to determine the number of units/observations that lie above a given lower
class limit or below a given upper class limit in a frequency distribution. When you are
interested to get the number of units below a specified value then you consider a less than
cumulative frequency distribution. However, when you are interested to get the number of
units that lie above a specified value, then you consider using a more than cumulative
frequency distribution.
Example
Age of Frequency(fi) Cum.freq. Cum.freq.
workers. (less than) (more than)
0-10 2 2
10-20 8 10
20-30 12 22
30-40 18 40
40-50 28 68
50-60 22 90
60-70 6 96
70-80 4 100
Sum =
From the above table it can be noted that:-

Case 1. Cumulative frequency less than. There are 40 people less than 40 years, 90 people
less than 60 years, 10 people less than 20 years and so on.
Case 2. Cumulative frequency more than. …………… people are older than 50 years of age
Cumulative RELATIVE frequency distribution

Used when we want to find out the proportion / percentage of units falling below or above a
particular value score point. If we want to find out the proportion of units falling below the
upper limit of a specified class interval then we apply a less than cumulative relative freq.
distribution. However, if we want to establish the proportion of units falling above the lower
limit of a particular class interval then we apply a greater than cumulative relative freq.
distribution.
Example
Age of (fi) (less than (Less than (More than CF) (More than
workers. CF) Relative. CF) relative. CF)
0-10 2 2 2/100 = 2% 100
10-20 8 10 98
20-30 12 22 90
30-40 18 40 78
40-50 28 68 68/100 = 68% 60
50-60 22 90 32
60-70 6 96 10
70-80 4 100 4 4/100 = 4%
Sum = 100
Case-1. Relative cumulative frequency less than: 68% of the people are below 50 years
of age and so on
Case-2. Relative cumulative frequency more than: …………. of the people are above 40
years of age, and so on. 15
3.3 DATA PRESENTATION
After classifying and tabulating our data, our next task is to present it so that it is easily
understood by management.
There are 4 ways of representing statistical data; Text presentation (data presented
in wordy form), Tabular presentation, Diagrammatic presentation, Graphical
presentation
A. TEXT PRESENTATION (data presented in wordy form). This is when data is

presented in word form. For example; in 2010 the company sold 20,345 crates of sodas
etc.
B. TABULAR PRESENTATION. This is a process of summarizing raw data and displaying it

in compact form for further analysis. A good statistical table should have a title. Usually
placed above the table should be brief yet complete and reflects all the data contained in
the table.
Advantages of tabulation: easy to understand, easy comparison between different
classes of data, easy location of required data, space is saved
Row – column table
A two way table
Frequency and percentage tables
16
C. DIAGRAMMATIC PRESENTATION. This involves bar charts, pictograms and pie charts.
Advantages of diagrams
1. Easy and attractive means of representing the dat
2. Facilitate comparisons
3. Save time and labor
4. Effective impressions
5. Have great memorizing value as compared to mere figures
Limitations of diagrams
1. They don’t give accurate results but rather a rough idea
2. Comparisons of diagrams can’t be made if the unit is not common or the phenomenon is not
the same
3. They can be misused easily
4. It’s expensive to draw diagrams.
BAR CHARTS
A SIMPLE BAR CHART
Makerere CPA Pass rates for June 2016 Pass Rate
Sitting
Paper-1 61%
Paper-2 88%
Paper-3 59%
Paper-4 56%
Paper-5 72%
17
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Paper-1 Paper-2 Paper-3 Paper-4 Paper-5
THE COMPONENT/ SUB-DIVIDED BAR CHART

Makerere CPA Pass rates for various sittings Paper1 Paper2 Paper3
June.2011 80%
66% 81%
Dec-11 72%
65% 83%
Jun-12 44%
55% 61%
Dec-12 56%
66% 77%
Jun-13 61%
88% 59%
250%
200%
150%
100%
50%
0%
June.2011 Dec-11 Jun-12 Dec-12 Jun-13
THE MULTIPLE BAR CHART

Makerere CPA Pass rates for various sittings Paper1 Paper2 Paper3
June.2011 80%
66% 81%
Dec-11 72%
65% 83%
Jun-12 44%
55% 61%
Dec-12 56%
66% 77%
Jun-13 61%
88% 59%
18
Jun-13
Dec-12
Jun-12
Dec-11
June.2011
0% 20% 40% 60% 80% 100%
PICTOGRAMS use pictures or symbols to represent a number of units of data. The pictures
usually relate to the data shown
A PIE CHART is a circle, which is divided by radial lines into sections or subsections of
different angles so that the area of a particular section is proportional to size of the figure
represented.
Makerere University CPA Intake figures for June Number
2016
Paper-1 66
Paper-2 88
Paper-3 93
Paper-4 102
Paper-5 149
19
Paper-1, 66
Paper-5, 149
Paper-2, 88
Paper-4, 102 Paper-3, 93
D. GRAPHICAL PRESENTATION OF DATA.

These include Histograms, line graphs, time series graphs etc
THE HISTOGRAM
It’s a plot of class frequencies against class boundaries. Its body comprises of bars of equal
width whose height represents the class frequency. Unlike in a bar graph, the bars in a
histogram are attached to each other and the width of the rectangle corresponds to the class
width and the height to the class frequencies.
Paper5 Test Marks Mid-Point (X) Frequency

10-14 12 8
15-19 17 28
20-24 22 27
25-29 27 12
30-34 32 4
35-39 37 1
Total 80
A histogram representing test marks

30
25
20
Frequency
15
10
0
10.-14 15-19 20-24 25-29 30-34 35-39
Test Marks
20
We can also construct a histogram for UNEQUAL CLASS WIDTHS by calculating a frequency
density. Frequency density = (frequency / class width)
Example A group of students ran a race. Their times are recorded in the table:
Time (seconds) 16 ≤ t < 18 18 ≤ t < 19 19 ≤ t < 20 20 ≤ t < 22 22 ≤ t < 26

Frequency 6 5 8 7 5
a) Draw a histogram to illustrate the data

b) Find the mode, median, SD, Quartile deviation, coefficient of quartile deviation, & the
coefficient of Skeweness (SK)
Solution
THE FREQUENCY POLYGON GRAPH

It’s a plot of class frequencies against class marks or mid points. Successive points are joined
by straight lines there by forming a line graph.
Two class intervals are added to the distribution one at the beginning and the other at the end
of the distribution for purposes of closing the polygon. From the above table, the frequency
polygon is as follows:-
30
25
20
15
10
0
14.-15 15-19 20-24 25-29 30-34 35-39
THE CUMULATIVE FREQUENCY CURVE (OGIVE)

The OGIVE is of two types. The less than Ogive and, the greater than Ogive
The cumulative freq distribution of statistics results

Age of (fi) (less than (Less than (More than CF) (More than
workers. CF) Relative. CF) relative. CF)
0-10 2 2 2/100 = 2% 100 Fill in
10-20 8 10 Fill in 98 Fill in
40-50 28 68 68/100 = 68% 60 Fill in
70-80 4 100 Fill in 4 4/100 = 4%
Sum = 100
21
Required: - Draw the “less than” and “more than” cumulative frequency curves from
this distribution. The less than Ogive. Here the less than cumulative frequencies are
plotted against the upper class limits /boundaries of each class
The more than Ogive

Here the more than cumulative frequencies are plotted against the lower class limits
/boundaries of each class as shown in the next graph.
Use the previous table to draw the morethan O-Give
A LINE GRAPH
This may be used to show changes in some economic variable, say, steel production over time.
In other words, if out of the two variables, one happens to be time (months, years, etc.), we
get a line graph over time or simply time series graph or histogram. A time series
expresses behavior of an economic variable over time.
22
Example
The table below shows daily temperatures for Kampala City, recorded for 6 days, in degrees
Fahrenheit.
Temperatures In Kampala City
Day Temperature
1 43° F
2 53° F
3 50° F
4 57° F
5 59° F
6 67° F
The data from the table above has been summarized in the line graph below.
Temperature in Kampala city
Z Chart
Z-charts: A chart often used in industry and constructed by plotting on it three series:
monthly, weekly, or daily data, the moving annual total, and the cumulative total dating from
the beginning of the current year
Z-chart: How to do it
1. Gather data on a regular basis for a given activity. This may be the sales per month, words
written by an author per week or calls handled by a customer response centre per day.
2. Identify the major review period to be considered. When you are gathering data on a
monthly cycle, then this is likely to be 12 months. If your minor review period is one day, then
the major review period is more likely to be something like one month or four weeks.
2. For each minor review period (one month in the examples here), build the longer-term view
by totaling the data for the past major review period (12 months in the examples here).
Note that the rolling total for 12 months ago will require data from 12 months previous to that.
Thus the longer-term view really does look back over a longer period.
3. For the current review (month, in the example here), sum the previous values over the past
major review period (12 months here).
4. Build the Z-chart, as in the diagram above. 23
The table below shows how these measurements can all be put into a simple table. You can
use a spreadsheet to automatically build the third and fourth column and also to subsequently
build the final Z-chart.
This Total since Rolling 12

month March month total
March 5 5 46
April 2 7 45
May 5 12 47
June 12 22 46
July 3 25 48
August 5 30 47
September 7 37 49
October 2 39 50
November 4 43 51
December 2 45 50
January 7 52 58
February 3 55 55
Read about the Lorenz chart

http://fcweb.limestone.on.ca/~stridef/World%20Issues/Unit%203%20-%20Lesson%204%20-
%20Lorenz%20Curve.pdf
24
HAPPY HOUR: NOW TRY OUT THESE MODEL EXAMINATION NUMBERS
Question-1
According to the 1990 Uganda Statistical Abstract, 9,000,000 people were employed in 1989
by the government of Uganda. Of these 3,000,000 were employed in the education sector,
3,500,000 were employed in the civil service and the rest were in the armed forces. In 1988
the number employees reported were 8,000,000 of which 2,500,000 were in the education
sector, 2,000,000 were in the civil service and the rest were in the armed forces. The same
abstract gave the figure for the total number of government employees in 1987 as 7,500,000
distributed as follows: education sector 3,200,000, civil service 2,000,000 and 1,200,000 in the
armed forces. Present the above information in a tabular format.
Question-2
It has been estimated that the level of late coming of employees at Roofing Limited is affecting
the company’s performance. Given below is a record of the time in minutes of 60 employees
who come late on a particular day.
45 55 20 15 16 17 05 09 10 15
57 12 46 30 32 48 49 59 20 21
26 30 31 19 14 09 03 01 22 23
30 37 38 36 35 42 44 48 47 46
15 18 21 32 31 30 38 39 40 41
17 32 33 36 22 34 35 51 43 31
I. Beginning with a class interval of 0-9, 10-19 construct a frequency distribution of 6 classes
with uniform class width.
II. It the company’s policy is to write a warning letter for those who come late after 30 minutes,
what percentage of employees will be served a warning letter on that day?
III. Construct a histogram and estimate the modal time of late coming
IV. Extract a cumulative frequency curve
Question-3
Weight in kilograms 60 - 65 - 70 - 75 - 80 - 85 -
Number 5 18 42 27 8 4
I. Construct a histogram and estimate the mode

II. Use a suitable diagram to estimate the median
III. Use the same diagram to estimate the 1st and 3rd quartile
25
Chapter-4
Measures of location/ averages / central tendency
4.1 INTRODUCTION
A measure of central tendency is a sample value around which the distribution of data is
centered. Or a single value which can neatly characterize the whole group of data. Includes,
the mean, mode and median
A Parameter and a Statistic: A parameter is a number describing something about a

whole population. E.g, population mean, population variance or population mode etc. A
statistic is something that describes a sample (eg sample mean, sample variance) and is used
as an estimator for a population parameter (because samples should represent populations!)
4.2 THE ARITHMETIC MEAN

In this syllabus, we are more concerned with grouped data. The formulae for computing an
arithmetic mean for grouped data is;
X 
  f .x 
f
Example1: The following set of raw data shows the salaries of CPA students from the
November 2019 class. The facilitator is arguing that on average everybody in the class gets 46
million per year. Is the lecturer correct?
Salaries (Millions) (fi)
0-10 2
10-20 8
20-30 12
30-40 18
40-50 28
50-60 22
60-70 6
70-80 4
Sum = 100
Solution
Salaries (Millions) f Mid-point (x) f .x
0-10 2 ….. ….
10-20 8
20-30 12
30-40 18
40-50 28
50-60 22
60-70 6
70-80 4
f  f .x 
x
 f .x 

 
f  26
4.3 ASSUMED/ WORKING MEAN: Is a method that uses an assumed mean in order to
calculate actual mean. This method does cut short the calculations that are more in the direct
method. The formula used for estimating mean of grouped data by using assumed mean
method is given below:
x  A
 f .d
f
Example: Assume any mean that lies between 40 -44 and calculate the actual mean using this
working mean method
Salaries f Mid-point Deviation f .d

(Millions) (x) (d)
0-10 2 …
10-20 8 ….
20-30 12 ….
30-40 18 ….
40-50 28
50-60 22
60-70 6
70-80 4
f  f .d 
x  A
 f .d   

     
f 
4.4 THE WEIGHTED MEAN: Is an average in which each quantity to be averaged is

assigned a weight. These weightings determine the relative importance of each quantity on the
average. Weightings are the equivalent of having that many like items with the same value
involved in the average.
n
w x i i
xw  i 1
n
w i 1
i
Note: when data values vary in importance, the analyst must choose the weight that best
reflects the importance of each data value in the determination of the mean.
Example: The following are 5 purchases of a raw material over the past 3 months.
Purchase Cost per Kilogram Number of Kilograms
($)(x) (w)
1 3.00 1200
2 3.40 500
3 2.80 2750
4 2.90 1000
5 3.25 800
Find the mean cost per kilogram. 27
Purchase Cost per Number of (WX)
Kilogram ($)(x) Kilograms (w)
1 3.00 1200
2 3.40 500
3 2.80 2750
4 2.90 1000
5 3.25 800
SUM= SUM=
5
w x i i
.............
xw  i 1
5
  2.96
w
............
i
i 1
4.5 GEOMETRIC MEAN (GM): Is a specialized measure for averaging proportional

increases in wages, goods, production etc? If n is large, we use logarithms instead of just
taking the nth root of the product.
GM = Antilog of 
 f . log X 
f
Where x is the midpoint of a particular class
Marks x f f .Logx
25 – 29 27 2 2log27=……..
30 – 34 32 4 …….
35 – 39 .. 7 …….
40 – 44 …. 10 ……
45 – 49 …… 8 ……
50 – 54 …… 6 …….
55 – 59 ……. 3 ……..
Sum =  fLogx  .............
Therefore,
  fLogx 
Geometric _ Mean _ of _ X  AntiLog   =………………………….
 f 
 
4.6 HARMONIC MEAN (HM): Used when data given consist of a set of rates such as
Shs/kilogram, speed/hour, and output/man U hour etc. It is the reciprocal of the mean of
reciprocals of the item values.
Harmonic _ Mean 
f
f 
  x 
28
Marks x f f
x
30-39 34.5 2
40-49 44.5 3
50-59 .. 11
60-69 .. 20
70-79 .. 32
80-89 .. 25
90-99 .. 7
f  100 f
  x   1.4368
100
Harmonic _ Mean   69.60
1.4368
ADVANTAGES OF THE ARITHMETIC MEAN

It’s easy to calculate, it can easily be manipulated to calculate other useful statistical measures,
and it uses the values of all the observations.
DISADVANTAGES OF THE ARITHMETIC MEAN

A few extreme values can distort the mean making it unrepresentative of the data set, it can’t
be read from the graph, when the data is discrete, it may produce a value that appears to be
unrealistic.
4.7 PROPERTIES OF ARITHMETIC MEAN

a) It can be calculated on any data set and thus it always exists
b) It’s very sensitive to extreme values and this makes the mean unrepresentative of the data
c) A set of numerical data has only one mean and so the mean is unique
d) It takes into account all observations in the data set
e) It can be used for further statistical treatment, e.g. mean for several sets of data can be
combined into an overall mean for all data
f) It is relatively reliable in the sense that means of several samples drawn from the same
population do not vary as widely as other statistics used to estimate the population mean
g) The sum of the deviations of a set of numbers from their mean is zero, It cannot be
calculated for open intervals
h) The sum of squares of the deviations from the mean are minimal, i.e. sum of squares of
deviations from the mean is < sum of squares of deviations from any other observation.
4.8 CALCULATING THE CORRECT MEAN FROM WRONG MEAN

Example: The mean age of a group of 100 employees at KCC was found to be 32 years. Later
it was discovered that that age 57 was misread as 37. Find the correct mean age
29
Solution
4.9 THE MEDIAN

It is a value of a random variable that divides an ordered data set into two equal parts. It is an
alternative average to the mean. Can be estimated using an interpolation formulae.
Interpolation is the mathematical technique for estimating an unknown value by utilizing

immediately surrounding known values.
 N  CFb 
For grouped data; the median = Lm  2  xc
 fm 
 
Where, Lm is the lower class limit for the median class
CFb is the cumulative frequency before you reach the median
Fm is the frequency of the median class
C is the width of the median class interval
Example: Compute the median for the following marks

Lengths (mm) Frequency ( f i ) Cumulative frequency less than
25 – 29 2 2
30 – 34 4 6
35 – 39 7 13
40 – 44 10 23
45 – 49 18 41
50 – 54 6 47
55 – 59 3 50
Total = 50
The location of the median class is obtained by dividing the total number of observations
by 2: Therefore ( N / 2)  (50 / 2)  25 th . The median lies in the 25th position. Then where
does the 25th term fall? This can be answered by determining the cumulative frequencies
for the data. When the table is closely examined, the 25th term falls in the fifth -class
interval since there are only 23 values in the first two class intervals. The median value lies
somewhere in the third-class interval (45-49).
Lm=45. And the cumulative frequencies up to but not including the median class interval is
CFb=23, Fm=18, C=5
30
Therefore;
Median =
The median can also be obtained from the graph (OGIVE) as follows
Space to Insert an O-Give
4.11 ADVANTAGES OF THE MEDIAN

It’s an appropriate alternative to the mean when extreme values are present at one or both
ends of the distribution or data set. E.g.: (158, 138,141,148,148,146,157,252), the mean was
previously 161 which is unrepresentative of the data set, but if the values are arranged in
ascending order the median is 148 which is at least a good representative measure for this
average, It can be used when certain end values of the distribution are difficult, expensive, or
impossible to obtain and the basic example here is life data, Can be used where non numeric
data is available unlike the mean that can’t be used in this case. E.g., the median size of a shirt
can be determined if all shirts are measured in terms like large, extra large, medium etc, Will
often assume a value equal to one of its original items which is considered as an advantage
over the mean.
4.12 DISADVANTAGE OF THE MEDIAN

The median is difficult to handle theoretically in more advanced statistical work so its use is
restricted to analysis at a basic level, In a grouped frequency distribution, the value of the
median within the median class is only an estimate whether it’s calculated or estimated from
the graph.
4.13 QUARTILES, DECILES AND PERCENTILES

The Nth percentile of an array of data is the value of the item such that N percent items lie
below it.
31
 % N  CFb 
For any grouped data, the Nth Percentile Pn is calculated as; L p    xc

 fp 

Quartiles are those observations which divide an ordered data set into quarters (4 equal parts).
Qi  Lq   4
 
 n i  Cf b 
 xC
 fi 
 
Quartile deviation = [(Q3 – Q1)/2] Also called the semi- inter-quartile range
Inter-quartile range = (Q3 – Q1)
Coefficient of quartile deviation = [(Q3 – Q1)/(Q3+ Q1)]
Likewise, an Nth decile is a value of the item below which N/10 items lie.
Di  Ld   10
 
 n i  Cf b 
 xC
fi 
 
Example: From this data, compute the quartile deviation
32
Practice question
Profits (Shs) 5- 9 10-14 15-19 20-24 25-29 30-34 35-39
Frequency 8 18 27 21 10 28 8
33
From the information given below, calculate the;
a) 90th and 10th percentile
b) Quartile deviation, inter-quartile range, & coefficient of quartile deviation =
c) What is the 10th decile in the data given
4.15 THE MODE

It is a value occurring most frequently in a series or group of items and around which the
other items are distributed most densely. Used to measure a representative average value in
terms of ‘popularity’ E.g. If you asked a shop attendant at Anisuma Traders that; what price
does an average television set sell here? He is likely to give you the price of the best selling TV
brand. This is the mode. Here mode is more representative than even the mean or median.
Other types of data where the mode is sometimes used as an average are shoe & cloth sizes,
number of defectives found on a production line, or size of a company by number of
employees.
 D1 
Mode = Lm    xC
 D1  D2 
Where, Lm is the lower class limit of modal class, D1 is difference between the largest
frequency & frequency immediately preceding it. D2 is the difference between the largest
frequency & frequency immediately following it. C is the modal class width.
Example: Compute the median for the following marks

Marks Frequency ( f )
20-30 6
30-40 (modal class) 18 (modal frequency
40-50 11
50-60 11
60-70 3
70-80 1
f  N  50
 12 
Mode = 30    x10  36.32
12  7 
34
We can also obtain the mode from the graph (Histogram)
Space to insert a histogram which we can use to obtain the mode
Histogram
4.16 CHARACTERISTICS OF THE MODE

a) It’s usually used as an alternative to the mean and median when the situation calls for
the most popular item in the data set
b) Easy to understand, not difficult to calculate and can be used when the distribution has
open ended classes
c) Although the mode usually ignores isolated extreme values
d) It’s thought to be too much affected by the most popular class when the distribution is
significantly skewed
e) Sometimes it might not exist
f) When the set of items all have different values. Or, might not be unique when there are
two or more values that do have the highest frequency
g) Unlike the mean and the median, it has no natural measure of dispersion to go with it
which is a particular disadvantage in most cases where further analysis is required
h) Like the median, the mode is not used for further statistical work.
The mode can also be obtained from the O-Give by graphically drawing a perpendicular
to the base from the point on the curve where the curve is steepest/ most nearly vertically.
But, the mode is often estimated from a histogram.
We can also estimate the mode from the formulae. Mode = 3 Median – 2Mean
35
Chapter-5
Measures of dispersion
5.1 INTRODUCTION
These show the extent to which the values in a distribution are spread around their summary
figures like the Mean, Median and Mode.
5.2 ABSOLUTE MEASURES OF DISPERSION (TYPE1)

A measure of dispersion is in absolute form when it states the actual amount by which the
value of an item on an average deviates from a measure of central tendency. These include;
Range, Inter-quartile Range, Semi-quartile range, Quartile deviation, Mean deviation, Variance
and Standard derivation. Such measures can’t be used to compare the variability of the two
distributions which are expressed in different units of measurement.
5.2.1 THE RANGE (HIGHEST VALUE – LOWEST VALUE)

Advantages
It’s easy to compute and requires very little calculations.
Disadvantages
It does not tell us the arrangement of the observations that do fall in between the two extreme
values, it may be misleading if either of the extreme values is an outlier, it can’t be used if we
are dealing with open end classes; it’s not suitable for mathematical treatment. Range for
grouped data is obtained by subtracting the lower limit of the smallest class from the upper
limit of the highest class.
5.2.2 QUARTILE DEVIATION (ALREADY COVERED)

5.2.3 MEAN DEVIATION. This gives the mean of absolute values of the deviation from
some measure of central tendency.
N
 f i xi  x  f x  x
MD  i 1 
N N
Example: Compute the mean deviation in the sales of this company
Class F x  mean f x  mean
X (F . X)
0-4 2 1 2 13.3 13.3
5-9 7 14 98 8.3 ----
10-14 12 23 276 ---- ---
15-19 17 21 357 ------ ---
20-24 22 15 330 ----
25-29 27 6 162 ----
Total 80 1225 ----
Mean 
 fx  1225  15.3 : Thus MD  5.1Sales
N 80
36
5.2.4 COEFFICIENT OF MEAN DEVIATION
= (MEAN DEVIATION / MEAN)
5.2.5 STANDARD DEVIATION / VARIANCE

The standard deviation is a statistical measure which expresses the average deviation about
the mean in the original units of the random variable. It’s the square root of variance.
Variance is defined as the mean of the squared deviations of individual observations from
their arithmetic mean.
If we consider a sample, the standard deviation is given by:-
 f x  x   fx   fx  n fx 2   fx 
2 2 2 2
 
 OR s    OR: s 
2 2 2
s
n 1 n  n  nn  1
If we consider a population, the variance is given by:-

 f  x   2  fx 2   fx 
2
 2
 OR  
2
 
N N  N 
Example: Consider a POPULATION of sales from Sadolin Paints in billions
Class X F (F . X) (F.X2)
0-4 2 1 2
5-9 7 14 98 …
10-14 12 23 276
15-19 17 21 357
20-24 22 15 330
25-29 27 6 162 …
Total 80 1225 Total =
First, obtain the mean as follows
x
 f .x 

 
f 
Then proceed to obtain the variance as follows
 fx   fx 
2 2
Var  S 2
    .......
N  N 
Note: If the examiner specifies that the data relates- to a sample; then we use n-1 as the
denominator as follows:
n fx 2   fx 
2
s  2
.........
nn  1
37
Computation of mean deviation, standard deviation and variance using the working
mean method
Advantages of the using the standard deviation

It uses all the observations, is closely related to the most commonly used measure of location-
the mean, it’s easy to manipulate mathematically.
Disadvantages: It’s complicated to define and calculate, its value can be distorted by
extreme values.
RELATIVE MEASURES OF DISPERSION (TYPE2)

These are obtained as ratios or percentages and are thus pure numbers independent of the
units of measurement. In order to compare the variability of two distributions (even if they
have the same units), we compute the relative measures of dispersion instead of the absolute
measures of dispersion.
COEFFICIENT OF VARIATION: Sometimes, we may need to compare the variability in two

samples of data from different random variables. A direct comparison of their respective
standard deviations/ variances can be misleading since the random variables might be
measured in different units. Standard deviation & variance comparisons are possible if the two
variables were measured in similar units. When units of the samples compared are different,
we use the coefficient of variation (CV) as a measure of relative variability. Expressed in
percentage form
s   
And for a sample, v   x100  and v   x100  for the case of a population.
x   
The bigger the value, the greater the dispersion
Example on coefficient of variation

The following are scores of two candidates who applied for the post of financial controller in
standard chartered Bank
Trump 12 115 6 73 7 19 119 36 84 29
Truth 47 12 76 42 4 51 37 48 13 0
Which of the two candidates performed better and which one is more consistent?
Tip: compute the mean & SD for ungrouped data: Use Mean = SUM(X)/N. And Variance =
Sum (X-Mean)2/N. Then proceed and compute CV = (SD/Mean)100%
38
USES OF MEASURES OF DISPERSION IN BUSINESS
To find out the reliability of an average, to control the variation of data from the central value,
to compare two or more sets of data from their central value, to obtain other statistical
measures for further analysis of data.
AN IDEAL MEASURE OF DISPERSION MUST HAVE THE FOLLOWING

CHARACTERISTICS
a) It should be rigidly defined, should be easy to calculate and easy to understand, should
be based on all observations
b) Should be amenable to further mathematical treatment
c) Should be affected as little as possible by fluctuations of sampling and it should not be
much affected by extreme observations.
ADJUSTING VALUES OF MEAN AND STANDARD DEVIATIONS (SD) FOR MISTAKES

Example1: The mean and standard deviation of a set of 100 observations were worked out as
40 and 5 respectively by a computer which by mistake took the value 50 in the place of 40 for
one observation. Recalculate the correct mean and correct variance
Example2: The number of employees and variance of wages per employee for two factories
are given below:
Factory A Factory B
N0. Of employees 50 100
Average wage per employee 120 85
Variance in wages per employee 9 16
I. In which factory is there greater variance in the distribution of wages per employee?
II. Suppose in factory B, the wages of an employee were wrongly written as 120 instead
of 100, what would be the corrected variance for factory B?
Example3: A sample of 10 members gave a mean of 13 and variance of 4. Later it was

discovered that the number 12 included in the sample should have been 21. Find the corrected
mean and variance?
39
MEASURES OF SKEWNESS AND KURTOSIS
SKEWNESS When a frequency distribution is not symmetrical, it is said to be asymmetrical or
skewed. Thus, skewness is a measure of symmetry, or more precisely, the lack of symmetry.
The direction of variation of the data set (skewness) can be measured by the (Pearson’s
coefficient of Skewness: SK).
SK = [3(Mean - Median)]/ Standard Deviation. Where (-3 < SK < +3)
OR SK = (Mean – Mode)/ Standard Deviation
Interpretation
If SK = 0, then the set of observations is symmetrical and normally distributed. For a
symmetrically distributed population or sample, the (mean = median = mode). Half of all
measurements are greater than the mean, while half are less than the mean.
If SK > 0, then the set of observations is positively skewed (positive skewness): Also
implying that a greater proportion of the observations are less than or equal to (as opposed to
greater than or equal to) the mean. The presence of extreme observations on the right hand
side of a distribution makes it positively skewed. Mean > Median > Mode
If SK < 0, then the set of observations is negatively skewed indicating that more observations
are greater than or equal to the mean. The presence of extreme observations to the left hand
side of a distribution make it negatively skewed and the relationship between mean, median
and mode is: Mean < Median < Mode.
40
For moderately skewed distributions, we can get the relationship between the mode, the
median and the mean. In case we have one of the 3, we can compute the other using the
formulae:- Mode ═ 3(median) – 2(mean): An example skewed distributions is the wage
distribution in which we find that most of the wages are skewed to the right.
Advantage of SK: It is independent of the scale. Because (mean-mode) and standard deviation
have same scale and it will be canceled out when taking the ratio. Disadvantage of SK: It
depends on the extreme values.
Example: Comment on the level of skewness using Karl Pearson coefficient of skewness
BOWLEY’S COEFFICIENT OF SKEWNESS

This measure is based on quartiles. For a symmetrical distribution, it is seen that Q1, and Q3
are equidistant from median (Q2). Thus (Q3 - Q2) - (Q2 - Q1) can be taken as an absolute
measure of skewness.
Properties of Bowley’s coefficient of skewness
Advantage: Skq does not depend on extreme values. Disadvantage: Skq does not utilize the
data fully.
Example: The following table shows the distribution of 128 families according to the number
of children.
41
Find the Bowley’s coefficient of skewness
Solution
42
KURTOSIS: This is another measure of the shape of a distribution. Whereas skewness
measures the lack of symmetry of the frequency curve of a distribution, kurtosis is a measure
of the relative peakedness of its frequency curve. Various frequency curves can be divided into
three categories depending upon the shape of their peak. The three shapes are termed as
Leptokurtic, Mesokurtic and Platykurhc as shown in Figure below.
43
For normal distribution, k ≈ 0.25 i.e. mesokurtic
A. I f K< 0.25, the distribution is platykurtic
B. I f K> 0.25, the distribution is leptokurtic
HAPPY HOUR: NOW TRY OUT THESE VERY SIMPLE NUMBERS

Question 1
a) Explain the difference between the following terms as used in quantitative analysis
I. Discrete and continuous variable
II. Sampling and a census
III. Systematic and quota sampling
IV. Data classification and data array
V. A class limit and class boundary
VI. Inclusive and exclusive classes
b) Given the following values of X (Ungrouped data)

84 92 73 67 88 74 91 74 70 88 82 70 73 88 92 91
Find; 1) Mean, 2) the mode, 3) the median, 4) variance, 5) coefficient of variation, 6) mean
deviation and Skewness
Question 2
Weight in kilograms 60- 65- 70- 75- 80- 85-
Number 5 18 42 27 8 4
I) Compute the arithmetic mean

II) Compute the geometric and harmonic means
III) Assume the mean to lie between (70--), use any assumed/ working mean to compute
the mean value
IV) Compute the median and modal weights
V) Compute the variance in weights
VI) What is the coefficient of variation in ages
VII) Comment on the level of skewness in weights
VIII) Percentage of people above 75 years
Question 3
a) A ‘standard deviation is a natural partner to a mean’ discuss the meaning of this statement
whilst highlighting any two properties of an arithmetic mean
b) A machine produces circular bolts and, as a quality control test, 250 bolts were selected
randomly and the diameter of their heads measured as follows:
44
Diameter of head (cm) Number of components
0.9747 - 0.9749 2
0.9750 - 0.9752 6
0.9753 - 0.9755 8
0.9756 - 0.9758 15
0.9759 - 0.9761 42
0.9762 - 0.9764 68
0.9765 - 0.9767 49
0.9768 - 0.9770 25
0.9771 - 0.9773 18
0.9774 - 0.9776 12
0.9777 - 0.9779 4
0.9780 - 0.9782 1
I. Determine whether the customer is getting reasonable value if the label on the
circular bolt advertises that the average diameter of the head is 0.97642 cm.
II. What is the quartile deviation in diameters of the bolts?
III. Buyers have threatened to return any bolts below 0.9765 cm. What
percentages of bolts were returned to the factory?
Question-4
The following data relates to sales of Besigye enterprises limited
Profits (Shs) 5- 9 10-14 15-19 20-24 25-29 30-34 35-39
Frequency 8 18 27 21 10 28 8
Find; 1) the mode, 2) the median, 3) variance, 4) coefficient of variation, 5) mean deviation,
6) upper quartile and 7) lower quartile, 8) Coefficient of quartile deviation 9) 80% percentile
Question-5
a) Differentiate between the following terms as used in measures of dispersion:
i) Platykurtic versus leptokurtic and mesokurtic.
ii) Coefficient of skewness and coefficient of variation
iii) Quartile deviation and mean deviation
b) The following data relates to sales of Matete enterprises limited
Profits (Shs) 10-19 20-29 30-39 40-49 50-59 60-69 70-79 80-89
Frequency 6 M 4 N 8 P 2 Q
Cumm: Freq A 10 B 18 C 30 D 35
Find;
I) Fill in the missing values
II) Compute the Geometric Mean and 2) Harmonic Mean 3) Weighted Mean 4) Mean
deviation
III) Estimate the semi- interquartile range
IV) Compute Bowleys measure of skewness and interpret the value
V) Construct an O-Give and estimate the median profits
VI) Construct a Histogram and estimate the modal profits
45
Question-6
The following data relates to sales of Besigye enterprises limited
Profits (Shs) 16-17 18-20 21-24 25-29 30-34 35-44 45-54 55-74
Frequency 4 73 185 104 34 33 22 26
Note the class width are not the same
Find;
I) Construct a histogram and find the mode
II) Compute the average sales
III) Compute the risk (Standard deviation) in sales
IV) Construct a line graph
Question-7
a) As part of a class assignment in Financial Accounting, Kamese has compiled the
following profits made by Mabreezy Ltd and Kyalya Ltd enterprises respectively (In
Millions of Uganda Shillings).
Year 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Mabreezy Ltd 12 115 6 73 7 19 119 36 84 29
Kyalya Ltd 47 12 76 42 4 51 37 48 13 0
By computing the arithmetic mean and a coefficient of variation for profits of each company,
which of the two companies:- 1) Performed better 2) Was consistent in profitability?
b) Plot line graphs for the two profits of two companies

c) Explain the difference between the following terms as used in quantitative methods
I) A geometric mean and harmonic mean
II) A mode and a median
III) A measure of central tendency and measure of dispersion
IV) A coefficient of variation and measure of Skewness: A data array and class limit
d) When do we use; A geometric mean, A harmonic mean

e) What are the qualities of a good measure of dispersion?
f) Discuss the advantages and disadvantages of the median and mode as measures of
central tendency
YOU MUST BE FULL OF HAPPINESS AFTER UNDERSTANDING THIS TOPIC. NOW TRY OUT
THESE VERY SIMPLE PAST PAPER NUMBERS
EXTRACT THESE NUMBERS FROM THE QUESTION BANK
Sitting Number Sitting Number Sitting Number
June 2016 November 2013 December 2010
November 2015 June 2013 June 2009
August 2015 November 2012 December 2009
June 2015 June 2012 June 2008
November 2014 December 2011 December 2008
June 2014 June 2010
46
Chapter-6
Introduction to probability
6.1 INTRODUCTION
All future events are uncertain to some degree. Example; That the present NRM government
will still be in power in a year’s time (given that it is not an election year) is likely, but far
from certain; that a conservative party, led by Ken Lukyamuzi will be in power in a year’s time
is highly unlikely, but not totally impossible. Probability theory enables the difference in
the uncertainty of events to be made more precise by measuring their likelihood on a scale.
Definition: Probability is the chance that an event will happen. Probabilities are
expressed as fractions or decimals. They range between 0 – 1 inclusive. Assigning a
probability of 0 means that something can never happen while 1 means that something will
always happen.
6.2 INTERPRETING PROBABILITIES

A. Probabilities are numbers between (0 and 1 inclusive). They give us an idea of whether or
not a physical event will occur.
B. Probabilities near 0 indicate that the event in question is unlikely to occur or rarely to occur
(example; chance/probability of someone swimming across the Atlantic ocean is 0)
C. Probabilities near 1 indicate that the event in question is likely to occur (Eg. Probability of
having at least one accident in Kampala next year) OR Probability of someone dying at
Mulago hospital during next year)
D. Probabilities near (½) indicate that the event in question has about the same chance of
occurring as it has of failing to occur.
A. TYPES OF PROBABILITY
A. CLASSICAL/ A PRIORI PROBABILITY:
Definition1: Either, A type of probability based on the assumption that the outcomes of an
experiment are equally likely. So we obtain the probability of an event taking place by
dividing the number of favorable outcomes by the total number of total outcomes.
Probability of the event A happening = Number of favorable outcomes (NA)

Total number of outcomes in sample
space (N)
Definition2: OR A method where the probability of an event is calculated by a

process of logic: No experiment or judgment is required. Probabilities involving coins, dice,
playing cards, etc., can fall into this category.
Example1: Using the classical probability concept

In 2012, a lecturer taught Business Quantitative Techniques and Operations Management and
has prepared a grade distribution summary per course for the next departmental meeting.
47
Course taught
Grade Business Quantitative Operations
Techniques Management
A 7 8
B 9 10
C 11 12
D 6 9
O 5 8
If a student, who was in the above class is randomly selected;

Required:
Find the probability that the student:-
i. Received grade A
ii. Was in the Business Quantitative Techniques class
iii. Was in the Business Quantitative Techniques class and received grade A
Solution
Course taught
Grade Business Quantitative Operations TOTAL
Techniques Management
A 7 8 15
B 9 10 19
C 11 12 23
D 6 9 15
O 5 8 13
Total 38 47 85
Note this is a classic example of classical/ aprior probability that is got by dividing the number
of favorable events out of the total events.
NoOfFavourableEvents
That is to say; Probability Pr obability 
TotalEvents
15
I. Received grade A : P( A)   0.176
85
38
II. Was in the Business Quantitative Techniques class P( A)   0.447
85
III. Was in the Business Quantitative Techniques class and received grade A
7
P( A)   0.0823
85
B. RELATIVE FREQUENCY APPROACH

Definition: A relative frequency probability is a type of probability in which we divide the
number of events in a population that meet a particular condition by the total number in the
population. Such computed probabilities indicate how often an event will occur compared to
other events.
Probability of the event A happening = Number of times event A has been

occurring
Total number of times the experiment
was performed 48
For example, to estimate the probability of rain on a given day in September in Kampala, look
at the last ten years’ records to find that it rained on 57 days. Then:
P (Rain) = (No. of days rain recorded/ Total no. of days (= 10 × 30) = (57/ 300) = 0.19
C. SUBJECTIVE/ PERSONAL PROBABILITY

Definition: A type of probability that expresses an individual’s degree of belief about the
chance that an event will occur.
Example1: Let the event be the achievement of an African Union by the year 2016. There is
no way that either of the first two approaches could be employed to calculate this probability.
However, an individual can express his or her own feeling on the likelihood of this event by
comparing it with an event of known probability.
Example2. Predicting that it will rain tomorrow based on previous weather data is an
experiment yielding subjective probability values.
OTHER EXAMPLES IN RESPECT OF THE THREE PROBABILITY TYPES

1. What is the probability of throwing a 6 with one throw of a dice? Answer: A priori approach
– there are six possible outcomes: 1, 2, 3, 4, 5 or 6 showing. All outcomes are equally
likely, therefore: P(Throwing a 6)= (1/6)
2. What is the probability of a southern by pass for road vehicles being completed by 2025?
Answer: The subjective approach is the only one possible, since logical thought alone
cannot lead to an answer and there are no past observations. My assessment is a small
one, around 0.02.
3. How could the probability of obtaining a head on one spin of a biased coin be calculated?
Answer: The a priori approach may be possible if one had information on the aerodynamic
behavior of the coin. A more realistic method would be to conduct several trial spins and
count the number of times a head appeared:- P (obtaining a head) = (N0 of observed
heads / N0 of trial spins)
4. What is the probability of drawing an ace in one cut of a pack of playing cards? Use the a
priori method. There are 52 possible outcomes (one for each card in the deck) and the
probability of picking any one card, say the ace of diamonds, must therefore be (1/52).
There are 4 aces in the pack, hence: P(Drawing an ace) = (4/52) = (1/13)
KEY TERMS USED IN PROBABILITY.

A. Experiment. It’s any physical action/process that is observed and the outcome is
noted.
Or. It’s the activity that produces an event. In a coin toss we say it is a coin toss
experiment producing two events: head and tail.
B. A random experiment: A process leading to two or more possible outcomes without
knowing which outcome will occur. Take an example of a business manager who is not
certain whether an investment in Juba will yield any profits.
Examples:
1. A coin is tossed and the outcome is either a head or a tail.
2. A six sided die is rolled and any figure on the die can come on top but you don’t know
which one will appear on top.
3. A customer enters a shop and either purchases a shirt or does not.
4. Etc.
49
So in each of the above random experiments we can specify the outcomes. So the different
possible outcomes are called basic outcomes. The possible outcomes of a random experiment
are called the basic outcomes and the set of all basic outcomes is called the sample space
usually denoted by S.
C. Sample space “S”. It is a set of all possible outcomes of a random experiment; e.g. in
tossing a coin, the sample space is either a tail or head. (H, T). In many cases, we are
interested in some subset of the basic outcomes and not the individual outcomes. For
example if we roll a die we might be interested in whether the outcome is even i.e. 2, 4
or 6. These are evens. So an event is a:-
D. An event. An event “E” is any subset of the basic outcomes from the sample space.
OR: It is one or more of the possible outcomes of doing something, e.g. for tossing a coin, the
event is a head or tail.
E. A definite event. Something that is certain to happen

F. Intersection of events: If A & B are two events in the sample space S, their
intersection is denoted by A∩B and is the set of all basic outcomes in S that belong to
both A and B. Hence, the intersection occurs if and only if both A and B occur.
G. Mutually exclusive events. Events are mutually exclusive if one and only one of
them can take place at a time. So the two events say A and B have no common
basic outcomes. So the intersection between these events is 0 since they can’t occur
together. Hence in a coin toss, you either get a tail or head but not both at the same
time. Hence, they are mutually exclusive.
H. Union of events: If A and B are two events in the sample space S, their union is
denoted by AUB and is the set of all basic outcomes in S that belong to at least one of
these two events. Hence the union occurs if and only if either A or B or both occur.
I. Complement of an event: If A is an event in the sample space S, the set of basic
outcomes of a random experiment belonging to S but not to A is called a complement
of A and denoted by A1
J. Independent events. These are events in which the probability of occurrence or non-
occurrence of one is in no way affected by the occurrence non-occurrence of the other.
K. Dependent events. These are events in which the probability of occurrence or non-
occurrence of one event is affected by the occurrence or non-occurrence of the other
event.
BASIC RULES & PRINCIPLES OF PROBABILITY

In order for us to develop the framework for assessing and manipulating probabilities, we need
to first set down some rules.
Rule1. The probability of an Event E, occurring must be at least or, at most, 1, i.e.
0  P( E )  1 . That’s to say; P( E )  0
Rule2. If all possible events are mutually exclusive, the sum of their probabilities is equal to 1.

n
i 1
P( Ei )  1
Question: Three individuals A, B and C have applied for the position of accounts assistant in
Nile Breweries. Candidate A and B have the same change of being hired. The probability of
candidate C being hired is twice that of candidate A. Find the probability that candidate hired
is: 1) A and 2) B
Answer 50
Rule3. The general additional rule
If two mutually exclusive events (A&B) occur, the probability that one or the other will
occur equals the sum of their individual probabilities,
P( A  B)  P( A)  P( B)
If they are non-mutually exclusive events then they can occur at the same time. The
probability of their union is: P( A  B)  P( A)  P( B)  P( A  B) .
The addition law can be extended to cover more events, e.g. for three events A, B and C: P(at
least one of A or B or C occurring) P(AUBUC) and we find that;
P( A  B  C )  P( A)  P( B)  P(C )  P( A  B)  P( A  C )  P( B  C )  P( A  B  C )
Unless you like learning formulae, do not bother to remember this as you can always solve
problems involving three events more easily using the Venn diagram.
Example-1: Three Newspapers, Monitor, New Vision and Observer are published in Kampala.
Of the adult population of Kampala, 65% read Monitor, 40% read New Vision, 25% read
Observer, 20% read both Monitor and New Vision, 20% read both Monitor and Observer, 10%
read both New Vision and Observer and 5% read all three. What is the probability that an
adult selected at random reads at least one of the magazines?
Solution
Definition of probabilities
P (M) = 0.65 P (N) = 0.40 P (O) = 0.25
P(M  N )  0.20 P(M  O)  0.20 P( N  O)  0.10
P( M  N  O)  0.05
The probability of reading at least one of the newspapers is the Union of all the three sets:
From the above diagram; total people who read at least one of the newspapers is equal to:-
30%+15%+15%+5%+5%+15%, i.e. 85%. So; 85% of the adult population reads at
least one of the newspapers
Example-2: A researcher from the NRM secretariat mentioned that 60% of Ugandans read
New Vision, 40% read Daily Monitor and 30% read the East African. Further, he claimed that
20% read both Daily Monitor and the New Vision, 15% read both the New Vision and the East
African whereas 10% read both Daily Monitor and the East African and 30% read the New
Vision only. Use your knowledge of basic/ introductory probability to answer the following
question. If a person is selected at random, find the chance that:-
51
I) That person reads NONE the three Newspapers (ANS 0.1)
II) That person reads at least one of the three papers (ANS 0.9)
III) Estimate the number of people who read at least two papers if the total sample used
in this survey was 500 (ANS 175)
IV) Estimate the chance that if a person reads at least two newspapers, he also reads all
the three papers (ANS 0.1429)
Answer
Example-3: December 2009: QN 2b: In order to establish the reading habits of CPA(U)
students, a sample of 50 students was selected and asked to name the newspapers they read
regularly. The results obtained showed that 25 read The New Vision (N), 16 read The Daily
Monitor (M), 14 read The Observer (O); 5 read both N and M, 4 read both M and O, and 6
read both N and O; and 2 read all the three. Required: Find the probability that a student
selected in this sample reads: (i) at least one of the newspapers. (ii) Only one of the
newspapers. (iii) Only the New Vision
52
Example-4; June 2010: QN 5a: In order to establish the competitiveness of Yellow
Telecom, Purple Telecom and Red Telecom mobile telephone companies in the Ugandan
market, a survey was conducted on a sample of customers selected from Kampala. The
results generated were as follows: 190 were Yellow Telecom customers, 205 were for Purple
Telecom and 260 for Red Telecom. 15 customers were for all the three networks, 55 were for
Yellow Telecom and Red Telecom, 100 for Yellow Telecom only, 135 were for Red Telecom
only. 180 of the people under study were not on any network. Required: Determine the: (i)
size of the sample. (ii) Probability that a customer is on Purple Telecom only. (iii) Probability
that a customer is on one network only. (iv) Probability that a customer is on two networks
only. (2 marks)
Example 5: In a sample of 50 girls and 50 boys, fair and dark hair occurs as follows.
Girls (G) Boys (B)

Fair hair (F) 33 22
Dark hair (D) 17 28
If a child is selected at random, from the sample, what is the probability of selecting a girl OR
child with fair hair?
Solution
50 55
P(G)  ; P( F ) 
100 100 ;
P(G  F )  P(G)  P( F )  P(G  F )  50 / 100  55 / 100  33 / 100  72 / 100
Example 6: At Steel Rolling Mills, the production manager has provided the following
chances/ probabilities in regard to the number of defects in monthly shipments to Juba.
Number of defectives 0 1 2 3 More than 3

Probability/ chance 0.29 0.36 0.22 0.10 0.03
a) What is the probability that there will be fewer than 3 defective parts in a shipment?
b) What is the chance that there will be at most 2 defective part in a shipment?
c) What is the chance that there will be at least 2 defective parts in a shipment?
Example: A department store manager at Sadolin paints has monitored the number of
complaints received per week about the poor customer service. The probabilities/ chances for
the number of complaints in a week, established by this review are shown in the following
table. If we let A be the event “There will be at least 1 complaint per week”. And B is the
event that “There will be fewer than 10 complaints per week.”
Number of complaints 0 1-3 4-6 7-9 10-12 More than 12

Probability 0.14 0.39 0.23 0.15 0.06 0.03
a) Find the probability of A. also; Find the probability of B

b) Find the probability of the complement of A
c) Find the probability of the Union of A and B
d) Find the probability of the intersection of A and B
53
6.3 THE MULTIPLICATIVE LAW OF PROBABILITIES
If they are independent events, it means that their respective probabilities are unconnected to
each other. For example, probabilities obtained when a die and coin are tossed together.
Here,
P( AandB)  P( A  B)  P( A) xP( B)
If A, B and C are independent events, then
P( AandBandC)  P( A  B  C )  P( A) xP( B) xP(C )
This continues up to any number of events in the exam question
Conditional probabilities: If A and B are events, the conditional probability of event

A given that event B has already occurred is given by P (A/B) and is found to be:-
P A  B 
P A / B  
PB  Provided that P (B) is greater than 0
PB  A
pB / A 
Similarly, P A provided that P (A) is greater than 0
USING PROBABILITY TREES TO SOLVE PROBABILITY NUMBERS

Example 7: a bag contains 6 white and 9 black balls. Three drawings of 1 ball each are made
such that 1) the balls are replaced before the next drawing and 2) the balls are not replaced
before the next drawing. Find the probability that all the black balls are drawn in each case.
(ANS: With replacement = 729/3375. Without replacement = 504/2730). Note: The number is
easily solved using a probability tree
54
Example 8: A simple question is given to three CPA QT students; Kapele, Kamese and Kavu.
The chances of them getting right are respectively; ½, 1/3, and ¼. Find the probability that
the problem will be solved by all the three students (ANS = ¾). And the probability that none
of the three students will solve it (ANS = ¼) Note: The number is easily solved using a
probability tree
Example 9: A bag contains 4 white and 6 black balls. Two balls are drawn at random. If the
balls are drawn at random one after the other without replacement, find the probability that
both balls are white (ANS= 2/15). Both balls are black (ANS=1/3). First ball is white and
second ball is black (ANS = 4/15). One ball is white, the other is black (ANS = 8/15). Note:
The number is easily solved using a probability tree
55
Example 10: After failing in the the presidential elections, Mabreezy applied for a Job in two
different organizations (URA and MTN). The probability that he gets a job in the first
organization (URA) is (2/5). And the probability that he gets a job in the second organization is
(4/5). Required: Assuming that getting both Jobs is independent, find the probability that she
gets:
Probability of getting in URA = (2/5)
Probability of getting in MTN = (4/5)
i) Chance of getting both Jobs = Prob (URA) x Prob (MTN)
P( A  B)  (2 / 5) x(4 / 5)  8 / 25
ii) Only one Job = Either URA or MTN

= P( A  B)  P( A)  P( B)  P( A  B)
P( A  B)  (2 / 5)  (4 / 5)  (8 / 25)  22 / 25
P( A  B)  0.88
iii) Neither Job P( A  B)1  1  p( A  B)

 1  0.88  0.12
Example 11; The just concluded election was a highly contested one with counter
accusations of voter bribery on each side. According to the observers, Rukungiri District was
the epi-center of this voter bribery episode. Records from the Electoral Commission indicate
that 50% of the total registered voters voted for Museveni, 30% voted for Besigye and 20%
voted for John Patrick Amama Mbabazi. The observers maintained that 50% of those who
voted for Mr. Museveni were simply bribed, 82% of those who voted for Besigye were bribed
and 65% of those who voted for Mbabazi were bribed.
Use your knowledge of basic probabilities (Multiplicative law and conditional probabilities) to
answer the following questions
I) If a voter is selected at random, determine the probability that he was bribed
(ANS: 0.626)
II) If a voter is selected at random, and found not to have been bribed, find the
probability that he / she voted for Besigye (ANS: 0.1444)
III) If a voter is selected at random and found to have been bribed, find the
probability that he/ she did not vote for Museveni (ANS: 0.6006)
56
Example 12: An industry manufactures a particular type of light bulb from three departments
A1, A2 AND A3. 30% are manufactured by A1, 45% by A2, and 25% by A3. It was found that
the bulbs manufactured by A1, 1% were faulty. And 1% were also faulty in A2 and 2% were
faulty in A3. Suppose a bulb is selected and found to be faulty, find the probability that:-
I) It is one of A3’s bulbs (ANS: 0.4)
II) It is one of A1’s bulbs (ANS: 0.36)
III) It is one of A2’s (ANS: 0.24)
Example 13: In a week, a company either has good sales of 200units or bad sales of 100
units. In the first week, the probability of good sales is 0.6. In the second week, the probability
of good sales is 0.8 if the sales in the first week were good or 0.3 if the sales in the first week
were bad. What is the chance of total sales over both weeks being? (I) 400 units (answer;
0.48) (II) 200 units (answer; 0.0.28) (III) 300 units (answer; 0.24)
Example 14: in a factory, a certain type of sweets is packed into boxes of four different
production lines X1, X2, X3 and X4. Records show that a small percentage of boxes are not
packed properly for sale. 1% of X1, 3% from X2, 2.5% from X3 and 2% from X4 are
improperly packed. If the percentages of total output that have come from the production lines
are 35% from X1, 20% from X2, 24% from X3 and 21% from X4, determine the chance that a
box selected at random from the entire output is;
1) Not properly packed (ANS 0.0197)
2) Not properly packed and comes from X1(ANS 35/197))
3) Not properly packed and comes from X2 (ANS 60/197)
57
THE BAYE’S THEOREM
Previously, we noted that the conditional probability of an event is a probability obtained with
the additional information that some other event has already occurred. We used P(B|A) to
denoted the conditional probability of event B occurring, given that event A has already
occurred. The following formula was provided for finding P (B|A):
In addition to the above formal rule, the textbook also included this "intuitive approach for
finding a conditional probability": The conditional probability of B given A can be found by
assuming that event A has occurred and, working under that assumption, calculating the
probability that event B will occur.
In this part we extend the discussion of conditional probability to include applications of Bayes'
theorem (or Bayes' rule), which we use for revising a probability value based on additional
information that is later obtained. One key to understanding the essence of Bayes' theorem is
to recognize that we are dealing with sequential events, whereby new additional information is
obtained for a subsequent event, and that new information is used to revise the probability of
the initial event. In this context, the terms prior probability and posterior probability are
commonly used.
A prior probability is an initial probability value originally obtained before any additional
information is obtained.
A posterior probability is a probability value that has been revised by using additional
information that is later obtained.
Bayes' Theorem formulae

The probability of event A, given that event B has subsequently occurred, is
That's a formidable expression, but we will simplify its calculation. See the following example,
which illustrates use of the above expression, but also see the alternative method based on a
more intuitive application of Bayes' theorem.
Example 15: The ministry of Health is conducting a study on cigars smoking. They went to
Mbale and discovered that 51% of the adults are males. This implies that 49% are females.
9.5% of sampled males smoke cigars, whereas 1.7% of females smoke cigars. One adult is
randomly selected. Find the probability that the selected subject is a male
Solution
Let's use the following notation:
58
Let's now apply Bayes’ theorem by using the preceding formula with M in place of A, and C in
place of B. We get the following result:
BAYES' THEOREM GENERALIZED- EXPANSION OF THE BAYES THEOREM

Example 16: An aircraft emergency locator transmitter (ELT) is a device designed to transmit
a signal in the case of a crash. The Altigauge Manufacturing Company makes 80% of the ELTs,
the Bryant Company makes 15% of them, and the Chartair Company makes the other
5%. The ELTs made by Altigauge have a 4% rate of defects, the Bryant ELTs have a 6% rate
of defects, and the Chartair ELTs have a 9% rate of defects (which helps to explain why
Chartair has the lowest market share).
1) If an ELT is randomly selected from the general population of all ELTs, find the
probability that it was made by the Altigauge Manufacturing Company.
2) If a randomly selected ELT is then tested and is found to be defective, find the
probability that it was made by the Altigauge Manufacturing Company.
Solution
We use the following notation:
A = ELT manufactured by Altigauge
B = ELT manufactured by Bryant
C = ELT manufactured by Chartair
D = ELT is defective
D = ELT is not defective (or it is good)
59
a. If an ELT is randomly selected from the general population of all ELTs, the
probability that it was made by Altigauge is 0.8 (because Altigauge manufactures 80% of
them).
b. If we now have the additional information that the ELT was tested and was found
to be defective, we want to revise the probability from part (a) so that the new
information can be used. We want to find the value of P(A|D), which is the probability
that the ELT was made by the Altigauge company given that it is defective. Based on the
given information, we know these probabilities:
Here is Bayes' theorem extended to include three events corresponding to the

selection of ELTs from the three manufacturers (A, B, C):
Example 17: Suppose a day’s production schedule calls for 9000 items. Three machines each
with a daily capacity of 4000 items are available and the probability that an item is defective is
1, 2 and 4 percent of the machines A, B and C respectively. On a given day, 4000 items were
produced on A, 4000 on B, and 1000 on C. One item is selected at random and found to be
defective. What is the probability that it was produced on machine A?
Solution
A: Item was produced on machine A
B: Item was produced on machine B
C: Item was produced on machine B
D: Item is defective
Obviously, we want to solve P (A/D)
It is given that P (A) = P(B) = P(C) = 4000/9000 = 4/9
Also; P (D/A) = 1%, Also; P (D/B) = 2%, Also; P (D/C) = 4%
Thus, applying BAYES theorem,
ANS= 1/4
60
Example 18: Up to now, Mr. Amama Mbabazi’s camp still maintains that all figures from tally
sheets were wrongly entered at Central tallying center in Nambole. They claim that Electro
commission hired only three data entry specialists to enter all information from all tally sheets
and declaration forms. According to this camp, Data entry specialist 1 processed 30% of all
declaration forms; specialist 2 processed 45% of all declaration forms while specialist 3
processed only 25%. The proportions of incorrectly entered declaration forms by data entry
specialist 1, 2 and 3 are 0.3, 0.5 and 0.2 respectively. Assuming that a random declaration
form is found to have been incorrectly entered, what is the probability that it was processed by
data entry specialist 1?
Example 19: If a machine is correctly set up, it will produce 90% acceptable items. If it
incorrectly set up, it will produce only 40% acceptable items. Experience shows that the 70%
of the setups are correctly done. Find the probability that a setup has been correctly done if
refer the set up; 1) the first item produced is accepted (ANS = 21/25), 2) the first two items
produced are accepted (ANS = 189/205), 3) the first three items produced are acceptable,
acceptable, and unacceptable respectively (ANS = 63/95)
61
Example 20: There are three plants in a factory producing the same product independently.
Experience shows that 2%, 4% and 5% of the items produced by the respective plants are
defective. The daily outputs of the respective plants are 2000, 3000 and 5000 units
respectively. If an item is selected at random from the outputs of the three plants, and is
found to be defective, find the probability that it was produced by the second plant. (ANS =
12/41)
Example 21: Multiple Industries has four plants scattered around the city producing the same
homogeneous item at all plants. The first plant produces 30 per cent of the total production,
second plant 25 per cent, third plant 35 per cent and the fourth plant 10 per cent. The firm
has a single warehouse in the city for storing the finished product of all the plants without any
distinction. From the past performance records on the proportion of defectives, it has been
found that 5 per cent, 10 per cent, 15 per cent and 20 per cent from the items produced at
plants 1, 2, 3, and 4 respectively are defective. Before the shipment of the items to a dealer,
one unit is selected and found defective. What is the probability that the item was produced in
plant 3?
Example 22: When visiting a friend, John may go by road, air or rail. The probabilities of
using a road, air or rail are 0.3, 0.8 or 0.6 respectively. The corresponding probabilities of
arriving on agreed time are 0.2, 0.8 & 0.1 respectively. Find the probability of having used the
road given that he arrived on time. (ANS = 0.0789)
YOU MUST BE FULL OF HAPPINESS AFTER UNDERSTANDING THIS TOPIC. NOW

TRY OUT THESE VERY SIMPLE PAST PAPER NUMBERS
June 2014 June 2010
62
Chapter-7
Decision-making
7.1 INTRODUCTION
Decision making is a process of selecting from a set of alternative courses of action, the best
option that is considered to meet the objectives of a decision problem more satisfactorily than
as judged by the decision maker. A good decision is always based on logic, uses data, applies
techniques of quantitative analysis and considers all possible alternatives.
7.2 DEFINITIONS FOR TERMS ENCOUNTERED IN DECISION MAKING

(a) Decision Alternatives. These refer to the list of choices available to a decision maker.
For example, an investor may have to decide whether to purchase shares or bonds or both,
which would represent three decision alternatives.
(b) States of Nature. These are the future environmental conditions that influence the
desirability of decision alternatives. For example, whether the stock market is rising, falling or
steady would be likely to influence the desirability of investing in shares or bonds, and so
would represent three possible states of nature facing an investor. For any given decision
problem, the states of nature should be exhaustive and mutually exclusive.
(c) Uncertainty. Under conditions of uncertainty, we assume that the possible states of
nature are known to the decision-maker, but their probabilities are unknown. Most decisions in
business and management are of this type, and probabilities often have to be estimated from
past experience, or the states of nature have to be regarded as equally likely.
(d) Risk. Under conditions of risk, we assume that the possible states of nature are known
and that the probabilities that they will occur are also known.
(e) Payoff Table. A payoff table has columns that represent the possible states of nature and
rows that represent the decision alternatives. Each cell of the table gives the payoff (usually a
revenue, profit or loss) associated with the given decision alternative and state of nature.
(f) Decision Tree. Is a diagrammatic representation of a decision problem, which shows the
decision alternatives, the possible states of nature, the state of nature probabilities (if known)
and the payoffs. It is an alternative way of presenting the information in a payoff table.
7.3 DECISION-MAKING UNDER RISK

Under conditions of risk, we assume that decisions are taken with full knowledge of the
probabilities associated with the possible states of nature. Let us then suppose that
we know that:
The probability of a relatively weak pound over the next five years is 0.2
The probability of a relatively stable pound over the next five years is 0.5
The probability of a relatively strong pound over the next five years is 0.3.
As the three states of nature are exhaustive and mutually exclusive, these three probabilities
must sum to one (i.e. it is a certainty that the pound will be relatively weak, stable or strong).
These probabilities enable us to calculate the expected monetary values of each decision
alternative.
Expected Monetary Value (EMV): Given a set of n possible monetary values (say, x1, x2, x3,
…..xn) and n probabilities associated with each monetary value (say, (say, p1, p2, p3, …..pn),
the expected monetary value (EMV) is calculated as the sum of the products of the monetary
values and their associated probabilities (i.e.
EMV  ( p1 x1  p2 x2  .....pn xn )
. For i  1,2,3......n 63
With regard to decision-making problems, the x's are the payoffs resulting from each decision
alternative for a given state of nature and the p's are the state of nature probabilities.
EXPECTED VALUE OF PERFECT INFORMATION

The expected value of perfect information is the difference between the total pay off that
would occur if we knew which stakes of nature would occur and the expected monetary payoff
from the best decision alternative when there is no information available about the
occurrence of the states of nature. That’s to say; The Expected Value of Perfect
Information = (EVPI) = (EVwPI)- Maximum EMV
I.e. It’s the difference between the payoff under certainty and the payoff under risk
Where EVwPI is the expected value with perfect information?
Decision nodes that depict decision points (represented by square boxes)

Chance nodes that depict points at which states of nature occur (represented by
circles)
Payoff nodes that depict the final payoffs (represented by vertical lines)
Example: JPAM after bitterly losing an election wants to construct a milk plant in his home
village of Kinkizi. Below are the pay offs and associated chances of getting such pay offs
State of the Economy

High (0.3) Moderate (0.5) Low (0.2)
Large plant 200,000 100,000 -120,000
Decision Small plant 90,000 50,000 -20,000
Alternatives No plant 0 0 0
Advise Mr. JPM on the best investment strategy and compute the expected value of perfect
information
Solution
Invest in the large plant 64

The Expected Value of Perfect Information = Expected value with perfect information –
Maximum EMV.
Expected value of perfect information =……………………………………………………
Multistage problems involve a sequence of several decisions and outcomes. It is possible for a
decision to be immediately followed by another decision. Decision trees are best for showing
the sequential arrangement. Suppose they will first decide whether to pay $4000 to conduct a
market survey. Survey results will be imperfect. Then they will decide whether to build a large
plant, small plant, or no plant. Then they will find out what the outcome and payoff are
Conduct the survey. If the survey results are positive, then build the large plant (EMV =
$141,840). If the survey results are negative, then build the small plant (EMV = $16,540)
Example; Mr. Ogga has won a jackpot of UGX 2 billion. He has to decide where to invest his
money in one of the following real estate projects: Building apartments, Building offices or
Building a warehouse. The annual payoffs are in UGX’ millions are presented in the table
below.
65
State of the Economy
Good Economy Bad Economy
Apartments 50 30
Decision Offices 100 -40
Alternatives Ware House 30 10
Assuming there is a 60% chance that the economy would be good in the above scenario, use
your knowledge of decision theory (especially decision trees) to advise Mr. Ogga where to
invest
Example 2; A Japanese automobile manufacture currently produces its bestselling U.S model
in Japan, but the relative strength of the Japanese Yen versus the U.S dollar has been making
the car very expensive for the U.S market. To ensure a lower and more stable price, the
company is considering the possibility of manufacturing cars for U.S consumers at one of its
American plants. Its payoff table, in millions of dollars, is given below:
Alternative State of nature

Weak dollar Moderate dollar Strong dollar
Probability = 0.3 Probability = 0.5 Probability = 0.2
Japanese production 10 15 25
American production 20 18 16
I. Make a decision tree summary of the table above.

II. Compute the expected payoff for each production. (Japan 15.5; US 18.2)
III. Compute the expected payoff with perfect information. ($20M)
IV. Compute the expected value of perfect information ($1.8M)
66
Example 3; A farmer intends to drill a well for his herd. In the past, only 80% of the wells
drilled in this area were successful at 12 meters of depth. On finding no water at 12M, some
farmers drilled further up to 18 meters but only 30% struck water at 18M. The prevailing cost
of drilling the first 10M is Sh500, 000 and drilling an extra meter costs Sh60,000. The farmer
knows that if he does not get his own well, he will have to pay one million shillings to get
water from a neighbor’s well. Draw an appropriate decision tree and determine the farmer’s
strategy under the Expected Value (EMV) approach.
Example 4; A company is considering a short manufacturing run for either of two new
products, a temperature sensor or a pressure sensor. The market for each product is known if
the products can be successfully developed. However, there is some chance that it will not be
possible to successfully develop them. Revenue of $1,000,000 would be realized from selling
the temperature sensor and revenue of $400,000 would be realized from selling the pressure
sensor. Both of these amounts are net of production cost but do not include development cost.
If development is unsuccessful for a product, then there will be no sales, and the development
cost will be totally lost. Development cost would be $100,000 for the temperature sensor and
$10,000 for the pressure sensor.
Example 5; A company has a wonder product of which it expects great things. At the
moment the company has two courses of action open to it: test-market the product or
abandon it. If they test it, it will cost Sh10 million and the response could be positive or
negative with respective probabilities of 0.6 and 0.4. If the response is positive the company
could either abandon the product or market it full scale. If it markets the product full scale the
outcome might be low, medium or high demand, and the net pay-offs are: a loss of Sh20
million, profit of Sh20 million and Sh100 million respectively. The outcomes have
corresponding probabilities of 0.2, 0.5 and 0.3. If the result of the test marketing is negative
and the company goes ahead and markets the product, estimated losses would be Sh60
million. If, at any point, the company abandons the product, there would be a net gain of Sh5
million shillings from the sale of scrap. You are required to draw an appropriate decision tree
and determine the company’s strategy under the Expected Monetary Value (EMV) approach.
Example 6: A trust officer, Mugisha for a major banking institution is planning the investment
of a $ 1 million family trust for the coming year. The trust officer has identified a portfolio of
stocks and another group of bonds that might be selected for investment. The family trust can
be invested in stocks or bonds exclusively, or a mix of the two. This trust officer prefers to
divide the funds in increments of 10 percent; that is, the family trust may be split 100 percent
stocks /0 percent bonds, 90 percent stocks /10 percent bonds, 80 percent stocks /20 percent
bonds, and so on. The trust officer has evaluated the relationship between the yields on the
different investments and general economic conditions. Her judgment is as follows:
1) If the next year is characterized by solid growth in the economy, bonds will yield 12
percent and stocks 20 percent.
2) If the next year is characterized by inflation, bonds will yield 18 percent and stocks 10
percent.
3) If the next year is characterized by stagnation, bonds will yield 12 percent and stocks 8
percent.
I) Formulate a payoff table where payoffs represent the annual yield, in dollars,
associated with the different investment strategies and the occurrence of various
economic conditions
II) Suppose that a leading economic forecasting firm projects P (solid growth) = 0.4, P
(inflation) =0.25, and P (stagnation) = 0.35. Use the expected value criterion to select
67
the appropriate strategy.
III) What is the expected value with perfect information?
Example 7: Define the term ‘expected monetary value’ and explain its usefulness in
decision making
Mukwano Industries is considering its strategy for the next five years. It has three possible
courses of action, 1, 2 and 3, which are described below:-
1. Expand the business by re-locating to a larger site. This is estimated to have a 70% chance
of success, which will increase the company’s profits by UGX 40 million, but there is a 30%
chance of failure, which will decrease the company’s profits by UGX 60 million.
2. Take over a small rival company. The probability of successfully taking over the rival
company is estimated to be 0.5, and the rival company’s financial statements suggest that
there would then be a 60% chance of increasing profits by UGX 50 million and a 40% chance
of increasing profits by just UGX 10 million. However, if the takeover attempt fails (probability
= 0.5), profits are estimated to fall by UGX 15 million.
3. Make no new investments. In this case, it is estimated that there is a 50% chance that
profits will remain unchanged, a 20% chance that profits will increase by UGX 10 million and a
30% chance that profits will fall by UGX1 million. Using a decision tree and calculating the
Expected Monetary Values, advise the company on the best strategy to follow.
c) One of the company’s directors argues that the company should adopt strategy3. He points
out that strategy 3 will at worst lead to a UGX 1 million fall in profits, while adopting strategy 1
could result in a UGX 60 million fall in profits and adopting strategy 2 could result in a UGX 15
million fall in profits. Comment on the director’s choice of strategy.
Solution
The expected monetary value of a business decision is the average return that can be
expected, taking into account probabilities. For any business strategy, the EMV is calculated by
multiplying the estimated value of the possible outcomes by their associated probabilities and
then summing. It is a useful measure in business as it allows decision-makers to compare
alternative decisions - the highest EMV being one criterion employed by decision-makers to
choose among alternative strategies.
a)
68
Re-location: EMVI = (0.7 x 40) + (0.3 x (-60)) = 10
Takeover: EMVII = 0.5 [(0.6 x 50) + (0.4 x 10)] + (0.5 x (-15)) = 9.5
No new investment: EMVIII = (0.5 x 0) + (0.2 x 10) + (0.3 x (-1)) = 1.7
Based on the EMV criterion, the best strategy is Strategy I, i.e. to expand by re-location.
b) The director is clearly risk-averse and is adopting the ‘maximin’ criterion, i.e. choosing
Strategy III with the highest minimum outcome. This may be sensible in this example, as
making no new investment avoids the risk (present in the ‘relocation’ and ‘takeover’
strategies) of experiencing substantial falls in profit which could place the company in a very
difficult financial position. There may also be non-pecuniary factors which the directors
should take into account, such as the preferred location of the company and the level of
employment that may be created by each strategy. The environmental effects (externalities)
of the three strategies might also be important.
Example: Quality Chemicals Ltd is introducing a new HIV vaccine which requires testing at a
cost of Shs.20 million. The diagram below illustrates the decisions taken:
69
Required:
Study the decision tree and evaluate the expected values of each unit course and give your
conclusion.
Solution
A. Drug Testing positive and marketing it
Expected
EMV= (0.8x0.2 x(600  20)  0.8x0.5x(300  20)  0.8x0.3x(400  20)  104 M
B. Drug testing positive and abandoning it
Expected EMV= 0.8 x(80  20)  48M
C. Drug testing negative and marketing it
Expected EMV= 0.2 x(400  20)  84 M
D. Drug testing negative and abandoning it

Expected EMV= 0.2 x(90  20)  14 M
Quality chemicals limited should take a course of action that has the highest EMV; That is to
say, a drug tests positive and the company proceeds to market it.


June 2014 June 2010
70
Chapter-8
Permutations & Combinations
8.1 INTRODUCTION
What is a permutation? A permutation of n objects taken r at a time is a selection of r
objects from a total of n objects where (r ≤ n), where ORDER MATTERS. The word
permutation means a particular sequence or order of arrangement of things. For example, BAC
and CBA are both permutations of the first three letters of the alphabet. Permutation problems
are concerned with the number of possible sequences into which things can be arranged.
There is a basic principle governing such problems: "If one operation can be done in m ways,
and if a second operation can be done in n ways, then the two operations can be done in
succession in m times n different ways."
Example1. There are three different colored buses (red, yellow and green) which run
between two places. If I want to use a different colored bus for each direction, in how many
different ways can I make the double journey?
Applying the basic principle, we see that the first part of the trip can be done in three ways
(red, yellow and green), while the second part of the trip can be done in only two ways
(excluding the color already used). Thus the total number of different possible ways is 3x2= 6
8.2 PERMUTATIONS OF DIFFERENT ITEMS (DISSIMILAR ITEMS)

If we have a group of different items, we can now calculate the total number of permutations
quite easily. Suppose there are 4 different things which we arrange in a row. Any one thing
can be put first, so the first place can be filled in 4 different ways. After the first place has
been filled, only 3 things remain, and so the second place can be filled in 3 possible ways.
Then the third place can be filled in 2 possible ways and the fourth place in only 1 way. The
basic principle tells us that the total number of different arrangements is therefore;
4x3x2x1 = 24ways. OR 4ǃ. The symbol ( ǃ ) is read as factorial. So; 4ǃ = 4x3x2x1 = 24ways
There are 6 seats available in the Akamba executive bus compartment. In how many ways can
six persons be seated?
Permutations of some items

The number of ordered sets of (r) elements taken from a set of (n) different elements without
replacement
n!
n
Pr 
Is; (n  r )!
Example1 of permutation of some items selected from n items

In how many ways can the first two positions in a race be filled if there are four competitors?
Solution
Let the competitors be A, B, C and D. Then the different permutations are
AB BA CA DA
AC BC CB DB
AD BD CD DC
Here 4P2 = 12. This can be proved by the box method. Assume we have 2 boxes to fill- one for
the first position, and one for the second. The first box can be filled in 4 ways (A, B, C, D),
71
leaving 3 items over. The second box can now only be filled in 3 ways. As any item in the first
box can be associated with any item in the second box 4P2 = 4 x 3 = 12 ways
Example2 of permutation of some items selected from n items

In how many different ways can 11 men football team be selected from a squad of 20 men if
positions are not ignored?
PERMUTATIONS OF N THINGS NOT ALL OF WHICH ARE DISSIMILAR

If out of n things, r1 are of one type, r2 are of another type and so on, the number of
permutations would be;
n!
n
Pr 
r1! r2 !.....
Example of permutation of n things not all of which are dissimilar

In how many ways can the letters of the word (STATISTICS) be arranged?
Solution
8.3 COMBINATIONS
When dealing with permutations, we are concerned principally with the order or sequence in
which things occur. Other problems occur in which we need to calculate the number of groups
of a certain size irrespective of their sequence. These are called combinations
The number of ways (combinations) of n things taken r at a time without following a particular
order is;
n!
n
Cr 
(n  r )! r!
Example1- of selecting combinations of n things taken r at a time

In how many ways can a committee of 2 people be selected from a group of 4 people?
Solution
This is a similar problem to what we have seen so far except for the fact that order is not
taken into account. Thus terms such as AB and BA are the same here. The different
combinations are;
AB BC CD
AC BD AD
Hence we have 4C2 = 6. This figure is easily got from the calculator.
Example2 of selecting combinations of n things taken r at a time

In how many different ways can 11 men football team be selected from a squad of 20 men if
positions are ignored?
72
What is the probability of getting 3 white balls in a draw of 3 balls from a box containing 5
white and 4 red balls?
Solution
Hint: Required probability =

5
C3 / 9C3  5 / 42

From a sales force of 150 persons, one will be chosen to attend the special sales meeting. If
52 are single and 72 are college graduates and the three fourths of the 52 ( that are single)
are college graduates, what is the chance that a sales person selected at random will be
neither single nor college graduate?
Solution


June 2014 June 2010
Chapter-9 73
Discrete probability distribution
9.1 INTRODUCTION
In the previous chapters, we began our development of probability to represent situations with
uncertainty outcomes. In this chapter, we use those ideas to develop probability models with
an emphasis on discrete random variables. Probability models have extensive application to a
number of business problems and some of these applications are covered here.
Let’s first understand the definitions of the following terms

A random variable: It is a variable that takes on numerical values determined by the
outcome of a random experiment. It is usually denoted by capital letters and the
corresponding lower case letter to denote possible values of the random variable.
Example: Computer Empire has five computers on its shelf and from past experience, the
manager knows that the probabilities of selling one through five computers are equal and at
least one computer will be sold. We can use random variable X to denote the outcome. This
random variable can take on specific values; x  1, x  2, x  3,.......,x  5 . Each has a
probability of 0.2 and then the random variable X qualifies to be called a discrete random
variable.
Discrete random variable: It is a random variable which can take on one countable number
at a time.
Example:
Number of computers in a store at any one moment
Number of defective products in a sample of 20 items from a large shipment to Juba
A probability distribution is a graph, table or formula that specifies the probabilities

associated with all the possible values of a variable. There are two types of probability
distributions. Discrete and continuous probability distributions. Let’s start our discussion
with the Discrete probability distributions
9.2 DISCRETE PROBABILITY DISTRIBUTION

This one assumes that the outcome of a random variable under study can take on only specific
(usually integer) values. Examples; outcome of a car to have only 0, 1, 2, 3 or 4 flat tyres. A
machine can produce only 0, 1, 2, 3 or 4 defective products in a given production run. It can
be seen from these examples that the outcome of every trial (experiment) has only discrete
(integer) values. There are two common types of discrete probability distributions; Binomial
and Poisson probability distributions. BUT, these two discrete probability
distributions are not part of the CPA syllabus. As such, attention is skewed towards
the continuous probability distributions.
Example1 of a discrete probability distribution

The manager of Nandos has noted the following probabilities of selling sand witches. This is
called a discrete probability distribution for the company.
Number of sand witches (x) Probabilities p(x)

0 0.1
1 0.2
2 0.4
74
3 0.3

The same probability distribution can be plotted on a graph.
9.3 THE AVERAGE/ MEAN/ EXPECTED MONETARY VALUE, VARIANCE AND

STANDARD DEVIATION OF A DISCRETE PROBABILITY DISTRIBUTION
Expected Monetary Value (EMV) (Mean)

EMV  E ( x )   x. p  x 
.
Variance
V ( x)  x . p x   E  x 
2 2
.
The square root of variance gives the standard deviation
Example: A factory manager at Uganda Breweries Limited is considering whether to replace a

chemical mixing machine which is constantly breaking down. A review of past records indicates
the following probability distribution for the number of machine breakdowns in a week.
N0 of breakdowns 0 1 2 3 4
Probability 0.10 0.26 0.42 0.16 0.06
Required
(i) Find the: mean, Variance and Standard Deviation of the number of weekly breakdow
Mean = EMV  E ( x)   x. px 

Variance = V ( x)  x . px   E x 
2 2
Standard deviation= Square root of variance
N0 0 1 2 3 4
P(x) 0.10 0.26 0.42 0.16 0.06 Total
xP (x) 0 0.26 0.84 0.48 0.24 1.82
x2 0 1 4 9 16
x 2
. px  0 0.26 1.68 1.44 0.96 4.34
75
EMV  E ( x)   x. p x   1.82
V ( x)   x 2 . px   E x   (4.34  1.82 2 )  1.03
2
Standard Deviation = VAR = 1.03  1.014

(ii) It is estimated that each breakdown costs the company $1,500 in lost output. Find the
mean and standard deviation of the weekly costs to the company from the breakdowns of
this machine.
Mean = EMV  E ( x)   x. p x 
Variance = V ( x)  x . px   E x 
2 2
Standard deviation= Square root of variance
N0 0 1 2 3 4
Costs (x) 0 1,500 3,000 4,500 6,000
P(x) 0.10 0.26 0.42 0.16 0.06 Total
xP (x) 0 390 1260 720 360 2730
x2 0 2250000 9000000 20250000 36000000
 x 2 . px  0 585000 3780000 3240000 2160000 9,765,000
EMV  E ( x)   x. p x   2,730
V ( x)   x 2 . px   E x   (9765000  2730 2 )  2,312,100
2
Standard Deviation = VAR = 2,312100  1,520 .6
HAPPY HOUR: TRY THESE NUMBERS

Example 1:
a) A discrete random variable X has the following probability distribution
Number of votes (In Millions) 1 2 3 4 5
Probability M 2M 3M 4M 5M
I) Determine the value of M

II) Evaluate the probability (P<X<4)
III) Calculate the mean, variance and standard deviation in votes
b) A discrete random variable X is represented by the probability distribution
1  k
 forX  1,2,3,....6
P ( X  x)   kx
0....Elsewhere
Find:
I) The value of K
II) The expectation of K
III) The variance in the distribution
IV) The probability P( X  3) /( X  4)
76
c) A discrete random variable X is represented by the probability distribution
2 X 2 K ::: forX  0,1,2,3,....6

P( X  x)  
0....Elsewhere
Find:
I) The value of K
II) The expectation of K
III) The variance in the distribution
IV) The probability P( X  3) /( X  4)
Example 2: Ruhama constituency was a very tricky voting battle ground for all candidates in
the previous presidential election. Being a very wise presidential candidate, Mr. Mabreezy had
complied a table of possible votes and the respective chances of getting those votes in that
district as follows:-
Number of votes (In Millions) 3 5 2.5 6

Probability 0.1 0.2 0.3 0.4
Compute the mathematical expected number of votes. Help Mr. Mabreezy compute the
Variance and Standard deviation in votes obtained

June 2014 June 2010
77
9.4 THE BINOMIAL PROBABILITY DISTRIBUTION
The binomial probability distribution is applicable in situations where an experiment, consisting
of a certain number of trials satisfies all of four conditions. These are:
1. The number of trials must be fixed and finite. This number is usually denoted by n.
2. Every trial must result in one or other of only two mutually exclusive possible
outcomes, which, for convenience, we usually label "success" or "failure". Examples
are:
(a) When we roll a die we get a six or we do not get a six.
(b) A product is either defective or not defective.
(c) A tossed coin comes down heads or tails.
(d) A child is either a boy or a girl.
We must, of course, define which event is the success before the term is used.
3. The probability of a success or failure at each trial must remain constant throughout
the experiment. The probability of a success is usually denoted by p, and that of a
failure by q.
4. The outcome of every trial must be independent of the outcome of every other trial.
For example, if a coin is unbiased, then the probability of obtaining a head on the tenth
time it is tossed remains 1/2, even if the previous nine tosses have all resulted in
heads.
Provided these four conditions all hold, the binomial distribution enables us to work out the
probability of any given number of successes or failures in a specified number of trials.
9.5 Calculating binomial probabilities

If we perform our trial n times, the probabilities of 0, 1, 2, 3 …... up to n successes are given
by successive terms in the binomial expansion (q  p) n where p is the probability of success
and q the probability of failure at any one trial. Note that p and q must remain constant from
trial to trial. From our formula for the binomial expansion, the general term in the expansion
of (q  p) n is n C r q nr p r . This gives us the probability of exactly r successes in n trials of the
experiment. If we call x the number of successes, then we have:
P( X  r ) nCr q nr p r For r  0,1,2,3,4..........NOTE: q  (1  p)

Example-1
A medical survey was conducted in order to establish the proportion of the
population which was infected with cancer. The results indicated that 40% of the
population was suffering from the disease. A sample of 6 people was later taken
and examined for the disease. Find the probability that the following outcomes
were observed;
a) Only one person had the disease
a) Exactly two people had the disease
b) At most two people had the disease
c) At least two people had the disease
d) Three or four people had the disease
78
79
80
81
82
83
84
85
86
NOTE: WE CAN ALSO READ THESE FIGURES DIRECTLY FROM BINOMIAL TABLES
GIVEN IN AN EXAM
PART OF THE BINOMIAL TABLE
U can also handle this using the POISSON tables
HAPPY HOUR: NOW LET’S TRY OUT THESE NUMBERS

Question 1 on application of binomial probability distributions.
The probability that a match will break on being struck is 0.04. What is the probability that, out
of a box of 50: (a) none will break; (b) more than 2 will break?
Question 2 of an application of binomial probability distributions.

It has been found that, on average, 5% of the eggs supplied to a supermarket are cracked. If
you buy a box of 6 eggs, what is the probability that it contains 2 or more cracked eggs?
Question 3 of an application of binomial probability distributions.
A retail sales manager will accept delivery of a large consignment of goods if a random sample
of 10 items contains no defectives. If 3% of the producer's total output is defective, what is
the probability that delivery of a consignment will be accepted? How would the situation
change if the random sample were of only 5 items?
Question 4 of an application of Poisson probability distributions. A book contains 100

misprints distributed randomly throughout the 100 pages. What is the probability that a page
observed at random contains at least 2 misprints? Assume a Poisson distribution. 87
Question 5 of an application of Poisson probability distributions.
At Oscar industries, there is a small chance of 1 out of 500 for any pens manufactured to be
defective. The pens are supplied in packets of 10. Use the Poisson distribution to establish the
approximate number of packets containing: 1) no defectives, 2) two defectives, 3) three
defectives.
There is another very common use for the Poisson distribution, and that is as an approximation
to the binomial distribution. If n is large, the calculations for the binomial distribution are often
very tedious, even with a calculator. Fortunately, if we put m = np, the Poisson terms can, in
certain circumstances, be used instead of the binomial terms. The conditions for using the
Poisson as an approximation to the binomial are that:
1) n should be large;
2) p should be small, or q = (1-p) should be small.
Question 6
a) Distinguish between a binomial and Poisson probability distribution

b) What are the conditions for a Poisson approximation to a binomial distribution
c) What is the probability of picking three defective items from a lot of 100 items when it’s
known that the probability of an item being defective is only 5%?
Question 7
a) In a family with 10 children, if the probability of having a male child is the same as that of
having a female child, what is the probability that: (a) 6 of the children will be boys, (b)
none will be a girl, (c) at most, 2 will be boys?
b) Calculate the probability that, for 6 telephone lines: (a) at least 1 of the lines is engaged
and (b) all 6 lines are engaged. When the probability of 1 line being engaged is 1/4.
Question 8
a) Define the following: 1) A probability distribution, 2) Discrete probability distribution and

continuous probability distribution.
b) The average number of misprints per page in a company’s annual reports is 0.4. If a
page is selected at random, find the probability that it will contain more than two
misprints.
c) On average, one in every four new clients approached by an insurance salesman
purchases an insurance policy. Suppose that the salesman approaches five new clients.
Find the probability that exactly two will buy a policy; find the probability that at least
two will buy a policy.
d) Suppose that a salesperson succeeds in making a sale with one in every three potential
customers contacted. During a particular week, ten new potential customers are
contacted: Find the probability that the salesperson will succeed in making:
I.More than two sales
II.At least two sales
III.On average, three people per hour enter a furniture shop. In any given hour, find
the probability that more than three people will enter the shop.
88
Chapter-10
Continuous probability distribution
10.1 INTRODUCTION
In the previous subsection, we developed discrete random variables and their respective
probability distributions. Here we extend the probability concepts to continuous random
variables and probability distributions. Many economic and business measures such as sales,
investment, consumption, costs, and revenues are represented by continuous random
variables. In addition, measures of time, temperature, distance and weight also fall in this
category. Probability statements for continuous random variables are specified over
ranges. Example: The probability that sales of UTL will be between 140 and 196 millions for
March is a typical example.
10.2 THE NORMAL PROBABILITY DISTRIBUTION

In this subsection, we present the normal probability distribution which is a continuous random
variable probability distribution used often in business and economic scenarios.
If we take a continuous random variable, say the height of students, weights of people,
length of items cut by a machine, and plot the probability density function, one fact emerges
that most of these density curves resemble in shape. This common shape is bell shaped,
uni-modal with most of the density concentrated near the mode and very little
density lying towards the lower and upper extremes.
This relative frequency curve includes the relative frequency of all the values of the variable x,
from  ∞ to +∞ and the area between the curve and the x-axis is one unit and so it is a
theoretical relative frequency distribution or a probability distribution. This uni-modal
symmetrical bell-shaped curve is of great theoretical and practical importance in statistical
work. It is called the normal distribution or sometimes the Gaussian distribution after the
scientist who developed its use for examining random errors of observation in experimental
work.
89
When we consider the relative frequency curves of continuous variables, we discover a similar
pattern in the measurements of a great many natural phenomena. For example, the frequency
curve obtained from the set of heights of 80 employees, used earlier in the course, is uni
modal with small frequencies in the tails. Later, we calculated the mean, standard deviation
and median of this set of data. Since we were dealing with a comparatively small sample, the
values of these measures were empirical, and it is reasonable to assume that the theoretical
relative frequency distribution (the probability distribution) deduced from the data would be
symmetrical and normal. The same type of frequency distribution is found in the populations of
dimensions of items from a factory production line and errors of observation in scientific
experiments. This wide range of application of the normal distribution accounts for its
importance in statistical inference.
10.3 PROPERTIES OF A NORMAL DISTRIBUTION CURVE

1. The curve is symmetrical, uni-modal and bell shaped.
2. All the values of y are greater than zero and approach zero as x approaches  ∞ to +∞
3. It can be proved that the area between the curve and the x-axis is one unit.
4. It can be proved that:
(i) The mean, mode and median are all equal to the parameter μ.
(ii) The standard deviation is equal to the parameter σ.
5. The probability associated with a particular range of x values is described by the area
under the curve between the limits of the x range. ( example; (x1≤ x ≤x2))
6. If Z = (x - µ)/σ., then it can be proved that z has the same normal distribution for every
pair of values of the parameters μ and σ. This distribution is called the standard normal
distribution, and z is called the z-score or the standardized value of the variable.
10.4 THE STANDARD NORMAL DISTRIBUTION

Since z is normally distributed, the range of z is  ∞ to +∞ and the distribution has the
properties (a), (b), (c), (d) and (e) listed above. In addition, it can be proved that μ = 0 and σ
= 1. The figure below shows the shape and location of the standard normal distribution.
Since σ = 1, the ordinates drawn in Figure above are 1, 2 and 2.5 standard deviations on each
side of the mean. You can see that values of z more than 3 standard deviations from the mean
are very unlikely to occur, and that 50% of the values of z lie below zero (the mean), and 50%
above zero. It can be calculated that:
I. About 68% of the distribution lies within 1 standard deviation of the mean.
II. About 95% of the distribution lies within 2 standard deviations of the mean.
III. About 99% of the distribution lies within 2.5 standard deviations of the mean. 90
Since z is a continuous variable we cannot find the probability that z takes any exact value but
only the probability that it lies in a given range of values. In other words, the probability that
the value of z lies between any two given values is equal to the area under the curve between
the ordinates at these two values.
10.5 USING THE STANDARDIZED NORMAL DISTRIBUTION TABLE TO SOLVE A

VARIETY OF PROBABILITIES
Any set of normally distributed data can be converted to its standardized form and the desired
probabilities can then be determined from a table of the standardized normal distribution.
Example: Salaries of MPs are normally distributed with mean m = 23.56 Million and standard
deviation, σ = 4.55 Million. What proportion of MPs would have salaries of 14.45 Million or
less?
That is the proportion of MPs with salaries of 14.46 or less is 2.28%.
Example 1: Carrots entering a processing factory have an average length of 15.3 cm, and
standard deviation of 5.4 cm. If the lengths are approximately normally distributed, what is the
maximum length of the lowest 5% of the load? (I.e., what value cuts off the lowest 5 %?)
91
Example 2: X is a normally normally distributed variable with mean μ = 30 and standard
deviation σ = 4. Find a) P(x < 40), b) P(x > 21), c) P (30 < x < 35)
Solution
Note: What is meant here by area is the area under the standard normal curve.
Illustration of the probability on the normal curve
a) For x = 40, the z-value z = (40 - 30) / 4 = 2.5

Hence P(x < 40) = P(z < 2.5) = [area to the left of 2.5] = 0.9938
b) For x = 21, z = (21 - 30) / 4 = -2.25

Hence P(x > 21) = P(z > -2.25) = [total area] - [area to the left of -2.25]
= 1 - 0.0122 = 0.9878
c) For x = 30 , z = (30 - 30) / 4 = 0 and for x = 35, z = (35 - 30) / 4 = 1.25

Hence P(30 < x < 35) = P(0 < z < 1.25) = [area to the left of z = 1.25] - [area to the
left of 0] = 0.8944 - 0.5 = 0.3944
Example 3: A radar unit is used to measure speeds of cars on a motorway. The speeds are
normally distributed with a mean of 90 km/hr and a standard deviation of 10 km/hr. What is
the probability that a car picked at random is travelling at more than 100 km/hr?
Solution
Let x be the random variable that represents the speed of cars. x has μ = 90 and σ = 10.
92
Illustration on the probability on the normal curve
We have to find the probability that x is higher than 100 or P(x > 100)
For x = 100 , z = (100 - 90) / 10 = 1
P(x > 90) = P(z >, 1) = [total area] - [area to the left of z = 1]
= 1 - 0.8413 = 0.1587
The probability that a car selected at a random has a speed greater than 100 km/hr is
equal to 0.1587
Example 4: The time taken to assemble a car in a certain plant is a random variable having a
normal distribution of 20 hours and a standard deviation of 2 hours. What is the probability
that a car can be assembled at this plant in a period of time a) less than 19.5 hours? b)
Between 20 and 22 hours?
Solution
Illustration on the probability on the normal curve
a) P(x < 19.5) = P(z < -0.25) = 0.4013

b) P(20 < x < 22) = P(0 < z < 1) = 0.3413
Example 5: The annual salaries of employees in a large company are approximately normally
distributed with a mean of $50,000 and a standard deviation of $20,000.
a) What percent of people earn less than $40,000?
b) What percent of people earn between $45,000 and $65,000?
c) What percent of people earn more than $70,000?
Solution
Illustration of the probability on the normal curve
93
a) For x = 40000, z = -0.5
Area to the left (less than) of z = -0.5 is equal to 0.3085 = 30.85% earn less than
$40,000.
b) For x = 45000 , z = -0.25 and for x = 65000, z = 0.75

Area between z = -0.25 and z = 0.75 is equal to 0.3720 = 37.20 earn between $45,000
and $65,000.
c)For x = 70000, z = 1
Area to the right (higher) of z = 1 is equal to 0.1586 = 15.86% earn more than
$70,000.
HAPPY HOUR: NOW TRY OUT THESE SIMPLE NUMBERS

Question 1: A large group of students took a test in Physics and the final grades have a mean
of 70 and a standard deviation of 10. If we can approximate the distribution of these grades by
a normal distribution, what percent of the students a) scored higher than 80? b) Should pass
the test (grades≥60)? c) Should fail the test (grades<60)?
Question 2: The time taken to assemble a car in a certain plant is a random variable having a
normal distribution of 20 hours and a standard deviation of 2 hours. What is the probability
that a car can be assembled at this plant in a period of time a) less than 19.5 hours? b)
between 20 and 22 hours?
Question 3: A professor has determined that the final averages in his statistics course are
normally distributed with a mean of 72 and a standard deviation of 5. He decides to assign his
grades for his current course such that the top 15% of the students receive an A. What is the
lowest average a student must receive to earn an A?
Question 4: The amount of tip the waiters in an exclusive restaurant receive per shift is
normally distributed with a mean of $80 and a standard deviation of $10. A waiter feels he
has provided poor service if his total tip for the shift is less than $65. Based on his theory,
what is the probability that he has provided poor service?
Question 5:
a) With illustrations, what do you understand by the following terms
I. Standard error of the mean
II. Two tailed test
a) Daily power consumption at Nile Breweries is normally distributed with mean 10,000
kilowatts and a standard deviation of 2,000 kilowatts. What is the chance that the
consumption of electricity on a given day is:
I) Greater than 13,000 kilowatts
II) Less than 8,000 kilowatts
III) Between 7,500 and 14,000 kilowatts
b) A total population of 700 students sat an examination for which the pass mark was 50.
Their marks are normally distributed. 28 students scored below 40 marks while 35
scored above 60 marks. 1) Compute the mean and standard deviation in marks. 2) By
using the computed mean and standard deviation, what is the chance that any student
passed the examination? 3) If the pass mark is reduced to 48, find the increase in the
number of students who passed 94
c) To investigate the average amount spent each year on home improvements in Uganda,
a random sample of 900 homeowners is selected. The sample mean is Uganda shillings
420 Millions and the standard deviation is estimated to be Uganda shillings 240 Millions.
I) Calculate 95% confidence intervals for the population mean.
II) What sample size would be required to estimate the population mean to within
Uganda shillings 10 Million with 95% confidence?
Question 6
a) List 4 properties of the normal distribution curve
b) Explain the difference between:
I. Type I and Type II error
II. Confidence interval and confidence limits
III. Null and alternative hypothesis
c) Kamese Jimmy, an employee of “Tuvawala enterprises” has estimated that the daily
demand for his products is approximately normally distributed with average sales of
twelve per day and standard deviation of four products. How many products must
Kamese stock in the morning so that not more than one chance in five of running out of
stock during the day?
d) His boss is not amused about his dwindling daily sales and has instructed him to write a
report about sales performance. After careful analysis, he has discovered that twenty
three percent of total daily sales are greater than sixty but only twenty one percent of
total daily sales are below forty. What average and standard deviation in the distribution
of sales should Kamese include in her report?
Question-7: The CPA registrar; has summarized examination results as shown below
Result Percentage of candidates

Passed with a distinction 10%
Passed without distinction 60%
Failed 30%
The registrar knows that a candidate fails if he/she obtains less than 40 marks (out of 100),
while he must obtain at least 75 marks in order to pass with a distinction. Determine the mean
and standard deviation in the distribution of marks, assuming this to be normal

June 2014 June 2010
95
CUMMULATIVE NORMAL DISTRIBUTION TABLE (CHOICE- A)
96
NORMAL DISTRIBUTION TABLE (CHOICE B)
97
STUDENT’S T- DISTRIBUTION TABLE
98
Chapter-11
Estimation and hypothesis testing
11.1 INTRODUCTION TO SAMPLING THEORY

Consider a problem of determining the average salary of the whole population of URA
employees. For this URA population, consider the hypothetical situation of drawing every
possible sample of size n from this population. Assume that there are; k such samples. If the
sample mean is computed for each of these k samples, then there will be k-sample means
produced.
If a frequency distribution of these k sample means is now constructed, and the mean and
standard deviation of this distribution of the k sample means is found, the following features
will emerge.
I. The sample means are themselves random variables since the values are likely to vary
from sample to sample.
II. The means of these k sample means will be found to be equal to the true population
mean; µx. Therefore, it is possible to say that the average of the sample means equals the
population mean
III. The standard deviation of these k-sample means will be equal to; x x 
  / n
. This is

called the Standard Error (SE) of the sample means. The SE assumes that the
population from which the sample is drawn is large and that the population standard
deviation is known.
IV. If the distribution of these k sample means will have the shape of the normal
distribution, this distribution of the sample means is called the sampling distribution of the
sample means (x bar).
11.2 CONFIDENCE INTERVALS

Suppose Ssebagala and Sons electrical center wants to find out the average life of some
electric bulbs in a large shipment from Phillips Corporation in England. Testing every bulb to
failure will certainly give the exact average life of these bulbs, but the knowledge the manager
would have obtained would be useless since there will be no bulbs left. The correct procedure
is for the company quality controller to take out a sample of n bulbs, test them to failure and
record their average life ( x‾ ) [which is the sample mean]. Then, he can take (x‾) as an
estimate of the population mean (µ). BUT, how much confidence can the manager have on his
estimate? If, he established that the mean life of sample bulbs (x‾) = 2,500hrs, surely the
population mean life (µ) can be greater than, OR less than 2,500. It is for this reason that the
manager needs to take an interval, i.e, a range of values about the sample mean in which the
population mean is likely to lie.
This process is known as interval estimation. BUT how wide is this interval? is usually a big
problem. To solve this problem we specify a confidence interval or level of confidence.
This refers to the degree of confidence we have. That is to say; the probability that the actual
population parameter lies within a specified range about the sample statistics. The population
parameter is what we are trying to estimate (µ in this case) and the sample statistic is what we
know by measurement (x‾) in this case. It can also be noted that apart from the interval
estimate, a point estimate also exists in estimating values of the true population. A point
99
estimate is when a single sample statistic is taken as the estimate of the unknown population
parameter.
Confidence interval for a single POPULATION MEAN µ [Large sample & population
standard deviation is known]
We shall use a simple formula below;
  
xz [x  z   xz ]
n Which expands into; n n
The one which has the negative sign is the lower confidence limit and one with a plus sign is
the upper confidence limit.
There are three key assumptions made;
A. The populati
B. The sample analyzed is large, and
C. The sample is a random sample.
11.3 THE MEANING OF A LARGE SAMPLE

In this type of analysis, a sample which contains more than 30 items is counted as large.
Different techniques must be used in the analysis of small samples, as we consider later in the
unit. Because of the time factor in examinations, you will sometimes be given questions
involving smaller samples and, in such cases, either you will be told that the sample is taken
from a normal population or you should state, as an extra assumption, that the population is
normal.
11.4 THE MEANING OF A RANDOM SAMPLE

A random sample is a sample that has been selected so that any other sample of the same
size was equally likely to have been selected, or it is a sample that has been selected so that
each individual item is equally likely to have been chosen. (These two definitions are
equivalent.) Note that we always assume that we are analyzing a random sample, although in
practice, because of the nature of the population or a restriction on the cost or time allowed
for the survey, some other method of selection has been used. If, in an exam, you are asked
to comment on the result of your analysis, you should state that a bias may have been
introduced by the method of selection.
The new Kampala City Authority is trying to establish satellite centers in the remote outskirts of
Kampala to reduce traffic congestion. A researcher from the authority has taken a sample of
100 taxi commuters from Gayaza and established that the sample mean time to reach Kampala
daily was is approximately 35.8minutes. UTODA already knows that the population standard
deviation in commuting time is 20minutes. Help the authority researcher set a 95% & 90%
confidence limits for the time taken to reach Kampala.
Solution
Interpretation
The probability is 0.95 that the actual time taken to commute from Gayaza to Kampala daily
lies between 31.88 minutes and 39.73 minutes
After several complaints of under pay far below the legal Ugandan minimum wage, the
workers MP along with a team of researchers from several labor unions undertook a study on
sample of 100 employees and found that average pay was just UGX35,600 and population 100
standard deviation in pay of UGX12,500. If the salaries are assumed to be normally distributed,
help the MP obtain a 95% and 99% confidence limits for the true population mean pay and
interpret the results.
Solution
Example3 on confidence interval for a single mean (large sample &

A manager from Ssebagala and Sons electrical center has taken a sample of 100 fluorescent
tubes imported in a consignment from Phillips Corporation-UK and established that their mean
life is 20.5 hours with a standard deviation of 1.6 hours. Help the manager establish (a) a 95%
confidence interval; (b) a 99% confidence interval for the average length of life of those bulbs.
Interpret the result in each case. Solution
Walk to work campaigns have been ignited by the escalating cost of living yet salaries are
stagnant says DP president; Norbert Mao. To prove this, a Makerere Don from the Institute of
Social Research has carried out a study on a random sample of 1,600 households Arua town.
He has established a sample mean salary of UGX25, 000 with a standard deviation of UGX12,
000. (a) Calculate a 95 per cent confidence interval for the mean salaries. AND (b) what
sample size would be required to estimate the mean to within ± UGX 400 with a 95 per cent
confidence interval?
Solution
Confidence interval for a single POPULATION MEAN µ [Large samples & population
standard deviation is UNKNOWN]
Previously, we have been assuming that the sample standard deviation is known. This
assumption helped to compute the standard error of the sample mean. However, in most
cases, the population standard deviation is unknown and needs to be estimated from the
sample standard deviation. When this is the case, probabilities for the sampling distribution of
the sample mean are derived, not from the z distribution, but from a distribution called the
student’s t distribution. Definition: this is a similar to the z distribution, bell shaped and
symmetrical about the mean of zero but doesn’t have a constant standard deviation (which the
z has). Its standard deviation is based on sample standard deviation and varies inversely with
the size of the sample used. Thus, when the population standard deviation is UNKNOWN and is
estimated by the sample standard deviation, the relationship between the sampling distribution
of the sample mean and its standardized values is given by the student’s t-distribution
translation formulae.
x  x s s
t( n1)  [ x  t( n1)    x  t( n1) ]
( s x / n ) AND the confidence Interval is; n n
Example1 on confidence interval for a single POPULATION MEAN µ [Small samples

& population standard deviation is UNKNOWN]
A manager at Nakumatt has analyzed the value of purchases made before mid-day. He took a
small sample of just 25 people entering the store and discovered that sample mean purchases
amounted to UGX165, 450 and standard deviation in purchases of UGX38,600. Set up a 95%
confidence interval for the actual mean value of purchases made at this store before mid day.
Solution
Confidence interval for a single POPULATION PROPORTION 101

If the data of the random variable under consideration is nominal/ ordinal- scaled, the
appropriate measure of central location is the proportion. In the same way, the population
mean can be estimated from the sample mean, a population p
estimated based on the point estimate of its sample proportion p.
So the confidence interval for a population proportion is given by;
pq pq
[p  z   p z ]
n n WHEN the sample size is large (n GREATER than 30)
pq pq
[ p  t( n1)    p  t( n1) ]
n n WHEN the sample size is small (n LESS than or equal to
30)
. p is the single sample proportion. q  (1  p)
Example1 on confidence interval for a single POPULATION PROPORTION

samples]
A professor at the University of Nairobi conducted a study on 120 traders and discovered that
42 of them felt that local by-laws were still hampering the progress of syndicated economic
trading among the east African states now that economic and political federations are at center
stage. Construct the 95% confidence limits for the true population proportion of traders who
believe that local by laws are still hampering effective trading.
Solution
Example2 confidence interval for a single POPULATION PROPORTION

samples]
In the aforementioned example, now assume that only 20 traders were interviewed. Construct
the 95% confidence limits for the true population proportion of traders who believe that local
by laws are still hampering effective trading. Solution
Confidence interval for DIFFERENCE between POPULATION Proportions 1- 2

[large samples]
Assuming that population standard deviations are known, and both sample sizes are large
(that is to say; both n1 + n2 are greater than 30) we use this confidence interval;
 pq pq 
( p1  p 2 )  z 1 1  2 2 

 n1 n2 
If we asssume that population standard deviations are UNKNOWN, and both sample sizes are
small (that is to say; both n1 + n2 are less than 30) we use this confidence interval;
 pq pq 
( p1  p 2 )  t  1 1  2 2 

 n1 n2 
Mind about the (n1+n2-2) degrees of freedom
Confidence interval for DIFFERENCE between POPULATION Proportions 1- 2

[large samples]
Crown Beverages recently conducted a taste study test for its Mirida fruity soda. The tests
were conducted concurrently in large super markets in Kampala and Jinja. 240 consumers 102
were surved in Kampala and 180 indicated that they liked the flavor. In Jinja, 400 consumers
participated and 320 liked the new flavor. Construct a 95% confidence interval for the
difference between the actual proportions of Kampala and Jinja consumers who liked the
flavor.
Solution
Now repeat this problem assuming that only 16 consumers were surveyed in Kampala and 12
liked the drink. And in Jinja only 10 were surveyed and 8 liked the drink. Solution
Confidence interval for DIFFERENCE between POPULATION MEANS µ1- µ2 [Large

samples & population standard deviation is known]
Assuming that population standard deviations are known, and both sample sizes are large
(that is to say; both n1 + n2 are greater than 30) we use this confidence interval;
 2  2
( x 1  x 2 )  z 1  2 
 n1 n2 
 
If we asssume that population standard deviations are UNKNOWN, and both sample sizes are
small (that is to say; both n1 + n2 are less than 30) we use this confidence interval;
 s2 s 2 
( x 1  x 2 )  t( n1 n2 2)  1  2 
 n1 n2 
 
Mind about the (n1+n2 - 2) degrees of freedom
Example1 on confidence interval for DIFFERENCE between POPULATION MEANS

µ1- µ2 [Large samples & population standard deviation is known]
Just before the CHOGM, the Uganda Tourist Board conducted a study to determine average
length of stay of British and German visitors to Bujagali and Murchison falls. A sample of 60
British and another sample of 90 German visitors were interviewed at the respective sites and
their respective mean lengths of stay were 15.2days and 11.7days respectively. Assume that
population standard deviations in lengths of stay for British tourists is 4days and that of
Germans is 6 days, help construct a 90% confidence interval for the estimate of the difference
in mean lengths of stay between British and German tourists in the reported sites. Solution
Now repeat the same question assuming that only 15british and 10 German tourists had been
interviewed. Interpret your findings Solution
11.5 HYPOTHESIS TESTS/ SIGNIFICANCE TESTS

Previously, you have leant how to set confidence intervals around sample statistics to
estimate the true value of the population parameter from which the sample is drawn. Another
area of inferential statistics involves establishing whether a claim OR generalization made
about the true value of a population parameter is valid. This is what is commonly known as
hypothesis testing.
Examples of claims made about population parameters

I. DFCU claiming that its average return on investment is 25% p.a
II. A PWC tax auditor is convinced that that more than 15% of all Serena Hotel tax returns
are incorrectly completed
III.
IV.
Mukwano industries claiming that one in every 4 households uses their detergents
YANA tyres company claiming that the average tread life of their tyres is 100,000km
103
To test these claims, sample data is gathered and analyzed. On the basis of sample findings,
the hypothesized value of the population parameter is either accepted as probably true or
rejected as probably false. The process of testing the validity of a claim about the true
value of any population parameter is called HYPOTHESIS TESTING.
Hypothesis tests will be conducted on the following 4 population parameters which are usually
measures of location
( x )
A. Hypothesis tests on a single population mean
B. Hypothesis tests on a single population proportion ( )
C. Hypothesis tests on a difference between two population means (1  2 )
D. Hypothesis tests on a difference between two population proportions (1   2 )
Process / steps in hypothesis tests. To adequately carry out the tests, we follow a
standard five step procedure.
1. Formulate the null & alternative hypotheses
2. Determine the area of acceptance/ rejection (ie chose the level of
significance)
3. Compute the corresponding sample statistic
4. Compare the sample statistic to the area of acceptance
5. Conclude; i.e. accept or reject the null hypothesis
Steps in detail
The null hypothesis, H0: the sample belongs to the population.
The alternative hypothesis, H1: the sample does not belong to the population.
Definitions: A research hypothesis is a prediction of the outcome of a study. The
prediction may be based on an educated guess or a formal theory.
The null hypothesis states that the true population parameter value is equivalent to the
hypothesized population parameter (Stated as H0). The alternative hypothesis states that
the true population parameter is not equivalent to the hypothesized population parameter
(Stated as H1). It is always expressed in a manner which negates the null hypothesis.
The null hypothesis can be stated in one of the three ways depending on the nature of
hypothesis tested.
1. A TWO-SIDED hypothesis test (Sometimes called Non-directional

hypothesis)
When a claim is made that the population parameter IS EQUAL to a specified value and the
test is to verify the specified value only, then the null hypothesis is formulated as a TWO
SIDED.
H0: Population parameter = specified value (like 10.9km)

H1: Population parameter  specified value (like 10.9km)
The null hypothesis will be rejected in favor of the alternative hypothesis if evidence points
towards the true population parameter being either significantly less, or greater than the null
hypothesized value.
104
2. A ONE-SIDED hypothesis test [UPPER TAILED hypothesis test]
When a claim is made that the population parameter IS LESS THAN OR EQUAL to a
specified value and the test is to verify the specified value only, then the null hypothesis is
formulated as a one-sided (UPPER TAILED) test.
H0: Population parameter  specified value (like 10.9km)
H1: Population parameter  specified value (like 10.9km)
towards the true population parameter being only significantly greater than the null
hypothesized value.
3. A ONE-SIDED hypothesis test [LOWER TAILED hypothesis test]

When a claim is made that the population parameter IS GREATER THAN OR EQUAL to a
specified value and the test is to verify the specified value only, then the null hypothesis is
formulated as a one sided (LOWER TAILED) test.
H0: Population parameter  specified value (like 10.9km)
H1: Population parameter  specified value (like 10.9km)
towards the true population parameter being only significantly less than the null hypothesized
value.
Note. When formulating the NULL & ALTERNATIVE hypothesis, it’s a must that the equality
sign should always appear in the null hypothesis. This equates the population parameter to a
specified hypothesized value (two sided test) or a range of hypothesized values which either
begins (if the test were one sided or lower tailed) or ends (if the test were one sided-upper
tailed) with the specified value. The strict inequality sign must appear only in the alternative
hypothesis.
4. Determining the area of acceptance and rejection

The sampling distribution of any sample statistic is used to test the validity of a claim as stated
in the null hypothesis. The central value of a sampling distribution is assumed to be the null
hypothesized population parameter. The area of acceptance is a range of values of the
sample statistic centered about the null hypothesized population parameter that would lead to
an acceptance of the null hypothesis for values of the sample statistic which falls within limits.
Similarly, area of rejection is that range of sample statistic values that would lead to a
rejection of the null hypothesis for values of the sample statistic which fall within limits.
For a two tailed test, the rejection area is both below and above the null hypothesized
value of the population parameter.
105
For a upper tailed test, the rejection area is above the null hypothesized value of the
population parameter.
For a lower tailed test, the rejection area is below the null hypothesized value of the
population parameter.
Types of errors: Because the sample measures are variables, we run a risk of making the
wrong decision in two ways. We may reject Ho when it is true, making what is called a Type I
error, or we may accept Ho when it is false, making what is called a Type II error.
Accept H0 Reject H0
H0 true Correct decision Making a type I Error
H0 false Making a type II Error Correct decision
Thus; a TYPE I ERROR is the probability of rejecting the null hypothesis when in fact it’s true
(it is represented by the area under the sampling distribution over the area of rejection). A
TYPE II ERROR is the probability of accepting the null hypothesis when in fact it is false. The
type I error is called the level of significance and is represented by the symbol  (alpha).
The level of significance is the area (probability) in the “tails” of a sampling distribution and
used to find cut off points that separate the area of acceptance from the area of rejection. This
level of significance in fact defines the likelihood of rejecting the null hypothesis when it is in
fact true (that’s to say; type I error).
106
The figure below illustrates the relationship between the level of significance and the area of
acceptance for any sampling distribution for a two tailed test
The level of significance (α - alpha) appears only in the upper tail of a sampling distribution for
a one sided (upper tailed) hypothesis test. Conversely, for a one sided (lower tailed)
hypothesis, the level of significance is contained only in the lower tail of the sampling
distribution. The shaded areas represented above show the level of significance.
There is no direct algebraic connection between the values of the risks of making the two
errors, but as one risk is decreased the other is increased. We use the Type I error in
hypothesis tests and the Type II error to find the power of the test, which is outside the scope
of this course. Having formulated the hypotheses, you must decide on the size of the risk of
making a Type I error that you are prepared to accept. For example, you may decide that a
5% risk is acceptable (making the wrong decision once in 20 times) or you may require only a
1% risk (wrong once in a 100 times). You then divide all the possible values of the sample
measure into two sets, putting (say) 5% of them in the critical or rejection region and 95% in
the acceptance region. The percentage chosen is called the level of significance. Next calculate
the value of the measure for the sample you are using, see which region it belongs to and
state your decision.
Example1 on hypothesis test for as SINGLE POPULATION MEAN [Large samples &
population standard deviation is known]
Dr. Tanga Odoi, the chairman Makerere University Staff Association claims that on average a
Ugandan university lecturer receives UGX2.85millions per month which is too little to
counteract the rising cost of living. Ministry of education officials wish to test this claim and
have conducted a study on a sample of 250 lectures from all the government universities
throughout the country. They have discovered that sample mean monthly pay is
UGX3.03millions. Assuming that the population standard deviation of monthly pay is
UGX1.00million, test the claim at 5% significance level that the average monthly pay is indeed
UGX2.85millions
Solution
Step1. Stating the hypotheses
H0: Mean salary = 2.85 millions
H1: Mean salary  2.85 millions
107
Step2. Determine the acceptance and rejection regions (using the level of significance)
This is a two tailed test;   5% . This is a TWO TAILED TEST.
Illustration
Step3. Compute the necessary statistic

Here the z statistic is the most appropriate (ie n  30 and population standard deviation
is given). (Please note that this is very important!!!)
 x  x 
z   2.846
 ( x / n ) 
Step4. Compare the computed Z statistic value to the critical z value(s)
From the normal tables; get the critical z values (there are +1.96 & -1.96). It can be seen that
the computed Z value lies in the rejection region as indicated above.
Step5. Conclude (reject criteria).

Reject the H0 at 5% significance level and conclude that the average monthly pay for MUK
lectures is not UGX2.85millions. The professor’s claim cannot be supported on the basis of the
sample evidence received
Example2 on hypothesis test for a SINGLE POPULATION MEAN [Large samples &
Test the above claim at 5% that the monthly pay is more than UGX2.85millions
Solution
NOTE: This is a one sided test (UPPER TAILED TEST) because the claim that the monthly
wage is strictly greater than UGX2.85millions.
Step1. Stating the hypotheses

H0: Mean salary  2.85 millions
H1: Mean salary  2.85 millions
H1: reflects the view that the average monthly pay is STRICTLY greater than the hypothesized
value of 2.85 millions (which is what the ministry officials want to establish).
Now complete the rest of the steps
Step2. Determine the area of acceptance/ rejection (i.e. chose the level of significance)
108
Step3. Compute the corresponding sample statistic
Step4. Compare the sample statistic to the area of acceptance

Step5. Conclude; i.e. accept or reject the null hypothesis
Test the above claim at 5% that the monthly pay is LESS THAN UGX2.85millions
Solution
Example3 on hypothesis test for as SINGLE POPULATION MEAN [Large samples &
Question1. Explain the difference between the following terms:
a. Interval estimate and a point estimate
b. A confidence interval and a random sample
c. A z and t-test
d. Small and large sample
Question2. A sample of 100 fluorescent light tubes from the Short Life Tube Company has a
mean life of 20.5 hours and a standard deviation of 1.6 hours. Test:
(a) At the 1% level whether the sample comes from a population with mean 23.2 hours.
(b) At the 5% level whether it comes from a population with mean 20.8 hours.
(c) At the 5% level whether it comes from a population with mean less than 20.8 hours.
Solution
Example4 on hypothesis test for a SINGLE POPULATION MEAN [Large samples &
I. Explain any 5 properties of a normal distribution
II. Explain the difference between the following terms
a. Type I and type II errors
b. Null and Alternative hypotheses
c. One tailed and two tailed hypothesis tests
d. Lower tail and upper tail tests.
III. A sample of 150 students had an average IQ of 112 with a standard deviation of 9.
(a) At what level of significance would you accept that this sample is taken from a student
population with average IQ of 110?
(b) At the 5% level would you accept that the student population had an average IQ greater
than 113?
Solution
Example3 on hypothesis test for DIFFERENCE BETWEEN POPULATION MEANS

[Large samples & population standard deviation is known]
Alice, an employee of NWSC has always claimed that delivery times of our local courier service
providers were different. She decided to use both Posta Uganda and EMS over a period of 3
months to deliver water bills to similar destinations. Delivery times were recorded. POSTA was
used 60times and the sample mean delivery time was found to be 42minutes. EMS was used
48times and the sample mean delivery time was found to be 38minutes. Assume that the
population standard deviation of the delivery times for POSTA 14minutes and that of EMS is
10minutes, test at 5% significance level that there is no difference between mean delivery
times between the two courier services.
Solution
H0: 1  2  0
H1: 1  2  0
109
Note the z statistic to be used is given as;
 
 
 x1  x 2 
z   1.73
 1   2
2 2

 n1 n2 
Assume that Alice wished to test whether POSTA’s mean delivery time is greater than EMS
delivery time. Test this claim at 5% level of significance.
H0: 1  2  0 [POSTA’s time is less or equal to EMS]

H1: 1  2  0 [POSTA’s time exceeds EMS]
Example1 on hypothesis test for as SINGLE POPULATION PROPORTION [Large

samples & population standard deviation is known]
The SAMONA marketing manager sampled 360 females who used skin care products. 108
respondents indicated that they used the company products exclusively. Test this claim at 1%
significance level that the market share of SAMONA products is 35%
Solution
H 0 :   0.35
H1 :   0.35
Note the z statistic to be used is given as; p is the sample proportion (108/360) and  is the
population proportion; 0.35
 
 p  
z   1.989
  (1   ) 
 
 n 
Example2 on hypothesis test for DIFFERENCE BETWEEN POPULATION

PROPORTIONS [Large samples & population standard deviation is known]
MTN engaged Synovate research company to establish whether both teenagers and adults had
the same recall rate of their slogan “”MTN; Everywhere you go”. The research company
interviewed 640 teenagers and 420 young adults. 333 teenagers recalled the slogan and 260
adults also recalled correctly the slogan. Test at 10% level whether there is an equal recall rate
between teenagers and adults ( that is to say; the marketing campaign was equally effective to
both groups)
Solution
H 0 : 1   2  0
H1 : 1   2  0
Note the z statistic to be used is given as; p is the sample proportion (108/360) and  is the
population proportion; 0.35
110
 
 
 p1  p2 
z   3.175

 ˆ (1  ˆ )   
1 1  x x x x
 n n  ˆ  1 2 p1  1 ; p2  2
  1 2 
BUT;
n1  n2 ; n1 n2
If we are dealing with SMALL SAMPLES (n1+n2 is less than 30); we use the respective
statistics.
Example1 on small samples; the mean weekly sales of NOMI detergent was 146.3 Sackets
per store. After an extensive advertisement campaign, the mean weekly sales in 22 stores for
a particular week increased to 153.7 and showed a standard deviation of 17.2. Is the evidence
conclusive that the advertising campaign was successful? Test this at 5% significance level.
Solution
 x  x 
t   1.92
 x
s / n  . Mind about the (n-1) degrees of freedom
Example1 on use of the use of a t- table in hypothesis tests. One mean

Test the hypothesis that the mean is 40% on such a sample of values.
Sample A. 38%, 32%, 44%, 37% AND 34%
Solution
Example2 on small samples (difference between two means).

The operations manager of HIMA cement believes that their Kampala branch is performing
better than their Jinja Branch in terms of the number of orders received. The accountant wants
to verify this statement and has taken a random sample of 10 orders from Kampala and found
that they had a mean value of UGX13.16millions with a standard deviation of UGX2.58millions.
Orders received by the Jinja branch were also randomly selected and the average of the 13
orders sampled was UGX11.57millions and standard deviation 3.23millions. Establish whether
the operations manager’s belief is valid at 5% significance level.
Solution
 
 
 x1  x 2 
t ;
2 2 
 s s
 
 n1 n2 
 (n1  1) s12  (n2  1) s2 2 
AND s   ; this is called the pooled variance
2
 n1  n2  2 
Example3 on small samples (difference between two means).

Two groups of students are asked to take a test. The results are:
Group A scores 45, 87, 64, 92, and 38
Group B scores 40, 35, 61, 50, 47, and 32
Is there a significant difference in mean score between the two groups?
Solution
111
Chapter-12
Chi square tests
12.1 INTRODUCTION
Previously, we learned how to test hypotheses using data from one or two samples. We used
one-sample tests to determine if a mean was significantly different from a hypothesized value
and two-sample tests to determine if the difference between two means was significant. These
tests are known as parametric tests, because they involve testing the parameters of a
population, such as the mean and proportions. They use the parametric statistics of
samples that come from the population being tested. To formulate these tests, we
make assumptions about the population, for example, that the population is normally
distributed. There are certain kinds of data that cannot be tested in this way such
as: data which was not collected in a random sample and therefore does not have a
normal distribution; ordinal data; ranked data; and data from more than two
populations. In business, we often encounter data of this type, such as:
A. The results of a survey of which brand of washing powder consumers prefer

B. An analysis of the arrival of customers at supermarket checkouts
C. A survey of employees' attitudes towards performance appraisal in different
departments
D. A study of whether male staff have been more successful in passing professional
examinations than female staff.
For these types of data, it is necessary to use tests which do not make restrictive assumptions
about the shape of population distributions. These are known as non-parametric tests. Non-
parametric tests have certain advantages over parametric tests:
 It is not necessary to assume that the population is distributed in the shape of a normal
curve or in any other specific shape
 They can be used with data measured in ordinal or nominal scales, which is often the
type of data obtained from business surveys, particularly where self-completion
questionnaires are used
 Generally, they are quicker to do and easier to understand; sometimes, formal ranking
or ordering is not necessary.
But non-parametric methods are often not as precise as parametric tests, because they use
less information. In this study unit, we are going to consider one of the most commonly used
non-parametric tests, called the chi-squared test (pronounced "ki-squared"). Critical values
in the chi-squared test are denoted by the symbol χ2. The chi-squared test is principally used
to determine:
 If two population attributes are independent of each other, or

 If a given set of data is consistent with a particular distribution, known as the
"goodness of fit" test
We will consider each of these versions of the chi-squared test in turn.
112
12.2 THE CHI – SQUARED GOODNESS OF FIT
The chi-squared test can also be used to determine if there is a similarity – that is, a good fit
– Between the distribution of observed data and a theoretical probability distribution, such as
the normal distribution or the binomial or Poisson distributions. We first perform the chi-
squared test to establish whether there is a significant difference between our observed
distribution and the theoretical distribution we have chosen; this information then enables us
to decide whether the observed data is a sample from our hypothesized theoretical
distribution. The chi-squared goodness of fit test is therefore a useful tool for managers, who
often need to make decisions on the basis of statistical information. For example, a
maintenance manager at a factory may use information about the frequency of breakdowns to
decide how many engineers to deploy on each shift.
There is a series of steps which need to be undertaken to determine goodness of fit:-
(a) Formulating the null and alternative hypotheses

(b) Calculating the expected frequencies
(c) Calculating the chi-squared statistic of the sample
(d) Determining the appropriate number of degrees of freedom
(e) Ascertaining whether the sample statistic falls inside the acceptance region
We will now consider each of these in more detail, taking as an example data collected on the
amount of money consumers spend on chocolate each week.
Example 1: In one of the random surveys, IPSOS consultancy services asked clients one
question ‘In general, how do you rate the quality of services provided by Ugandan businesses?’
The distribution of responses to this question was as follows:
Excellent 8%
Pretty good 47%
Only fair 34%
Poor 11%
UMEME Management wants to establish whether the results of this consumer survey apply to
their company. They randomly interviewed 207 consumers as leave UMEME offices in various
parts of the city. The key question asked was ‘How do you rate the quality of customer service
at our branches’. The results obtained were
Response Frequency observed (f0)

Excellent 21
Pretty good 109
Only fair 62
Poor 15
Use the chi –squared goodness test to establish whether the observed frequencies of
responses from the company survey are the same as the frequencies that-would-be expected
based on the IPSOS national survey. Use a 5% level of significance
Solution
113
H0: The observed distribution is the same as the expected distribution
HA: The observed distribution is NOT the same as the expected distribution
Response Expected proportion Expected Frequency (fe)

(Proportion x Sample Total)
Excellent 8% 207 x 8% = 16.56
Pretty good 47% 207 x 47% = 97.29
Only fair 34% 207 x 34% = 70.38
Poor 11% 207 x 11% = 22.77
Total = 207
Response f0 fe   f  fe 2 
 f 
 e 
Excellent 21 16.56 (
Pretty good 109 97.29
Only fair 62 70.38
Poor 15 22.77
Total 207 207 6.25
  f 0  fe 
2

 2     6.25
 fe 

With four categories, K = 4: Thus, the degrees of freedom are V = K-1; V= 4-1 = 3
Therefore;
  ,v 2   0.05,3 2  7.815
Illustration of the Chi Squared distribution, the computed and critical values
Because the computed X2 value of 6.25 falls in the acceptance region, we do not reject the null
hypothesis. Therefore, the data gathered on 207 customers indicates that the distribution of
114
responses of customers is not significantly different from the distribution of responses to the
IPOS Survey.
Example 2: Dairy farmers in Mbarara want to know whether sales of their milk are uniformly
distributed over a full year, so that they may plan for milk production and storage. A uniform
distribution means that the observed frequencies are the same in all categories. The farmers
gather data from one large supermarket and record it as follows
Month Liters Month Liters Month Liters Month Liters sold

sold sold sold
January 1610 April 1590 July 1410 October 1564
February 1585 May 1540 August 1350 November 1602
March 1649 June 1397 September 1495 December 1655
Total = 18,447
Use a 10% level of significance to assess whether the data captured fit a uniform distribution
Solution
H0: The monthly figures of milk sales are uniformly distributed

HA: The monthly figures of milk sales are NOT uniformly distributed

18,447
fe   1,537.25litres
12
If the frequencies are uniformly distributed, the same number of liters are expected to
be sold each month as indicated above
Month f0 fe   f  fe 2 
 f 
 e 
January 1610 1537.25
February 1585 1537.25
March 1649 1537.25
April 1590 1537.25
May 1540 1537.25
June 1397 1537.25
July 1410 1537.25
August 1350 1537.25
September 1495 1537.25
October 1564 1537.25
November 1602 1537.25
December 1655 1537.25
Total 18,447 18,447 74.37
  f 0  fe  2

 2     74.37
 fe 
115
With four categories, K = 4: Thus, the degrees of freedom are V = K-1; V= 12-1 = 11
Therefore;   ,v   2 0.01,11  24.725
2
Because the computed X2 value = 74.37 is greater than the critical Chi Squared value, there is
enough evidence to reject the null and accept the alternative that the distribution of milk sales
is not uniform
Example 3: Kifefe, a teller within DFCU acacia branch believes that the distribution of random
arrivals at his branch are Poisson distributed. He attempts to test this hypothesis by recording
the information below relating to observed frequencies of arrivals on a one minute interval
within the bank.
Number of arrivals Observed frequencies
0 7
1 18
2 25
3 17
4 12
5 5
Solution
H0: The frequency distribution is Poisson

HA: The frequency distribution is NOT Poisson

To determine the expected frequencies, the supervisor must obtain the probabilities of each
category of arrivals and then multiply them by the total of the observed frequencies. These
probabilities are obtained by determining labda and then using the Poisson distribution table or
computing them manually, (We have seen this previously). Since it is the mean of the Poisson
distribution, lambda can be determined from the observed data by computing the mean of the
data. in this case the supervisor computes the weighted mean by summing the product of
number of arrivals with frequency of those arrivals and dividing that sum by the total number
of observed frequencies
116
Number of arrivals Observed frequencies Arrival x Observed
frequencies
0 7 0
1 18 18
2 25
3 17
4 12
5 5
84 192
192
The mean =    2.3
84
With the value of lambda, and the Poisson distribution table / computed Poisson
probabilities, the supervisor can determine the probabilities of the number of arrivals in each
category. The expected probabilities can also be computed manually (Seen this already for X =
0, 1, 2, 3, and 4). Using these probabilities and the total of 84 from the observed data, she
computes the expected frequencies by multiplying each expected probability by the total 84.
Number of arrivals Expected probabilities Excepted frequencies

0 0.1003 8.42
1 0.2306 19.37
2 0.2652 22.28
3 0.2033 17.08
4 0.1169 9.82
5 0.0837 7.03
84

She uses the expected frequencies and the observed frequencies to compute the calculated
Chi-squared value
Number of f0 fe   f  fe 2 
arrivals
 f 
 e 
0 7 8.42 0.24
1 18
2 25
3 17
4 12
5 5 7.03 0.59
84 84 1.74
  f 0  fe 2 
  
2
  1.74
 fe 

The degrees of freedom for a goodness of fit test involving a Poisson distribution are always
K-2 and K is the number of recorded arrivals.
Thus, the degrees of freedom are V = K-2; V= 6-2 = 4 117
Therefore;   ,v   2 0.05, 4  9.488
2
(e) Ascertaining whether the sample statistic falls inside the acceptance
region
Because the computed X2 value = 1.74 is not greater than the critical Chi Squared value of
9.488, so the supervisor’s decision is not to reject the null hypothesis. That is, she fails to
reject the hypothesis that the distribution of the bank arrivals is actually Poisson.
Example 5: Suppose that a researcher has developed an overall attitude scale to determine
how employees feel towards their company. In theory, the scores can vary from 0 to 50. The
researcher pretests the measurement instrument on a randomly selected group of 100
employees. He tallies the scores and summarizes them into the six categories shown below.
Score category Frequency (f)

10- under 15 11
15- under 20 14
20- under 25 24
25- under 30 28
30- under 35 13
35- under 40 10
100
By using a significance level of 0.05, use the Chi squared goodness of fit to test the hypothesis
that the pretest scores are approximately normally distributed.
Solution
H0: The attitude scores are normally distributed
HA: The attitude scores are NOT normally distributed
Number of arrivals Expected probabilities Excepted frequencies

0 0.1003 8.42
1 0.2306 19.37
2 0.2652 22.28
3 0.2033 17.08
4 0.1169 9.82
5 0.0837 7.03
84
118
To compute the calculated Chi squared value, the researcher must determine the expected
frequencies. For a normal distribution, he computes the expected frequencies by using the Z
table, the mean and the standard deviation. Thus, we need to first compute the mean and the
standard deviation
Score category Frequency (f) X FX FX2

10- under 15 11 12.5 137.5 1,718.75
15- under 20 14 17.5
20- under 25 24
25- under 30 28
30- under 35 13
35- under 40 10
100 2,490 67,125
X 
  f .x   ..  24.9
f ...
 fx   fx 
2 2
2     7.194 ( SD)
N  N 
Therefore, we use these two statistics to compute the expected probabilities for each category
of score. Use the normal probability distribution Z scores as indicated below
X 
Z
SD
Score category Expected probability

10- under 15 Z = (10-24.9)/7.194 =-2.07; P= 0.4808
Z = (15-24.9)/7.194 =-1.38; P= 0.4162
Expected probability = 0.4808- 0.4162 = 0.0646
15- under 20 Z = (15-24.9)/7.194 =-1.38; P= 0.4162
Z = (20-24.9)/7.194 =-0.68; P= 0.2517
Expected probability = 0.4162- 0.2517 = 0.1645
20- under 25 0.2557
25- under 30 0.2571
30- under 35 0.1581
35- under 40 0.0629
The six probabilities do not have a sum of 1.000. Even though observed frequencies were
obtained only for the six categories, getting a score less than 10 or greater than 40 was also
possible. Because 0.500 of the probabilities lie in each half of a normal probability distribution
and utilizing the sum of the expected probabilities on each side of the mean, 24.9, the
researcher can obtain the probability of <10 category as -.5- (0.0646+0.1645+0.2517)=
0.0192. Similarly, he can obtain the probability of the >40 category; 0.5 –
(0.004+0.2571+0.1581+0.0629) = 0.0179
119
He obtains the expected frequencies by multiplying each excepted probability by the total
frequency (100) as shown below
Score category Expected probability Expected frequency

<10 0.0192 1.92
10- under 15 0.0646 6.46
15- under 20 0.1645
20- under 25 0.2557
25- under 30 0.2571
30- under 35 0.1581
35- under 40 0.0629
>40 0.0179 1.79
1.000 100.00
As the <10 and >40 categories have expected values of less than 5, each must be collapsed
(combined) with the adjacent category. As a result, the <10 category becomes part of the 10-
15 category and the >40 category becomes part of the 35-40 category.
Score category Expected probability Expected frequency

10- under 15 0.0838 8.38
15- under 20 0.1645 16.45
20- under 25 0.2557 25.57
25- under 30 0.2571 25.71
30- under 35 0.1581 15.81
35- under 40 0.0808 8.08
1.000 100.00
The Chi squared value can be calculated as follows.
Category f0 fe   f  fe 2 
 f 
 e 
10- under 15 11 8.38 0.82
15- under 20 14 16.45 0.36
20- under 25 24 25.57
25- under 30 28 25.71
30- under 35 13 15.81
35- under 40 10 8.08 0.46
2.44
  f 0  fe 2 
 2     2.44
 fe 

The degrees of freedom for a goodness of fit test involving a Normal distribution are always
K-3 and K is the number of recorded score bands.
Thus, the degrees of freedom are V = K - 3; V= 6 - 3 = 3
Therefore;   ,v   2 0.05,3  7.815
2
120
region
Because the computed X2 value = 2.44 is not greater than the critical Chi Squared value of
7.815, so the researcher does not reject the null hypothesis that these categories are normally
distributed
CHI-SQUARED AS A TEST OF INDEPENDENCE

The Chi Square goodness of fit test is used to analyze the distribution of the frequencies for
categories of one variable like age or number of bank arrivals, to determine whether the
distribution of these frequencies is the same as some hypothesized or expected distribution.
However, the goodness of fit test cannot be used to analyze two variables simultaneously. A
different Chi square test, the Chi-square test of independence, can be used to analyze the
frequencies of two variables with multiple categories to determine whether the two variables
are independent. Many a time, this type of analysis is desirable. For example, a market
researcher might want to determine whether the type of soft drink preferred by a consumer is
independent of the consumer’s age. An organizational behaviorist might want to know whether
absenteeism is independent of job satisfaction.
The Chi square test of independence can be used to analyze any level data measurement, but
it is particularly useful in analyzing nominal data.
Example 1: A researcher is interested in determining whether a type of gasoline preferred is

independent of a person’s income. He used a simple questionnaire and asked respondents one
question about gasoline preference and another question about income. The researcher tallied
the responses and obtained the following results
Type of gasoline
Income Regular Premium Extra premium Totals
Less than 85 16 6 107
USD30,000
30,000-49,999 102 27 13 142
50,000-99,999 36 22 15 73
More than 15 23 25 63
100,000
Total 238 88 59 385
Using a level of significance of 10%, use the chi square test of impendence to ascertain
whether the type of gasoline preferred is independent of income level.
121
Solution
H0: The type of gasoline preferred is independent of income

HA: The type of gasoline preferred is independent of income

The expected frequencies are computed as (Row total x column total)/ Grand total
We compute expected frequencies for each observed value in the table as follows
Expected frequency for value 85 = F85 = (107 x 238)/ 385 = 66.15

Expected frequency for value 16 = F16 = (107 x 88)/ 385 = 24.46
Expected frequency for value 6 = F6 = …
Expected frequency for value 36 = F36= …
Expected frequency for value 25 = F25= …

We use the expected frequencies and the observed frequencies to compute the calculated Chi-
squared value
f0 fe   f  fe 2 
 f 
 e 
85 66.15 5.37
16 24.46
6 ..
102
27
13
36
22
15
15
23
25 24.42
Total 70.78
  f 0  fe 2 
 2     70.78
 fe 

The degrees of freedom for a Chi Square test of independence are computed as 122
v  df  r  1c  1.
v  df  4  13  1  6
Therefore;   ,v   2 0.01,6  16.812
2
region
Because the computed X2 value = 70.78 is greater than the critical Chi Squared value of
16.812, the researcher rejects the null hypothesis. That’s to say, the type of gasoline preferred
is not independent of income.
Question 3: DFCU Bank undertook a study on staff preferences for a car leasing scheme in
three locations of Mbarara, Jinja and Kampala.
Preference Mbarara Jinja Kampala Total N0. Of

N0.of staff N0.of staff N0.of staff staff
Scheme1 76 66 61 203
Scheme2 42 35 49 126
At 5% significance level, confirm whether car leasing scheme is independent of location.
Question 4: It is thought that the ages of the club’s members may be related to their
preferred sports. To investigate this, the following table was constructed:
Preferred sports
Age Swimming Badminton Squash
Under 20 20 20 10
20 and under 40 10 30 20
Over 40 10 10 30
Calculate the  statistic to test the hypothesis that age and preferred sport are independent.
2
Use a 5% level of significance and state your conclusions carefully.
Question 5: Is the transportation mode used to ship goods independent of the type of
industry? Suppose that data shown represent frequency counts of types of transportation used
by the publishing and computer hardware industries. Analyze the data by using the chi- square
test of independence to determine whether the type of industry is independent of the
transportation mode at 5% level of significance.
Transportation mode
Air Train Truck
Industry Publishing 32 12 41
Computer hardware 5 6 24 123
Below is the chi-squared distribution table that appears in your examinations
124
Chapter-13
Control charts
13.1 INTRODUCTION
It’s essential that all products produced in a factory meet the requirements of those who use
them. Every product possesses a number of elements that jointly describe its fitness for use.
These elements are called quality characteristics.
13.2 QUALITY CHARACTERISTICS MAY BE OF SEVERAL TYPES:-

A. Physical; length, weight, volume and viscosity
B. Sensory; taste, appearances and color.
C. Time orientation; reliability, maintainability and service ability.
In most companies, it’s difficult to produce products that have no defects. If a product is to
meet the customers’ fitness for use criteria, it should be produced by a process that’s stable
(repeatable). The production process should operate in little variability around the target
dimensions of the products’ quality characteristics. Statistical Process Control is a powerful
collection of problem solving tools useful in achieving process stability and improving the
capability through reduction of variability.
The 7 major tools that are commonly used to reduce variability in production processes and
products include; the control chart, Histogram, Check sheet, Pareto chart, Cause and effect
diagram, Defect concentration diagram and the Scatter diagram. For this particular paper, we
shall concentrate more on the control charts.
Variability in any production process can be due to;

 Chance causes of variability in products and production processes OR
 Assignable causes of variability in products and production processes.
In any production process regardless of how its well designed or carefully maintained, a certain
amount of inherent or natural variability will always exist. This natural variability (OR
background noise) is the cumulative effect of many small essentially unavoidable causes. A
production process operating with only chance causes of variability in products is said to be in
statistical control.
Other kinds of variability due to; improperly adjusted machines, operator errors, defective raw
materials may occasionally be present in products. These are usually large and easily
noticeable. We refer to these sources of variability that are not part of the chance caused
pattern as assignable causes. A production process operating in the presence of assignable
causes of variation as operating out of control
The major objective of statistical quality control is to quickly detect the occurrence of
assignable causes of process shifts so that investigations of the process and collective action
may be undertaken before any non-conforming units are manufactured. Here we use control
charts to do the investigation.
13.3 CONTROL CHARTS

The CONTROL CHART is a graphical display of the quality characteristic that has been
measured or computed from the sample Vs. the sample number or Time. The chart contains
the Center Line (CL) that represents the average value (mean) of the quality characteristic
corresponding to the in control state. Two other horizontal line; Upper Control Limit (UCL) and 125
the Lower Control Limit (LCL) for which if all sample points lie within/ between them then the
production process is in control. If a product plots outside the control limits then there is
evidence that the process is out of control and investigation is required to find and eliminate
the assignable causes responsible for this behavior. It’s customary to connect the sample
points on the control chart with straight lines so as to easily visualize how the sequence of the
points has evolved overtime. Even if all points plot inside the control limits. If they behave in a
systematic/ non random manner then this is an indication that the process is out of control.
Example.
13.3.1 USES OF CONTROL CHARTS

a) Show Changes in Data Pattern. I.e, Permits Seeing Trends - Make Corrections Before
Process Is Out of Control.
b) Show When Changes in Data Are Due to;
I. Special or Assignable Causes; Represent Problems to be Corrected, In
Chart, Data Outside Control Limits or Trend
II. Chance or Common Causes; Random Variations
c) They are proven techniques for improving quality. A successful control chart shall
reduce scrap and rework
d) Prevention of defects. Helps to keep the process in control with a philosophy of do it
the first time.
e) They prevent unnecessary process adjustments. They can detect chance causes and
assignable causes of variability in production process.
f) They provide diagnostic information which allows the implementation of changes in
production processes
g) Provide information about process capability and information about process parameters
and their stability over time.
13.4 TYPES OF CONTROL CHARTS

 Variable control charts. These are control charts for measurements like weight, length
etc. ( X bar and Range Charts).
 Attribute control charts. These are control charts for attribute compliance ( P and C
charts)
13.5 CALCULATING MAJOR LINES IN A CONTROL CHART

 Average Value: take the average of the sample data 126
 UCL: Multiply the Standard deviation by three. Then add that value to the Average
Value.
 LCL: Multiply the Standard deviation by three. Then subtract that value from the
Average Value.
13.6 Types of control charts

There are generally two types of control charts; 1) control charts for measurements and 2)
control charts for compliance items. In this section, we discuss two common types of control
charts for measurements; the X bar chart and the range charts. We shall also briefly discuss
the two types of attribute compliance; the P charts and C charts.
Each control chart has a centerline, an upper control limit and a lower control limit. Data are
recorded on the control chart, and the control chart is examined for disturbing patterns or for
data points that indicate a process is out of control. Once it is determined that a process is out
of control, measures can be taken to correct the problem causing the deviation.
The X bar chart
It is a graph of sample means computed over a series of small random samples over a period
of time. These means are average measurements of some product characteristic. For example,
the measurement could be the volume of fluid in a liter of alcohol, the thickness of a metal
sheet etc. The sample means are plotted on a graph that contains a centerline and upper and
lower control limits (UCL & LCL).
The followings steps are used to produce an X bar chart.

Step 1 Decide on the quality to be measured
Step 2 Determine the sample size
Step 3 Gather 20 to 30 samples
Step 4 Compute the sample average X bar for each sample
Step 5 Compute the sample range R for each sample
Step 6 Determine the average sample mean for all samples X bar bar
Therefore;
Upper Control Limit (UCL) X  A2 R
Center Line (X Bar Bar)
X=
X 
k
Lower Control Limit (LCL) X  A2 R
Alternatively, you can still use control limits by computing standard deviations for each sample
Upper Control Limit (UCL) X  A3 S
S=
S
k
Lower Control Limit (LCL) X  A3 S
127
Example 1: Given the data below: Is the process in control?
Number of samples
Observations 1 2 3 4 5 6 7 8 9 10
1 5 0 -15 -15 -5 10 10 -5 5 20
2 25 10 -10 0 5 -15 15 -5 5 -10
3 -25 0 0 -25 5 -10 5 5 -5 0
4 0 5 -10 -5 0 -10 -10 -15 -5 -5
5 15 0 0 5 -10 5 -15 10 10 5
The data given below are in 0.0001 units. Plot both the X bar and Range charts
Solution
First, compute the mean and ranges of all the 10 samples given as follows
Number of samples
Observations 1 2 3 4 5 6 7 8 9 10
Sample mean (X Bar) 4 3 -7 -8 -1 -4 1 -2 2 2
Sample range (R.) 50 10 15 30 15 25 30 25 15 30
Sample standard
deviation (S)
Upper Control Limit (UCL) X  A2 R =  1  (0.5768 x24.5)  13.13

X=
X 
4  3  ....  2  2  10
  1
k 10 10
Lower Control Limit (LCL) X  A2 R =  1  (0.5768 x24.5)  15.13
R Bar
=R=
 R  50  10  ....  15  30  245  24.5
k 10 10
Note: A2 value corresponding to a sample of size 5 is obtained directly from the tables:
When using Standard deviation approach, A3 is also obtained directly from tables
Plot of the X bar control chart
128
The Range (R.) Chart
Upper Control Limit (UCL) =D R 3
Center Line (R Bar)

=
R
k
Lower Control Limit (LCL) = D4 R
Note: D3 and D4 corresponding to sample size are obtained directly from the table
From the previous number,
Upper Control Limit (UCL) = D R  (0 x 24.5)  0
3
Center Line (R Bar)

=
 R  24.5
k
Lower Control Limit (LCL) = D4 R  (2.114 x24.5)  51.79  52
Now you can plot the range chart using the new control limits
Plot of the Range control chart
You can also plot a P- Chart by using the following controls

P- Chart represents proportions chart
Upper Control Limit (UCL) PQ
=P3
n
Center Line (R Bar)
=
p
k
Lower Control Limit (LCL) PQ
=P3
n
129
Example; Given the following data. Construct a proportion chart (P)
Sample 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
N 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50
N0 out of 4 3 1 0 5 2 3 1 4 2 2 6 0 2 1 6 2 3 1 5
compliance
Proportion (P) 0.08 0.06 0.10
Solution
You can also plot a C- Chart by using the following controls

C- Chart represents number of non-conformance per item or unit chart
Upper Control Limit (UCL) =C  3 C

Center Line (R Bar)
=
C
k
Lower Control Limit (LCL) =C  3 C 130
TABLE OF CONTROL CHART CONSTANTS
X-bar Chart for sigma R Chart Constants S Chart Constants
Constants estimate
Sample A2 A3 d2 D3 D4 B3 B4
Size = n
2 1.880 2.659 1.128 -- 3.267 -- 3.267
3 1.023 1.954 1.693 -- 2.574 -- 2.568
4 0.729 1.628 2.059 -- 2.282 -- 2.266
5 0.577 1.427 2.326 -- 2.114 -- 2.089
6 0.483 1.287 2.534 -- 2.004 0.030 1.970
7 0.419 1.182 2.704 0.076 1.924 0.118 1.882
8 0.373 1.099 2.847 0.136 1.864 0.185 1.815
9 0.337 1.032 2.970 0.184 1.816 0.239 1.761
10 0.308 0.975 3.078 0.223 1.777 0.284 1.716
11 0.285 0.927 3.173 0.256 1.744 0.321 1.679
12 0.266 0.886 3.258 0.283 1.717 0.354 1.646
13 0.249 0.850 3.336 0.307 1.693 0.382 1.618
14 0.235 0.817 3.407 0.328 1.672 0.406 1.594
15 0.223 0.789 3.472 0.347 1.653 0.428 1.572
16 0.212 0.763 3.532 0.363 1.637 0.448 1.552
17 0.203 0.739 3.588 0.378 1.622 0.466 1.534
18 0.194 0.718 3.640 0.391 1.608 0.482 1.518
19 0.187 0.698 3.689 0.403 1.597 0.497 1.503
20 0.180 0.680 3.735 0.415 1.585 0.510 1.490
21 0.173 0.663 3.778 0.425 1.575 0.523 1.477
22 0.167 0.647 3.819 0.434 1.566 0.534 1.466
23 0.162 0.633 3.858 0.443 1.557 0.545 1.455
24 0.157 0.619 3.895 0.451 1.548 0.555 1.445
25 0.153 0.606 3.931 0.459 1.541 0.565 1.435
HAPPY HOUR: TRY OUT THIS NUMBER

Question
a) The major objective of statistical quality control is to quickly detect the occurrence of
assignable causes of process shifts so that investigations of the process and collective
action may be undertaken before any non-conforming units are manufactured. Here
there is wide use of control charts. Explain the various uses of control chart in modern
product quality management.
b) Kapere (U) limited is a food processing company that makes potato chips and cheese
chips. Although their products are packaged and sold by weight, the company has been
taking sample bags of these chips and counting the number of chips in each bag.
Shown below is the number of chips per bag for five samples of seven bags of chips.
Sample 1 Sample 2 Sample 3 Sample 4 Sample 5

25 22 30 32 25
23 21 23 26 23
29 24 22 27 29
31 25 26 28 27
26 23 28 25 27
28 26 27 25 26
27 29 21 31 24
Use the data above to construct an X bar control chart and R control chart and briefly discuss
the results. 131
Chapter-14
Linear algebra and calculus
14.1 BASIC ALGEBRA

Algebra is a branch of mathematics in which, instead of using numbers, we use letters to
represent numbers. We all know that 2 + 3 = 5. Suppose, though, that we substitute letters
for the first two numbers, so that: 2 = a, and 3 = b. We can then write: a + b = 5. Therefore,
basic algebra can help us find the unknown when given certain equation.
Some of the more commonly encountered functions in the business world include:
Linear, Polynomials, Multivariate, Logarithmic, Discrete vs continuous, step function and
exponential functions.
14.2 APPLYING BASIC LINEAR ALGEBRA TO BUSINESS

For purposes of this examination, you must understand how to formulate and solve two simple
equations. For instance, given the two equations:
3x + 4y = 52 AND 5x + y = 30. Find x and y

Solution
3x + 4y = 52 Look for terms that can be eliminated. The equations do
5x + y = 30 not have any x or y terms with the same coefficient.
5 (3x + 4y) = 5 (52) In order to use the elimination method, you have to
5x + y = 0 create variables that have the same coefficient—then
you can eliminate them. Multiply the top equation by 5.
15x + 20y = 260
5x + y = 30
15x + 20y = 260 Now multiply the bottom equation by −3.
-3 (5x + y) = −3 (30)
15x + 20y = 260

−15x – 3y = −90
15x + 20y = + 260 Next add the equations, and solve for y.
−15x – 3y = – 90
17y = 170
y = 10
3x + 4y = 52 Substitute y = 10 into one of the original equations to
3x + 4(10) = 52 find x.
3x + 40 = 52
3x = 12
x=4
The solution is (4, 10) You arrive at the same solution as before.
Note: The examiner is testing whether you can apply this knowledge of formulating two
simple equations and then solving them.
Example1; the sum of two numbers is 10. Their difference is 6. What are the two numbers?
x + y = 10 Write a system of equations to model the situation.
x – y = 6 x = one number
y = the other number 132
x+y= Add the equations to eliminate the y-term and then
10 solve for x.
+ x–y=
6
2x = 16
x =8
x + y = 10 Substitute the value for x into one of the original
8 + y = 10 equations to find y.
y =2
Example2; A theater sold 800 tickets for Friday night’s performance. One child ticket costs
$4.50 and one adult ticket costs $6.00.The total amount collected was $4,500. How many of
each type of ticket were sold?
Example3; When Kapele’s wife priced the pens at 110/-, 30 pens were bought. But when she
reduced the price to 50/-, the demand increased by 20 units. Required: If you assume that
the relationship between the price of these pens and quantity demanded is linear of the form
P= a + bQ, determine the equation for this relationship and compute the price when 22 pens
are bought
133
Solution
Desirable linear equation:P= a + bQ
PriceQuantity
110/-30 Units
50/-50 Units
The two linear equations are110 = a + 30b
50 = a + 50b
On solving simultaneously; a=200 & b = - 3
Final linear equation= P= 200 - 3Q
When 22 pens are bought; the price is; P= 200 – 3 (22)= 134/-
Example4; A shirt is on sale for $15.00 and has been marked down 35%. How much was the
shirt being sold for before the sale?
Solution
Let’s start with defining p to be the price of the shirt before the sale. It has been marked
down by 35%. This means that 0.35p has been subtracted off from the original price.
Therefore, the equation (and solution) is,
So, with rounding it looks like the shirt was originally sold
for $23.08.
Question 1: Solve the following simultaneous equations

x  y  0 x  y  1 3x  4 y  20 2 y  4 x  6  y  21 x  4
    
2 x  y  3 x  2 y  7 4 x  5 y  10 4 y  8 x  12 2 y  x  6
3x  2 y  47.5 4 x  y  7 100 x  100 y  500 4m  2n  2 3a  2b  14
    
5x  4 y  2713 x  2 y  2 200 x  100 y  100 5m  n  8 a  3b  10
1
.
Question 2: Applying linear equations in real scenarios

CPA Paper-6 students have appreciated the services of Mr. Mabreezy and want to buy him a
gift. If each member of the class contributes 6 dollars towards the purchase of the gift, they
will still be short by 16 dollars. On the other hand if each member contributes 9 dollars, they
will have a surplus of 11 dollars, after the purchase of the gift. Find the number of the
students in the class. Find the cost of the gift.

a. Excited NRM top officials want to buy a present for their candidate YKM7 for winning a
highly contested presidential election. If each top NRM official contributes Shillings
6,000,000 towards the purchase of the gift, they will still be short by 16,000,000. On the
other hand if each member contributes Shillings 9,000,000, they will have a surplus of
Shillings 11,000,000, after the purchase of this expensive gift. Find the number of top
officials in this meeting. Find the cost of this expensive gift.
134
b. One of the NRM top officials insists that their candidate deserves customized BMWs and
Mercedes Benzes- Model 2016. When they contacted Spear Motors Limited, they
discovered that 3 BMWs and 5 Mercedes Benzes were costing a whopping Shillings
900,000,000. While, 4 BMWs and 10 Mercedes Benzes were costing Shillings 1,
400,000,000. Utilize your basic knowledge simple linear equations to find the price of each
car.

a. In a certain Algebra class there is a total of 350 possible points. These points come from 5
homework sets that are worth 10 points each and 3 hour exams that are worth 100 points
each. A student has received homework scores of 4, 8, 7, 7, and 9 and the first two exam
scores are 78 and 83. Assuming that grades are assigned according to the standard scale
and there are no weights assigned to any of the grades is it possible for the student to
receive an A in the class and if so what is the minimum score on the third exam that will
give an A? What about a B?
b. We want to build a set of shelves. The width of the set of shelves needs to be 4 times the
height of the set of shelves and the set of shelves must have three shelves in it. If there
are 72 feet of wood to use to build the set of shelves what should the dimensions of the
set of shelves be?

a. A calculator has been marked up 15% and is being sold for $78.50. How much did the
store pay the manufacturer of the calculator?
b. Two cars are 500 miles apart and moving directly towards each other. One car is moving at
a speed of 100 mph and the other is moving at 70 mph. Assuming that the cars start
moving at the same time how long does it take for the two cars to meet?
c. Two boats start out 100 miles apart and start moving to the right at the same time. The
boat on the left is moving at twice the speed as the boat on the right. Five hours after
starting the boat on the left catches up with the boat on the right. How fast was each boat
moving?
d. An office has two envelope stuffing machines. Machine A can stuff a batch of envelopes in
5 hours, while Machine B can stuff a batch of envelopes in 3 hours. How long would it take
the two machines working together to stuff a batch of envelopes?
14.2 APPLYING BASIC QUADRATIC EQUATIONS TO BUSINESS

All the equations we have studied so far have been of a form where all the variables are to the
power of 1 and there is no term where x and y are multiplied together – i.e. xy does not
feature in the equation. (The types of equation studied so far are known as linear equations.
By contrast, a quadratic equation is one that contains the square of the unknown number, but
no higher power. For example, the following are quadratic expressions in x: Either; 4x2+ 7x OR
2x2 - x +1 OR ½x2- 2
Many a time, quadratic equations are in the form of ax2+bx+c

It is possible to have examination questions that demand expression and three quadratic
equations and then you are required to solve them.
Example1: Gashumba has a small shop at Garden city shopping mall those sales ladies
leggings (q). He has cost his latest leggings from China as follows:
5 leggings cost Shs. 13,525
2 leggings cost Shs. 12,880 135
8 leggings cost Shs. 14,280
Assuming Gashumba’s cost of the leggings is a quadratic function in terms of q, obtain that
particular quadratic function and also establish the cost of 12 pants based on the same cost
equation
Solution
Desirable quadratic equation: Cost  a Q 2  bQ  C
Quantity Cost
5 Units 13,525/-
2 Units 12,880/-
8 Units 14,280/-
The three quadratic equations are 13525  25a  5b  c.........i
12880  4a  2b  c..........ii
14280  64a  8b  c............iii
When equations (i) and (ii) are 645  21a  3b...........iv
solved simultaneously, we get:-
When equations (ii) and (iii) are 1400  60a  6b..............v
solved simultaneously, we get:-
When equations (iv) and (v) are a  6.11
solved simultaneously, we get:- b  172.22
When the values of (a) and (b) 13525  25(6.11)  5(172.22)  c.
obtained above are substituted in c  12,511 .2
equation (i) above; we get (c) as:-
Cost  6.11 Q 2  172 .22Q  12511 .2
Thus; the cost of 12 leggings is Cost  6.11(12 2 )  172.22(12)  12511 .2  15457 .68 / 
HAPPY 10 MINUTES: NOW TRY OUT THIS NUMBER

Question1: An NRM top official has suggested that they also need to be appreciated with
brand new vehicles since they equally did a good Job during the presidential campaigns. Due
to cost considerations, they have settled for Nissan Murano- model 2016. The Manager at
Motor Care (U) Limited has informed the happy officials that the cost of shipping the vehicles
can vary depending on the quantity shipped. Below is a schedule of shipping costs (C ) Vis-Vis
quantity of vehicles ordered (Q)
Quantity of vehicles ordered (Q) 10 40 100

Shipping cost of the vehicles (C ) 2,900 800 2,000
Note: The Shipping costs are in Billions of Uganda Shillings

If the shipping costs for the vehicles follow a quadratic equation in the form of:-
COST  aQ 2  bQ  k
Use your knowledge of basic calculus to estimate the quadratic cost equation and also estimate
costs required to ship in 80 vehicles
14.4 SOLVING QUADRATIC EQUATIONS

Example 1: Solve the equation: x2 - 9x + 20 = 0 by factorization
Solution
First, factorize the quadratic equation x2 - 9x + 20 = 0
Find two numbers which add up to 9 and multiply to give 20. These numbers are 4 and 5. 136
(x - 4) (x - 5) = 0
Now find the value x so that when these brackets are multiplied together the answer is
0. This means either (x - 4) = 0 or (x - 5) = 0
So x = 4 or x = 5. You can check these answers by substituting 4 and 5 in to the
equation: x2- 9x + 20: Substituting 4 gives: 42 - 9 × 4 + 20 = 16 - 36 + 20 = 0
Substituting 5 gives: 52 - 9 × 5 + 20 = 25 - 45 + 20 = 0
Remember these 3 simple steps and you will be able to solve quadratic equations.
Example 2: Solve x2 + 5x + 6 = 0.
This equation is already in the form "(quadratic) equals (zero)" but, unlike the previous
example, this isn't yet factored. The quadratic must first be factored, because it is only
when you MULTIPLY and get zero that you can say anything about the factors and
solutions. You can't conclude anything about the individual terms of the unfactored
quadratic (like the 5x or the 6), because you can add lots of stuff that totals zero.
So the first thing I have to do is factor:
x2 + 5x + 6 = (x + 2)(x + 3)
Set this equal to zero:
(x + 2)(x + 3) = 0
Solve each factor:
x + 2 = 0 or x + 3 = 0
x = –2 or x = – 3
The solution to x2 + 5x + 6 = 0 is x = –3, –2
We can also use quadratic formulae

Example3: Solve for x in the following equation. x2 +2x -8 = 0
Example4: Solve for x in the following equation. 3x2 -10x +5 = 0
137
HAPPY HOUR: NOW TRY OUT THESE NUMBERS
Solve these quadratic equations
1 x 2  2x  8 2 3  7x  2x 2  0 3 22t  t 2  21
4 t 2  7  6t 5 5x 2  2  9x 6 6 p 2  5  17 p
7 y2  2y  4 8 2r 2  6r  3 9 3q  2q 2  7
10 20  3r 2  5r 11 5  2x  x 2 12 3t 2  14t  5
14.5 CALCULUS DIFFERENTIATION

Differentiation: Used to find maximum or minimum points of curves of certain business
functions for example cost, revenue and profit functions.
Derivative notations (How derivatives are usually presented)
f’(x) y’
dy d
[ f ( x)]
dx dx
RULES OF DIFFERENTIATION
1. The Constant Function Rule
The derivative of a constant function f (x) = k, where k is a constant, is zero.
Given f (x) = k, f (x) = 0
The Linear Function Rule

The derivative of a linear function f (x) = mx + b is equal to m, the coefficient of x. The
derivative of a variable raised to the first power is always equal to the coefficient of the
variable, while the derivative of a constant is simply zero.
Given f (x) = mx + b, f  (x) = m
138
Example
2. The Linear Function Rule
If y = a + bx
dy
b
dx
dy
e.g. y = 10 + 6x then dx  6
The Power Function Rule

The derivative of a power function f (x) = kxn, where k is a constant and n is any real number,
is equal to the coefficient k times the exponent n, multiplied by the variable x raised to the n –
1 power.
Given f (x) = kxn, f  (x) = k  n  xn – 1
Example
The Rules for Sums and Differences

The derivative of a sum of two functions f (x) = g(x) + h(x), where g(x) and h(x) are both
differentiable functions, is equal to the sum of the derivatives of the individual functions.
Similarly, the derivative of the difference of two functions is equal to the difference of the
derivatives of the two functions.
Given f (x) = g (x)  h (x), f  (x) = g (x)  h (x)
139
Example
The Product Rule

The derivative of a product f (x) = g(x)  h(x), where g(x) and h(x) are both differentiable
functions, is equal to the first function multiplied by the derivative of the second plus the
second function multiplied by the derivative of the first. Given f (x) = g(x)  h(x), f (x) = g(x) 
h (x) + h(x)  g (x).
d du dv
(uv)  v u
Or; for f (x) = (u v). f (x) = dx dx dx
Example
Find the derivative of y = (6x3-x) (10-20x)
Solution
The Quotient Rule

The derivative of a quotient f (x) = g(x)  h(x), where g(x) and h(x) are both differentiable
functions and h(x)  0, is equal to the denominator times the derivative of the numerator,
minus the numerator times the derivative of the denominator, all divided by the denominator
squared.
h( x)  g ( x)  g ( x)  h( x)
If , f ( x)  g ( x) / h( x).thenf ( x) 
[h( x)]2
du dv
v u
d  u  dx dx
 
Or for f (x) = (u/ v). dx  v 
2
v
Example
Find the derivative of 140
Solution
The Chain Rule

Given a composite function, also called a function of a function, in which y is a function of u
and u in turn is a function of x, that is, y = f (u) and u = g(x), then y = f [g(x)] and the
derivative of y with respect to x is equal to the derivative of the first function with respect to u
times the derivative of the second function with respect to x:
dy dy du
 x
dx du dx
Example
Find the derivative of the function: f (x) = (4x3 +7)5
Solution:
Combination of both rules

Here both the quotient and product rules could be required to solve a particular business or
economics problem.
Example7
Find the derivative of the function: f (x) = 4x (3x - 1) / (2x-5)
Solution:

Differentiate the following functions
IDENTIFYING TURNING POINTS: RELATIVE EXTREMA

A relative extremum is a point at which a function is at a relative maximum or relative
minimum. To be a relative extremum, the point must be a critical point. A critical point is a
point in the domain of a function, where the derivative equals zero or is undefined.
Testing for relative extrema

For a relative minimum at x = a, we need f‘(a) = 0 and f” (a) > 0. 141
For a relative maximum at x = a, we need f‘(a) = 0 and f” (a) < 0.
For f” (a) = 0. The test is inconclusive. I.e. point of inflexion.
DIFFERENTIATION IN ECONOMICS
Application I
• Total Costs = TC = FC + VC
• Total Revenue = TR = P * Q
•  = Profit = TR – TC
• Break-even:  = 0, or TR = TC
• Profit Maximisation: MR = MC
Application I: Marginal Functions (Revenue, Costs and Profit)
Example 1: A firm faces the demand curve P=17-3Q

(i) Find an expression for TR in terms of Q
(ii) Find an expression for MR in terms of Q
Solution
TR = P.Q = 17Q – 3Q2
d TR
MR   17  6Q
dQ
Example 2: A firms total cost curve is given by TC=Q3- 4Q2+12Q

(ii) Find an expression for AC in terms of Q
(iii) Find an expression for MC in terms of Q
(iv) When does AC=MC?
(v) When does the slope of AC=0?
(vi) Plot MC and AC curves and comment on the economic significance of
their relationship
Solution
142
Example 3: A company manufactures and sells X units of a product per month. The monthly
cost and demand equations are given respectively as below
x
C  72,000  6 x and P  200  respectively and 0  x  6,000
30
Required
Find;
(i) The sales revenue function
Solution
Given that price function= x
P  200 
30
It should be noted that sales revenue = (Price x Quantity)
Thus, the sales revenue x  x2 
P  (200  ) x  Re venu  200 x  
30  30 
(ii) The production level that will maximize the company’s profits and state the resulting
profit.
Solution
Sales Revenue=  x2 
Re venu  200 x  
 30 
Cost function = C  72,000  6 x
Profit function= Total Revenue – Costs
Profit function=  x2 
200 x    (72,000  6 x)
 30 
Profit function= x2
194 x   72,000
30
OR; Multiplying through by 30; Pr   x 2  5820 x  2,160,000
At maximum profit d Pr
 0; Thus;2 x  5820  0; x  2,910
dX
On the second derivative, d 2 Pr
 2  0; Thus _ Maximum
dX 2
Hence; 2,910 Units maximize the company’s profits
Maximum Profit= Pr  (2,910 ) 2  5820 (2,910)  2,160,000
= 6,308,100/- 143
Question1: The daily profit, P, of an oil refinery is given by P = 8x − 0.02x2, where x is the
number of barrels of oil refined. How many barrels will give maximum profit and what is the
maximum profit?
Question 2: Given that a firm has a total revenue function of TR= 3300q - 26q2 and the
firms total costs are represented by q3-2q2+420q+750. Assuming that q > 0; find the
profit maximising function and also compute the maximum profit.
Question3: A manufacturer knows that if x hundred products are demanded in a particular

week; the total cost function would be; TC = 14 + 3x and the corresponding revenue
function would be; TR = 19x – 2x2
I. Derive the total profit function.
II. Find the profit break even points
III. Calculate the level of demand that maximises the profit of the company and hence the
maximum profit.
Question4.
If a company has a demand function; p = x2 – 24x+ 117. Establish the number of units of
production and the price at which total revenue will be maximized and hence find the total
maximum revenue.
November 2014
June 2014
December 2011
June 2010
December 2008
144
Chapter-15
Linear programming
15.1 INTRODUCTION
Linear Programming: Problem Formulation
Linear Programming is a mathematical technique for optimum allocation of limited or scarce
resources, such as labor, material, machine, money, energy and so on to several competing
activities such as products, services, jobs and so on, on the basis of a given criteria of
optimality.
The term ‘Linear’ is used to describe the proportionate relationship of two or more variables
in a model. The given change in one variable will always cause a resulting proportional change
in another variable. The word ‘programming’ is used to specify a sort of planning that
involves the economic allocation of limited resources by adopting a particular course of action
or strategy among various alternatives strategies to achieve the desired objective. Hence,
Linear Programming is a mathematical technique for optimum allocation of limited or scarce
resources, such as labor, material, machine, money energy etc among competing products or
activities.
15.2 AREAS WHERE LINEAR PROGRAMMING IS USED

There are numerous examples of linear programming applications to such problems, including:
• Determining optimal schedules for airlines for planes, pilots, and ground personnel
• Developing financial plans
• Determining optimal blends of animal feed mixes
• Determining optimal diet plans
• Identifying the best set of worker–job assignments
• Developing optimal production schedules
• Developing shipping plans that will minimize shipping costs
• Identifying the optimal mix of products in a factory.
15.3 COMPONENTS OF LINEAR PROGRAMMING MODELS

1. Objective function: is a mathematical expression that can be used to determine the total
profit (or cost, etc., depending on the objective) for a given solution.
2. Decision variables: represent choices available to the decision maker in terms of amounts
of either inputs or outputs.
3. Constraints: are limitations that restrict the alternatives available to decision makers. The
three types of constraints are less than or equal to (≤), greater than or equal to (≥), and
simply equal to (=).
4. Parameters: Numerical values or constants.
5. Feasible solution space The set of all feasible combinations of decision variables as
define by the constraints.
15.4 STRUCTURE OF LINEAR PROGRAMMING MODEL.

The general structure of the Linear Programming model essentially consists of three
components.
I. The activities (variables) and their relationships
II. The objective function and;
III. The constraints
The activities are represented by x1 , x2 ,.....xn . These are known as Decision variables.
145
The objective function of an LPP (Linear Programming Problem) is a mathematical
representation of the objective in terms a measurable quantity such as profit, cost, revenue,
etc.
Optimize (Maximize or Minimize) Z = c1x1+ c2x2+ ……. cnxn. Where Z is the measure of
performance variable. x1, x2,…xn are the decision variables. And c1, c2, ……. cn are the
parameters that give contribution to decision variables. The constraints are the set of linear
inequalities and/or equalities which impose restriction of the limited resources.
15.5 ASSUMPTIONS OF LINEAR PROGRAMMING

Certainty. In all LP models it is assumed that, all the coefficients of the decision variables are
known with certainty. Coefficients such as unit profit contributions, prices and the amount of
resources required per unit output are known constants. The available resources are also
assumed to be known with certainty.
Divisibility (Continuity). It’s assumed that the decision variables can take on fractional
values. Non-integer values of decision variables are acceptable. Therefore, linear programming
allows a production programme which uses 600 units of electrical components and 76 2 man
3
1
hours of labor time to produce 80 units A and 33 2 units of B. It’s practically feasible to have
fractional values in the resource allocation and production activities in many business
situations.
Additivity. The value of the objective function for the given value of decision variables and
the total sum of resources used, must be equal to the sum of the contributions (Profit or Cost)
earned from each decision variable and sum of the resources used by each decision variable
respectively. The objective function is the direct sum of the individual contributions of the
different variables
Linearity. All relationships in the LP model (i.e. in both objective function and constraints)
must be linear.
Non-negativity: negative values of decision variables are unacceptable. Linear programming
(LP) techniques consist of a sequence of steps that will lead to an optimal solution to problems
there are four components provide the structure of a linear programming model.
15.6 GUIDELINES/ STEPS FOR FORMULATING LINEAR PROGRAMMING MODEL

I.Identify and define the decision variable of the problem.
II.Define the objective function.
III.State the constraints to which the objective function should be optimized (i.e.
Maximization or Minimization)
IV. Add the non-negative constraints from the consideration that the negative values of
the decision variables do not have any valid physical interpretation.
V. Draw the graph and find the coordinates of the extreme points for each constraint and
hence obtain the maximum/ minimum values.
15.7 USING GRAPHS TO SOLVE LINEAR PROGRAMMING PROBLEMS (LPP)

Note1. This method is suitable when we have only 2 decision variables. In case, there are
more than 2 decision variables, we use the Simplex Algorithm method.
Note2. A company can be interested to Maximize; profits, revenues, output etc. At the same
time, a company can be interested to minimize; costs, man hours, inputs etc.
Note3. A LPP can have mixed inequality signs
146
Note4. A LPP can have a unique optimal solution, an infinite number of optimal
solutions, an unbounded solution, or even NO solution at all.
MAXIMIZATION PROBLEMS
Example1. MAXIMIZATION Problem (Generating a linear programming problem &
solving it using the GRAPHICAL METHOD)
Max z = 3x + 4y subject to the following constraints:
Compute the values of X & Y that maximize the above Linear Programming Problem (LPP)
Solution
The three inequalities in the curly braces are the constraints. The area of the plane that
they mark off will be the feasibility region. The formula "z = 3x + 4y" is the
optimization equation (Objective function). We need to find the (x, y) corner points of
the feasibility region that return the largest values of z.
The step is to solve each inequality for the more-easily graphed equivalent forms:
It's easy to graph the system. Obtain coordinates for each constraint indicated above
147
To find the corner points -- which aren't always clear from the graph – we pair the lines
(thus forming a system of linear equations) and solve:
y = –( 1/2 )x + 7 y = –( 1/2 )x + 7 y = 3x
y = 3x y=x–2 y=x–2
–( 1/2 )x + 7 = 3x –( 1/2 )x + 7 = x – 2
3x = x – 2
–x + 14 = 6x –x + 14 = 2x – 4
2x = –2
14 = 7x 18 = 3x
x = –1
2=x 6=x
y = 3(–1) = –3
y = 3(2) = 6 y = (6) – 2 = 4
corner point at (2, 6) corner point at (6, 4) corner pt. at (–1, –3)
So the corner points are (2, 6), (6, 4), and (–1, –3).
Somebody really smart proved that, for linear systems like this, the maximum and
minimum values of the optimization equation will always be on the corners of the
feasibility region. So, to find the solution to this exercise, I only need to plug these
three points into "z = 3x + 4y".
(2, 6): z = 3(2) + 4(6) = 6 + 24 = 30

(6, 4): z = 3(6) + 4(4) = 18 + 16 = 34
(–1, –3): z = 3(–1) + 4(–3) = –3 – 12 = –15
Therefore, the company can produce 6 Units of X and 4 Units of Y to
maximize a profit of 34 shillings
Then the maximum of z = 34 occurs at (6, 4)
Example2. MAXIMIZATION Problem (Generating a linear programming problem &

A calculator company produces a scientific calculator and a graphing calculator. Long-term

projections indicate an expected demand of at least 100 scientific and 80 graphing calculators
each day. Because of limitations on production capacity, no more than 200 scientific and 170
graphing calculators can be made daily. To satisfy a shipping contract, a total of at least 200
calculators much be shipped each day. If each scientific calculator sold results in a $2 loss, but
each graphing calculator produces a $5 profit, how many of each type should be made daily to
maximize net profits?
Solution
The question asks for the optimal number of calculators, so my variables will stand for that:
x: number of scientific calculators produced
y: number of graphing calculators produced
Since they can't produce negative numbers of calculators, I have the two constraints, x > 0
and y > 0. But in this case, I can ignore these constraints, because I already have that x > 100
and y > 80. The exercise also gives maximums: x < 200 and y < 170. The minimum shipping
requirement gives me x + y > 200; in other words, y > –x + 200. The profit relation will be my
optimization equation: P = –2x + 5y.
148
So the entire Linear Programming Problem is:
Maximize P = –2x + 5y, subject to:
100 < x < 200
80 < y < 170
y > –x + 200
The feasibility region graphs as:
When you test the corner points at (100, 170), (200, 170), (200, 80), (120, 80), and (100,
100), you should obtain the maximum value of P = 650 at (x, y) = (100, 170). That is, the
solution is "100 scientific calculators and 170 graphing calculators".
MINIMIZATION PROBLEMS
Example1. MINIMIZATION Problem (Generating a linear programming problem &
In order to ensure optimal health (and thus accurate test results), a lab technician needs to
feed the rabbits a daily diet containing a minimum of 24 grams (g) of fat, 36 g of
carbohydrates, and 4 g of protein. But the rabbits should be fed no more than five ounces of
food a day. Rather than order rabbit food that is custom-blended, it is cheaper to order Food X
and Food Y, and blend them for an optimal mix. Food X contains 8 g of fat, 12 g of
carbohydrates, and 2 g of protein per ounce, and costs $0.20 per ounce. Food Y contains 12 g
of fat, 12 g of carbohydrates, and 1 g of protein per ounce, at a cost of $0.30 per ounce. What
is the optimal blend?
Solution
Since the exercise is asking for the number of ounces of each food required for the optimal
daily blend, my variables will stand for the number of ounces of each:
x: number of ounces of Food X
y: number of ounces of Food Y
Since I can't use negative amounts of either food, the first two constrains are the usual ones:
x > 0 and y > 0. The other constraints come from the grams of fat, carbohydrates, and protein
per ounce:
149
Fat: 8x + 12y > 24
Carbs: 12x + 12y > 36
Protein: 2x + 1y > 4
Also, the maximum weight of the food is five ounces, so:
x+y<5
The optimization equation will be the cost relation C = 0.2x + 0.3y, but this time I'll be finding
the minimum value, not the maximum.
After rearranging the inequalities, the system graphs as:
(Note: One of the lines above is irrelevant to the system. Can you tell which one?)
When you test the corners at (0, 4), (3, 0), and (1, 2), you should get a minimum cost of sixty
cents per daily serving, using three ounces of Food X only.

Question1. MAXIMIZATION Problem (Generating a linear programming problem &
A manufacturer produces two types of models of a product; M1 and M2. Each model of the type
M1 requires 4 hours of grinding and 2 hours of polishing; whereas each model of M2 requires 2
hours of grinding and 5 hours of polishing. The manufacturer has 2 grinders and 3 polishers.
Each grinder works for 40 hours a week and each polisher works 60 hours a week. Profit on M1
model is Shs.3.00 and on model M2 is Shs.4.00.Whatever produced in a week is sold in the
market. Using the graphical method, how should the manufacturer allocate his production
capacity to the two types of models, so that he makes maximum profit in a week?

Kindo makes necklaces X1 and bracelets X2. Necklace have a profit margin of 24/= and
bracelets 18/=. Necklace takes 2 hours for stone cutting, 7 hours for setting, and 6 hours for
polishing. Bracelets take 5 hours for stone cutting, 7 hours for setting and 3 hours for
polishing. The Jeweller has 40 hours for stone cutting, 70 hours for setting and 48 hours for
polishing. How many necklaces and bracelets should be produced so as to maximize profits 150
A producer has to decide how many units of good X1 and good X2 to produce in order to
maximize revenue from their sale. One unit of good X1 requires 6 units of labor, 3 hours of
machine work and 2 units of raw materials, while one unit of X2 requires 5 units of labor, one
unit one hour of machine work and 4 units of raw materials. The available resources are: 90
units of labor, 36 hours of machine time and 60 units of raw materials. A unit of X1 sells at $5
while a unit of X2 sells at $3. Formulate a linear programming model and determine the
combination of the two products that maximizes the producer’s revenue.

A firm is engaged in producing two products. A and B. Each unit of product A requires 2 kg of
raw material and 4 labor hours for processing, whereas each unit of B requires 3 kg of raw
materials and 3 labor hours for the same type. Every week, the firm has an availability of 60 kg
of raw material and 96 labor hours. One unit of product A sold yields Rs.40 and one unit of
product B sold gives Rs.35 as profit.
Formulate this as an Linear Programming Problem to determine as to how many units of each
of the products should be produced per week so that the firm can earn maximum profit.
MINIMIZATION PROBLEMS
Question1. MINIMIZATION Problem (Generating a linear programming problem &
At the Uganda Wildlife Education center, the game warden wants all his animals to get a
minimum of 36 milligrams (mg) of iodine, 84mgs of iron, and 16mg of zinc each day. One feed
of y1 provides 3mgs of iodine, 6mg of iron and 1mg of zinc. A second feed y2 provides 2 mg of
iodine, 6mgs of iron and 4mgs of zinc. The first type of feed costs $20, the second costs $15.
Using the graphical method, advise the game warden on the least cost combination of feeds
that guarantees daily requirements.

The agricultural research institute suggested the farmer to spread out at least 4800 kg of
special phosphate fertilizer and not less than 7200 kg of a special nitrogen fertilizer to raise the
productivity of crops in his fields. There are two sources for obtaining these – mixtures A and
mixtures B. Both of these are available in bags weighing 100kg each and they cost Rs.40 and
Rs.24 respectively. Mixture A contains phosphate and nitrogen equivalent of 20kg and 80 kg
respectively, while mixture B contains these ingredients equivalent of 50 kg each. Write this as
an LPP and determine how many bags of each type the farmer should buy in order to obtain
the required fertilizer at minimum cost.

A person requires 10, 12, and 12 units chemicals A, B and C respectively for his garden. A
liquid product contains 5, 2 and 1 units of A,B and C respectively per jar. A dry product
contains 1,2 and 4 units of A,B and C per carton. If the liquid product sells for Rs.3 per jar and
151
the dry product sells for Rs.2 per carton, how many of each should be purchased, in order to
minimize the cost and meet the requirements?
H THE SIMPLEX ALGORITHM METHOD AND THE DUAL

An algorithm is a systematic procedure or set of rules for finding a solution to a problem. The
simplex algorithm is a computational method that (1) seeks out basic feasible solutions for a
system of linear equations and (2) tests the solutions for optimality. This method solves LPP
which have more than 2 decision variables.
H1 STEPS WHEN USING THE SIMPLEX ALGORITHM METHOD

1. Set up the problem. That is, write the objective function and the constraints.
2. Convert the inequalities into equations. This is done by adding one slack variable for
each inequality.
3. Construct the initial simplex tableau which will work as a framework for the algorithm.
The initial simplex tableau represents the first basic solution when x1 and x2 equal to 0. It is
composed of the coefficient matrix of the constraint equations and the column vector of
constants set above a row of indicators which are the negatives of the coefficients of the
decision variables in the objective function and a zero coefficient for each slack variable.
The constant column entry in the last row is also 0.
4. The most negative entry in the bottom row identifies a pivot column.
5. Calculate the quotients. The smallest quotient identifies a pivot row. The
element in the intersection of the column identified in step 4 and the row
identified in this step is identified as the pivot element. The quotients are computed
by dividing the far right column by the identified column in step 4. A quotient that is a zero,
or a negative number, or that has a zero in the denominator, is ignored.
6. Perform pivoting to make all other entries in this column zero apart from the
pivot element
7. When there are no more negative entries in the bottom row, we are finished;
otherwise, we start again from step 4.
8. Read off your answers. Get the variables using the columns with 1 and 0s. All other
variables are zero. The maximum value you are looking for appears in the bottom right
hand corner.
Example10. Maximization Problem using the SIMPLEX ALGORITHM

Niki holds two part-time jobs, Job I and Job II. She never wants to work more than a total of
12 hours a week. She has determined that for every hour she works at Job I, she needs 2
hours of preparation time, and for every hour she works at Job II, she needs one hour of
preparation time, and she cannot spend more than 16 hours for preparation. If she makes
Shs.40 an hour at Job I, and Shs.30 an hour at Job II, how many hours should she work per
week at each job to maximize her income?
Solution: In solving this problem, we will follow the algorithm listed above.
1. Set up the problem. That is, write the objective function and the constraints.
Since the simplex method is used for problems that consist of many variables, it is not practical
to use the variables x, y, z etc. We use the symbols x1, x2, x3, and so on.
Let x1 = The number of hours per week Niki will work at Job I. 152
and x2 = The number of hours per week Niki will work at Job II.
It is customary to choose the variable that is to be maximized as Z.
The problem is formulated the same way as we did in the last chapter.
Maximize Z = 40x1 + 30x2
Subject to: x1 + x2 ≤ 12
2x1 + x2 ≤ 16
x1 ≥ 0; x2 ≥ 0
2. Convert the inequalities into equations. This is done by adding one slack variable
for each inequality. For example to convert the inequality x1 + x2 ≤ 12 into an equation, we
add a non-negative variable y1, and we get; x1 + x2 + s1 = 12. Here the variable s1 picks
up the slack, and it represents the amount by which x1 + x2 falls short of 12. In this
problem, if Niki works fewer that 12 hours, say 10, then s1 is 2. Later when we read off the
final solution from the simplex table, the values of the slack variables will identify the unused
amounts. We can even rewrite the objective function Z = 40x1 + 30x2 AS – 40x1 – 30x2 + Z =
0.
After adding the slack variables, our problem reads
Objective function: – 40x1 – 30x2 + Z = 0

Subject to constraints: x1 + x2 + s1 = 12
2x1 + x2 + s2 = 16
x1 ≥ 0; x2 ≥ 0
3. Construct the initial simplex tableau. Write the objective function as the bottom
row.
Now that the inequalities are converted into equations, we can represent the problem into an
augmented matrix called the initial simplex tableau as follows;
x1 x2 s1 s2 Z C
1 1 1 0 0 12
2 1 0 1 0 16
–40 –30 0 0 1 0
Here the vertical line separates the left hand side of the equations from the right side. The
horizontal line separates the constraints from the objective function. The right side of the
equation is represented by the column C (which are the solution quantities). The reader needs
to observe that the last four columns of this matrix look like the final matrix for the solution of
a system of equations. If we arbitrarily choose x1 = 0 and x2 = 0, we get;
S1 S2 Z C
1 0 0 12
0 1 0 16
0 0 1 0
Which reads; s1 = 12, s2 = 16 and Z = 0
The solution obtained by arbitrarily assigning values to some variables and then solving for the
remaining variables is called the basic solution associated with the tableau. So the above
solution is the basic solution associated with the initial simplex tableau. We can label the basic
solution variable in the right of the last column as shown in the table below.
x1 x2 s1 s2 Z
1 1 1 0 0 12 s1
2 1
–40 –30 0
0 1
0
0
1
16
0
s2
Z
153
The most negative entry in the bottom row identifies a column.
The most negative entry in the bottom row is –40, therefore the column 1 is identified.
x1 x2 s1 s2 Z
1 1 1 0 0 12 s1
2 1 0 1 0 16 s2
–40 –30 0 0 1 0 Z
Question. Why do we choose the most negative entry in the bottom row?
Answer. The most negative entry in the bottom row represents the largest coefficient in the
objective function – the coefficient whose entry will increase the value of the objective function
the quickest.
The simplex method begins at a corner point where all the main variables, the variables that
have symbols such as x1, x2, x3 etc., are zero. It then moves from a corner point to the
adjacent corner point always increasing the value of the objective function. In the case of the
objective function: Z = 40x1+ 30x2, it will make more sense to increase the value of x1 rather
than x2. The variable x1 represents the number of hours per week Niki works at Job I. Since
Job I pays: Shs. 40 per hour as opposed to Job II which pays only Shs. 30, the variable x1 will
increase the objective function by Shs. 40 for a unit of increase in the variable x1.
5. Calculate the quotients. The smallest quotient identifies a row. The element
in the intersection of the column identified in step 4 and the row identified in this step is
identified as the pivot element.
As mentioned in the algorithm, in order to calculate the quotient, we divide the entries in the
far right column by the entries in column 1, excluding the entry in the bottom row.
x1 x2 s1 s2 Z
1 1 1 0 0 12 s1 12 ÷ 1 = 12
2 1 0 1 0 16 s2
–40 –30 0 0 1 0 Z
The smallest: of the two quotients, 12 and 8, is 8. Therefore row 2 is identified. The
intersection of column 1 and row 2 is the entry 2, which has been highlighted. This is our
pivot element.
Question: Why do we find quotients, and why does the smallest quotient identify a row?
Answer When we choose the most negative entry in the bottom row, we are trying to
increase the value of the objective function by bringing in the variable x1. But we cannot
choose any value for x1. Can we let: x1 = 100? Definitely not! That is because Niki never
wants to work for more than 12 hours at both jobs combined. In other words: x 1 + x2 ≤ 12.
Now can we let x1 = 12? Again, the answer is no because the preparation time for Job I is two
times the time spent on the job. Since Niki never wants to spend more than 16 hours for
154
preparation, the maximum time she can work is 16 ÷ 2 = 8. Now you see the purpose of
computing the quotients.
Question Why do we identify the pivot element?

Answer As we have mentioned earlier, the simplex method begins with a corner point
and then moves to the next corner point always improving the value of the objective function.
The value of the objective function is improved by changing the number of units of the
variables. We may add the number of units of one variable, while throwing away the units of
another. Pivoting allows us to do just that. The variable whose units are being added is called
the entering variable, and the variable whose units are being replaced is called the
departing variable. The entering variable in the above table is x1, and it was identified by
the most negative entry in the bottom row. The departing variable y2 was identified by the
lowest of all quotients.
6. Perform pivoting to make all other entries in this column zero.

Pivoting is a process of obtaining a 1 in the location of the pivot element, and then making all
other entries zeros in that column. So now our job is to make our pivot element a 1 by
dividing the entire second row by 2. The result follows.
x1 x2 s1 s2 Z
1 1 1 0 0 12
1 1/2 0 1/2 0 8
–40 –30 0 0 1 0
To obtain a zero in the entry first above the pivot element, we multiply the second row by –1
and add it to row 1. We get;
x1 x2 s1 s2 Z
0 1/2 1 –1/2 0 4
1 1/2 0 1/2 0 8
–40 –30 0 0 1 0
To obtain a zero in the element below the pivot, we multiply the second row by 40 and add it
to the last row.
x1 x2 s1 s2 Z
0 1/2 1 –1/2 0 4 s1
1 1/2 0 1/2 0 8 x1
0 –10 0 20 1 320 Z
We now determine the basic solution associated with this tableau. By arbitrarily choosing x2
= 0 and s2 = 0, we obtain: x1 = 8, s1 = 4, and z = 320. If we write the augmented matrix,
whose left side is a matrix with columns that have one 1 and all other entries zeros, we get the
following matrix stating the same thing.
155
x1 s2 Z C
0 1 0 4
1 0 0 8
0 0 1 320
We can restate the solution associated with this matrix as x1 = 8, x2 = 0, s1 = 4, s2 = 0 and z

= 320. At this stage of the game, it reads that if Niki works 8 hours at Job I, and no hours at
Job II, her profit Z will be $320. Recall from Example 1 in section 3.1 that (8, 0) was one of
our corner points. Here s1 = 4 and s2 = 0 mean that she will be left with 4 hours of working
time and no preparation time.
7. When there are no more negative entries in the bottom row, we are finished;
otherwise, we start again from step 4.
Since there is still a negative entry, –10, in the bottom row, we need to begin, again, from
step 4. This time we will not repeat the details of every step, instead, we will identify the
column and row that give us the pivot element, and highlight the pivot element. The result is
as follows.
x1 x2 s1 s2 Z
0 1/2 1 –1/2 0 4 y1
1 1/2 0 1/2 0 8 x1 8 ÷ 1/2 =

16
0 –10 0 20 1 320 Z
We make the pivot element 1 by multiplying row 1 by 2, and we get
x1 x2 s1 s2 Z
0 1 2 –1 0 8
1 1/2 0 1/2 0 8
0 –10 0 20 1 320
Now to make all other entries as zeros in this column, we first multiply row 1 by –1/2 and add
it to row 2, and then multiply row 1 by 10 and add it to the bottom row.
x1 x2 s1 s2 Z
0 1 2 –1 0 8 x2
1 0 –1 1 0 4 x1
0 0 20 10 1 400
Z
We no longer have negative entries in the bottom row, therefore we are finished.
Question. Why are we finished when there are no negative entries in the bottom row?
Answer. The answer lies in the bottom row. The bottom row corresponds to the following
equation. 156
0x1 + 0x2+ 20s1 + 10s2 + Z = 400 or
Z = 400 – 20s1 – 10s2
Since all variables are non-negative, the highest value Z can ever achieve is 400, and that will
happen only when s1 and s2 are zero.
8. Read off your answers.

We now read off our answers, that is, we determine the basic solution associated with the final
simplex tableau. Again, we look at the columns that have a 1 and all other entries zeros.
Since the columns labeled s1 and s2 are not such columns, we arbitrarily choose s1 = 0, and
s2 = 0, and we get;
x1 x2 Z | C

0 1 0 | 8


1 0 0 | 4 

0 0 1 | 400 
The matrix reads x1 = 4 , x2= 8 and z = 400. The final solution says that if Niki works 4
hours at Job I and 8 hours at Job II, she will maximize her income to Shs. 400. Since both
slack variables are zero, it means that she would have used up all the working time, as well as
the preparation time, and none will be left.
H2 MARGINAL VALUE AND SHADOW PRICING

The value of the indicator under each slack variable in the final tableau expresses the
marginal value or shadow price of the input associated with the variable. That’s to say,
how much the value of the objective function would change as a result of a 1-unit
change in the availability of the input.
Thus; the company’s profits would increase by (4/3) Shs for a unit increase in the constant of
constraint 2. Likewise, profits would also increase by (1/2) Shs for a unit increase in the
constant of constraint3. And by 0 for a 1-unit increase in the constant of constraint 1(OR-The
addition of still another unit would add nothing to the profit function-marginal value is 0).
Example1 MAXIMIZATION Problem (Using the SIMPLEX METHOD)

Maximize Z = 100A + 300B + 200C
Subject to: A + B + C ≤ 100; 40A + 20B + 30C ≤ 3, 200; A + 2B +C ≤ 160. And the non -
negative constraints. Solve the linear programming problem using the simplex method and
interpret the results
H3 MINIMIZATION PROBLEMS
If the simplex algorithm is used for minimization problems, the negative values generated by
the surplus variables present a special problem. The first basic solution will consist of totally
negative numbers and so will not be feasible. Consequently, still other variables, called artificial
variables must be introduced to generate an initial feasible solution. It is frequently easier to
solve minimization problems by using the DUAL.
Every MINIMIZATION problem in LPP has a corresponding MAXIMIZATION problem and

every maximization problem has a corresponding minimization problem. Any original problem
is called the PRIMAL and the corresponding problem is called the DUAL. Therefore, when
given any minimization problem (PRIMAL), first convert it into a corresponding maximization
problem (DUAL). 157
EXAMPLE11. CONVERSION OF PRIMAL TO DUAL
Convert the following minimization problem into its dual.
Minimize C= 12x1 + 16x2
Subject to: x1 + 2x2 ≥ 40; x1 + x2 ≥ 30; x1 ≥ 0; x2 ≥ 0
Solution
To achieve our goal, we first express our problem as the following matrix.
1 2 40
1 1 30
12 16 0
Observe that this table looks like an initial simplex tableau without the slack variables. Next,
we write a matrix whose columns are the rows of this matrix, and the rows are the columns.
Such a matrix is called a transpose of the original matrix.
We get;
1 1 12
2 1 16
40 30 0
The following maximization problem associated with the above matrix is called its dual.
Maximize Z = 40y1 + 30y2
Subject to: y1 + y2 ≤ 12;
2y1 + y2 ≤ 16;
y1 ≥ 0; y2 ≥ 0
We have chosen the variables as y's, instead of x's, to distinguish the two problems.
EXAMPLE12. CONVERSION OF PRIMAL TO DUAL

Solve graphically both the minimization problem and its dual, the maximization problem. Check
whether you get the same answer.
Solution: Our minimization problem is as follows.

Minimize Z = 12x1 + 16x2
We now graph the inequalities.
158
We have plotted the graph, shaded the feasibility region, and labeled the corner points. The
corner point (20, 10) gives the lowest value for the objective function and that value is 400.
Now it’s DUAL is;

Subject to: y1 + y2 ≤ 12; 2y1 + y2 ≤ 16; y1 ≥ 0; y2 ≥ 0
We graph the inequalities;
Again, we have plotted the graph, shaded the feasibility region, and labeled the corner points.
The corner point (4, 8) gives the highest value for the objective function, with a value of 400.
We observe that the minimum value of the minimization problem is the same as the maximum
value of the maximization problem; they are both 400. This is not a coincident. We state the
duality principle. The Duality Principle: The objective function of the minimization problem
reaches its minimum if and only if the objective function of its dual reaches its maximum. And
when they do, they are equal. Our next goal is to extract the solution for our minimization
problem from the corresponding dual. To do this, we solve the dual by the Simplex method.
Example13. Use the simplex method to solve the minimization problem below by solving its
dual. Minimize Z = 12x1 + 16x2
Solution: The dual is as follows:

Subject to: y1 + y2 ≤ 12; 2y1 + y2 ≤ 16; y1 ≥ 0; y2 ≥ 0
Then solve this using the simplex method which we have seen already.
159
Question1
a. Write short notes on the following:
I. Decision variables
II. Constraints
III. Objective function
IV. A dual and primal
V. A pivot element
VI. Binding and non binding constraints
c. Explain very well the assumptions made when solving linear programming problems.
d. What factors should be put in consideration when constructing a linear programming
problem?
e. Explain very well business areas in which linear programming has found enormous use
Question2. CONVERSION OF PRIMAL TO DUAL

Given the primal problem below, determine the dual and solve the original primal function
using the graphical method.
Max Z = 20x1+40x2
Subject to; 9 x1 + 18x2 ≤ 9000; 10x1 + 10x2 ≤ 6000; 21x1 + 10.5x2 ≤ 10500
Question 3. CONVERSION OF PRIMAL TO DUAL

For the given problem, formulate the dual; solve the dual using the simplex algorithm. Use the
final dual tableau to determine the optimal values of the primal in the objective function and
decision variables.
Minimize c  14 y1  40 y 2  18 y3
Subject to:
2 y1  5 y 2  y3  50 ; y1  5 y 2  3 y3  30 ; y1, y 2, y3  0 And the non - negative constraints.
Question 4. CONVERSION OF PRIMAL TO DUAL

Use the procedure in 4 above to also attempt this one
Minimize c  240 y1  120 y 2
Subject to:
4 y1  8 y 2  56 ; 2 y1  2 y 2  24 ; 3 y1  y 2  18
y1, y 2  0 And the non - negative constraints.

June 2014 June 2010
160
Chapter-16
Correlation & regression analysis
A INTRODUCTION
CORRELATION ANALYSIS. When two variables are related in such a way that the change in
one variable is accompanied by a change in another then the two variables are said to be
correlated. Thus; Correlation Analysis is a statistical measure of the strength and
direction of the relationship between two or more variables. The strength of the relationship
is given by the correlation coefficient that ranges between [(– 1 and +1 that’s to say (- 1 ≤
rxy ≤ +1)].
B SCATTER DIAGRAM.
This is another visual display of data. It shows the association between two variables acting
continuously on the same item. The scatter diagram illustrates the strength of the correlation
between the variables through the slope of a line.
A. Scatter diagrams make the relationship between two continuous variables stand out
visually on the page in a way that the raw data cannot.
B. Scatter diagrams may be used in examining a cause-and-effect relationship between
variable data (continuous measurement data).
C. They can also show relationships between two effects to see if they might stem from
a common cause or serve as surrogates for each other. They can also be used to
examine the relationship between two causes
Example of a scatter diagram

Assuming I want to get a graphical representation of the relationship between number of QT
revision sessions attended and actual QT marks obtained. I take time and compile this data.
Student A B C D E F G
N0 of Sessions 3 5 6 7 3 2 1
QT Marks 44% 68% 81% 88% 67% 44% 29%
Scatter diagram for QT marks against number of revision sessions would appear as below.
100
90
80
70
QT Marks in %
60
50
40
30
20
10
0
0 1 2 3 4 5 6 7 8
Number of QT Revision Sessions attended
161
Note: Since most of the scatter points are lying on the estimated regression line, we can use
this scatter diagram to estimate say; QT marks when a student attends 20 revision sessions
Scatter diagrams show relationships, but do not prove that one variable causes the other!
C THE CORRELATION COEFFICIENT ( rxy)

If all points on the scatter diagram lie very close to a straight line then there is a strong
relationship between x and y than when they are scattered away from the line. To measure
the strength and direction of the correlation or relationship between x and y, we utilize a
coefficient of correlation popularly called the Pearson or product-moment correlation
coefficient (PMCC).
The (PMCC) is either;

 xy   n X Y  Where (- 1 ≤ rxy ≤ +1)
rxy     1 n = No. of observations
 x  nX    y  nY  
  
2 2 2 2
_

x = Mean of x values
Or _
n xy    x  y
 2
y = Mean of y values
rxy 
n x    x n y    y  
2 2 2 2
y = The dependant variables
Or x = The independent variables
rxy 
 x  x y  y       3
 x  x 2   y  y 2 
   
The strength and the direction of the relationship between two variables on the basis of
correlation coefficient are summarized in the table below.
Degree of relationship. Positive (+ve) Negative(-ve)

Perfect +1 -1
Very high 0.75 – 1 -0.75 – -1
High 0.50 – 0.75 -0.50 – -0.75
Low 0.25 – 0.50 -0.25 – -0.50
Very low 0.00 – 0.25 0.00 – 0.25
Absent/no relationship 0.00 0.00
Example1. Using any of the formulae above, compute the Pearson or product-moment
correlation coefficient to establish the strength & direction of the relationship between
advertising expenditure and company sales: Interpret your results. Also, test whether the
computed correlation is significant at 5%:
Month Jan Feb Mar April May June July August

Adverting expenditure 15 30 31 40 17 39 24 25
($000)
Sales 28 37 42 45 24 50 32 35
The Pearson’s product moment correlation coefficient can be calculated as follows. 162
-Let X and Y represent the advertising costs and sales made respectively.
-We can use the table.
Month X Y XY X2 Y2
Jan 15 28 420 225 784
Feb 30 37 1110 900 1369
March 31 42 1302 961 1764
April 40 45 1800 1600 2025
May 17 25 425 289 625
June 39 50 1950 1521 2500
July 24 32 768 576 1024
August 25 35 875 625 1225
Total 221 294 8650 6697 11316
From the formulae,

n xy    x y
rxy   0.96
n x   x n y   y 
2 2 2 2
The above result shows that there is a very strong positive relationship between
the advertising expenditure and the sales revenues got for the last 8 months.
This is how the scatter diagram appears to display the relationship between advertising
expenses and sales revenues.
D THE RANK CORRELATION COEFFICIENT

Previously, we have expressed both x and y in numerical form. However, there are situations
where we are unable to express the variables in numerical form in order to carry out
correlation analysis but we have enough information to rank these variables as 1st, 2nd, 3rd, etc.
Here, we employ the Spearman’s rank correlation coefficient method to compute the
correlation coefficient. Examples of variables that can’t be expressed in numerical form include
efficiency, honesty, and intelligence of employees. Etc.
The correlation coefficient computed is a measure of correlation between two sets of rankings.

6  d 2 
rs  1   2


 
 n n  1  WHERE  1  rs  1
Where, d = difference between ranks, n = number of pairs of observations, When rs = 1, it

indicates a perfect negative relationship between variables. And rs = +1, it indicates a perfect
positive relationship between the variables. Note: This rank correlation analysis is used in
qualitative analysis where data is presented in such form as profession, education,
intelligence quotient, etc.
Example on Rank Correlation Coefficient

Two interviewers in KPMG separately gave the following points to 7 candidates who had
applied for the post of chief accountant in EABL. a) Using the spearman’s rank correlation
coefficient, do you think both interviewers have similar opinions on the candidates? b) Is the
established correlation greater than 0 at 2.5% significance level?
163
Interviewer-1 Interviewer-2 (Y)
(X)
8 56
5 44
11 79
13 72
10 70
5 54
18 94
15 85
2 33
8 65
Solution
X Y Rank of X Rank of Y d = (rank X – rank Y) d2 = (rank X – rank Y)2
8 56 4.5 4 0.5 0.25
5 44 2.5 2 0.5 0.25
11 79 7 8 -
1.0 1.00
13 72 8 7 1.0 1.00
10 70 6 6 0.0 0.00
5 54 2.5 3 -
0.5 0.25
18 94 10 10 0.0 0.00
15 85 9 9 0.0 0.00
2 33 1 1 0.0 0.00
8 65 4.5 5 -
0.5 0.25
∑d2 = 3

6 d 2 
rs  1   2
 6 x3
  1  0.98
 
 n n  1  10(10 2  1)
Solution (Interpretation of findings is very important)
E THE COEFFICIENT OF DETERMINATION (r2)

This refers to the amount of variation in the dependent variable that is brought about or
accounted for by the changes in the independent variable. Example; rs = 0.98 , the coefficient
of determination would be r2 = (0.98)2 x 100% =0.9604x100=96%. This implies that,
the variation in marks of QT is 96% brought about by the number of revision sessions
attended. The remaining 4% is due to other factors such as intelligent quotient etc. Note: Did
you know that it’s possible to obtain a correlation coefficient yet it doesn’t make sense in
reality? E.g. if rxy = 0.91 between QT Marks and Ages of 30 students. This is a typical example
of spurious correlation.
I REGRESSION ANALYSIS
The correlation coefficient computed earlier tells us the extent to which the two variables x & y
are linearly related BUT it doesn’t tell us how to find a particular straight line that represents
the relationship! The problem of establishing which particular line best fits the points of the
scatter diagram comes under the heading Linear Regression Analysis. Thus, Regression is
the statistical method, which measures the average relationship between two or more 164
variables and then help us to estimate or predict the unknown value of one variable from the
known value(s) of related variable(s).E.g., the marketing manager can find the average
relationship between money spent on adverts and sales revenues made and then be able
to estimate the level of sales at a given advertising budget. The actual estimation is done by
using a rule, which specifies the estimate of one variable when given the value of another
variable.
It’s important that: -

-The manager has to identify which variable depends on the other so as to be able to
distinguish dependent and independent variable.
-The independent variable is also known as the regressor or predictor or explainator variable
while the dependent variable is known as the regressed or explained variable.
J WHY STUDY REGRESSION AND CORRELATION? (IMPORTANCE)

1) To learn how business decisions are based on the relationships between 2 or more
variables.
2) To use scatter diagrams to visualize the relationship between two or more variables.
3) To use regression analysis to establish the relationship between two or more variables.
4) To use “least square” estimating equations to predict future values of dependent
variables.
5) To understand the coefficient of variation as a measure of strength of the relationship
between two variables.
6) To learn the limitations of regression and correlation analysis and interpret its use in
business.
K LEAST SQUARES REGRESSION LINE

Least squares analysis is a process whereby a regression model is developed by producing
the minimum sum of squared error values.
This is the y on x linear regression line and is represented as: y  a  bx . The values of a and
b are obtained as follows:-
n xy    x  y   xy  n X Y
b b
 
n  x 2   x 
2
or  x 2  nx  ;
2
a  Y  bX
The straight line obtained from this method of least squares is known as the line of best fit.
Example on fitting a line of best fit
A farmer wishes to know the linear relationship between the amount of fertilizers applied and
the harvest he gets assuming that he keeps other factors constant.
The table below gives the amount of fertilizers and the output from each garden.
Garden Amount of fertilizers Output (Y)

(x)
A 10 20
B 12 25
C 14 22
D 16 26
E 18 26
We can find the relationship between the harvests and the amount of fertilizers applied as
follows:- 165
-We assume that the harvests depend on the amount of fertilizers applied and therefore let X
be the amount of fertilizers applied and Y represent the output respectively.
-The required relationship can be summarized in the form of the general equation.
Y = a + bx
-The values of the constants are obtained from the method of least squares formulae above
with the help of the table below.
Garden Amount of Output (Y) XY X2

fertilizers (x)
A 10 20 200 100
B 12 25 300 144
C 14 22 308 196
D 16 26 416 256
E 18 26 468 324
TOTAL 70 119 1,692 1,020
Substituting in the formulae,

n xy    x  y 
b  0.65
 
n  x 2   x 
2
a  Y  b X  23.8  (0.65 x14)  14.7

So the least squares regression line is: Y  14.7  0.65 X
Note:-
1-the vertical intercept of 14.7 units shows the level of output when there are no fertilizers
applied at all:
2-the slope (regression coefficient) of 0.65 shows that output increases by 0.65 units for every
additional unit of fertilizer applied to the garden:
3-the above regression equation can also be used to predict the output when say 25 units of
fertilizers are applied to the garden:
ie,. Y=14.7 + 0.65x
so, Y=14.7 + (0.65 x 25)
Y=14.7 + 16.25
Then, Y= 30.95 units.
Limitation of correlation & Regression Analysis

The values used to calculate both the correlation coefficients and regression lines are affected
by the number of pairs of observations used. Larger samples are more reliable than smaller
samples.
L DIFFERENCES BETWEEN CORRELATION AND REGRESSION ANALYSIS

A. In regression analysis, only one variable is taken to be random (usually Y) while the
other is taken to be fixed. In correlation analysis both variables are random.
B. In regression analysis the aim is to establish an equation between the fixed and
random variables so as to predict a random variable where as in correlation analysis we
compute a single number to show the strength of a linear relationship between two
variables.
166
C. In regression analysis, our interest is to predict while in correlation analysis, the
interest is to measure the strength of a linear relationship.
D. In regression analysis the fixed (independent variable) may cause the random
(dependent) variable while in correlation analysis there is no cause effect relationship.

Question
a) Differentiate between the following terms as used in quantitative methods
I) Correlation and regression analysis
II) A scatter diagram and a linear regression line
III) Correlation coefficient and coefficient of determination
IV) Spearman and Pearson correlation coefficient
V) Dependent and independent variable
VI) Extrapolation and interpolation
b) Explain the relevance of correlation analysis in modern business management

c) Explain the difference between correlation and regression analysis
d) Given the following variables X and Y
X 10 20 30 20 40
Y 98 102 115 212 200
1) Find the correlation between X and Y

2) Interpret the results you have obtained
3) Find the least squares regression line
4) Interpret the value of a and b you have obtained
5) Forecast the value of Y when X is 55
6) Forecast the value of X when Y is 188
Question
a) When do we use a spear man correlation coefficient
b) Five candidates who applied for a post of chief cook at Sheraton Kampala Hotel were
interviewed by two panelists and the following are the points awarded by each panelist.
Candidate A B C D E
1st Panelist 5.7 5.8 5.9 5.6 5.6
2 Panelist
nd
5.6 5.7 6.0 5.5 5.8
Rank the marks-compute and interpret the Spearman’s coefficient of rank correlation
Question
a) Correlation is not necessarily causation. Discuss
b) List four (4) differences between correlation and regression
c) The following data gives actual sales of Hima cement factory in each of the 8 regions of
Uganda together with the forecast of sales by two different methods.
Region A B C D E F G H
Actual Sales 15 19 30 12 58 10 23 17 167
Forecast -1 13 25 23 26 48 15 28 10
Forecast -2 16 19 26 14 65 19 27 22
Required
Calculate the spearman’s rank correlation coefficient between;
1) Actual sales and forecast-1
2) Actual sales and forecast-2
3) Which forecasting method would you recommend next year and why?
d) Extract a simple linear regression line (Y = a + bX) between actual sales (Y) and Sales
forecast-1 (X)
Question
The following data relates to costs derived by a cost accountant (Anita) at Java Café.
Week 1 2 3 4 5 6 7 8 9 10
Output (X) ‘000 20 2 4 23 18 14 10 8 13 8
Total costs (Y) 60 25 26 66 49 48 35 18 40 33
a) Find the least squares regression line for the costs on output
b) What are the fixed costs of the factory?
c) In a given week, Café Javas wants to produce 25,000 units, what would be the total
cost to be incurred?
d) Compute the Pearson correlation coefficient and interpret your results
Question
Uchumi and Tusky’s have gone fwhaaa-Bankrupt! You have been hired as a takeover
consultant by the regional supermarket Giant Nakumatti. As part of your preliminary research
work, you have discovered and recorded the following data relating to turnover, number of
stores and regions within the east African region
UCHUMI
Region A B C D E F G H I
Stores (X) 4 6 7 2 9 11 15 5 8
Turnover (Y)’ Billions 45 61 89 12 48 76 67 76 80
TUSKY’S
Region A B C D E F G H I
Stores (X) 7 8 4 7 8 14 18 14 15
Turnover (Y)’ Billions 65 67 44 80 67 60 80 90 95
a) Find the least squares regression lines for each supermarket store
b) Assume the investors want to open up 50 stores in each region, predict the turnover for
each supermarket and comment on the likely accuracy of the answer
c) For each supermarket, what is the amount of turnover that is not affected by the
number of stores?
d) Predict the number of stores an investor needs to open up for each supermarket if he
wants to realize a total turnover of 200 billion in each case.
e) By using the Pearson’s correlation coefficient, which supermarket would you invest in
and why? 168
Question
a) List four (4) differences between correlation and regression
b) The Electronics World Chain of stores selling equipment wanted to study the
relationship between store sales volume y (measured in millions of shillings) and the
number of households x, in the store’s area in one month. The data for 15 stores is as
follows:-
Store 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
N0 of 161 99 135 120 164 221 179 214 204 101 231 206 248 107 205
Households
(X)
Sales 15.7 9.3 13.7 12.4 15.4 24.2 20.2 23 20.7 13.5 22.5 19.5 24.3 11.5 19.8
Volume
(Y)
I. Calculate the Spearman’s rank correlation coefficient.

II. Comment on the results.
c) A fitted regression of sales (in million shillings) onto advertising is given by sales =
268+7.37Ads equation.
I. What sales will each extra Shs 1 million of advertising generate?
II. What sales would the firm average with zero advertising?
III. If the firm spends Shs 10 million on advertising, what would be the expected sales?
IV. What advertising cost would generate average sales of Shs 562.8 million?
Question
a) Correlation doesn’t necessarily mean causation. Discuss this statement and highlight the
importance of correlation and regression analysis in modern business management.
b) A lecturer wanted to wanted to compare the two methods of teaching (A) and (B). She
selected a random sample of 24 students, which she grouped into 12pairs so that the
students in pair have approximately equal scores in an intelligent test. In each pair, one
student was taught by method A and the other by method B and then examined at the
end of the semester. The marks obtained were recorded as follows.
Pair 1 2 3 4 5 6 7 8 9 10 11 12
A 58 68 48 38 70 48 64 70 50 66 32 54
B 84 80 42 62 56 64 48 50 42 32 52 60
Find the rank correlation coefficient and comment on the result obtained
c) To investigate the relationship between annual food and transport expenditure and
annual household disposable income, a sample of 100 households was taken and the
following sums were calculated from the sample data: y  752 , 
y 2  6,220 , 
 x  2,486 ,  x 2
 63,622 , and  xy  19,600 . Where y represents annual
expenditure on food and transport (in ‘000 UG. Shillings) and x represents annual
household disposable income (in ‘000 UG. Shillings).
Taking annual expenditure on food and transport as the dependent variable and
household disposable income as the independent variable, calculate and interpret the
equation of the least-squares regression line
169
Chapter-17
Forecasting: Time series analysis
A INTRODUCTION
A TIME SERIES is a collection of figures or values about some event belonging to different
time periods. OR, a series of recorded observations arranged in a chronological order. These
time periods may be in years, months, days or even hours.
Examples of time series

I. The annual production of coffee in Uganda over the last 28 years.
II. The monthly sales of quick save supermarket for the last 6 months.
III. The weekly deposits received by Crane Bank for the last three months.
IV. The daily output of a factory for the last month.
Relevance of time series

1. It helps in the analysis of past behavior of the phenomenon/ happening under
consideration
2. Helps decision makers predict or forecast the behavior of the phenomenon under study
in future, which is very essential for business planning
3. It enables decision makers to make comparative studies of the values in different
phenomenon under study over a given period of time.
B FORECASTING TECHNIQUES IN TIME SERIES

Since time series are predominantly used in making forecasts. We need to identify the various
methods used in making such forecasts. These are: 1) quantitative and 2) qualitative
techniques.
A) Quantitative techniques/ methods

Quantitative techniques are such techniques that are used to analyze past data of the item to
be forecast. The techniques rely solely on various assumptions that enable the use of the data
to forecast future performances. Such methods include; 1) simple average method, 2)
weighted average method, 3) moving average, 4) least squares, and 5) exponential
smoothing.
B) Qualitative techniques/ methods

These techniques on the other hand are forecasting techniques that are used when the data is
very scarce and no assumptions are reliably made. These methods employ human judgment
and experience to turn qualitative estimates. Such methods include; 1) the Delphi method of
forecasting, 2) the market research method, 3) historical analog.
A. QUANTITIVE FORECASTING METHODS

1. Simple average method
This refers to the method of averaging periodic data. It gives the same weight to each periodic
data. However, it is inappropriate in inflationary periods because in such situations, recent data
should be given more weight to adjust recent values
2. Weighted average method

This is an improvement on the simple average method. Here, periodic data is given different
weights.
170
3. Method of moving averages
This is another form of averaging and forms the link between several periodic data which
makes it to adjust to recent values especially if the data periods involved in each average are
many. Unfortunately, this method does not use all the available data and as such some
information is unrepresented.
4. Method of least squares

This method obtains the average by obtaining the center of data plotted on a graph. Its
average becomes more accurate when the data is not scattered much otherwise it gives a
misleading average. However, it renders itself further to mathematical treatment because of its
concept of minimizing deviations
5. Exponential smoothing
This is the best and desirable method. It has in built mechanisms for assigning weights in a
desirable way. It involves assigning small weights to older data and vice versa. The problem is
that the initial weight assigned is done subjectively.
B. QUALITATIVE FORECASTING METHODS
4. The Delphi method

This technique is mainly used for long term forecasting, designed to obtain expert consensus
for a particular forecast, without the problem of submitting to pressure to conform to a
majority view. This method involves the use of a panel of experts who independently answer a
sequence of questionnaires in which the responses to one questionnaire are used to produce
the next questionnaire. Thus, any information available to some experts and not others is
passed on to all, so that their subsequent judgments are refined as more information and
experience becomes available.
5. Market research approach

This method uses opinion surveys, analyses of market data, questionnaires and other
investigations to gauge the reaction of the market to a particular product, design, price etc.
Market research is often very accurate for the relatively short term, but longer term forecasts
based purely on surveys are likely to be suspect because people’s attitudes and intentions
change.
6. Historical analogy
In this method, when past data on a particular item is not available, analysis is made on items
similar to that under forecast and then such analyses are used to establish the life cycle and
expected patterns of the item under consideration/ forecast. However, a lot of care is needed
in using analogies which relate to similar items in different time periods.
C CHARACTERISTICS OF BUSINESS DATA THAT AFFECT TIME SERIES

These are the features that affect any business data
a) Trend characteristics. This refers to a smooth, regular, long-term movement of a time
series showing the tendency of a steady growth, fall or stagnation over a long period of
time. The trend is very common in most time series relating to economic and business
events. E .g-An upward movement is observed in time series relating to population,
incomes, production, sales volume, bank deposits etc. -A down ward trend is observed in
time series relating to deaths, epidemics etc.
171
b) Seasonal variations. These are short-term fluctuations in the time series data for a
period of less than a year. These involve patterns of change within a year that tend to be
repeated from season to season. This type of variation occurs especially in data that is
recorded weekly, monthly or quarterly. Practically, all businesses and economic series/
data have recurring seasonal patterns.
The seasonal variations may be attributed to two major causes: -Natural factors. This is
climatic changes in different seasons of the year bring about fluctuations in some events.eg
sale of soft drinks will be higher in hot seasons and lower in cool seasons. -Man made
conventions. This is where habits, fashions, customs and other conventions of people in
society cause variations in the time series data during a year e.g. Sales of Christmas cards
are high in December etc.
c) Long term cyclic variations: Such variations in business data can be thought of as due
to underlying economic causes outside the scope of the immediate business environment.
Examples include standard trade cycles or minor recessions. Cyclic patterns usually reflect
periods of boom, depression or recovery. Because of the multiplicity of factors that bring
about these cycles, it is very difficult to predict such characteristics/ patterns. For example,
a depression may occur because of unprecedented increase in petrol prices or stock price
crush. This characteristic also does have a fixed period pattern
d) Random/ irregular fluctuations: These are disturbances due to everyday

unpredictable influences such weather, illnesses, transport breakdowns and others. Such
fluctuations are difficult to isolate because they do not have a fixed pattern such as those
possessed by the above characteristics. These characteristics happen randomly and do not
have a fixed period of occurrence, for-example a fall in output due to a cargo train accident
or to any unusual causes cannot be predicted. These causes are not often known.
D APPLICATION OF QUANTITIVE FORECASTING METHODS

A. LEAST SQUARES METHOD: This method assumes that observations provided follow
a linear trend. The trend line equation is given by: t = a + bx. And after getting the
trend line, it can be used to obtain the trend values which will be plotted on the same
graph. Disadvantage: Unsuitable for highly seasoned data. The same trend line can be
used to forecast/ predict values in future periods when all the previous information is
available.
The trend line equation is given by: t = a + bx. Where,

n xy    x  y 
b
 
n  x 2   x 
2
AND a  Y  bX
Example of estimating a trend for the three methods listed above

The following data was compiled by the Entebbe airport Civil Aviation Authority in regard to
the number of outward passengers going through the airport. a) Compute a time series trend
using the method of least squares and forecast passengers going through the airport in
year-3 second quarter.
172
Year (s) Year1 Year2 Year3
Quarter 1 2 3 4 1 2 3 4 1 2 3 4
Passengers
(Millions) 220 140 108 210 122 242 246 160 260 ? ? ?
Solution
n xy    x  y 
b
 
n  x 2   x 
2
AND a  Y  bX
We put the various variables in a table
X Y (X.Y) X2
Year1 1 220 220 1
2 140 280 4
3 108 324 9
4 210 840 16
Year2 5 122 610 25
6 242 452 36
7 246 1722 49
8 160 1280 64
Year3 9 260 2340 81
Total 45 1708 8068 285
On substitution,
n xy    x  y 
b  7.869 AND a  Y  b X  229.1
 
n  x 2   x 
2
Therefore, the trend equation is given by: t = 229.1 + 7.869x. We can then forecast the
passenger numbers in year3 second quarter as follows: t = 229.1 + 7.869 (10) = 150.41
Approximately 150 passengers would go through the airport in the second quarter of year3
THE MOVING AVERAGE METHOD

This forecasting method works on a principle that the average forecast for the next period in
time is an average of several preceding periods. A moving average forecast uses a number of
the most recent actual data values in generating a forecast.
Advantages: The best technique so far for estimating a trend. Simple to use and it is
applicable to changing conditions Disadvantages. No trend values are obtained for the
beginning and end points of a series. The method is difficult to use for predicting or
forecasting future values which is the main objective of time series analysis.
173
4-Moving 4-Moving Centered Moving
Year Qtr Data Total Average Average (Trend)
Year1 1 2.2
2 5
3 7.9 18.3 4.575
4 3.2 4.75 4.66
Year2 1 2.9 19
2 5.2
a. The first 4 point moving average is obtained by: (2.2+5+7.9)/4 = 4.575

Then proceed and obtain other moving averages
b. In order for us to obtain a four-moving average forecast for the 3rd quarter of year two, we
obtain the average of the last four values in our series (7.9+3.2+2.9+5.2)/4=__________
EXPONENTIAL SMOOTHING
For a moving average (MA), each value in the average was given an equal weight. In a three
point moving average, each value is given a weight of 1/3. Exponential smoothing is a way of
constructing an ‘average’ which gives more weight to recent values of the variable. The
smoothed series is given by the equation:
New smoothed value = (1 – α) [Previous smoothed value] + α [most recent actual value]
Ie. St = (1 – α) St-1 + αxt.
Where α is between 0 and 1. The value of α (alpha) is chosen by the forecaster. The larger its
value, the heavier the weighting being given to the recent values. It’s value may be selected
after testing out several values and measuring which is the best. In practice α is usually in the
174
range 0.1 to 0.4le
E TIME SERIES DECOMPOSITION

Time series models that is to say, moving averages, exponential smoothing and least squares
method have been illustrated using data that is not seasonally adjusted to include factors like
price inflation, trend, seasonal, random and other variations. As method earlier, business data
is affected by;
A. Trend variations (T)
B. Seasonal variations (S)
C. Cyclic variations ( C)
D. Random variations (R)
Thus, to make reasonably accurate forecasts, it is important for us to make adjustments to the
business data so that provisions for the four components are taken care of.
The decomposition of time series data implies ensuring that the time series forecasts
made are accurately computed by making adjustments for trend, cyclical, and random
variations.
175
F METHODS USED IN DECOMPOSING TIME SERIES DATA
There are two methods used in decomposing time series data; additive and multiplicative
models.
A. ADDITIVE MODEL
The model assumes that the effects of the various factors/ components of the time series are
additive in nature. Sales = ( T+S+C+R)
B. MULTIPLICATIVE MODEL
The model assumes that the effects of the various factors/ components of the time series are
multiplicative in nature. Sales = (T x S x C x R). The factors are interdependent on each other
and inter related
The S, C & R can be expressed in percentages or proportions.
When business data is compiled, it is imperative to separate the seasonal variations from such
business data (De- Seasonalize). A seasonal index will show us whether there is a fall/ rise in
data and it is expressed in percentages.
The seasonal variations are measured using the additive model where as the
seasonal indices are measured using the multiplicative model.
To accurately measure the variations and seasonal indices, the time series data is split into
quarters and hence we obtain quarterly averages for each quarter.
Example:
The table below indicates profits made by Sejusa Corporation.
Year Quarter-1 Quarter-2 Quarter-3 Quarter-4
2011 97 106 96 100
2012 101 104 80 118
2013 116 111 93 114
2014 120 101 101 106
I. Compute the seasonal variations

II. Compute the seasonal indices
Solution
Step2: Obtain the average of each quarter
Year Quarter- Quarter- Quarter-

Quarter-1 2 3 4
2011 97 106 96 100
2012 101 104 80 118
2013 116 111 93 114
2014 120 101 101 106
Total 370 422 434 438
Average 105.5 108.5 109.5
176
92.5

Step2: Obtain the average of all averages
 92.5  105 .5  100 .5  109 .5 

    104
 4
Step3: The seasonal variations for each quarter is the deviation of each quarterly average from
the average of averages
Quarters Seasonal variations

Quarter-1 92.5-104=-11.5
Quarter-2 105.5-104=1.5
Quarter-3 108.5-104=4.5
Quarter-4 109.5-104=5.5
II) Seasonal indices are obtained as percentage variations of quarterly averages above using
the multiplicative model below
Quarters Seasonal Indices

Quarter-1 (92.5/104) x 100 = 09%
Quarter-2 (105.5/104) x 100 = 101%
Quarter-3 (108.5/104) x 100 = 104%
Quarter-4 (109.5/104) x 100 = 105%
G FORECASTING USING TREND AND SEASONAL VARIATIONS

In this case, seasonal variations will be computed by first determining the trend estimates
using the known general regression equation.
The trend variations are obtained for each quarter using the formulae:
(Actual sales/ Trend estimates) x 100
The quarterly indices are then obtained by getting the average of all annual seasonal indices.
Example:
The table below indicates profits made by Sejusa Corporation.
Year Quarter-1 Quarter-2 Quarter-3 Quarter-4

2011 20 32 62 29
2012 21 42 75 31
2013 23 39 77 48
2014 27 39 92 53
I. Compute the trend line using least squares method

II. Compute the estimates for sales for each year
III. Compute the seasonal percentage variations for each quarter’s actual sales from
estimates and the average seasonal variations
IV. Compute the average seasonal variations/indices
V.
VI.
Compute the sales forecast
Forecast sales for the year 2015 177
Solution
1. Compute the trend line using least squares method
The trend line equation is given by: t = a + bx.
X Y (X.Y) X2
2011 1 20
2 32
3 62
4 29
2012 5 21
6 42
7 75
8 31
2013 9 23
10 39
11 77
12 48
2014 13 27
14 39
15 92
16 53
Total 136 710 6,661 1,496

n xy    x  y 
b  1.84 AND a  Y  b X  28.74
 
n  x 2   x 
2
2. The estimated sales

The estimated sales are the trend variations obtained by substituting yearly quarterly values
into the above trend linear equation
2011 Quarter-1 sales are obtained as:

Y  28.74  1.8 x  (28.74  [1.84 x1])  30.58
Y  28.74  1.8x  (28.74  [1.84 x5])  37.94
Y  28.74  1.8x  (28.74  [1.84 x9])  44.94
Y  28.74  1.8x  (28.74  [1.84 x13])  52.66
178
3. Percentage seasonal variations
Are computed using this formulae: (Actual sales/ Trend estimates) x 100
A. Seasonal variations for 2011 quarter-1
(20 /30.58) x 100 = 65.0%
A. Seasonal variations for 2012 quarter-1
(21 /37.94) x 100 = 55.0%
All other percentage seasonal variations are given in this table
Y Trend Seasonal
X estimates variations
2011 1 20 30.58 65
2 32 32.42 99
3 62 34.26 181
4 29 36.10 80
2012 5 21 37.94 55
6 42 39.78 106
7 75 41.62 180
8 31 43.46 71
2013 9 23 45.30 51
10 39 47.14 83
11 77 48.98 157
12 48 50.82 94
2014 13 27 52.66 51
14 39 54.50 72
15 92 56.34 163
16 53 58.18 91
4. Average seasonal variations / indices

Year Quarter- Quarter- Quarter-
Quarter-1 2 3 4
% % % %
2011 65 99 181 80
2012 55 106 180 71
2013 51 83 157 94
2014 51 72 163 91
Total 222 360 681 336
Average 56% 90% 170% 84%
5. Sales forecast
The above average seasonal averages are used to forecast the yearly quarterly sales as shown
in the table below using the formulae:
Seasonally adjusted forecast = (Trend estimate x seasonal variation)

179
Y Trend Seasonal Seasonally
X sales estimates Indices adjusted
quarters % forecasts
2011 1 20 30.58 56 17.12
2 32 32.42 90 29.18
3 62 34.26 170 58.24
4 29 36.10 84 30.32
2012 5 21 37.94 56 21.24

6 42 39.78 90 35.8
7 75 41.62 170 70.74
8 31 43.46 84 36.51
2013 9 23 45.30 56 25.37

10 39 47.14 90 42.43
11 77 48.98 170 83.27
12 48 50.82 84 42.69
2014 13 27 52.66 56 29.49

14 39 54.50 90 49.05
15 92 56.34 170 95.78
16 53 58.18 84 48.87
6. Sales forecast for 2016

Once the trend line and average seasonal indices have been determined, they can be used to
forecast future sales. This is what is known as extrapolation of seasonal variations
As such, sales forecasts are computed as follows:
Trend Seasonal Seasonally adjusted forecasts

X estimates indices
quarters %
2015 17 60.02 56 33.61
18 55.67 90 10.02
19 108.29 170 20.58
20 55.02 84 46.24
Example 2
The sales of a leasing company have a trend line that is represented by the equation
(Y = 55,000 + 1250t), where t =1 in the first quarter of 2006. The four quarterly seasonal
indices of sales are 107, 100, 82 and 111
Required:
1. Approximate how sales have been increasing in each quarter up to the year ended
2008 180
2. Forecast the quarterly sales for the year ended 2009
Solution
Part-1
Trend Seasonal Seasonally
T estimates Indices adjusted
quarters % forecasts
2006 1 56,250 107 60,188
2 57,500 100 57,500
3 58,750 82 48,175
4 60,000 111 66,000
2007 5 61,250 107 65,538

6 62,500 100 62,500
7 63,750 82 52,275
8 65,000 111 72,150
2008 9 66,250 107 70,888

10 67,500 100 67,500
11 68,750 82 6,375
12 70,000 111 77,700
Part-2
Trend Seasonal Seasonally adjusted forecasts
T estimates indices
quarters %
2009 13 71,250 107 76,238
14 72,500 100 72,500
15 73,750 82 60,475
16 75,000 111 83,250
Trend estimates are obtained by substituting quarterly values in the trend equation
HAPPY HOUR: TRY OUT THESE NUMBERS

Question
The following data was compiled by the Entebbe airport Civil Aviation Authority in regard to
the number of outward passengers going through the airport.
Year 2008 2009 2010

Quarter 1 2 3 4 1 2 3 4 1 2 3 4
Passengers
(Millions)-y 2.2 5 7.9 3.2 2.9 5.2 8.2 3.8 3.2 5.8 9.1 4.1
Trend (t) 4.66 4.78 4.84 4.95 5.06 5.18 5.36 5.51
Required
I.
II.
Using the additive model, compute the seasonal variations for each of the 4 quarters
Obtain the seasonally adjusted outward passenger movements.
181
III. Forecast/ project passenger movements for all quarters of 2011
IV. Plot both the trend values and seasonally adjusted values on the same graph.
Question2
Question3
182
Chapter-18
Net work scheduling & analysis
18.1 INTRODUCTION
NETWORK SCHEDULING is a task of breaking down a project into its constituent activities
and deciding on their time relationships. The technique is a method of minimizing the trouble
spots such as production, delays and interruptions by determining critical factors and
coordinating various parts of the overall job.
There are two basic planning & control techniques that do utilize a network to complete a
predetermined project or schedule.
A. Critical Path Method (CPM)

B. Programme Evaluation & Review Technique (PERT)
A project is a combination of interrelated activities all of which must be executed in a certain

order for its completion. For example; the Bujagali hydro power project requires; 8,000
workers, 1,000 construction professionals, 100 medical personnel, 4 full service dining halls,
27,000 meals per day, 40bed field hospital nearby.
The work involved in a project can be divided into three phases corresponding to the
management functions of planning scheduling and control.
The planning phase. This involves setting the objectives of the project and the assumptions
to be made. It also involves the listing of tasks or jobs that must be performed to effectively
complete a project under consideration. In this phase, men, materials, machines, estimated
duration and costs required for the project are also determined.
The scheduling phase. This consists of laying the activities according to the precedence order
and determining:-
I. The start and finish times for each activity
II. The critical path on which the activities require special attention and
III. The slack and float for the non critical paths
The controlling phase. This comes after planning and scheduling phases.
It involves;
I. Making periodical progress reports
II. Reviewing the progress
III. Analyzing the status of the project
IV. Management decisions regarding updating, crashing and resource allocation
18.2 BASIC TERMS IN NETWORK PLANNING

In order to effectively use the Critical Path Method and the Programme Evaluation & Review
Technique, one should be able to understand the following terms.
A. A Network is the graphic representation of logically and sequentially connected arrows

and nodes representing activities and events of a project (sometimes called arrow
diagrams).
183
B. An Activity. Some action, which is time consuming effort necessary to complete a
particular part of a project. As such, each activity has a beginning and ending point. It
is represented by an arrow. The illustration below is an activity on arc.
It should be noted that the above activity A has been placed on an arrow/ arc. Thus, it
is termed as activity on arc. But, it is also possible to draw activities G and H on nodes
as illustrated below.
G H
C. Event. Beginning and ending points of an activity (represented by nodes). An event is

a point in time and doesn’t consume any resources. It is represented by a numbered
circle.
G
5 6
Tail event Head event
D. Dummy activity. Certain activities which neither consume time nor resources but are
used simply to represent a connection or link between the events. Shown with a dotted
line.
-
1 2
The purpose of a dummy activity is;

I. To maintain uniqueness in the numbering system as every activity may have
distinct set of events by which the activity can be identified.
II. To maintain proper logic in the network.
18.3 RULES GOVERNING NETWORK CONSTRUCTION

I. Try to avoid arrows which cross each other
II. Use straight arrows
III. No event can occur until every activity preceding it has been completed
IV. An event cannot occur twice. That is to say, there should be no loops
V. An activity succeeding an event cannot be started until that event has occurred
VI. Use arrows from left to right. Avoid mixing two directions, vertical and standing arrows
may be used if necessary
VII. Dummies should be introduced if it is extremely necessary
VIII. The network has only one entry point called the start event and one point of
emergence called the end or terminal event.
18.4 ACTIVITY ON ARC (AOA)

Example1.
A manager of a certain company has provided the following information in relation to
launching a new brand of soft drink
I. Draw the network to show the relationship between activities and events 184
II. List all the possible routes/ paths of the network and their associated durations
Activity label Description Preceding Duration

activities (days)
A Decide on test market area - 1
B Agree on marketing strategy - 2
C Agree on production specification - 3
D Decide on the brand name B 1
E prepare adverting package A 2
F Prepare an advertising package E 3
G Design product packaging D 2
H Production of the test batch C 5
I Package and distribute G, H 10
J Monitor media support F,D 3
Illustration of the network
A list of all the possible routes/ paths

Routes / paths Duration
Route 1 AEFJ
Route 2 BDJ
Route 3 BDGI
Route 4 CHI
Example2.
A manager at Spear Motors has compiled the following Jobs whose precedence relationships
are given below.
Activities/ Jobs 1-2 1-3 2-3 2-5 3-4 3-6 4-5 4-6 5- 6-
6 7
Duration 15 15 3 5 8 12 1 14 3 14
I. Draw the activity network to show the relationship between activities and events 185
Solution

Route 1 1-2-5-6-7
Route 2 1-2-3-4-6-7
Route 3 1-3-4-6-7
Route 4 1-3-6-7
Route 5 1-3-4-5-6-7
18.5 Activity on Node (AON)

Example3. Construct the network to satisfy the following precedence requirements
Activities A B C D E F G H
Immediate predecessor - A A A B,C C,D D,F G,E
Completion time 3 6 7 5 13 8 11 6
I. Draw the activity on node network to show the relationship between activities and
events
A list of all the possible routes/ paths 186

Route 1 ABEH
Route 2 ACEH
Route 3 ACFGH
Route 4 ADFGH
Route 5 ADGH

Immediate predecessor - - A B,A D,C C E F,G
events

Immediate predecessor - - - A B,C C A,D E,F
events
187
18.6 COMMON ERRORS WHEN DRAWING NETWORKS

A. Looping (cycling)
A looping (cycling) error arises when one draws an endless loop. A loop can be formed if
an activity were represented as going back in time
B. Dangling
This implies disconnecting an activity before the completion of all the activities in a network
diagram.
C. Redundancy
This occurs if a dummy activity is the only activity emanating from an event and which can
easily be eliminated.
PRACTICE EXERCISES
Question1
Construct a network for the project whose activities and their precedence relationships are
given below.
188
Activities A B C D E F G H I
Immediate - A A - D B,C,E F D G,H
predecessor
Question2
given below.
Activities A B C D E F G H I J K
Immediate - - - A B B C D E H,I F,G
Qu
predecessor
est
ion3
given below.
Activities A B C D E F G H I J K
Immediate - - A A I,J,K B,D B,D F A G,H F
predecessor
Question4
given below.
Activities A B C D E F G H I
Immediate - - A,B B B A,B F,D F,D C,G
predecessor
18.7 TIME ANALYSIS

18.7.1 INTRODUCTION
Once the network of a project is constructed, the time analysis of the network becomes
essential for planning various activities of a project. An activity time is the forecast of the time
an activity is expected to take from its starting point to its completion (under normal
conditions).
Time analysis technique establishes the overall completion time of projects by calculating what
is known as the critical path. 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 in a network and it is possible for
the critical path to run through a dummy. The following paragraphs give step by step, the
procedure for establishing the critical path.
Example6.
Given the following network diagram,
I. Compute the activity durations for each route/ path
II. What is the critical path and total project duration?
189
Routes / paths Duration (Days)
Route 1 ABE 6
Route 2 ADF 12 (This is the critical path)
Route 3 ACZF 7
It should be noted that a dummy activity (Z) consumes no time in terms of activity duration.
Thus a 0 is appended to any dummy activity.
18.7.2 THE EARLIEST START TIMES (EST) AND LATEST START TIMES (LST) OF AN
ACTIVITY
A critical path can also be obtained by calculating Earliest Start Times (EST) and Latest Start
Times (LST) of an activity.
A. EARLIEST START TIMES (EST)

FOR ACTIVITY ON ARC
Once the activities have been timed it is possible to assess the total project time by calculating
the EST's for each activity. The EST is the earliest possible time at which a succeeding activity
can start and the method of calculation will be apparent from the following example.
Assume the following network has been drawn and the activity times estimated in days.
190
The EST's can be inserted as follows:-
Notes on calculation of EST (termed the FORWARD PASS):

(a) The EST of a head event is obtained by adding onto the EST of the tail event, the
linking activity duration starting from Event 0, time 0 and working forward through the
network.
(b) Where two or more routes arrive at an event the LONGEST route time must be taken,
e.g. Activity F depends on completion of D and E. E is completed by day 5 and D is not
complete until day 7; therefore F cannot start before day 7.
(c) The EST in the finish event No. 5 is the project duration and is the shortest time in
which the whole project can be completed.
B. LATEST START TIMES (LST)

To enable the critical path to be isolated, the LST for each activity must also be established.
The LST is the latest possible time at which a preceding activity can finish without increasing
the project duration. Using the example above the LSTs are as follows:-
Notes on calculating LST (termed the BACKWARD PASS)

(a) Starting at the finish event No. 5, insert the LST (ie day 9) and work backwards
through the network deducting each activity duration from the previously calculated
LST.
(b) Where the tails of activities B and C join event No. 1, the LST for C is day 3 and the
LST for B is day 1. The lowest number is taken as the LST for event No. 1 because if
event no. 1 occurred at day 3 then activities B and C could not be completed by day 7
as required and the project would be delayed.
191
18.7.3 THE CRITICAL PATH
Examination of the above figure shows that one path through the network (A, B, D, and F)
has EST's and LST's which are identical. This is the critical path which it should be noted is the
chain of activities which has the longest duration. The critical path can be indicated on the
network either by a different color or by two small transverse lines across the arrows along the
path as follows:
Critical path implications

The activities along the critical path are vital activities which must be completed by their
EST's/LST's otherwise the project will be delayed. The non critical activities (in the example
above, C and E) have spare time or float available, ie C and/or E could take up to an
additional two days in total without delaying the project duration. If it is required to reduce
the overall project duration then the time of one or more of the activities on the critical path
must be reduced perhaps by using more labor, or more or better equipment or some other
method of reducing job times.
Note that for simple networks the critical path can be found by inspection, ie looking for the
longest route, but the above procedure is necessary for larger projects and must be
understood. The procedure is similar to that used by most computer programs dealing with
network analysis.
C. EARLIEST START TIMES (EST)

FOR ACTIVITY ON NODE
In an examination question, you may be told to compute the following

• Earliest Start Time (EST)
• Earliest Finish Time (EFT)
• Latest start time (LST)
• Latest Finish Time (LFT)
We use the FORWARD PASS to identify earliest times (EST and EFT)
EST Rule: All immediate predecessors must be done before an activity can begin
– If only 1 immediate predecessor, then: EST = EFT of predecessor
– If >1 immediate predecessors, then EST = Max {all predecessor EFT’s}
EFT Rule: EFT = EST + activity time
192
Node Notation
Example:-
General Foundry Inc has 16 weeks to install a complex air filter system on its smokestack. It
may be forced to close if not completed within 16 weeks due to environmental regulations.
They have identified 8 activities as listed below.
Time estimates for the company
I) Draw an Activity on Node diagram

II) Estimate the:-
• Earliest Start Time (EST)
• Earliest Finish Time (EFT)
• Latest start time (LST) 193
• Latest Finish Time (LFT)
III) Compute the slack for each activity
IV) From the slack values, what is the critical path?
SOLUTION
Part-1: ACTIVITY ON NODE DIAGRAM
PART-2: FORWARD PASS: EARLIEST START AND FINISH TIMES
BACKWARD PASS: Identifies latest times (LST an LFT)

LFT Rule: If activity is the immediate predecessor to only 1 activity, then: LFT = LST
of immediate follower. If activity is the immediate predecessor to multiple activities,
then LFT = Min {LST of all immediate followers}
LST Rule: LST = LFT – activity time
194
PART-3: SLACK TIME AND CRITICAL PATH(S)
• Slack is the length of time an activity can be delayed without delaying the project.
Slack = LST – EST
• Activities with 0 slack are Critical Activities
• The CRITICAL PATH is a continuous path through the network from start to finish
that include only critical activities
Project schedule and slack times
Critical Path and Slack Time
195
TOTAL SLACK TIME VS. FREE SLACK TIME
• Total slack time is shared by more than 1 activity
Example: A 1 week delay in activity B will leave 0 slack for activity D
• Free slack time is associated with only 1 activity
Example: Activity F has 6 week of free slack time
18.7.4 FLOAT FOR ACTIVITY ON ARC

Float or spare time can only be associated with activities which are non-critical. By definition,
activities on the critical path cannot have float.
There are three types of float, Total Float, Free Float and Independent Float. To
illustrate these types of float, part of a network will be used together with a bar diagram of the
timings, thus:
196
(a) Total Float. This is the amount of time a path of activities could be delayed without
affecting the overall project duration. (For simplicity the path in this example consists
of one activity only, ie Activity K).
Total Float = Latest Head time - Earliest Tail time - Activity Duration
Total Float = 50 - 10 - 10
= 30 days
(b) Free Float. This is the amount of time an activity can be delayed without affecting the
commencement of a subsequent activity at its earliest start time, but may affect float of
a previous activity.
Free Float = Earliest Head Time - Earliest Tail time - Activity Duration
Free Float = 40 - 10 - 10
= 20 days
(c) Independent float. This is the amount of time an activity can be delayed when all
preceding activities are completed as late as possible and all succeeding activities
completed as early as possible. Independent float therefore does not affect float of
either preceding or subsequent activities.
Independent float = Earliest Head time - Latest Tail time - Activity Duration
Independent float = 40 - 20 – 10 = 10 days
PRACTICE EXERCISES
a) A project consists of a series of tasks labeled A, B ………..H, I with the following
constraints. A  D, E ; B, D  F ; C  G; B  H F , G  I . Note that W  X , Y implies that
X and Y can not start until W is completed.
197
Activity A B C D E F G H I
Preceding activity - - - A A B, D C B F, G
Normal time (Days) 23 8 20 16 24 18 19 4 10
I. Construct the network using the information above
II. Compute all the Earliest Event Time and Latest Event Time for each event
III. Compute the Total and Free floats for each activity and hence the critical path
Solution
Total Float = Latest Head time - Earliest Tail time - Activity Duration
198
Task TOTAL FLOAT
A 1-2 0
B 1-3 8
C 1-4 18
D 2-3 0
E 2-6 20
F 3-5 0
G 3-6 24
H 4-5 18
I 5-6 0
18.8 COST CONSIDERATIONS IN CRITICAL PATH METHOD

18.8.1 INTRODUCTION
In order to include the cost aspects in project scheduling, we must first define the cost
duration relationships for the various activities in a project.
The total cost of a project comprises of direct and indirect costs.
A. Direct costs are those which are directly dependent upon the amount of resources in
the execution of individual activities. Eg. Manpower loading, materials consumed, etc.
Direct costs increase if the activity duration is shortened/ reduced.
B. Indirect costs are costs associated with overhead expenses such as managerial
services, indirect supplies, general administration etc. the indirect cost is computed per
day, per week, or per month. The indirect costs decrease if the activity duration is to be
reduced.
It should be mentioned that some ways of reducing project duration are cheaper than others
and as such the analysis of costs in networks is of paramount importance. It helps in
identifying the cheapest way of reducing a project overall duration.
When a project is completed early, it often attracts a bonus. Again, when a project is
completed late, it often attracts a penalty. Therefore, any one handling extensive projects
needs to analyze costs to find out whether it is worthwhile to be late and pay extra costs or
rush a project and get a bonus.
The network diagram is used to identify activities whose duration should be shortened so that
the completion time of a project can be shortened.
18.8.2 PROJECT CRASHING

The process of reducing the activity duration by putting on extra effort. Some activities times
can be shortened (by adding more resources, working overtime, etc.). The following terms are
often encountered in crashing or speeding up of projects.
A. Normal cost
Normal cost is equal to the absolute minimum of direct cost required to perform an activity. It
is the cost associated with a normal time estimate for a particular activity. Has an associated
normal time.
199
B. Normal time
This is a time estimate for a particular activity when resources (that is to say; men, machines
etc) are used in the most efficient manner possible.
C. Crash cost
It is the minimum direct cost required to achieve the crash performance time. Such crash costs
are usually higher than the normal cost because of extra wages, overtime premiums and extra
facility among other factors
D. Crash time
This represents the minimum activity duration time that is possible and any attempts to further
crash would only raise the activity cost without reducing the time.
E. Cost slope
This is the average cost of shortening an activity by one time unit (day, week, month etc). The
cost slope is generally assumed to be linear and calculated from;
Cost slope = Crash cost – Normal cost

Normal time – Crash time
18.8.3 STEPS IN PROJECT CRASHING

Step1. Compute the crash cost per time period OR cost slope and also indicate the available
time to crash each activity (difference between the crash time and normal time)
Step2. Using current activity durations, draw the network, find the critical path and identify
the critical activities
Step3. Select the activities on the critical path to crash following the criteria below:-
If there is only one critical path
• Select the activity on this critical path that;
– Can still be crashed
– Has the smallest crash cost per period
If there is more than one critical path

• Select one activity from each critical path such that;
– Each selected activity can still be crashed
– The total crash cost of all selected activities is the smallest
– Note - the same activity may be common to more than one critical path
Step4. Update all activity times. If the desired due date has been reached, stop. If not, return
to Step 2
Note – When there are multiple options all with the same crash cost, look at
whichever is least complicated and would affect the fewest number of people
200
Example7.
The table below provides cost and estimates for a seven activity project
NORMAL CRASH
Activity Preceding Duration Cost Duration Cost
activity (weeks) $ (weeks) $
A 1-2 - 2 10 1 15
B 1-3 - 8 15 5 21
C 2-4 A 4 20 3 24
D 3-4 B 1 7 1 7
E 3-5 B 2 8 1 15
F 4-6 C,D 5 10 3 16
G 5-6 E 6 12 2 36
Find;
I. Draw the network from the above table and identify the critical path
II. Find the total normal cost and duration of the entire project
III. Crash the project activities so that the overall project completion time is reduced to 9
weeks
Solution
Part-I-
The critical path is B, E and G OR (1-3-5-6)
Part-II-
The total normal cost is obtained by summing all the normal costs of the project. And as such,
the total normal cost of the project is $82,000
The duration of the critical path is the total normal duration of the entire project. It is 16
Weeks
Part-III-
Note. Only activities on the critical path can be crashed. Thus their total duration is called the
crash duration.
201
Activity OR Crash Available Narration
Cost Per Crashing
Week $’000 Time
= (CC - NC) / (NT - CT) (NT - CT)
A 1-2 5 1
B 1-3 2 3 To be crashed first
C 2-4 4 1
D 3-4 0 0
E 3-5 7 1
F 4-6 3 2 To be crashed third
by 2 weeks
G 5-6 6 4 To be crashed
second by only 2
weeks
the critical activities
The network has been drawn on the previous page and the critical path is B, E and G OR (1 –
3 – 5 - 6). These are the activities to be crashed
Step3. Select the activities on the critical path to crash.

Note that since there is only one critical path, we select an activity that has got the least crash
cost per time period.
Note: we are supposed to reduce the total project duration from 16 to 9 weeks
1st Crashing: Since activity (1- 3; B) has the least crash cost per week, we crash it by all the
3 weeks at a rate of $2 (‘000) per week. Thus, total new project duration and associated new
total costs are given below.
Original project duration = 16 Weeks

New project duration = 16 – 3 =13 Weeks
New project total costs = 82 + (3x2) = $88
The table below is necessary in identifying the possible routes and their associated new
durations
Possible OR Before 1st 2nd 3rd

routes Crashing Crashing Crashing Crashing
1-3-5-6 BEG 16 13 11 9
1-3-4-6 BDF 14 11 11 9
1-2-4-6 ACF 11 11 11 9
2nd Crashing:
After the first crash, we still maintain a critical path of B, E and G (1 – 3 – 5 – 6) (duration is
13 weeks which is the longest). We are done with activity B (1-3). As such, we only have E, G 202
(3-5-6) on the critical path. Therefore, we crash activity G (5-6) since it has the lowest crash
cost per week. We reduce G by only 2 weeks to create a total duration of 11 weeks which is
parallel to other routes. If we reduce it by more than 2 weeks, we would have created other
two new critical paths before completing the G crash.
First crashing project duration = 13 Weeks

New project duration after second crashing = 13 – 3 =11 Weeks
New project total costs = 88 + (6x2) = $100 (’000)
3rd Crashing:
Note: Duration must be reduced to 9weeks. We have only managed to reduce it to 11 weeks.
At the third crashing attempt, we can see that all possible routes are now critical (all have the
same maximum durations of 11weeks)
Step3 clearly highlights that; if there is more than one critical path;
– The total crash cost of all selected activities is the smallest and a note that the
same activity may be common to more than one critical path.
Possible routes OR
1-3-5-6 BEG G (5-6) has the lowest cost slope. We crash it by
2weeks at $6 per week creating an extra cost of
($6 x2) = $12
1-3-4-6 BDF F (4-6) is common to both routes and has the
1-2-4-6 ACF lowest cost slope, we crash it by 2 weeks at $3
per week creating an extra cost of ($3 x2) = $6
Thus total new costs =$12 + $6 = $18
First crashing project duration = 11 Weeks

New project duration after second crashing = 11 – 2 = 9 Weeks
New project total costs = 100 + (6x2)+(3x2) = $118 (’000)
Conclusion
The project duration can be reduced up to 9 weeks only but the total project costs has shot up
to a total of $118,000.
We can use the new project durations and the new project costs to plot a time cost trade
off curve as illustrated below.
New time (weeks) 16 13 11 9

New costs (‘$000) 82 88 100 118
203
Example8. Incorporating over head costs
NORMAL CRASH
Activity Duration Cost $ Duration Cost $
(weeks) (‘000) (weeks) (‘000)
1-2 8 100 6 200
1-3 4 150 2 350
2-4 2 50 1 90
2-5 10 100 5 400
3-4 5 100 1 200
4-5 3 80 1 100
Given that indirect costs for the following project are $70 (‘000), crash the project and
determine the optimal duration and associated costs for completing this project.
Solution
Activity Crash Available

Cost Per Crashing
Week $’000 Time
= (CC - NC) / (NT - CT) (NT - CT)
1-2 50 2
1-3 50 2
2-4 40 1
2-5 60 5
3-4 25 4
4-5 10 2
the critical activities 204
The critical path is (1 – 2 – 5). These are the only activities to be crashed
Original project direct costs = $580
Original project indirect costs = $1,260 (18 x $70)
Total original project costs = $1,840 ($580 + $1,260)
Step3. Select the activities on the critical path to crash.

Note that since there is only one critical path, we select an activity that has got the least crash
cost per time period.
Note: An optimal duration refers to the duration which yields the minimum total cost where as
a minimum duration is the one for which no more crashing is possible
1st Crashing: Since activity (1- 2) on the critical path has the least crash cost per week,
we crash it by all the 2 weeks at a rate of $50 (‘000) per week. Thus, total new project
duration and associated new total costs are given below.

New direct project duration = 18 – 2 =16 Weeks
New direct project total costs = 1,840 + (2x50) = $1,940
New indirect total costs = 2 x $70 = $140
New total project costs = $1,800 ($1,940 - $140)
Note. Crashing a project increases direct costs and reduces indirect costs
The table below is necessary in identifying the possible routes and their associated new
durations
Possible Before 1st 2nd 3rd
routes Crashing Crashing Crashing Crashing
1-2-5 18 16 12
1–2–4- 13 11 11
5
1–3–4- 12 12 12
5
2nd Crashing:
After the first crash, we still maintain a critical path of (1 – 2 – 5) (duration is 16 weeks which
is the longest). We are done with activity (1-2). As such, we only have (2-5) on the critical
path. Therefore, we crash activity (2-5) since it has the lowest crash cost per week on this
205
critical path. We reduce (2-5) by only 4 weeks to create a total duration of 12 weeks which is
parallel to other routes (like1-3-4-5).

New direct project total costs = 1,800 + (4x60) = $2,040
3rd Crashing:
Note: Duration must be reduced as much as possible. We have only managed to reduce it to
12 weeks.
At the third crashing attempt, we can see that two new paths are now critical (1-2-5) and (1-3-
4-5).
Step3 clearly highlights that; if there is more than one critical path;
– The total crash cost of all selected activities is the smallest and a note that the
same activity may be common to more than one critical path.
Possible routes
1-2-5 (2-5) has the lowest cost slope and still
has balance duration of only 1week. We
crash it by 1weeks at $60 per week
creating an extra cost of ($60 x1) = $60
1–3–4-5 (4-5) has the lowest cost slope of $10.
Thus, we crash it by 1weeks at $10 per
week creating an extra cost of ($10 x1) =
$10
Total extra costs $70 (60 + 10)
resulting from
simultaneous crash
Second crash project duration = 12 Weeks

New direct project total costs = 1,760+ (1x70) = $1,830
Conclusion
The project duration can be reduced up to 11 weeks only but the total project costs has shot
up to a total of $1,760.
Example8. Dealing with bonuses for early and penalties for late completion of
projects
For national and regional security, the Government of Uganda has signed a contract with a
Chinese arms manufacturer to produce weapons that will supplement the current space based
defense system. The contract stipulates that the weaponry be ready within a year (50 working
weeks). There is a penalty of $100,000 per week for late delivery; there is also a bonus of
$50,000 per week for early delivery.
206
Several tasks are necessary to complete the project. First, the implications of the contract have
to be detailed by personnel, materials and systems departments. A joint team will carry out
this task, which is expected to take two weeks. Once this is complete the next three tasks can
be undertaken concurrently: ordering raw materials (three weeks), allocating and hiring
personnel (14 weeks), planning the modification and adaptation of equipment (12 weeks).
After the raw materials have been ordered it will take a further 20 weeks for them to be
delivered. Once the modifications have been planned it will take a further 18 weeks for them
to be carried out. When materials, equipment and personnel are available it will take two
weeks to train the quality control personnel and three weeks to train the production personnel.
These activities can take place simultaneously. When completed the production phase can start
and this will last for 18 weeks, followed by a further five weeks of testing. The final
preparations stage then takes three weeks. A few of the activities can be crashed, as shown in
the following table:
Activity Normal Crash

Time (weeks) Cost ($,000) Time (weeks) Cost ($,000)
Hire personnel 14 100 10 200
Plan 12 450 9 630
equipment
mods
Modifications 18 2,600 12 2,900
Production 18 1,800 14 2,400
Required
1. How long will the project take to complete, without crashing any activities? What are
the critical activities?
2. The time for the project can be reduced by crashing activities but this will affect bonus
or penalty payments specified in the contract. What schedule of activities is in the best
financial interests of the company?
Solution
18.9 THE GANTT CHART

Gantt charts are bar graphs that help plan and monitor project development or resource
allocation on a horizontal time scale.
Example
207
Task How long? What other task does it
depend on?
Choose which bag you want to take 5 minutes none
Find the bag 5 minutes Depends on choosing which
bag I want to take
Find the books you need for the day 5 minutes none
Get someone to make you a packed 10 minutes none
lunch
Get a letter from a parent/carer giving 20 minutes none
you permission to take part in a school
trip
Put the books and lunch in your bag 2 minutes Depends on finding the
books, and getting someone
to make the lunch
Close up the bag 1 minute Depends on putting the books
and lunch in the bag
Put the permission letter in your pocket 1 minute Depends on getting the letter
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
Tasks
Choose bag
Find bag
Find books
Get packed lunch made
Get letter
Put things in bag
Close bag
letter in pocket
18.10 USING COMPUTERS IN NETWORK ANALYSIS

18.10.1 ADVANTAGES OF COMPUTER-BASED PROJECT PLANNING AND CONTROL
The use of project management software is almost universal and there are few doubts as to
the benefits to its effectiveness. The most obvious advantages include the following:
• Speed Computerized project planning software offers the obvious example of speed. One
good package can produce the same planning information as a whole team of specialist
planners and in a much quicker time. Good software gives particular time savings in re-
planning after tradeoff analysis.
• Cost Modern high-performance software is initially expensive. The package itself can cost
several thousand pounds and the necessary staff training and development can cost much
more. However once the software is in place and staff have developed proficiency in the use of
the various systems, the potential cost savings can be very significant. One skilled planner can
effectively perform the same work as a whole team of skilled planners working on paper would
have formerly required.
• Capacity Good packages offer enormous capacity and even very large projects with
thousands of activities and resources can be accommodated. 208
In most cases, the limiting factor is the processing capacity and speed of the computer that
used.
• Reliability Modern software is extremely accurate and reliable. The programs are very
carefully checked and tested to ensure that the various calculations and presentations are
accurate. Any programme is only as reliable as the information that is input to it and the
possibility of human error remains.
• Combined analysis Most modern packages offer combined analysis functions.
The operator can use the software to plan and control time and cost simultaneously.
Information on resources can be stored within the system and automatic tradeoff scenarios
can be generated. This level and complexity of analysis is simply not possible on a manual
basis.
18.10.2 DISADVANTAGES OF COMPUTER-BASED PROJECT PLANNING AND

CONTROL
Project managers generally take project planning software for granted and regard it as a
standard tool for what they do. Project planning software, along with IT in general, has
reached all levels of many industries. There are, however, some disadvantages associated with
the use of project planning software.
• Reliance The use of advanced software generates an automatic reliance and a consequent
risk. The prudent project manager will ensure that all computerized data and records are
adequately protected and backed up but a surprising number of project managers do not take
adequate precautions.
System information can easily be lost or corrupted as a result of faults in the IT system, or
inadequate protection from external tampering or malicious viruses. Smaller companies which
run projects only occasionally are often guilty of not taking adequate precautions to protect
their project data.
• Over emphasis on system detail Very large project plans require a great deal of
administration and attention to detail. The project manager or planner may find that he or she
is spending an increasing amount of time in maintaining the various linkages and dependencies
within the program rather than managing the actual project. This tendency may not be a
problem as long as it is adequately controlled. It does mean, however, that the project
manager, as an expensive professional, may be spending some of his or her time on what is
really an ancillary support activity.
• Information dump Modern packages can store and process very large amounts of data and
allow aspects of the project plan to be looked at from numerous different aspects. Even the
most basic packages offer a multitude of different report formats and styles. There is often a
tendency for people to produce reports and printouts that contain too much information,
simply because it is so easy to do so. The project manager needs to exercise a degree of
restriction and control over the output so that the important facts and figures are not confused
within a mass of less important information.
• Potential misdirection Modern packages can produce very detailed and professional looking
reports. This can be a danger as people have a natural tendency to accept well presented
material as being accurate. As discussed above, the system is only as accurate as the
information that is input to it. Incorrect data will lead to incorrect reports and the degree or
quality of report presentation does not alter this fact.
18.10.3 GENERAL FACTORS FOR CONSIDERATION

IT is so widespread in modern commerce and industry that virtually everybody takes it for
granted. Project planning software is a specific IT application and is somewhat specialized in
209
its use. A company that is thinking of purchasing a project planning system, or replacing an
existing system, should consider a number of issues.
• Lead in time Modern packages are very complex and it can take a long time for staff to
develop full proficiency in the use of the system. Even with intensive training and testing
support, it can take three to six months for a new system to be installed and commissioned up
to a level where it is reliable.
• Transition. Transition may be a problem where a company is changing from one system to
another. People who have built up a detailed knowledge of one system tend to have natural
reluctance to switch to a new system, the main reason being the effort involved. Software
designers tend to use some common approaches but the detailed design of systems tends to
be quite different. It is still common to find word processing staff using systems such as Word
Perfect within networks that were converted to run on Microsoft operating systems several
years ago.
There is also the problem of parallel integration. Phasing out an old system and replacing it
with a new system is itself a project and is subject to all the (hopefully now familiar!) issues of
planning complexity and risk.
• Training Staff training can be a time consuming and expensive process, especially where
large numbers of staff are involved. Training on new systems has the effect of reducing the
availability of resources for use in existing systems and this can have an adverse effect on the
overall profitability of the organization. Intensive retraining can also be a source of stress
among staff and acts as a catalyst for conflict. Different departments often feel that they are
being unfairly pressurized into retraining earlier or more quickly than other departments.
• Updates The maintenance in proficiency in modern software is an ongoing process. Software
manufacturers introduce frequent updates and these can often involve considerable additional
functions and adaptations. The manufacturers of the more complex programs hold regular
training seminars so that users can stay up to date with the latest developments to the system.
Such seminars are useful and they do allow staff to use the software to its full potential, but
keeping up to date with the latest developments consumes time and money that could be used
elsewhere.
• Networking On large projects, it is standard practice to run some kind of configuration
management system (CMS). A CMS is a centralized information management system and is
usually based on a central server, which runs on centralized network software serving a
number of remote users. The project planning software used on a networked system is
accessible by a number of different people, even if the access is restricted to read-only.
This multiuser approach means that the project manager and planners have to have an
awareness of networks and the corresponding security and access implications that are
involved.
• Wider compatibility The logical extension of the CMS system is to link the central network to
external consultants and even (in some cases) to external contractors and suppliers. There are
obvious security considerations to allowing contractors and other external organizations to
have direct access to the project database, but there are also significant potential advantages.
This kind of approach is used already on large-scale term maintenance contracts. A particular
engineering company might win a contract to carry out maintenance and repairs on a power
station. Using a CMS, repair requests can be emailed directly to the contractor for online
estimating and programming. In some cases, work re-measurement and payments are also
made online.
18.10.4 SYSTEM CRITICAL SUCCESS FACTORS

Project planning software can be considered in terms of a number of critical success factors.
These are the characteristics of the system that determine how useable it is in terms of 210
delivering project success criteria. There are a number of generally accepted program success
factors.
• The system should be useable. Most people have encountered software that is not ‘user-
friendly’. Some programs seem to be very difficult to understand and do not give the user any
support or advice where problems are encountered.
Some of the early DOSbased estimating and planning packages were particularly bad for this.
The only way to learn to use them was to be shown how by an expert. The package itself and
the supporting documentation did not contain the amount of information required for a new
user to learn how to use the system. Modern packages are more approachable, but large
project-planning packages can still appear to be daunting to the uninitiated. Ideally, even the
most complex packages should be readily approachable and should offer support and
assistance to new users. Staff will respond much more readily to such software than they
would to unapproachable systems.
• The systems should use familiar displays. The most successful complex packages produce
displays and outputs that are compatible with what users expect. Most people expect a cost
report to have a certain type of appearance. It should typically show the budget limit for a
particular work package, how much has been spent to date and what level of progress has
been made in incurring these costs. The most effective displays and reports are those that
present this standard information as clearly and as succinctly as possible.
• The system should be CMS compatible. Large and complex projects often use a configuration
management system where all information is centralized and distributed to all members of the
project team through remove PCs. The amount of information that each person can obtain and
the level of access that is permissible is restricted by individual authority clearance levels. The
higher the clearance the greater the amount of information the person has access to within the
DMS. A centralized version of the project master schedule (PMS) is particularly important since
most members of the project team have a considerable interest in it.
• The system should be extendable. Complex software increasingly incorporates a degree of
over design. A common complaint about Microsoft Word is that it incorporates a great many
functions and applications that nobody ever uses. This overdesign results in the use of large
amounts of ROM so that the microchip makers have to keep developing new and more
powerful processors. The end result is that PCs become obsolete very quickly as only the latest
chips can run the latest version of the software. This end result is not intentional on the part of
Microsoft. They are simply designing systems that have built in capacity for extension. Many of
the functions that are not used are in fact designed to lead in to the next version of the
software. The potential for extension is an important aspect of software design. The software
should not be limited to what people want now. It should also be forward looking and try to
incorporate next generation ideas so that innovation supply stays ahead of demand.
PRACTICE EXERCISES
Question1.
The normal cost/duration and other relevant information for a project is given below.
Activity Normal Normal Minimum duration Cost per

duration total if accelerated day
(days) cost $ (days) accelerated
1-2 3 140 1 110
2-3
2-4
2
3
200
160
1
1
175
125 211
2-5 2 300 1 200
3-6 2 250 1 175
4-6 6 400 1 70
5-6 5 230 1 70
6-7 5 230 1 90
There is a bonus of $100 per day for every day that is saved below the contract period of
15days and a penalty of $200 for each day after the 15days.
Required
a) Calculate the normal duration and the normal cost of the project
b) Calculate the minimum cost of completing the project in 15days
c) State the optimum plan for the company to attempt
d) Revert to the normal programme and normal costs and state what action you would
recommend to ensure completion by the original date, if after the tenth day, the actual
situation was as follows;
I. Activities completed at normal cost; 1-2; 2-3; 3-6; 2-4; 2-5
II. Activities not yet started 4-6; 5-6; 6-7
What is the revised cost of the project in these circumstances?
Question2
Delco is a Swedish small engineering company that has been contracted by BATA shoe
company to design and produce a machine that will help increase mass production of BATA
shoes. The activities required in the design and manufacture of the machine are listed below
together with duration and costs.
Activity Description Preceding Duration Cost ($)

activity (weeks)
A Draw up estimate of costs - 2 400
B Agree estimate A 1 0
C Purchase internal machinery B 4 200
D Prepare design drawings B 6 450
E Construct main frame D 3 700
F Assemble machinery C,E 3 200
G Test machinery F 4 600
H Determine model type D 2 0
I Design outer casing D 3 250
J Construct outer casing H,I 8 600
K Final assembly G,J 2 450
L Final check K 2 200
In addition to the above cost figures, overheads of $250 per week will be incurred for the
duration of the project.
Required
I. What is the critical path and duration of the project?
II. What would be the effect of a strike at the factory supplying the internal machinery
thereby delaying its delivery for 4 weeks?
III. What would be the effect if the test on the machinery had been done incorrectly and
212
had to be redone taking another 4 weeks?

Question3
Each year, the “ICPAU” association prepares and distributes an annual programme. The
progarmme gives dates of meetings and a list of speakers with summaries of their talks. Also
included is an up-to-date list of paid up members. The activities to be carried out to complete
the preparation of the programme are as follows.
Activity Description Immediate Estimated

predecessor time
(days)
A Select dates for the programme - 4
B Secure agreement from speakers and prepare A 12
summaries of their talks
C Obtain advertising material for the programme A 11
D Mail membership renewal notices - 20
E Prepare list of paid up members D 6
F Send membership list to printer and read proofs B,C,E 7
G Print and assemble programme F 10
H Obtain computer printed address labels of members E 5
I Send out programmes G,H 4
Required
a. Draw a network for the scheme of activities set out above. indicate full information on
earliest and latest event times and indicate the critical path
b. Draw a bar chart for the scheme and indicate the total float of each activity
c. If each activity requires one member of office staff of the association, so that the
activities may be completed in the estimated times, what is the minimum number of
staff that should be allocated to the scheme?
d. What would be the effect on the total time if one of the allocated staff was taken ill for
the duration of the scheme and not replaced?
Example4
The following time- cost table (time in weeks and cost in shillings) applies to a project. Use it
to arrive at the network associated with completing the project in minimum time with minimum
cost.
Activity Normal Crash
Time Cost Time Cost
1-2 2 800 1 1400
1-3 5 1000 2 2000
1-4 5 1000 3 1800
2-4 1 500 1 500
2-5 5 1500 3 2100
3-4 4 2000 3 3000
3-5 6 1200 4 1600
4-5 5 900 3 1600
213
Example5
Activity Required Normal Time Normal Cost Crash Time Crash Cost
Predecessor
A (None) 8 $4000 6 $6000
B (None) 5 1500 4 2000
C (None) 6 2500 4 3000
D A 4 1800 3 2000
E A, B 6 1000 5 1200
F C 7 2000 5 3000
G A 5 3000 3 6000
H D, E, F 8 4500 5 9000
I C 9 6000 4 10000
J D, E, F 6 6000 4 8000
K G, H 4 2000 3 2600
L D, E, F 6 3000 3 9000
M I, J 4 8000 2 12000
Required
The scheduled completion time is 25 weeks. You must pay a $1,500 penalty for every week
you are late. You get a $1,000 bonus for every week you are early.
1. Show your network diagram.
2. Find the normal completion time and the critical path.
3. Determine the schedule that minimizes your total cost for this project, including any penalty
or bonus.
NOW THAT YOU ARE HAPPY, TRY OUT THESE PAST PAPER EXAM NUMBERS
Sitting Number Sitting Number

June 2016 November 2013
November 2015 June 2013
June 2015 November 2012
November 2014 June 2012
June 2014 December 2011
214
Chapter-19
Index numbers
19.1 INTRODUCTION
Bank of Uganda, consumer groups, families, Trade unions, several company directors, business
leaders and other economists are very much interested about changes in prices. Such changes
in prices are usually explained by a consumer price index (CPI), a producer price index etc.
The cost of living index is the most common type of index. It comprises the cost of so many
types of goods: like foods, fuel, transport, etc. instead of using the miscellaneous and often
confusing prices of all these purchases, we use an index number which summarizes them for
us. If the index number for 1996 is 182.1 compared with 165.3 in 1995, we can calculate that
the cost of living has risen by:-
182 .1  165 .3 
 165 .3  x100  10.2% This is rather easier than having to cope and deal with the range
of individual price increases involved.
19.2 INDEX NUMBERS

An index number is a special type of average which provides a measure of how much
something has changed from one time to another or how something compares with something
else. In brief, an index is used to represent changes betweens two or more groups of figures
taken from either different time periods or geographical locations.
Or
An index number shows the average percentage change of a variable from one point of time to
another. Here, they reduce a group of numbers to a mere usable and easily understood
figures which can easily be compared. Even where degree of confidentiality is required, an
index number may be used, e.g. a company may not be willing to disclose to the public its
level of profits so it opts for index numbers.
Example; In 2001, the average hourly wage was Ug. Shs. 20,000 and in 2006 the average
hourly wage is Ug.Shs. 35,000. What is the index for hourly earnings in Uganda for 2006
basing on 2001?
Solution:
Hourly earnings in 2006 x 100%  35,000 shs  x100  175
 20,000 shs 
Hourly earnings in 2001
Interpretation:
In Uganda, the hourly earnings did increase by 75% (175-100) between 2001 and 2006.
Example; The Uganda Bureau of Statistics reported that annual Banana production dropped
from 30,529,000 in 2002 to 5,100,000 in 2006. What’s the index for banana production during
the period?
 2006 Pr oduction   5,100,000 

 2002 Pr oduction  x100   30,529,000  x100  16.7%
 
Interpretation: 215
Between the periods mentioned, Banana production dropped by 83.3%.
The index number shows us that there has been a production drop of 83.3% since the base
period. An index number for a single price change like this is called; a price relative.
The particular time period of 2002, which we've chosen to compare against, is called the base
period.
The variable for that period, in this case the 30,529,000 is then given a value of 100,
corresponding to 100%.
Example; Monthly profits of firm X
Month Profit Profits based on

Jan=100
Jan 512 100
Feb 520 102
March 530 104
April 531 104
May 546 107
The right hand column shows Jan set to the standard figure of 100 and all the profit values set
to percentages of this standard. The figures in the right hand column are called index numbers
of profits and in this case, Jan is called the base month. Against which all others are compared.
So the index number illustrates changes by expressing the items in time series as percentages
of item at a chosen base period.
19.3 TYPES OF INDEX NUMBERS

There are three main types of index numbers.
I. Price index numbers. These do measure changes in commodity prices
II. Quantity index numbers. These measure changes in volume of goods produced,
consumed and distributed
III. Value index numbers. These measure changes in total value of goods produced and
sold. The value is given by; (Price x Quantity)
19.4 CLASSIFICATION/ TYPES OF INDEX NUMBERS OR METHODS OF

CONSTRUCTION OF INDEX NUMBERS
Index numbers can be classified into two (2) major categories:-
I. Un-weighted index numbers
II. Weighted index numbers
A. UNWEIGHTED INDEX NUMBERS

This is a category of index numbers that do not consider the relative importance of the
commodity for which they are computed:
Types of Un-weighted index numbers

In principle, there are three categories of index numbers.
a) Simple index numbers or relatives
b) Simple aggregate index numbers
c) Simple average index relatives
216
a) SIMPLE PRICE INDEX OR RELATIVE
This is a ratio of the (value) of one variable of a single commodity (price x quantity) in a given
year to its value in another year commonly referred to as base year.
A simple index number can be;
I. Price relatives
II. Quantity relatives
III. Value relatives
Case1. Price relative: This is the ratio of the price of a commodity in one period to the price in
another period.
 pn 
Price relative =   …expressed as a ratio
 p0 
 pn 
Price relative =   x100 …expressed as a percentage
 p0 
Case2. Quantity/ volume relatives: Index number expressed as a ratio of the quantity of a
single commodity in a given period to its quantity in another period
 qn 
Price relative =   …expressed as a ratio
 q0 
 qn 
Price relative =   x100 …expressed as a percentage
 q0 
Case3. Value relatives: Index numbers expressed as the ratio of the value of a given single
commodity (That’s to say; price x quantity) in a given period to the value in another period
called the base year
  Pn Qn 
I   x100
  P0Q0 
Example1 on the computation of price, quantity and value relatives
Year Price of beans / kg in Quantity of Beans in Kgs

Nakasero Market
2005 1,000 5,000
2006 1,200 3,800
2007 1,100 4,200
By taking 2005 as the base year, find

a) Price relative for both 2006 and 2007. (110%; 120%)
b) Quantity relative for both 2006 and 2007. (84%; 76%)
c) Value relative for 2006 and 2007 (92.4%; 91.2%)
b) SIMPLE AGGREGATE PRICE INDEX OR RELATIVE

This is a ratio of the sum of the commodities prices, quantities or values in a current year to
the corresponding value in the base year. Again, simple aggregate index numbers are also
divided into; 217
I. Simple aggregate price index
  Pn 
I   x100
  P0 
II. Simple aggregate volume/ quantity index number
  Qn 
I   x100
  Q0 
III. Simple aggregate value index numbers
  Pn Qn 
I   x100
  P0Q0 
Example on the computation of simple aggregate price index
Commodity Base Year Current Year

Price (P0) Quantity (Q0) Price (Pn) Quantity (Qn)
Blue band 2400 120 3200 160
Milk 1400 240 1800 200
Bread 1600 200 1000 320
Tea 4000 80 6000 80
Sugar 1200 400 800 600
Compute
a) Simple aggregate price index (121%)
b) Simple aggregate volume/ quantity index number (131%)
c) Simple aggregate value index numbers (123%)
C. SIMPLE AVERAGE OF RELATIVE PRICE INDEX. (MEAN OF RELATIVES INDEX)

This is an index number got by finding the average of the relatives of the commodities. The
simple average of relatives are divided into three parts
I. Simple average of price relatives

  Pn  
    
I   P0 
x100 
 n 
 
 
II. Simple average volume/quantity relatives
218
  Qn  
    
I   Q0 
x100 
 n 
 
 
III. Simple average of value relatives
  Pn Qn  1
I  x
  P0Q0  n
n is the number of commodities
The advantage here is that prices of all the commodities don’t have to be in the same units,
although the prices of each individual item must be in the same units. This is a useful measure
when we are dealing with results from different countries.
The major disadvantage of this index is that the relative importance of the items included is
not considered, i.e. items are not weighted in accordance with their relative importance.
Example 3 on the computation of average relatives
Commodity Base Year Current Year

A 500 80 800 70
B 420 50 450 100
C 300 120 210 180
D 900 75 750 140
Compute
a) Price average relative (1.05)
b) Volume average relative (1.58)
c) Value average relative (0.37)
Note: A simple aggregate index is easy to construct and easy to understand but it does not
meet a criterion of adequacy called the UNITS TEST. The test requires that index yield the
same results regardless of the units for which the prices of items are quoted. We can avoid
this problem by using the RATIO OF PRICES of a given item rather than the actual prices
themselves. See previous example
Fixed base index numbers

In fixed base method, a particular year is generally chosen arbitrarily and the prices of the
subsequent years are expressed as relatives of the prices of the base year. Sometimes instead
of choosing a single year as the base, a period of a few years is chosen and the average price
of this period is taken as the base year’s price. The year which is selected as a base should be
normal year or in other words, the price level in this year should neither be abnormally low nor
abnormally high. If an abnormal year is chosen as the base, the price relatives of the current
year calculated on its basis would give misleading conclusions. For example, a year in which
war was at its peak, say the year 1965, is chosen as a base year, the comparison of the price
level of the subsequent years to the prices of 1965 is bound to give misleading conclusions. 219
The reason is that the price level in the year 1965 was abnormally high. In order to remove
this difficulty associated with the selection of a normal year, the average price of a few years is
sometimes taken as the base price. The fixed base method is used by the Government in the
calculation of national index numbers. This is what we have been using in the previous
numbers
Chain based method

In this method, there is no fixed base period. The year immediately preceding the one for
which price index has to be calculated is assumed as the base year. Thus, for the year1994 the
base year would be 1993, for 1993 it would be 1992 for 1992 it would be 1991 and so on. In
his way there is no fixed base. It goes on changing. The chief advantage of this method is that
the price relatives of a year can be compared with the price level of the immediately preceding
year. Business-men mostly interested in comparison of this type rather than in comparison
relating to distant past. Yet another advantage of the chain base method is that it is possible
to include new items in an index number or to delete old times which are no more important.
In fixed base method it is not possible. But chain base method has drawback that comparison
cannot be made over a long period
In Chain Base,
Link relative of current years
or
Example: Find index numbers for the following data taking 1980 as base year.
Years
Price
Solution
Link Relatives
Year Price Chain Indices
220
Example: Illustration of computing index numbers
Commodities 1961 1962 1963 1964 1965

A 100 (9/10)*100=90 (8/9)*100=89 (9/8)*100=112.5 (11/9)*100=122
B 100 (5/4)*100=125 (3/5)*100=60 (6/3)*100=200 (4/6)*100=67
C 100 (8/6)*100=133 …=62.5 =80 =150
D 100 (13/11)*100=118 …=115 =93 =86
E 100 (8/6)*100=133 …=125 =140 =86
Total of link 500 599 451.5 625.5 511
relative
Mean of link 100 120 90 125 102
relative
Chain relative 100 120 108 135 137.7
II- WEIGHTED INDEX NUMBERS

These are index numbers that take into consideration the relative importance of the
commodities for which they are being computed. These are categorized into two types:
Case1. The weighted price index

These are index numbers that show the relative changes in the commodity prices. They
include;
a) Weighted index
b) Weighted average price index
c) Laspeyre’s price index
d) Paasche’s price index
e) Edeworth- Marshall’s price index
f) Fisher’s ideal index
g) Dorbish- Bowley’s index
a) A weighted index
Example: The table below represents the weights and indexes of five items
Item Weight (W) Index (I)

Food 304 124
Tobacco & alcohol 129 126
Housing 331 127
Transport 120 119
Goods & services 116 128
221
Find: 1) the index number of all items [ANS=125.115] 2) index number of nonfood items
[ANS=125.60] 3) index number of all food items [ANS=124]
Note: For such a number, use the formulae;
Index 
WI
W
Solution
Item Weight (W) Index (I) WI
Food 304 124 …
Tobacco & alcohol 129 126 …
Housing 331 127 …
Transport 120 119 ….
Goods & services 116 128 …
Total 125,115
IndexNo 
 WI  ... 
 W ...
Example: Given the following information: construct a cost of living index
Group Group Index (GI) No per 1965 Weight

with 1960 as the base year
Food 152 48
Fuel & lighting 110 5
Clothing 130 15
House rent 100 12
Miscellaneous 80 20
Solution
Group Group Index (GI) No per 1965 Weight GI x W

with 1960 as the base year
Food 152 48 …
Fuel & lighting 110 5 …
Clothing 130 15 …
House rent 100 12 …
Miscellaneous 80 20 …
Total 100 12596
CostofLivi ng 
 GroupIndexXweight  ...  125.96
W ...
222
b) WEIGHTED AVERAGE PRICE INDEX
This is computed using weights of any typical year and not necessarily the base year or current
year.
  Pn wi 
Weighted average price index = I    x100
  P0 wi 
Where; wi is the weights of the ith commodity
Example 3 on the computation of weighted average price index
Commodity Base year Price Current year Commodity

(P0) Price (P0) weights
A 500 800 70
B 420 450 100
C 300 210 180
D 900 750 140
Compute
The weighted average price index (148.2%)
c) LASPEYRES PRICE INDEX OR BASE WEIGHTED AGGREGATIVE INDEX.

If we use quantities or values consumed in the base period/year as weights then it is called a
Laspeyres price Index OR base weighted aggregative index. Developed after a scientist called
Etienne Laspeyres
Note: The disadvantages of simple aggregative index can be avoided by constructing a

weighted aggregative index. Here, the price of each.
Item is weighted in terms of the importance of each item in the index.
Steps involved.
Step 1: Assign a weight to the price of each item by multiplying the price by the weight
selected.
Step 2: Add these products to get the weighted aggregates.
Step 3: Divide each of these totals by the weighted aggregates of the year selected as base.
Use the formula:
  PnQ0 
I   x100 Where Qo is the quantity consumed in the base period and other
  P0Q0 
variables as defined before.
Note: This kind of index reflects changes in prices alone – the same quantities of goods are
priced at two different terms.
d) PAASCHE’S PRICE INDEX OR CURRENT WEIGHTED AGGREGATIVE INDEX.

We can use given year quantities to weigh both the base year prices and the given year prices
and construct a weighted aggregative index with given year weights.
This is called the Paasche’s price Index. OR current weighted aggregative index.
223
  PnQn 
I   x100
  P0Qn 
Example 4 on the computation of the Laspeyres and Paasche price Index
Commodity Base Year (2011) Current Year (2012)

A 800 15 1400 15
B 350 25 200 27
C 2000 12 1500 15
D 3000 18 2000 20
Compute
a) The Laspeyres price index (81.01%)
b) Paashe’s price index (79.77%)
It can be noted that the Laspeyres method results in a more logical answer as compared to the
simple index. But it should be noted that this method uses the amounts consumed in the base
period as weights. Therefore, it assumes that the eating habits of a typical consumer didn’t
change from 2011 to 2012. Thus only price fluctuated casing a decrease in the index from 100
in the base period to 81.01% in the 2012. Another major Disadvantage of the Laspeyres index
is that the weights become out of date as the pattern of demand changes. This index tends to
overstate the changes in prices as it takes no account of the fall in consumption when prices
rise.
Interpretation:
A market basket of food selected in 2012 costs 20.23% less than the same basket of foods at
2011 prices.
This method has an advantage of using current consumption figures as weights; that is the
consumption pattern is always up to date. However, this method poses a practical problem as
it needs to be revised every year using new present year weights. This consumes a lot of time
and involves expensive survey work. It tends to understate the changes in prices as most
people tend to buy less of those commodities which have gone up in prices. Because of the
above short comings, the Laspeyres method is most commonly used when assessing relative
changes in price levels/ computation of index numbers (with some modifications).
e) EDEWORTH- MARSHALL’S PRICE INDEX

This is a weighted index number computed using the sum of quantities in the base year and
the current year as the weights
That is to say; W = (Q0 + Qn)
  Pn (Qn  Qn ) 
Edeworth- marshall’s price index = I    x100
  P0 (Qn  Qn ) 
Example 4 on the computation of the Edeworth- marshall’s price index

224
Price Quantity Price Quantity Pn(Q0+Qn) P0(Q0+Qn)
(P0) (Q0) (Pn) (Qn)
A 800 15 1400 15 42000 24000
B 350 25 200 27 10400 18200
C 2000 12 1500 15 40500 54000
D 3000 18 2000 20 60000 90000
152900 182200
Compute
The Edeworth- marshall’s price index
  Pn (Qn  Qn )  152900
I   x100   83.92%
  P0 (Qn  Qn )  182200
e. THE FISHER’S IDEAL INDEX.

In order for an index number to be acceptable or not, it should satisfy the Factor Reversal test
which states that if the prices and quantities used in an index number are exchanged, then the
product of the two index numbers should be an index of the two index numbers. In practice,
none of the commonly used index numbers satisfies this test the fisher’s ideal index satisfies
this test.
FISHER’S IDEAL PRICE INDEX = ( PAASCHE) x( LASPEYRES) x100
FISHER’S IDEAL PRICE INDEX = (81) x(79.77)  80.38%
f) DORBISH- BOWLEY’S INDEX

This is the arithmetic mean of Laspeyre’s and Paasche’s price indices
  Pn Q0    Pn Qn  1
 x  81x79.77 x
1
DORBISH- BOWLEY’S INDEX= I   
  P0Q0    P0Qn  2 2
DORBISH- BOWLEY’S INDEX = 80.39%
18.4 PUBLISHED INDEX NUMBERS

There are two major types of published index numbers.
A. THE REAL INCOME (WAGE) INDEX NUMBER

Real income / wage index = Real income/wage of the current year x 100%
Real income/wage of the base year
However, one has to note that in inflationary situations, real income/ wage has to be deflated
in order to make an allowance for the effect of changing price levels. In deflating real income, 225
we shall obtain the deflated income and this shall be taken as the new real income for index
number computation purposes.
B. COST OF LIVING INDEX NUMBER

Cost of living index numbers also known as consumer price index (CPI) are special purpose
index numbers that are constructed to measure relative changes in the cost of living.
  Pn w 
Cost of living index= I    x100
  P0 w 
These index numbers are generally constructed for each week. Where the average of the
weekly index numbers is taken as the index number for the month and the average of the
monthly index numbers gives the cost of living for the whole year.
18.5 SHIFTING THE BASE YEAR

The base year of index numbers can be shifted in the following circumstances
a) When comparisons need to be made between index numbers having different base
periods or years
b) If the base year is too distant in the past and hence no longer useful for comparison
purposes
In such circumstances, the new index number known is the re cast index number computed
as;
Recast index number = Old index number of that year x 100%
Index number of the new base year
19.6 ADVANTAGES / IMPORTANCE OF USING INDEX NUMBERS

a) They provide a comprehensive way of comparing changes in numerical series from one
period to another.
b) Provide a simple way of comparing overall changes in different variables though such
variables may be reported in different units.
c) They allow for construction of composite series, for measuring changes in some
economic transaction such as price, volume or value.
d) Converting data into indexes makes it easier to assess the trend in a series composed
of exceptionally large numbers.
e) An index number is the most convenient way of expressing a change in a
heterogeneous / different group of items. The consumer price index for example
encompasses over 400 items. Only the conversion of these many diverse goods to one
index number enables the government and other concerned authorities informed about
the overall movement in prices.
f) They are used in deflating nominal values to obtain real ones. E.g. if today’s dollar will
buy 100/130 of 77% of what it could buy in the base year. If the average weekly
earnings were $115 in the base year and are now $125 in current period, then today’s
real earnings per week is 125 x 77% = $96.25, which is $18.75 less than in base year.
226
19.7 IMPORTANCE OF COST OF LIVING INDEX NUMBERS
a) They are purposely used in determining remunerations so as to maintain the same
standard of living in other years as that of the base year
b) Their reciprocal is used in measuring the purchasing power of money
c) They are also used to find real wages by the process of deflation
d) They give guidelines to the government on its decisions about the income and general
economic policy, wage policy and price policy among others
19.8 BASIC PROBLEMS WHEN CONSTRUCTING INDEX NUMBERS

a) Availability and comparability of data. This is a very serious problem when constructing
index numbers. Assume we want to make a comparison of commodity prices in 2004
with those of 1928. Here commodities like mobile phones, computers and DVDs would
be ruled out since they were not available in 1928. This makes the construction of an
index difficult and biased.
b) The selection of items to be included in the comparison. It is physically impossible to

include all commodities and all services when you are constructing a general-purpose
index. So how do you judge that this should be included and not this? Here we just
take samples in such a way that items included adequately reflect the overall
phenomenon being described.
c) The choice of time periods that are to be compared. Here, we choose a relatively
recent period of good economic activity, e.g. if we are to compare prices of goods from
1982 to 1992: 1982 acts as our base year and 1992 acts as our given period.
d) The choice of appropriate weights. The choice of the relative importance of each item
is very difficult. Here we weigh the prices of items in such a way as to account for their
relative importance in practice.
e) The choice of a sustainable formula. There are various ways in which relative changes
in prices of quantities can be described. However, choice on which is best is not easy.
19.9 LIMITATIONS OF INDEX NUMBERS

a) Index numbers are computed using many methods, which may give different results.
This at times creates confusion
b) Most index numbers are computed from sample data. this increases the risk of bias and
reduces on the accuracy of index numbers
c) Index numbers can be misused in order to draw desired conclusions. This means that
computed index numbers are as good as the person computing them.
d) Some index numbers are based on selected items. They simply depict the broad trend
and not the complete reality
HAPPY HOUR: NOW TRY OUT THESE MODEL EXAMINATION NUMBERS

Question-1
a) Explain the difference between a fixed base index number and a chain base index
number
b) Discuss three problems often encountered when computing index numbers
c) Given the following sales in a week for Beyonce enterprises
227
Weeks 1 2 3 4 1 2 3 4
Sales per week 20 35 50 30 35 25 100 50
Compute the chain indices for the above data
d) Uganda National Bureau of Statistics recorded the prices of the following commodities in
rural collection centres as:
I) Using 2001 as the base year, the Laspeyre’s Price index number for 2005 is 199%. Find
the value of x to the nearest kilogram.
II) Calculate Paasche’s Price index number and comment on the answer
III) Compute the Laspeyers Quantity index number and interpret your answer.
Question 2
a) Define an index number and explain the factors that must be considered when one is
constructing index numbers.
b) How does a price relative differ from a value relative?
c) Given the following information,

A 800 15 1400 15
B 350 25 200 27
C 2000 12 X 15
D 3000 18 2000 20
I) If the Laspeyres price index number was computed as 81.01%, compute the Price of
commodity C in the new year (missing value X)
II) Compute a Simple average of value relatives
III) Compute a Paashe’s quantity index
Question 3
a) Write short notes on the following:
I) Cost of living index number
II) Weighted index numbers
III) Fixed base index numbers
IV) The chain based method
b) Study the following table and answer the following questions
228
I) If the ratio between Laspeyre’s and Paasches’ index numbers is 28:27, find the
missing figure on the table.
II) Now, find the Pasche Quantity Index and Interpret your results
III) Compute the value relative Index number
Question 4
a) Distinguish between fixed-base and chain-base index numbers in time series.
b) The table below shows the average monthly wage (Shs ‘000’) for workers in each grade in
2000 and 2002 for a certain factory:
Required:
I) Obtain the values of x, y and z.
II) Obtain an index number for the monthly wage of the whole factory in 2002.
c) A recent survey carried out by a researcher on the prices of various goods and services
and the spending of an average person in Mbarara Town revealed the data indicated in
the table below:
Required:
I) Calculate the cost of living index in Mbarara Town, giving the answer to 1 decimal place.
II) Comment on the answer in (d) (i) above.
Question 5
a) List any four (4) basic problems encountered when constructing Index numbers
b) Explain the distinction between the following terms:
I. A price Index number and Quantity Index Number
II. A weighted and Un-weighted Index number
III. A fixed base and chain based index number
c) Given the following information,

229
A 800 15 1400 15
B 350 25 200 27
C 2000 12 1500 15
D 3000 18 2000 20
Compute:-
I. The Laspeyres price index
II. The Laspeyres Quantity index
III. Paashe’s quantity index
IV. Fishers ideal price Index
Question 6
a) Explain the importance of “weighting” in index number construction.
b) The following table shows the quantities of materials (steel, cement and timber) used
in the construction industry in 1995 and 2000, together with the proportions of total
value in 1995.
Quantities used Proportion of total value

1995 2000
Materials Qo Qn
Steel(Million tons) 150 120 0.42
Cement(million tons) 110 100 0.08
Timber(million tons) 60 62 0.50
I. Calculate the simple (Unweighted) aggregate quantity index for these materials in
2000, using 1995 as the base year.
II. Calculate an (Unweighted) arithmetic mean of the relatives, quantity index for 2000
using 1995 as the base year.
III. Using the proportions of the total value as weights, calculate the weighted arithmetic
mean of relatives’ quantity index for 2000 with 1995 as the base year.
230
Question 7
a) The following table is a summary of consumption patterns of a number of residents
from Bugolobi suburb for two periods.
Expenditure Prices (Shs)

(Shs)
2004 2003 2004
Sugar Kilos 100,000 1,200 1,400
Clothing Meters 180,000 12,000 18,000
Salt Kilos 10,000 200
150
Soap Bars 20,000 900
800
Matooke Bunches 80,000 4,500 3,000
Milk Litres 40,000 700
800
Note:
Quantity is a ratio of expenditure in a given year to price per unit in the same year. Construct
a suitable index number of prices from the above information concerning consumption patterns
for the above residents from 2003 to 2004.
NOW THAT YOU ARE HAPPY, TRY OUT THESE PAST PAPER EXAM NUMBERS
Sitting Number Sitting Number

June 2016 Q4a November 2013 7b
November 2015 3ab June 2013 2b,c
June 2015 None November 2012 4c
November 2014 4cd June 2012 5c
June 2014 4a December 2011 3a,b,c
231
SYLLABUS FROM THE CPA GUIDE
A. INTRODUCTION o-give, Lorenz and Z charts)
1. Meaning of statistics (viii) Basic principles of drawing each type of graph;
(a) Functions and limitations of statistics interpretation and
(b) Types of statistics: descriptive and inference statistics limitations of graphs
B. STATISTICAL DATA, PRESENTATION AND MEASURES 2. Measures of location / central tendency
1. Statistical data (a) Measures of averages: arithmetic mean,
(a) Primary and secondary weighted mean, mode, median,
(i) Sample and population harmonic and geometric means
(ii) Discrete and continuous (b) Characteristics, merits and demerits of each type
(iii) Quantitative and qualitative of average
(iv) Variable and attribute (c) Computation of the mean using actual values
(b) Sources of data and /or working mean
(c) Methods of data collection: interview, questionnaire, (d) Determination of the mode and median using
direct observation, census both computational and graphical
and sample survey methods
(d) Merits and demerits of each method of data (e) Interpretation of calculated values of each
collection measure
(e) Response errors (f) Computation of weighted mean, harmonic mean
(f) Sample frame and geometric mean
(g) Sampling techniques: random sampling (simple (g) Relationship between mean, mode and median
random and stratified), quasi Note: Knowledge of logarithms either from
random sampling (systematic and multistage sampling), mathematical tables or calculator
non- random sampling is assumed.
(cluster and quota sampling) 3. Measures of dispersion
(h) Advantages and disadvantages of each technique (a) Meaning of dispersion and variation
of sampling (b) Measures of dispersion: range, quartile deviation
(i) Methods of data presentation: (interquartile range), semi quartile
(i) Tables (row and column, two-way, frequency and deviation (semi-interquartile range), decile and
percentage distribution percentile range, mean
tables) deviation, quartile coefficient of dispersion, standard
(ii) Basic principles of table design deviation and variance
(iii) Interpretation of data presented in each type of (c) Characteristics, merits and demerits of each
table measure of dispersion
(iv) Limitations of each type of table (d) Determination of quartile, decile and percentile
(v) Pictograms, charts (simple and compound bar by computation and graphically
charts, and pie charts) (e) Computation of quartile coefficient of dispersion
(vi) Methods of construction of each type of chart; type (f) Computation of mean deviation, standard
of data suitable for deviation and variance using the
each type of chart; limitations of each type of chart working mean method
(vii) Graphs (line graph, histogram, frequency polygon (g) Properties of standard deviation and coefficient
and curves, including of variation
(h) Interpretation of calculated values of each
measure
4. Measures of skewness
(a) Positive, symmetric and negative skewness
(b) Relationship among mean, mode and median in
a skewed distribution: 3
(mean – median) = (mean – mode)
(c) Computation of measures of skewness using Karl
Pearson’s and Bowley’s
formulae
(d) Graphical illustration of the concept of skewness
and kurtosis
(e) Comparison of skewness with measures of central
tendency and dispersion
(f) Interpretation of skewness and degree of
skewness
232
C. PROBABILITY AND DISTRIBUTIONS critical (acceptance and rejection) regions, and level of
1. Probability theory significance
(a) Basic concepts of probability: event, outcome, sample, sample (b) Methods of testing (normal Z score, Student’s t and chi-
space and equiprobable square)
(b) Classical definition of probability (c) Reading values from tables (one tailed and two tailed)
(c) Properties of probability theory: (probability limits, total (d) Computation of the statistic and determination of the level
probability and complimentary probability); types of events of significance at a
(independent, dependent, and particular percentage
mutually exclusive events) (e) Testing of population mean and population proportion
(d) Rules used in probability theory that is addition and (f) Testing goodness of fit, use of contingency tables, and
multiplication computation of expected
(e) Conditional probabilities (prior and posterior, including use of values for chi-square test
probability trees and Bayes’ theorem) as methods of computation 3. Control charts
(f) Probability frequency distribution and the computation of (a) Definition of the concept ‘control chart’
expected values (expectation) (b) Advantages and disadvantages of control charts
Note: Application of the knowledge of set theory in probability is (c) Drawing control charts
assumed. (d) Interpretation and making conclusions from control charts
2. Permutations and combinations E. INDEX NUMBERS
(a) The factorial notation 1. Meaning, uses, importance and limitations of index numbers
(b) Application of permutations and combinations in probability 2. Price and quantity relatives
3. Probability distribution 3. Considerations in the construction of index numbers
(a) The concept of probability distribution 4. Types of index numbers: simple and weighted (price and
(b) Normal distribution – properties and standardization quantity)
(c) Computation of probabilities using normal distribution tables 5. Computation, comparison and interpretation of Laspeyres
(d) Binomial distribution probability properties and use of binomial and Paasche price and
formula quantity indices
(e) Computation of the mean, variance and standard deviation of 6. Time series relatives: fixed base and chain relatives
a binomial distribution 7. Published indices: cost of living, consumer, stock and retail
(f) Use of binomial distribution tables to find probabilities indices
(g) Approximating a binomial distribution to a normal distribution F. REGRESSION AND CORRELATION
(h) Properties and computation of probabilities using poisson 1. Correlation
distribution (a) Definition and importance of correlation analysis
D. ESTIMATION AND HYPOTHESIS TESTING (b) Types of correlation: linear or non-linear, positive, negative,
1. Estimation zero and no
(a) Definition of concepts: estimation, estimator (point estimator correlation
and interval (c) Scatter diagrams
estimator), standard error of the mean, and population and (d) Methods of calculating coefficient of correlation: product
sample mean moment (Karl
(b) Confidence limits and intervals Pearson’s) and rank correlation (Spearman’s formula).
(c) Estimation of the population mean from a large sample using (e) Interpretation of correlation co-efficient
normal distribution (f) Merits and demerits of each method
(d) Estimation of the mean from a small sample using Student’s t (g) Link between correlation and causation
distribution 2. Regression
(e) Use of contingency tables for chi-square distribution (a) Definition of terms: regression, regression equation,
(f) Estimation of the population proportion from a large sample regression coefficients
2. Hypothesis and significance tests (b) Distinction between correlation and regression
(a) Definition of basic concepts used in hypothesis testing: null (Ho) (c) Uses of regression line
and the alternative (Ha) hypotheses, errors in hypothesis testing (d) Definition of dependent and independent variables
(type (I) and type (II) errors), (e) Regression equation: y on x and x on y
(f) Calculation of regression equation using least squares
method
(g) Interpretation of regression line and prediction of other
values using the line
Note: This excludes non-linear regression and non-linear
correlation.
233
G. TIME SERIES AND FORECASTING (f) Definition of terms: cost and marginal cost, revenue and
1. Time series marginal revenue, and
(a) Definition of terms: forecasting, time series profit functions
(b) Examples of time series and their uses / (g) Application of differentiation in maximization and
importance minimization of revenue, cost
(c) Components of time series: secular trend, and profit functions in production functions including
seasonal variation, cyclic variation and irregular determination of their
variation values at any given level
(d) Seasonal adjustment and deflating time Note: This excludes logarithmic differentiation, exponential
series differentiation,
2. Methods of computing trend derivatives of parameters, implicit functions and partial
(a) Moving average method and exponential differentiation.
smoothing I. DECISION THEORY
(b) Computation of trend by least squares 1. Types of decision making
method 2. Decision rules and decision trees
(c) Graphical representation of data 3. Computation of expectation or expected values
(d) Estimation of production or output using a 4. Construction and interpretation of results from decision trees
trend line 5. Advantages and disadvantages of decision trees
(e) Limitations and merits of moving averages 6. Determination of redundancy
and least squares methods J. LINEAR PROGRAMMING
3. Forecasting 1. Definition of terms: linear programming problem, objective
(a) Steps involved in forecasting function, constraints,
(b) Methods of forecasting: quantitative and feasible solution and optimum solution
qualitative, regression analysis 2. Assumptions applied in linear programming
and depersonalization method, exponential 3. Advantages and limitations of linear programming
smoothing, interpolation and 4. Formulation of linear programming model
extrapolation and their interpretation 5. Graphical and simplex methods of solving linear
(c) Construction and uses of the Z chart programming problems
H. LINEAR ALGEBRA AND CALCULUS 6. Advantages and disadvantages of graphical and simplex
1. Algebra methods
(a) Linear, quadratic and simultaneous equations 7. Optimization (maximization and minimization) in linear
in 2 or 3 variables programming
(b) Methods of solving quadratic equations by 8. Terms associated with simplex method: slack, surplus,
factorization, completing the square shadow prices, primal, dual
and use of the quadratic formula and tableau
(c) Methods of solving simultaneous equations by 9. Solving primal and dual linear programming problems
elimination, substitution and 10. Interpretation of primal and dual solutions
matrix methods 11. Advantages of the dual method
(d) Formulating equations involving quadratic Note:
and simultaneous equations in 2 (i) This excludes non- linear programming, transportation,
variables or 3 variables assignment and nonlinear
2. Functions and graphs dynamic programming.
(a) Explicit functions in one variable of (ii) Basic knowledge of solving simple linear inequalities and
polynomial nature up to the third degree (i.e. interpretation of matrix
f(x) = a + bx + cx2+ dx3 +……….) representation is assumed.
(b) Graphical representation of functions K. NETWORK ANALYSIS
3. Calculus 1. Definition of terms: network, activity, event, dummy variable, critical
(a) Geometric interpretation of differentiation path, float, cost
slopes, dangler and lead time
(b) Basic rule of differentiation: y = xn ,
2. Drawing of networks and Gantt charts and their interpretation
Dx/dy = nx n-1 3. Advantages and limitations of network analysis
(c) Differentiation methods: sum, difference, 4. Determination of earliest start time (EST), latest start time (LST), and
chain rule (substitution), product total cost
and quotient rule of the functions (including 5. Techniques of network analysis: PERT (program evaluation and
differentiation of fractional and review technique) and
negative indices) CPM (critical path method)
(d) Finding a second derivative 6. Crashing of projects, cost analysis
7. Resource and cost scheduling
(e) Application of differentiation (or derivatives)
8. Use of computers in network analysis
in determining maxima and minima
values of functions
234
END AND
GOOD LUCK
235

Paper2-QT-2022 Lecture Notes-Hassan PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Paper2-QT-2022 Lecture Notes-Hassan PDF

Uploaded by

Copyright:

Available Formats

THE INSTITUTE OF

A. DESCRIPTIVE STATISTICS This is mainly concerned with collecting and summarizing

1.2 VARIABLES AND DATA

1.3 METHODS OF COLLECTING DATA

METHODS OF COLLECTING PRIMARY DATA

1 Personal interviews may be structured, semi- structured or unstructured

In a SEMI-STRUCTURED INTERVIEW, the interviewer has a set of prepared questions, but

In UNSTRUCTURED INTERVIEWS, the interviewer does not have a set of prepared

THE QUESTIONNAIRE METHOD

ADVANTAGES OF A MAILED QUESTIONNAIRE METHOD

DISADVANTAGES OF A MAILED QUESTIONNAIRE

WHEN ENUMERATORS ARE TASKED TO DO THE INTERVIEW

PERSONAL OBSERVATION METHOD OF DATA COLLECTION

METHODS OF COLLECTING SECONDARY DATA

EXTERNAL DATA SOURCES

2.2 REASONS FOR SAMPLING

Example on proportionate Stratified Random Sampling

Example: Ministry of Education Researchers used a multi-stage sampling design to survey

CLUSTER /AREA SAMPLING

Step-by-step Quota Sampling

When to Use Quota Samples

OBJECTIVES OF DATA CLASSIFICATION

A. ORGANIZING/ CLASSIFYING QUANTITATIVE DATA

EXAMPLE OF A DATA ARRAY (AGES OF 50 WORKERS FROM MTN)

11-19 20-29 30-39 40-49 Etc

Class boundaries (Real class limits / mathematical limits)

Class mark (class midpoint)

From the above table it can be noted that:-

Cumulative RELATIVE frequency distribution

A. TEXT PRESENTATION (data presented in wordy form). This is when data is

B. TABULAR PRESENTATION. This is a process of summarizing raw data and displaying it

Row – column table

A two way table

Frequency and percentage tables

THE COMPONENT/ SUB-DIVIDED BAR CHART

THE MULTIPLE BAR CHART

0% 20% 40% 60% 80% 100%

Paper-4, 102 Paper-3, 93

D. GRAPHICAL PRESENTATION OF DATA.

Paper5 Test Marks Mid-Point (X) Frequency

A histogram representing test marks

Time (seconds) 16 ≤ t < 18 18 ≤ t < 19 19 ≤ t < 20 20 ≤ t < 22 22 ≤ t < 26

a) Draw a histogram to illustrate the data

THE FREQUENCY POLYGON GRAPH

THE CUMULATIVE FREQUENCY CURVE (OGIVE)

The cumulative freq distribution of statistics results

The more than Ogive

Temperature in Kampala city

This Total since Rolling 12

Read about the Lorenz chart

I. Construct a histogram and estimate the mode

A Parameter and a Statistic: A parameter is a number describing something about a

4.2 THE ARITHMETIC MEAN

Salaries f Mid-point Deviation f .d

4.4 THE WEIGHTED MEAN: Is an average in which each quantity to be averaged is

4.5 GEOMETRIC MEAN (GM): Is a specialized measure for averaging proportional

ADVANTAGES OF THE ARITHMETIC MEAN

DISADVANTAGES OF THE ARITHMETIC MEAN

4.7 PROPERTIES OF ARITHMETIC MEAN

4.8 CALCULATING THE CORRECT MEAN FROM WRONG MEAN

4.9 THE MEDIAN

Interpolation is the mathematical technique for estimating an unknown value by utilizing

Example: Compute the median for the following marks