3 Ebook Research Methodology & Data Analysis

RESEARCH
METHODOLOGY
and
DATA ANALYSIS
2ndEdition
Zainudin Hj Awang
Universiti Sultan Zainal Abidin (UniSZA)

Terengganu Malaysia
E-Book Disclaimer Contents
This e-book is a sincere gift from the author. No
portion of this book may be reproduced or resales in
any form without permission from the author,
Professor Dr Zainudin Awang.
All rights reserved.

82research
RESEARCH METHODOLOGY AND DATA ANALYSIS 2ND EDITION
Author : Zainudin Awang

ISBN No : 978-967-363-422-4
Pages : 334
Price : RM 35.00
1/2
82research
2/2
Preface
Research Methodology and Data Analysis offers a comprehensive guide for

doing research project from the beginning to the end. The syllabus includes the
rationale for research, and the step by step procedure in writing a research
proposal. The book guides in forming the three most important elements in any
research project namely the research objectives, research questions, and
research hypotheses. These three elements formed the direction of the study.
The book also lays the selection of a proper research design including the
methods of sampling, methods of data collection, and methods of data analysis.
For its data analysis, this book introduces the second generation of statistical
analysis namely The Structural Equation Modeling (SEM). The author begins
data analysis by going through data mining, descriptive analysis, factor analysis,
and reliability analysis. Besides this, the book assesses the normality of the data
for each variable since this is crucial in determining the validity of any statistical
analysis employed later. Finally, the book provides proper guide to analyze
correlational and causal effects, the mediating or intervening effects, and also the
moderation effects among variables. More importantly, both types of variables
are discussed and analyzed namely “observed variables” and “latent constructs”.
The analysis to measure the effects of correlational & causal, the role of mediator
variable and the significance of moderator are carried out in SEM using AMOS
21.0.
Zainudin Bin Awang, PhD
i
Contents
1 INTRODUCTION TO RESEARCH 1
1
What is Research?
What is the Importance of Research? 1
The Needs Research 2
The Role of Research in Business 4
The Determinant for Conducting Research Project 4
The Types of Research 7
The Characteristics of Good Research 12
The Basic Ethics in Research 14
Exercises 14
2 THE RESEARCH PROCESS 19
Introduction
Process 1: Establishing the Problem for Research 18
Process 2: Identifying the Appropriate Research Variables 24
Process 3: Forming Research Objectives, Research Questions, and 25
Research Hypotheses
Process 4: Developing the Theoretical Framework for the Study 25
Process 5: Reviewing Literatures Related to the Problem 28
Process 6: Selecting an Appropriate Research Design 38
Process 7: Collecting Relevant Data 46
Process 8: Analyzing Data using Appropriate Statistical Methods 48
Process 9: Prepare the Final Report to the Management 51
Exercises 50
3 WRITING THE RESEARCH PROPOSAL 58

58
What is Research Proposal?
The Chapters and Sub Chapters in any Research Proposal 59
The Step by Step Guide in Writing a Research Proposal 60
Chapter 1: Introduction
ii
1.1 The Introductory Paragraph 60
1.2 The Problem Statement 60
1.3 The Objectives of the Study 62
1.4 The Research Questions 62
1.5 The Research Hypotheses 63
1.6 The Significance of the Study 67
Chapter 2: The Review of Literature Related to the Study

2.1 The Literature Review 68
2.2 The Definition of Terms 68
2.3 The Theorized Relationship among Variables in the Study 69
2.4 The Theoretical Framework of the Study 70
Chapter 3: The Research Methodology

3.1 The Target Population 72
3.2 The Sampling and Data Collection Procedure 73
3.3 The Measuring Instruments Employed in the Study 76
3.4 The Assumptions Required 76
3.5 The Procedure for Data Analysis 77
3.6 The Scope and Limitations of the Study 78
3.7 The Research Schedule 80
Exercises 81
4 THE SAMPLING DESIGN 82

82
Basic Sampling Terminology
Choosing the Appropriate Sampling Design 84
The Theoretical Tools to Assess the Quality of Sample 86
The Important Factors to Determine the Appropriate Sample Size 88
The Types of Sampling Design 88
Types of Probability Sampling Design 89
Types of Non-probability Sampling Design 95
Exercises 98
5 DATA, MEASUREMENT, AND SCALES 105

105
Types of Data Sources
The Primary Data Sources 105
The Secondary Data Sources 106
Types of Data 107
iii
The Qualitative Data 107
The Quantitative Data 109
Methods of Data Collection 110
Types and Levels of Measurement 112
The Comparison among Four Types of Scales 117
The Classification of Scaling 119
The Categorical Scale (Voting Scale) 119
The Comparison Scale (Ranking Scale) 124
Exercises 127
6 THE PRIMARY DATA SOURCES 136

136
Nature of Primary Data
Primary Data Collection Method 137
Types of Survey Method 140
The Personal Interview 140
The Observation Method 147
Exercises 150
7 THE SECONDARY DATA SOURCES 154

154
Nature of Secondary Data
The Usage of Secondary Data in Research 154
The Sources of Secondary Data 155
Procedure for Collection of Secondary Data 157
Evaluating the Quality of Secondary Data 162
Exercises 165
8 DESIGNING A QUESTIONNAIRE 167

167
What is Questionnaire?
The Purpose of Questionnaire 167
The Criteria of Good Questionnaire 168
The Process of Designing a Questionnaire 169
A Sample of a Complete Questionnaire 1 185
A Sample of a Complete Questionnaire 2 193
Exercises 199
iv
9 DATA MANAGEMENT 206
206
Data Management Process
The Process to Code a Question 212
Coding the Structured or Close-Ended Question 212
Coding the Unstructured or Open-Ended Question 215
The Codebook 217
Exercises 219
10 THE DATA ANALYSIS 223

223
Steps in Data Analysis Process
Step 1: Data Mining Procedure 223
Step 2: Descriptive Analysis for Demographic Variables 223
Running Data Reduction Procedure: The Exploratory Factor Analysis 229
The Reliability Analysis for the Measuring Items 234
The Analysis to Determine the Normality of the Data 235
Parametric Statistical Analysis 236
Testing the Hypothesis for Mean Differences between two 237
Groups: The Independent two sample t-test
Testing the Hypothesis for Mean Differences for More than two Groups: 238
The one way Analysis of Variance ANOVA
Testing the Inter-Relationship among Variables in the Study: The 241
Structural Equation Modeling (SEM)
Converting a Theoretical Framework into AMOS Syntax 245
Definition of Variables in AMOS Syntax 247
AMOS Output: The Standardized Model 249
AMOS Output: The Unstandardized Model 252
Testing for the Main Hypotheses in the Study 255
Exercises 260
11 STRUCTURAL EQUATION MODELING (SEM) 268

268
What is Structural Equation Modeling?
What is the Rationale for Employing SEM? 269
Modeling the Relationship between Constructs 269
Modeling the Causal Relationship 269
Modeling the Correlational Relationship 271
Modeling Complex Relationship among the Constructs 272
v
Modeling the Multiple Regression for Latent Constructs 272
Modeling the Mediator Variable in a Model 274
Modeling the Moderator Variable in a Model 275
Analyzing Structural Equation Modeling using AMOS Software 275
Analyzing the Measurement Model 276
Evaluating the Fitness of a Measurement and Structural Model 278
The Steps Involved in Validating the Measurement Model 281
Running the Confirmatory Factor Analysis 281
The Schematic Diagram of a Model 282
The Measurement Model Combining all Constructs 282
Assessing the Fitness of a Measurement Model 286
The Assessment of Normality for a Dataset 291
Analyzing the Structural Model 292
Interpreting AMOS Output for a Structural Model 296
Interpreting the output: The Standardized Regression Weights 296
Interpreting the output: The Unstandardized Regression Weights 298
12 ANALYZING THE MEDIATING VARIABLE 303

303
What is Mediating Variable?
Analyzing the Mediating Effects for the Observed Variables 304
Analyzing the Mediating Effects in for Latent Constructs 307
13 ANALYZING THE MODERATING VARIABLE 311

311
What is Moderating Variable?
The Interaction Effects of Moderating Variable 312
Modeling the Moderating Effects for Observed Variables 314
Analyzing the Moderating Effects for Observed Variables 315
Modeling the Moderating Effects for Latent Constructs 318
Analyzing the Moderator for Latent Constructs: The Multi-Group 319
CFA
Comparing the Group Effect for Moderator Variable 336
SUGGESTED SOLUTION FOR EXERCISES

Solution for Exercises in Chapter 1 339
vi
Bibliography
Index
vii
Chapter 1: Introduction to Research
CHAPTER INTRODUCTION TO RESEARCH

1
WHAT IS RESEARCH?
Research is a process of searching information concerning the underlying

problem. The research process involves identifying the problem thoroughly and
objectively, collecting and analyzing the relevant data in order to determine the
possible factors causing the problem. Thus, research activities are consistent
search for information with the objective to get a clearer picture concerning the
causes of the problem and to propose specific recommendations for the solution.
Why is research required? Is it because there is a “problem” to be solved?

Meaning: If there is no problem, then the research is not required? Not
necessarily since there is always a “problem” which requires research and further
research to be carried out. The problem could be:
a. The existing problem in the organization (declining sales, high complaints)
b. The anticipated problem in the future due to changes in lifestyles and

changes in socio-economic status would change the demand pattern
c. The theoretical gaps or knowledge gaps. Findings from previous research

might no longer hold due to lapse of time, thus new research is required.
WHAT IS THE IMPORTANCE OF RESEARCH?
Research activities provide the knowledge and skills needed for the manager in
the fast-paced decision-making environments. Research helps in decision-
making because the analysis of data provides the clear picture to the manager
concerning to the underlying problem, the associated factors, and offers the
possible solution. The researcher interprets the results from the analyzed data
and makes a thorough evaluation on the benefits and consequences for every
possible solution. The researcher also proposes certain recommendations for the
next course of action to be taken by the manager. Thus, the manager is able to
make a better decision with the help of research.
Prof Dr Zainudin Awang 1

UniSZA
Research provides valuable information that guides manager, employees and

other personnel of an organization in their daily work for the organization. The
information from research guides them to make an effective decision-making
which will benefit the organization in the long run. Decisions made blindly
(without prior research) could damage the organization in the future.
As an example, a research is required to identify the factors that cause the

demand for Malaysian made cars (Proton and Perodua) to decline drastically in
the past few years. The research undertaken for this specific case would be able
to determine the factors that cause the decreasing demand for the car.
From the result, the management could take certain corrective action to
improve the demand. For example, the outcome of a research suggested the
company to introduce new models, improve the quality of cars, improve the
service quality provided by its service centers, reduce the selling price, and
improve design of the car to suit younger generation. The specific
recommendation made to the company based on research findings would help
the management to take proper action so that the company could survive into the
future.
THE NEED FOR RESEARCH
In managing the organization, the manager as a decision maker would come

across many managerial problems that are interrelated in various areas such as
management, accounting, marketing, production, finance, etc. In helping the
manager to solve the management problems, the researcher would define the
research problem, determine the research objectives, identify the relevant
variables that may relate to the problem, collect data on all important variables,
and analyze them to finalize the root of a problem. Finally, the researcher will
write a full report based on the data analysis to be presented to the manager
concerning the problem and the possible course of action in order to tackle the
problem.
The essence of research is important to managers in which research helps them

to be more effective and efficient in making the decision for the organization. In
addition, research helps managers to identify the viable alternative among
several alternatives available for effective decision-making.

UniSZA
Knowledge on research would also help managers to utilize the available

information in more sophisticated methods in understanding the problem. The
manager also would be able to interact more effectively with every employee in
planning for the research and development project (R&D) for the organization.
Having adequate knowledge and experience in research would help

managers to look at the problem critically and objectively. Good manager could
sense the potential threat to his organization only by looking at certain statistics
from the news report. Among the statistical information available are the trend of
demand and supply of certain product, the rate of growth in GNP, the rate of
interest, the rate of inflation, and the government policies regarding the economy.
Sometimes, the statements and comments made by the expert in the related field
are also important since they deliver certain message regarding the specific
industry that might affect the organization.
A good manager would ask for a research report before making an

important decision. A good decision is the one made after taking into
consideration all data analysis, the interpreted findings, and recommendations. In
contrast, a bad decision is the one made without considering the information
available, the potential threat to the organization, and without listening to the
expert opinion concerning the underlying problem.
THE ROLE OF RESEARCH IN BUSINESS
Making the right decision is an ideal practice in any business whenever the
organization encounters a problem. A good organization primarily conducts
research to resolve the critical problems surrounding their business such as
competition, customer satisfaction, product innovation, customer complaints, and
new government policies affecting the industry.
Decision making process requires systematic and organized efforts to investigate

a specific problem in a business setting. The first step in understanding the
problem is to identify specifically the main issue that requires further
investigation. The next steps are to identify factors associated with the problem,
gather the relevant information, analyze data, interpret output and provide the
recommendation (based on the research findings) to the manager for his
decision-making. It simply means that the decision making process comprises a

UniSZA
series of steps designed and executed with the goal of getting the best solution
to the underlying problem faced by the organization.
Actually, the entire process in which the managers attempt to solve their
problem is the steps in conducting a research project. The research definition in
the previous section suggests that research involved the process of inquiry,
investigation, examination, and experimentation that need to be conducted
systematically, diligently, critically, objectively and in a logical order. Thus, the
results of research would be the findings, which would help the managers to deal
with the real situation.
Therefore, we can define business research as an organized, systematic
inquiry concerning the problem, and undertaken with the purpose of finding the
best solution to a problem. In addition, the research findings should be able to
clarify all ambiguities surrounding the problem. In research the ambiguities
surrounding the problem are addressed in term of research questions. The
research questions are unanswered questions from the management – and the
research is carried out with the intention to address those unanswered questions.
In other words, research would provide answers to all unanswered questions.
THE DETERMINANT FOR CONDUCTING RESEARCH PROJECT
Sometimes the manager has many alternatives to choose from in his decision-
making for the company. In this case, the manager requires more information
concerning each alternative in order to make the right choice. In other words, the
manager needs to carry-out a research. However, the determination of the need
for research centers on the following factors:
1. The availability of time frame.

2. The availability of data.
3. Nature of decision to be made.
4. The value of decision versus the cost of research.
1. The Availability of Time Frame
A systematic research takes time to accomplish. However, in many instances,

the manager needs to make decision immediately and hence it is made
without adequate information and a thorough understanding of the problems. If
the time allows, the manager should do the research first before making a

UniSZA
decision. Making a decision blindly without adequate information might land an

organization into trouble.
2. The Availability of Data
Adequate information is very important for a manager in making decision. The

role of research is to provide the required information to the manager to make
his decision. However, in certain cases, a manager has adequate information
to assist him and the research is not required. However, a good manager
knows that the research is still required to clarify some ambiguities
surrounding the industry in which the business is operating. Furthermore, if a
potential source of information exists, managers want to know how much it
costs to obtain the data.
For example, Era Timur, a company selling stationery items and providing
internet café plans to open a new branch in Kota Bharu. The manager wishes
to investigate the market potential for the business in the area. The relevant
information that would help the manager in making decision are the
characteristic of the population in the area, the number of potential customers
in the area, the number of schools, government departments as well as private
organizations in the area who are using the internet. And not to forget, the
number of existing competitors already operating in the area is also important.
Since this type of data is usually not readily available, the manager has to
carry out a specific research to collect and analyze the required information.
And not to forget, the manager also needs to know the perception of the
population in the area concerning the proposed business.
3. Nature of Decisions to Be Made
The value of research depends on the nature of the managerial decision to be

made. A minor decision that does not require a substantial investment may not
require the large amount of expenditure for research project. For example,
Pandu Yakin Sdn. Bhd., a car rental company wants to upgrade the
advertisement regarding some modifications in their services. The current
advertisement is not bad in achieving their targeted customers. The costs
required for research and upgrade the existing advertisement may be higher
than the benefits obtained.
Therefore, in this particular case, the company does not have to invest in
research. However, some organizations are willing to spend millions of ringgit

UniSZA
to do a small modification in their product because they are confident of long-

term benefits. For instances, the Telekom Malaysia (TM) re-branding and
positioning exercises for its products and services have incurred millions of
ringgit to the company in short term. However, the exercise would put the
company to gain more benefits in the future.
4. The Value of Decision versus the Cost of Research
We have discussed the various benefits of research. However, a substantial

expenditure is required for conducting good research. Thus, both cost and
benefit are important to justify a research project. A manager should identify
and weigh the potential values or benefits from the research and compare with
the expenditure they have to incur. Three questions would help managers
make the right decision in this case. First, is the rate of return worth the
investment? Second, will the information obtained from the research improves
the quality of a decision? Third, is the proposed research the best alternative
for the available fund? Figure 1.1 illustrates the simple checklist to determine
the requirement of a research.
Time constraints Availability of data
Is sufficient time
Yes Is the data available
Yes
available to do for analysis?
research?
No No
The importance
SHOULD NOT CONDUCT RESEARCH
No Is the decision very
critical for the
organization?
No
The Value of research Yes

Yes
CONDUCT Does the value of
RESEARCH research exceed the
cost of doing
research?
Figure 1.1: The Decision to Conduct or Not to Conduct Research

UniSZA
THE TYPES OF RESEARCH
In research, a distinction is needed not only about the type of research but also
about the most general category into which the implications of research might
fall. What is the distinction? This distinction is a convenient way to clarify the
research activity to shed light on the purpose and importance of the study. The
most basic distinction of research activities are categorized into applied
research (sometimes called action research) and basic research (sometimes
called pure research or fundamental research).
1. Applied Research or Action Research
Applied or action research is carried out for the purpose of solving an existing
problem. The researcher conducts an applied research when he wants to study
about the specific problem. For example, the demand for PROTON is
decreasing. The decrease in demand is reflected in the monthly sales of
PROTON cars nationwide. The management is trying to identify the factors that
contribute to the decreasing demand in order to take an appropriate step before
the problem gets worse.
In this case, the research helps the management to get a clearer picture of
what was happening in the automotive industry, the crucial factors behind it
and the correlation of each factor with the existing economic variables such
as interest rate, rate of inflation, and Gross Domestic Product (GDP). After
obtaining the data, the researcher would perform the data analysis and come
out with certain recommendation based on the output. The management can
use the information from the research report to help them in designing a new
marketing program, determine the pricing policy, and adopt certain
promotional strategy so that the downward trend in demand will change to an
upward trend.
Every business organization requires continuous research and development

program (R&D) to ensure its growth and survival in the industry. The
research will discover the strengths and weaknesses of an organization, its
product and service as well as that of its competitors. Since the socio-
economic variables changes over time, the management of business
organization should embark on continuous research and development in
order to keep pace with its competitors.

UniSZA
Examples of applied research or action research are:
a. The study to determine the level of acceptance among the Malaysians of

different ethnic background towards the One Malaysia Concept
introduced by the Prime Minister Dato’ Seri Mohd Najib Tun Razak.
b. The study to identify the main factors considered by the customers when
buying health products in the market
c. The study to assess the damage caused by the recent flood on

government schools in the country
d. The study to evaluate the performance of government departments in

giving the service to the people
2 Basic Research or Pure Research
One of the purposes for conducting research is to develop and evaluate the
concepts and theories. The researcher conducts basic research to expand the
body of knowledge in the particular area by determining the theoretical
explanation to the underlying problem. This type of research is not for solving
the problem at hand. Rather, the basic research is carried out to develop the
theoretical foundation concerning certain phenomena which take place and to
expand the limits of knowledge based on the findings.
One carries out basic research to confirm that certain phenomena conform to
a certain theory developed by the previous researchers. If the phenomenon
does not conform to the previous theory, then the researcher would propose
a new theory to explain what was happening. Although basic research does
not provide the solution to a specific problem, it does provide a
comprehensive understanding concerning the problem in general. Hence, the
organizations could apply the theory obtained from basic research to handle
their own problems.
As an example, a university professor is interested to investigate how the

level of service quality provided by the hospital contributes to patient
satisfaction for his academic interest. After doing literature search from the
published journals and gathering information from patients at several
hospitals, the professor found that factors like the quality of communication

UniSZA
by the frontline staff, the reputation of doctors, and the environment in the
hospital to be the most important criteria that contribute to the patient
satisfaction towards the hospital. Hence, whenever the management wants
to improve their customer satisfaction, they should specifically improve the
quality of those factors.
Examples of basic research or pure research are:
a. The influence of environmental awareness on consumer buying behavior

towards environmental friendly products or green products in the market.
b. The relationship between employee job satisfaction and the quality of

service they provide to their customers in service organizations.
c. The impact of customer satisfaction towards certain products and their

willingness to provide positive words of mouth and recommendation for
that particular product to their friends and families.
d. Modeling customer loyalty towards the service provided by certain

company based on their experience with the quality of service provided
and their perception on the corporate image of that company.
3. Historical Research
Historical research is the study that relates past events to the current events.
One significant difference between historical research and other types of
research is the type of data collected and the method of data collection.
Historical researcher collects data in a chronological order based on the events
occurring in the past.
Researchers doing historical research often accomplish this goal using

primary data sources (original documents or information from people who
have personal experience on the event) and secondary sources
(secondhand documents or information from people who have knowledge
about the event). For example, a historical researcher wants to examine the
economic achievement of Bumiputera entrepreneurs prior to the
implementation of National Economic Policy (NEP), ten years after the
implementation, and twenty years after the implementation of NEP.

UniSZA
In carrying out the historical research, firstly the researcher needs to define
clearly the problem to investigate. The problem of a historical research is
unlimited in scope since it consists of the events of the present day and the
events of the past. The past events have become a historical database that
is a vast collection of documents whereby much of them might be difficult to
verify their authenticity. This step is much like any researcher’s mental effort,
which is usually resulting from a personal interest in a particular area.
For example, one might be interested to learn the historical performance of

Nuri helicopters since their inception in the Malaysian Army some fifty years
ago. The researcher might want to study the rate of success and failure in
every operation conducted and also the cause of failure so that the
comparison can be made with other models.
Secondly, whenever possible, the researcher should formulate his

hypothesis in order to provide some direction for the study. His findings
based on the data gathered will determine the support for his hypothesis.
Thirdly, as in any research project, the researcher has to utilize a variety of

sources to gather the required data. The sources could differ quite
significantly between one another. Fourthly, the information obtained needs
to be evaluated both for its originality as well as for its accuracy. And this
task is quite cumbersome as well as time consuming. Lastly, the data
obtained need to be integrated to provide a coherent body of information
before it can be analyzed and interpreted as usual.
Define and refine State the research Determine relevant

the research research questions
objectives
problem
Interpret theFigure
results 1.2: The Collect
processdata
in Historical
and Research
Formulate relevant
and write report perform analysis research hypotheses
Figure 1.2: The General Flowchart of a Research

UniSZA
CHARACTERISTICS OF A GOOD RESEARCH
High-quality research is characterized by many characteristics and certain

characteristics tend to be related. Among the characteristics are:
1. It is based on the work of previous researchers in the area. This does not
necessarily mean that the new researcher duplicates their work, but the new
researcher refers the work of others and continues from where they stopped.
In other words, a new researcher continues from where the earlier researcher
finished their work in order to improve what has already been achieved. If the
new researcher does not continue from where they finish, he might be
repeating the same work all over again or he is re-inventing the same wheel!
2. It can be replicated in another set of population. If a research examines the

relationship between problem-solving ability and arithmetic reasoning of
primary school students, the same study should also be replicable to the
secondary school students for two reasons. First, one of the hallmarks of any
credible scientific finding is that it is applicable. Second, if the results of an
experiment can be replicated, they serve as a basis for further research in the
same area.
3. It is generalizable to other population possessing similar characteristics. In

other words, the conclusion from the study can also be applied to other
population if their characteristics are almost equal.
4. A good research is based on certain rationale and it relates to certain

theoretical foundation. Research does not stand alone as just interesting
questions without the logical theory. Major areas of research need to be
broken into smaller elements, and all these elements need to be tied together
in a framework with a theory.
5. Research is an activity undertaken for the betterment of society. Research

and development is an effort to find means and ways to improve the present
quality of products and services, and also the method of doing some other
thing in life.
Technically, a good research must have the following characteristics:
1. The research purpose is clearly defined. No ambiguity arises at all.
2. The outline of research process is explained in detail.

UniSZA
3. Research design is thoroughly planned and clearly understood.
4. High ethical standards of research are applied all the time.
5. Limitations are revealed and the scope of research is properly defined.
6. The method of data analysis and statistical tests to be employed is

adequately explained and justified.
7. Research findings are presented clearly and unambiguously.
8. Conclusions and recommendations regarding the problem under study are

clearly stated and explained.
BASIC ETHICS IN RESEARCH
Researchers should observe and follow the research ethics at all time during
their work. The ethics in research refer to a code of conduct or expected
societal norm of behavior of researchers while conducting research. Ethical
conduct applies to the researchers themselves, the organization involved in the
research project, the organization sponsoring the research, and the
respondents who participate in the research.
The basic principles of ethical research are:
1. Protection from harm - Researchers, respondents and individuals involved in

the study must be protected. If there is any doubt that there is certain risk
involved, then the research project should not be undertaken in the first place.
2. Maintenance of privacy - The work of researchers is confidential. The

researchers should not reveal the identity of their respondents involved in the
study. The data collected from respondents should be kept confidential at all
time and should only be used for the research at hand.
3. Coercion - The respondents are participating in the study at their own will.
Their response during data collection should be free from any influence.
Even, the respondents are free to participate or not participate in the study.

UniSZA
4. Informed consent - The respondents should be made to understand the

rationale for the research and they should agree to waive their right to privacy
when they participate in the research activities.
5. Confidentiality – The researcher needs to give assurance to respondents that

their identity would not be revealed, and the information given is treated as
private and confidential.
6. Sharing the findings – After the study has been completed, the report should
be presented to the audience. The report should also be made available to be
viewed by the respondents who participated in the study.

UniSZA
Exercises
True-False Questions
STATEMENT TRUE FALSE
A research that contributes towards the enrichment of the
1.
body of knowledge is called an action research.
Applied research obtains data to support theories or related
2.
hypothesis about social phenomena.
Researchers should observe the ethics seriously when
3.
conducting their research work.
A good manager is the one who is sensitive to the
4.
information affecting the business.
5. A good research is the one that expands the work of others.
Multiple Choice Questions
1. Research conducted to solve a particular problem is known as ………..
a. Pure research
b. Basic research
c. Applied research
d. None of the above
2. When we want to expand the limits of knowledge, without directly involving the
solution to any particular pragmatic problems, we conduct ………… research.
a. Basic
b. Applied
c. A telephone interview
d. A focus group
3. Business research helps organizations by:
a. Providing report about employees and morale problems

b. Improving efficiency with tools such as sales forecasts
c. Analyzing data that point to business opportunities
d. All of the above

UniSZA
4. When deciding whether business research should be conducted, each of the

following factors is a major consideration EXCEPT:
a. Time constraints
b. Availability of the data
c. Availability of company employees who are survey experts
d. Value of business research information in relation to its costs
5. Which statement best describes the nature of applied research?
a. Applied research requires the use of experimentation.

b. The primary purpose of applied research is to facilitate the managerial
decision process.
c. Applied research is conducted to verify the acceptability of a given theory.
d. Because applied research is exploratory, it does not require the use of
scientific methodology.
6. Which of the following deceptions is unethical?
a. The actual sponsor of the research is concealed.

b. The brand names of products to be compared are falsified.
c. Projective tests are used.
d. All of the above are unethical.
7. The subject’s right to privacy may be waived by:
a. Informed consent
b. Guaranteed anonymity
c. Confidentiality procedures
d. All of the above.
8. A research involves:
a. The identification of quality information, not its computerized processing.

b. An application of scientific methods to information collection and analysis.
c. The use of complex models rather than simpler quantitative.
d. None of the above.

UniSZA
9. Which of the following is NOT an example of problem-solving research?
a. Segmentation research
b. Product research
c. Pricing research
d. Forecasting research
Essay Questions
1. For each of the situation below, state whether the research should be
descriptive or causal. State the reasons for your choice.
a) To study the impact of advertising on the volume of sales.

b) To assess the reaction of society on the proposed highway in the area.
c) The profiling study on the customers of certain healthcare products.
d) To determine how interest rate influences the share price in the market.
2. Situation A:
A university lecturer is interested to investigate the factors that contribute to
absenteeism in the workplace. After gathering some information, the lecturer
identifies factors such as the flexibility of working hours, the amount of
monthly incentives given, and the working conditions as primary factors
influencing absenteeism.
Situation B:
An advertising agency wishes to conduct a research to find out the
effectiveness of its sales promotion on consumer demand of its product.
For each situation above, state whether it is a basic or an applied research

and give reasons for your answer.
3. List the guidelines to what is considered “a research” and “not a research”.
4. a. For each of the following situation, state whether it is a descriptive, a

historical or a causal research.
i. The relationship between the amount of time spent studying in a library
and the academic performance of university students.
ii. A study on the level of customer satisfaction regarding the quality of
service provided by the cafeteria of UiTM Shah Alam.

UniSZA
iii. A profile study of people who invest in the quick-rich scheme organized
through the internet.
iv. A study on the factors that influence the monthly expenditure of a family
living in Kuala Lumpur.
b. What is the distinction between basic research and applied research?
5. a. Explain the differences between descriptive research and causal research

design. Give one example of each type of the research design.
b. Identify five (5) characteristics of a high quality research.
6. Why is it important for managers to master the skills of research?
7. Briefly describe the purpose of descriptive research and exploratory research.
8. Miss Sarah, the owner of a small business (a woman’s dress boutique) has
invited a consultant to provide report concerning how her business is different
from similar businesses within a 40km radius with respect to the usage of
information technology for advertisement, the pricing strategy, the quality
customer service, level of customer satisfaction, the store appearance, and
also the product choices.
a. Determine whether it is an exploratory or a descriptive study.

b. Determine the scope of the study.
c. Identify the population of the study.
d. Suggest a suitable sampling method and data collection for the study.
9. a. State whether the research is basic or applied. Justify your answer.
Situation 1:
A researcher investigates five personality traits to see if they can explain
the purchasing behaviour of automobile buyers.
Situation 2:
A researcher examines how the heat processing and the use of flexible
pouches for storage help retain the freshness of fish.
b. State the differences between descriptive research and causal research.
10. What are the differences between applied research and basic research?

UniSZA
Chapter 2: The Research Process
CHAPTER THE RESEARCH PROCESS

2
INTRODUCTION
Research work requires a sequence of activities to be carried out from the

beginning until the final report is compiled and presented to the management.
Listed below are the processes required when a researcher is hired by an
organization for research and consultancy work.
1. Establishing the problem for research
2. Identifying the appropriate research variables
3. Forming research objectives, research questions, and hypothesis
4. Developing the theoretical framework for the study
5. Reviewing literatures related to the problem
6. Selecting an appropriate research design
7. Collecting relevant data from the population
8. Analyzing data using appropriate statistical methods
9. Preparing final report to the management
10. Presenting the results
PROCESS 1: ESTABLISHING THE PROBLEM FOR RESEARCH
The activities involved in the first step bring the researcher and the
management together because the manager needs the researcher to help his
decision making. Usually before the researcher undertakes a consultancy work,
the manager has to state what he believes to be the main issue surrounding the
problem. At this point, the researcher would make sure that the manager
understands the actual problem. The researcher then helps the manager to look
at the problem objectively, and justify whether it is really a problem or it is only a
symptom to a problem. It is important for the manager and the researcher to
agree on the problem to be researched so that the research processes would
provide information which is relevant to the problem.

UniSZA
The Tasks Involved In Establishing The Problem Are:
1. Analyzing the secondary data
The analysis of secondary data is an essential step in the problem

definition process. Working with previous data might provide important
clue concerning the underlying problem. Plotting secondary data against
time would help the research to identify the trend and its relationship with
certain variable. The secondary data is the information already collected
and compiled by certain organization for certain purpose.
One example of a secondary data is the volume of monthly sales of Proton

cars in the country collected by the Proton headquarter in Shah Alam. The
analysis of secondary data would show the trend of monthly sales for
proton cars against time. The researcher wants to determine the possible
factors that cause the decreasing trend. Among the factors that might
cause the demand to decline are the quality of cars, the quality of after
sale service, the price of competitor’s car in the market, and the change in
preference towards the imported cars.
2. Having discussion with the management
The researcher should always remember that the research will provide the
required information that helps the manager in making a decision. That is
why the researcher must discuss thoroughly with the manager before
begin to conduct a research. The discussion will reveal the actual problem
faced by the manager and hence what type of information required by the
manager from the research to facilitate his decision making for the
organization. In this case, the researcher and the manager need to agree
on the problem to be researched and the output required in solving the
problem.
3. Informal interview with the professional in the field
The informal interview with the professionals in the field is another method
that helps researcher to understand the research problem. The
professional is the person who has an expertise in the area related to the
study. For example the economists and the bankers are among the
experts who have broad knowledge concerning the economic outlook of
the country. The professional body in this industry is The Association of
Automobile Manufactures (AAM). The informal interview with them would

UniSZA
reveal the clearer scenario regarding the variables affecting the present
economy and the demand for certain products and service in the country.
The opinion from the experts would also help the researcher to identify the
main variables surrounding the problem. Thus, these variables should be
included in the research.
4. Doing a qualitative research
The qualitative research is an unstructured study and it is an exploratory in

nature that is based on a small sample. It may utilize some qualitative
techniques such as focus groups interviews, in-depth interview, and a pilot
survey. A qualitative research may not be conducted in a formal way.
Nevertheless, it offers the most sufficient technique that provides valuable
insights into the problem. For example, the focus-group interview with the
customers who have some 20 years experience using various cars
including Proton would reveal important information concerning the
strength and weaknesses of Proton as compared to its competitors in the
market.
The Steps Required When Doing Consultancy Works:
1. Understand the management’s expectation from research
The purpose of conducting research is to facilitate the management with

relevant findings. The researcher should try to meet the objectives of the
manager by clarifying what to achieve before the beginning of research. Both
parties must have a clear agreement on the problems and the objectives of
the research. This step provides the direction for a successful research
project. For example, the demand for a product is decreasing. The problem is
supported by the previous weekly and monthly sales figure.
The management’s objective is to decide whether to launch new product

package into the market, to launch heavy sales promotion for the existing
product, or to study the strength and weaknesses of competitors’ product
before making any decision. The research findings would indicate the
appropriate decision that the management should undertake.

UniSZA
2. Understand the background of the problem that research is required
What problem that triggers you to carry out research? The most important
step in problem identification process is to understand the background of the
problem. The failure to understand the actual problem will make research
project begins without clear objectives and direction.
Normally an exploratory study and informal interview with the expert in the
field might help both the researcher and the management to identify the
problem. The researcher should study the similar problem occurring in other
organizations, trying to figure out the similarities with the problem at hand in
terms of the variables involved. For example, the background of the problem
is the stiff competition in terms of product pricing, product range, service
quality, and product quality from foreign importers
3. Differentiate between the actual problem and the symptoms of a

problem itself
The actual problem may be identified only after gathering the background
information and conducting an exploratory study. For example, in measuring
the customer satisfaction towards a certain supermarket, a manager may
think many possible factors that influence customer satisfaction. Among them
are the products, the services, the location, the price, etc. The most critical
task of the researcher is to isolate and identify the most likely causes based
on the symptoms of the problem. However, certain factor that appears to be a
problem may be only symptoms of a deeper problem.
The iceberg principle serves as a useful analogy to describe the problem

definition process, the symptoms of the problem, and the causes of the
problem. This principle is like a sailor on the open sea who notices only the
10 percent of an iceberg that extends above the surface of the water, while
the other 90 percent is lying below the surface. The dangerous part of the
problem is the 90% portion of the iceberg below the surface, which is neither
visible nor understood by managers. For example, a marketing manager of a
Swimming Club in town is concerned about the possibility of losing
customers. From his observation on the club’s record, he found the number of
membership has declined since a new water park was built in the town a few
years ago. In this case, the declining in memberships is the symptom to a
problem.

UniSZA
By looking at the symptom, he has an opinion that the customers might prefer
the exclusive water park as compared to his traditional swimming pool.
However, after an exploratory study using the focus group interview with
customers, he found that the actual problem is the change in demographic
variables, the change in lifestyles, and hence the change in the preference of
customers. Children in this neighborhood have grown up and their preference
has switched to the water park. So, there is no demand for traditional
swimming pool from the generation of younger and outgoing lifestyles.
As another example, the decreasing sales volume is the symptom that can be
seen (daily, weekly, monthly, and annually). Nobody knows exactly the actual
problems that cause poor demand from customers until the research is
carried out. The research findings might discover the design of products is no
longer suitable for current lifestyle, and also the pricing package offered by
competitors is much better. These two factors cause customers to switch.
Remember: the manager cannot act on the symptom of the problem
(decreasing sales), but the manager can act on the factors that cause the
problem of decreasing sales (product design and pricing strategy).
4. Determine the unit of analysis for research (UOA)
The unit of analysis is the most important aspect in research and it is the
focus of the study. The researcher must decide clearly whether the research
will focus on individuals, households, organizations, certain departments
within an organization, certain geographical areas, or certain groups in a
population. Unit of analysis is the focus where the respondents evaluate its
performance which relates directly to the problem to be researched. In the
case of the swimming club, the unit of analysis is the swimming club itself.
The respondents are selected among the customers to evaluate the
performance of the swimming club in terms of its corporate image perception,
its services quality performance by its employees, etc. The respondents are
the existing customers and also the people who used to experience the
service provided by the swimming club.

UniSZA
PROCESS 2: IDENTIFYING THE APPROPRIATE RESEARCH VARIABLES
The research problem is normally associated with many factors, and those
factors are basically represented by variables. A variable is anything that
changes in value and represents a quantity that can differ in values,
magnitude, or strength. In other words, the variable is generally anything that
may assume different numerical values. Among the variables normally
associated with marketing research are perceived quality of product or
service, product image, product price, competitor’s price, socio-economic
status of respondent such as education, occupation, income, etc.
The Classification of Variables:
1. Dependent variable is the variable of main interest in any study.

Examples are the volume of transaction, the expected number of
customers, the perceived quality of service, and the customer’s loyalty
intention towards the product or service. Some textbooks define
dependent variables as the response variables.
2. Independent variable is the variable that causes the dependent variable

to change in values. Examples are the interest rate, the selling price, the
incentive provided, the amount of promotion, the quality of service
provided, the monthly income of customers, and the number of potential
customers in the target area. Some textbooks call independent variables
as the predictor variables.
3. Mediating variable or sometimes called an intervening variable is the

variable that mediates the relationship between the independent and
dependent variable. For example, in the service quality – customer loyalty
relationship, customer satisfaction is the one which stands between
service quality and customer loyalty. In this case, the service provided by
the company should be able to satisfy its customers, and these satisfied
customers would, in turn, become loyal customers to the company.
4. Moderating variable is the variable that moderates the performance of a

dependent variable. Examples are respondent’s socio-economic status. In
the case of service quality as an independent and customer satisfaction as
a dependent variable, the study is trying to ascertain whether the
UniSZA
respondents from different socio-economic status perceived differently

concerning their satisfaction towards the service quality provided by the
same organization.
The researchers must be careful in selecting the appropriate variables

when defining the managerial problem. The variables that are not directly
related to the problem should not be included in the study. The process of
identifying the relevant variables overlaps with the process of determining
the research objectives. Typically, each research objective would directly
reflect one or two variables that would be involved in the study.
Determine the measurement issues and scales for variables

The researcher needs to determine how the variables of interest in the
study would be measured and what scales to use. The researcher must have
a clear understanding concerning the types of measurement (nominal,
ordinal, interval, and ratio) since the types of analysis would depend on how
the acquired data is scaled. The lack of knowledge in the scaling process
could easily lead to using the wrong method of data analysis. The implication
could cause the misinterpretation of findings, conclusions, and
recommendations in the later process.
The researcher needs to review the literatures thoroughly in order to

ensure the appropriate variables to use. Observe the debates in the
literatures regarding the variables, the related issues concerning its scales,
its validity, and its reliability.
PROCESS 3: SETTING RESEARCH OBJECTIVES, RESEARCH QUESTIONS,

AND RESEARCH HYPOTHESES FOR THE STUDY
In doing the research project, the researcher has certain objectives to

accomplish. In the process of achieving the objectives, the researcher normally
discover some uncertainties regarding the variables, the factors influencing the
variables, and the nature of relationship among variables in the study. The
uncertainties that arise here are called research questions. A good researcher
manages to convert these uncertainties into research questions. The researcher
can assume the research questions as the questions he anticipates would be
raised by the audience of his research and must be clarified through this study.

UniSZA
1 Setting the research objectives
Research objectives set a list of what the research is going to achieve. They
provide evidence of the researcher’s clear understanding about the purpose
and direction of the proposed study. The research objectives should follow
from the definition of the research problems. Formally stated research
objectives provide the guidelines for determining which other steps must be
undertaken. The assumption is that if the objectives are achieved, the
decision maker will have the information needed to solve the problem.
Overall, research objectives are the elements of what your research is trying
to achieve. Most of the time the research objectives begin with the word “to”
Example 1: The objective of the study
1. The objective of this study is to identify the factors that contribute to the
declining demand for Proton among the Malaysian customers.
2. The objective of this study is to generate the items that constitute
service quality in higher education from the students’ perspective.
3. The objective of this study is to measure the strength of association
between socio-economic status of customers and their perception of
quality concerning the service provided by higher educational
institutions in the country.
4. The objective of this study is to assess the influence of service quality
provided by the university and the corporate image of a university on
students’ loyalty towards their university in doing postgraduate study.
The researcher can list several objectives to be achieved in his study.
Example 2: The objective of the study
Specifically the study is designed to achieve the following objectives:

1. To assess the effects of certain demographic variables on service quality,
corporate image, and students’ satisfaction.
2. To determine the influence of service quality provided at the university on
students’ satisfaction with their learning experience
3. To determine the influence of corporate image of the university on
students’ satisfaction with their learning experience
4. To determine the influence of service quality provided at the university on
students’ loyalty towards their university
UniSZA
5. To determine the influence of corporate image of the university on

6. To determine the influence of students’ satisfaction with the service
provided by the university on their loyalty towards that university in the
future
2. Forming the research questions
Research question is the “interrogative statements” the research is trying to

answer. It is an ambiguity shrouded the underlying research problem. The
researcher should identify the ambiguities so that it can be redefined into
specific research questions. In other words, research questions are the
ambiguities identified by the researcher so that his research project is directed
specifically to address those ambiguities. The research project is considered
successful if it manages to clear all ambiguities surrounding the problem. The
researcher utilizes the words “why, what, where, when, who, and how” to
develop the research questions.
The importance of a research question

Research questions will help clear the confusion that normally arises in any
research project. The importance of the research question lies in its roles that
help the researcher in the following manner:
a. Provides a clear direction in organizing the research project.

b. Delimits the research project through showing its boundaries.
c. Keeps the research project in focus at all times.
Doing a research project requires a complicated process and it is not

impossible for the researcher to slide along the way through the research
project. In this case, the research questions provide great assistance for the
researcher to bring the focus back on track. Specifically, the research
question is important to:
a. Provide a framework to carry out research project properly

b. Determine the data required by the research to answer the question
c. Provide a basis for designing a questionnaire to collect data
d. Provide a basis for the methods to be applied in data analysis
e. Provide a direction in writing the findings to address the question

UniSZA
Example 1: The research questions
This study attempts to address the following research questions:

1. How significant is the effect of respondent’s socio-economic status on
service quality, corporate image, and students’ satisfaction?
2. How significant is the effect service quality provided by the university in
influencing students’ satisfaction with the university?
3. How significant is the effect of corporate image of the university in
influencing students’ satisfaction towards the university?
4. How significant is the effect of service quality at the university in
influencing students’ loyalty towards that university in the future?
5. How significant is the effect of corporate image of the university in
influencing students’ loyalty towards that university in the future?
6. How significant is the students’ satisfactions with the university manage to
influence their loyalty towards that particular university? Or in other words,
the study is interested to answer questions like “Is there a significant
relationship exists between students’ satisfaction and their loyalty?”
3. Stating the research hypothesis

Hypothesis is defined as an intelligent anticipation or prediction of a
researcher concerning the research problem at hand. An intelligent researcher
manages to come out with several hypotheses regarding the problem at hand.
It shows that the researcher has a thorough knowledge regarding the
phenomena that he can predict th6e outcome of the study especially how
independent variables affect the dependent variables. A hypothesis normally
flows from the research questions. A good hypothesis provides a transition
from an interrogative statement into a declarative statement. Types of
hypothesis:
a. Null hypothesis is a status quo regarding the phenomena. The null

hypothesis serves as a starting point for observing the effect of the
independent variable on the dependent variable. It also serves as a
benchmark for the comparison between the “effects by chance” versus
the “real effect”.
The statement of status quo should read as follows:

UniSZA
a) Ho: There is no significant difference in the academic

performance among students due to the differences in study
methods applied.
b) Ho: There is no significant relationship between gender and
academic excellence of students in higher educational institutions.
c) Ho: There is no significant increase in the price of consumer items
in the market due to the increase in the cost of petroleum
b. Alternative hypothesis or sometimes called the research hypothesis is

a “declarative statement” made by the researcher regarding the
relationship between variables involved. The researcher makes this
intellectual proposition after some extensive reading of the literatures.
Thus, in the study, the researcher would support his proposition with the
proper methodology and data analysis. In most cases, researchers can
ignore the null hypothesis and only state their research hypothesis of the
study.
Example 1: The research hypotheses
a) H1: There is a significant difference in academic performance

among students due to the differences in study methods
applied.
b) H1: There is a significant relationship between gender and
academic excellence of students in higher educational
institutions.
c) H1: There is a significant increase in the price of consumer items
in the market due to the increase in the cost of petrol

UniSZA
Example 2:
Suppose that you are studying the effects of on-job training for
employees on their productivity and commitment towards the
organization.
For the above study you are required to develop the required research
question and its corresponding hypothesis:
Solution:
First of all, you need to determine the variables of main interest in the
study. In this case, the variable of interest is employees’ productivity and
their commitment. This study is trying to investigate the importance of
training on productivity and commitment. Thus, the dependent variables
are productivity and commitment and the independent variable is on-job
training. Now you can formulate the research question and its
corresponding hypothesis. In this case, the independent variable is on-job
training, while the dependent variables are productivity and commitment.
The unit of analysis in this study is an employee of the organization.
The research questions would be:
1. Does training provide significant effect on employees’ productivity?

2. Does training provide significant effect on employees’ commitment?
The theoretical framework for the study would be as in Figure 2.2

below.
Productivity
Training
Commitment

UniSZA
Figure 1: Theoretical Framework of the Study
From the theoretical framework, the hypothesis of the study would be:
The research hypotheses:
1. H1: Training does provide significant effect on employees’ productivity

2. H2: Training does provide significant effect on employees’ commitment
Example 3:
The management of UiTM is wondering why there are many students

who refuse to consume the food provided at dining halls at the residential
college. Every day, a large amount of food is not taken and wasted. The
government is paying the contractor at least RM8.50 per head for the
food and this has resulted in a waste of millions of ringgit every year.
Instead of eating the food provided at the dining halls, the students are
buying their food at the cafeteria or from the stalls nearby. The
management of UiTM wants you to study this matter. The findings of your
research should determine the main factors that contribute to the above
problem so that appropriate actions can be taken to rectify the situation.
At the same time you should provide some recommendations based on

your findings to the management so that the existing problem can be
resolved.
1. What is your research question?

First of all you need to determine the variables involved in the study.
The dependent variable is the students’ preference of eating at dining
hall.
The independent variables are the quality of food served and also the
quality of service provided by the food operator.
The research questions would be:
a) Does the quality of food provided by the caterer have an impact on

the students’ preferences to eat at the dining hall?
b) Does the service quality provided by the caterer have an impact on
the students’ preferences to eat at the dining hall?
UniSZA
The theoretical framework for the study would be:
Food
quality
Eating
at dining
hall
Service
quality
Figure 2.3: Theoretical Framework of the Study
2. What is your hypothesis?

From the theoretical framework, the hypothesis of the study would be:
The hypothesis for research question 1

a) H1: The Food Quality served by the customer has significant effects
on students’ preference to eat at dining halls.
The hypothesis for research question 2

b) H2: The Service Quality provided by the customer has significant
effects on students’ preference to eat at dining halls.
PROCESS 4: DEVELOPING THE THEORETICAL FRAMEWORK FOR THE

STUDY
This step will take place after the study identifies the appropriate variables
through extensive literature search. The theoretical framework is a schematic
diagram showing how the study believes the variables are related among
each other. From the schematic diagram, the study would debate the
rationale why certain relationships between variables are believed to exist in
reality. The nature and the direction of relationships among the variables are
also discussed and the discussions are supported with findings from past
UniSZA
researches and also certain theoretical insights. From the theoretical

framework, relevant hypotheses can be developed. The theoretical framework
and hypotheses provide proper direction for the study to proceed till the end
in term of the instrumentation to measure each variable, the methods of
sampling and data collection, and the procedure for data analysis to use in
order to test the stated hypotheses.
Example 2.1: The theoretical framework for the study:

1) Independent variable: Service Quality, Corporate Image
2) Dependent variable: Customer Loyalty
3) Mediating variable: Customer Satisfaction
Service
Quality
Customer Customer
Satisfaction Loyalty
Corporate
Image
Figure 2.1: The theoretical framework relating the four variables in the study
How the variables in the study relate to each other?
In the above diagram, the study believes that service quality performance
provided by the organization and corporate image of that particular organization
are two independent variables. These two variables are believed to influence the
mediating variable namely customer satisfaction. Finally, the study believes that
the satisfied customers would become loyal customers towards the organization
in the future. In this case customer loyalty is the intended dependent variable.
(Remember: The relationship among variables in a study and any argument made
regarding the relationship must be supported by the literatures in that particular field.
That is why literature review is one of the important chapters in any research)
UniSZA
PROCESS 5: REVIEWING LITERATURES RELATED TO THE PROBLEM
Literature survey is a comprehensive review of the published and unpublished

work from previous researchers in the areas related to the problem at hand.
Literature search will help the researcher to focus on the research problem
based on certain aspects found to be important in the published studies. For
example, the literatures would reveal the main variables involved in the
problem, their measurement, the validity, and reliability. Sometimes the
literatures published the instruments employed in the study. Using these
instruments, the researcher could customize them in order to suit the current
setting.
The literature is used deductively as a framework for advancing research

questions and their corresponding hypotheses. The review of literature is placed
at the beginning of the study to introduce the problem, the related variables, and
to provide meaningful direction for the research questions and their
corresponding hypotheses. The findings from the literature in similar studies
should be included in the report for the researcher to compare his findings with
the findings of previous researchers. As previously stated, the literature review is
important for the researcher to identify the variables involved, the measurements
applied, the methodology used, and the expected findings. Furthermore, the
findings from literature survey provide a solid foundation of the existing
knowledge in the research area. This foundation could be compared with the
findings of the present research.
The Purpose of Literature Review
The literature review in a research study serves the following purposes:
1. To ensure that all important findings in the previous studies are included as a
basis to support the development of current research.
2. To enable the researcher to understand and define the underlying problem

precisely so that his research would obtain maximum benefits.

UniSZA
3. To guide the researcher in selecting the variables for the study and to provide
the rationale for selecting those variables.
4. To help the researcher in developing a theoretical framework for the study,

and to formulate the research questions and the corresponding hypotheses.
5. To avoid the researcher from “reinventing the wheel”. The researcher might
be researching exactly what has been researched if he does not review the
literatures prior to doing his research.
6. To share with the reader the findings of previous studies that relate closely to
the current study. Thus, the replicability of the findings of the current research
is enhanced.
7. To provide the framework for establishing the importance of the current study
as well as a benchmark for comparing the findings of the current study. The
researcher can debate as to how his finding is consistent or not consistent
with the results obtained by previous researchers.
Steps Required In Conducting Literature Survey
There are three steps involved in conducting the literature survey;
(1) Identify the relevant sources for literature in the field of interest,
(2) Obtain and read all literatures which is relevant with the study,
(3) Write up the literature review based on the information obtained.
a) Identifying the relevant sources

The library is a rich storage base for secondary data including books,
journals, newspapers, magazines, conference proceedings, doctoral
dissertations, masters’ theses, government publications, and other reports.
However, using the modern technology, all published articles in the
newspapers, periodicals, conference proceedings, and other sources are
available on the online databases. There are three forms of databases
useful for searching the information for the research topic.
i. Bibliography index database. This database provides information

regarding the bibliographic citations such as the name of the author,
the title of the article, title of the book, source of publication, the
year the article is published, and the volume of its publication.

UniSZA
ii. Abstract database. This database provides an abstract from all

articles
iii. Full-text database. This database provides the full text of all articles
b) Extracting and gathering the relevant information
After making literature search, the researcher might have a long list of
articles from newspapers, magazines, books, and journals. Then, the
researcher can select certain articles from the group. In the selection
process, the researcher should go through the abstract of the articles. The
researcher should consider whether the articles will make useful
contribution to the literature. From the articles, the researcher should
systematically focus on the knowledge gap that is being researched, the
research design employed in the study and finally how the finding
consistence with the theory etc.
In reviewing the content of research studies, the researchers need to
consider what information to be extracted and to be summarized for the
literature review. A good literature review should include the following
elements:
i. Mention about the main problem addressed in the study.
ii. State the central purpose or the main focus of the study.
iii. State information about the population and sample of the study.
iv. Review the main findings of the study (LR) that relate to the current
study.
v. In some cases, the researcher needs to point out the technical
review depending on the forms of literature review (integrative,
theoretical, or methodological review). This process will facilitate
the writing up of the literature review with minimum disruption and
maximum efficiency.
c) Writing a literature review
Literature review involved the process of collecting and summarizing the

relevant studies done by previous researchers in the same area. In writing
a literature review, the researcher must present clearly and logically the
relevant research work done related to the study. The written literature
must bring together all relevant information in a logical order instead of
chronological order. In his review of literature, the researcher must
highlight the problem related to the study, the scope covered in the study,
UniSZA
the main variables involved, the relationship among variables in the

theoretical framework, and finally the findings of the study. The researcher
needs to report whether the findings are consistent or not with the general
theory and also with the findings of previous researchers. Remember, the
literature review must be exhaustive – meaning all theories related to the
study are discussed and all literatures (papers, journal articles, books)
related to the study are covered, discussed, and debated.
There are several methods of citing references and using quotations for
literature review. The American Psychological Association (APA) offers
detailed information regarding citations, quotations, and references of the
literature.
PROCESS 6: SELECTING AN APPROPRIATE RESEARCH DESIGN
The research design serves as a master plan of the methods and procedures
for the researcher to collect and analyze the data required by the research.
Determining the most suitable research design is a function of research
objectives and the specific requirements of the research. In this master plan, the
researcher must consider the type of data required, the design technique
(survey, observation, experiment, and case study), the sampling methodology
and procedures, the schedule and the budget.
Different research may require different types of research design. The

common types of research design are discussed as follows:
Non Experimental Research Design
Non-experimental research includes a variety of different methods that

describe relationships between variables. For example, the survey on
television-watching behavior of teenagers and adults, the researcher would
like to know what programs the teenagers watch on television. Non-
experimental research has two types; that are exploratory and descriptive.
UniSZA
1. The Exploratory Research Design
Exploratory research is conducted to clarify the nature of ambiguous

problems. Ambiguous means that the nature of the problem is unclear, in
which the objectives are vague and decision alternatives are difficult to
understand. The purpose of exploratory research is to explore the problem
or situation to gain better understanding about the dimensions of the
problem.
Exploratory research is not intended to provide conclusive evidence from

which to determine a particular course of action. Usually, exploratory
research is conducted with the expectation that subsequent research will
be required to provide such conclusive evidence. Therefore, exploratory
research is conducted for three interrelated purposes: (1) diagnosing a
situation, (2) screening alternatives, and (3) discovering new ideas.
Exploratory research focuses on collecting either secondary or primary

data, and using unstructured formal or informal procedures to interpret
them. This research design incorporates the fewest characteristics or
principles of the scientific method. It is often used simply to classify the
problems or opportunities. Hence, it is not intended to provide conclusive
information from which a particular course of action can be determined.
Among the examples of exploratory research techniques are; focus-group
interviews, experience surveys, and pilot studies. Exploratory research
may also use some forms of secondary data (e.g. online database).
Exploratory research can be somewhat intuitive and is used by many
managers to monitor the demand pattern for their products compared to
the competitors’ products in the market.
Example 2.7:
A Thai restaurant is considering to extend its operating hours and to

expand its product line with a breakfast menu. Exploratory research is
required to provide some understanding about the problem. However, an
exploratory study with a small number of customers found a negative
reaction to a spicy breakfast menu that is one of the menus offered by the
business. The findings from the study indicate that the restaurant should
not offer this menu for breakfast.
UniSZA
2.The Descriptive Research Design
The major purpose of descriptive research, as the name implies is to

describe the characteristics of a population or an existing phenomenon.
Descriptive research is not a stand-alone research; it serves as a basis for
other types of research.
Descriptive research uses a set of scientific methods and procedures to

collect data and create data structures that describe the characteristics
(e.g. gender, occupation, income, academic qualification, age categories,
political opinions, preferences towards PROTON, and purchase
intentions) of a target population. Descriptive research designs are
appropriate when the research objectives include determination of the
degree to which marketing variables are related to the actual market
phenomena. The researcher looks for answers to the “how”, “who”, “what”,
“when”, and “where” questions concerning the components of a market.
Unlike exploratory research, descriptive research is based on some

previous understanding of the nature of the research problem. Although
the researcher may have general understanding about the situation, he
may need conclusive evidence from the research to determine the course
of action. Many circumstances require descriptive research to explain the
nature of things. Frequently, the basic understandings from exploratory
research and the conclusive evidences of descriptive research are used to
determine the extent of differences and association in certain variables
among subgroups in a population. The study might also be interested in
estimating the proportion of a population that have certain characteristics.
Example 2.8:
En Hishamuddin, a manager of D’Cinta Enterprise in Machang, wants to

focus on cosmetic product by offering some additional items with the
particular brands. He believed that 12 to 15 year old girls cared a lot
about fragrances, lipstick, and mascara, but he lacked quantitative
evidence. He conducts a descriptive research to identify the
characteristics of his customers. This research found that 40% of the
young girls use facial cream, 30% use fragrance and 20% use lip-gloss

UniSZA
and another 10% use more than one product category. The study also
found that 60% of customers who use cosmetics are loyal to certain
brands, and another 40% do not care much about brand. In addition, 70%
of the customers who are brand loyal choose the brand advertised by their
favorite female artist.
The Experimental Research Design
In the experimental research is where participants are assigned to groups

based on some selected criterion often called the treatment variable. In
other words, experimental research is one way to test the presence of
cause-and-effect relationship.
1. Causal Research Design
Often, there is confusion on the interpretation of causal research design

process regarding to the term “relationship”. The basis of the term
“relationship” goes into two distinctions, the “cause and effect relationship”
and “associational relationship” between variables. Although both
distinctions mentioned are about the relationship between variables, the
difference between these two distinctions is “cause and effect relationship”
provides better understanding regarding the relationship between
variables. However, depending on the design of the study, if the study
wants to determine the cause of one or more variables, it is called a
causal study. However, if the study is interested in determining the
important variable associated with the problem, this study is called
correlational study.
For example, does smoking cause cancer? - refers to the question for
causal study.
Rather, are smoking and cancer related? - refers to the correlational
study.
Causal research is designed to collect raw data structures and information

that will allow the decision maker or researcher to model the cause-and-
effect relationships between two or more variables. Causal research is
most appropriate when the research objective includes the need to
understand the reasons why the market response changes when certain
economic variables change. It can be used to understand the functional
UniSZA
relationship between the causal factors and the effect predicted on the
market performance variable under investigation. This type of research
design allows decision makers to gain the highest level of understanding
in the research process. In addition, understanding the cause-effect
relationships among market factors allow the manager to make if-then
statements about the variables.
Causal research designs offer opportunity for identifying, determining, and

explaining causality among critical market factors, they tend to be
complex, expensive, and time consuming. Among the different data
collection techniques available, experimental designs hold the greatest
potential for establishing cause-effect relationship because they allow
researchers to investigate changes in one variable while manipulating one
or two other variables under controlled conditions.
2. Quasi-Experimental Research Design
In the quasi-experimental research, the respondents are pre-assigned to

groups based on some characteristic or quality these people bring to the
study. For example, differences in gender, level of education, age group,
program of study, economic status, and political preference are among the
characteristics.
The quasi-experimental method is most suitable when a researcher

cannot assign people to groups and test the effects of group membership
on some other outcome. Quasi-experimental research also called post-
hoc research or after-the-fact research since the actual research takes
place after the assignment of groups such as employed versus
unemployed, malnourished versus non-malnourished, male versus
female. Because assignment has already taken place, the researcher has
a degree of control over the cause of whatever effects are being
examined.
Table 2.1: The Comparison of Research Design

Exploratory Descriptive Causal
Objective: To discover the insights To describe the problem To determine the cause and
and new ideas about at hand and relate to the effect relationship between
the problem at hand. To characteristics of the the independent and
confirm a problem population under study dependent variables

UniSZA
Characteristics: Flexible, versatile, and Based on specific Manipulation of independent

does not based on a objectives, research variables, and measure the
proper research design. questions, and research effects on the dependent
hypotheses of the study variable
Methods: Expert interviews, pilot Structured data collection Laboratory experiments,
surveys, analyze method, proper statistical changes in the level of
secondary data, and data analysis procedure independent variables.
conduct qualitative changes in government
research. policy on import tax, etc
Example of The volume of car sales The volume of car sales is The rate of interest on car
possible findings:
is declining. However, related to specific factors loan, the import duty, and
the exact factor causing such as rate of interest, certain automobile policies
the problem is still economic growth, affect the sale of cars in the
unclear at this stage increase in population country
The major classifications of research design are exploratory, descriptive,

and causal research design. However, the distinctions among these
classifications are not absolute. For example, a research project may
involve more than one type of research design and serve for several
purposes. In other words, a combination of research design may be
employed depending on the nature of the problem. For instance, problem
definition begins with an exploratory research when little is known.
Exploratory research is the initial step in the overall research design
framework followed by descriptive or causal research.
Nevertheless, it is not necessary to begin every research design with

exploratory research. The decision depends upon the precision with which
the problem has been defined and the researcher’s degree of certainty
about the approach to the problem. Exploratory research may also follow
descriptive or causal research. Basically, exploratory research is conducted
to understand the research findings of descriptive or causal research. This
situation happens when the results are difficult to interpret into the real
situation.
Unlike exploratory research, descriptive research is based on some

previous understanding of the nature of the research problem. Although the
researcher may have general understanding about the situation, they need
conclusive evidence of the answer required to determine the course of
action. Many circumstances require descriptive research to explain the
UniSZA
nature of things. Frequently, the basic understandings from exploratory

research and the conclusive evidences of descriptive research are used to
determine the extent of differences and association in certain variables
among subgroups in a population. Besides, the study is also interested in
estimating the proportion of a population that has certain characteristics.
Cross Sectional Design versus Longitudinal Design
1. The Cross Sectional Design

Cross sectional design refers to the study in which data collection is
gathered from the sample only once. Cross-sectional design is classified
into single and multiple cross-sectional designs. In a single cross-sectional
design, only one sample is collected from the population, and the data is
obtained only once, while, in multiple cross-sectional designs, two or more
samples are collected from the population and the information obtained is
also once. One type of multiple cross-sectional designs is cohort analysis.
Cohort analysis is a study that consists of a series of surveys conducted at
appropriate time interval. Cohort refers to the sample of respondents who
experience the same event within the same time interval. For example the
respondents who enrolled in UiTM between the year 1990 and 2000 are
selected in the study. In cohort analysis, the data is collected from the age
cohort such as 10 to 15 years old, 16 to 20 years old to be the samples for
the study. However, the samples of this study are drawn independently.
2. The Longitudinal Design

In longitudinal design, the data is collected from the same sample on the
same variables at two different times. In other words, the same people are
studied at two different time and the same variables are measured twice.
This study is appropriate when the researcher wants to study the
phenomena at more than one point in time. Sometimes, the management
wants to assess the effectiveness of the “on the job training” provided to the
employees. In this case, the researcher needs to study at two time intervals.
The first is to measure the efficiency of the employees to perform the task
before they attend the training, and the second is to measure the efficiency
to perform the same task after they attend the training. Another example is
to study the increase in demand for national car before and after the
announcement of the automobile policy by the government which includes
the sales tax, import duty, and other incentives.

UniSZA
PROCESS 7: COLLECTING RELEVANT DATA
There are two fundamental approaches to gathering raw data. One is to ask
respondents on the variables of interest through the arranged interviews or
structured questionnaires. The other is to observe the respondents and record
the variables of interest using professional observers or high-tech devices.
Self-administered questionnaires, personal interviews, internet survey,
telephone interviews, and focus groups discussions are just some of the tools
researchers can use to collect data.
A major advantage of questioning techniques over observation is that the

questioning techniques allow the researcher to collect a wider array of data. Raw
data from a survey can pertain not only to the respondent’s current behavior but
also to his and her state of mind or intentions. In turn, observation methods can
be characterized as natural or contrived, disguised or undisguised, structured or
unstructured, direct or indirect, or human or mechanical. For example,
researchers might use trained human observers or a variety of mechanical
devices such as a video camera, tape recorder, audiometer, eye camera,
psycho-galvanometer to record behavior or events.
As the communication technology advances, researchers are moving

toward integrating the benefits of technology with existing questioning tools that
enable faster data acquisition and retrieval practices. The researchers can send
the questionnaire to respondents through e-mail, and the respondents can send
their response also through the e-mail. Every day, we can see the researcher
post the questions on TV screen and the respondents respond through Short
Messaging Services (SMS).
Census versus Sample
The objective of the most research projects is to obtain information about

the variables of the population. A population is any complete group of items or
people that the researcher wants to study. A sample is a subset or subgroup of
the population selected for the study. A census is an investigation of all elements
that make up the population. On the other hand, a sample study is a portion of
elements selected for the study. Presented below are the differences of criteria
between census and sample study.

UniSZA
Table 2.1 summarizes the conditions favoring the use of census over
sample. A research project usually has some budget and time constraints. A
census is appropriate when the budget amount is large and the time completion
is long. However, census is unrealistic if the population is large; for instance, the
population of certain consumer products in Malaysia. A census is also preferred
when high variance in terms of the characteristics of population. For example, a
study conducted to compare the usage of Proton cars versus Chevrolet in
Malaysia.
Table 2.1: The determination criteria of using census or sample

Criteria Census Sample Study
1. Budget requirement Large amount Small amount
2. Time frame requirement Long Short
3. Population size Small population Large population
4. Population characteristics Large variation Small variation
5. Cost for sampling errors High Low
6. Cost for non-sampling errors Low High
7. Nature of measurement Non-destructive Destructive items
8. Attention to individual cases No Yes
A census should be employed if the cost of sampling error is high. For

example, if the sample omitted a major group (students, for example) in the study
to determine the patronage of internet cafes, the results could be misleading
since one important group is not included in the study. A sample may be
preferred if the measurement process results in destruction of the elements
sampled. For example, product usage tests result in the damage to the product.
Sampling is also necessary to focus attention on individual case such as depth
interviews, focus group and so forth.
Pilot testing
A pilot testing is the survey conducted prior to the actual field study. The
purpose is to detect weaknesses in questionnaires and its instrumentation so that
correction can be made. From the pilot testing work, the researcher would know
the actual logistic requirement and budget for the actual research so that no
more interruption occurs when the actual research is carried out. Pilot testing is
very important since the researcher might have overlooked many aspects which
would affect the outcome of the research.

UniSZA
PROCESS 8: ANALYZING DATA USING APPROPRIATE METHOD OF

STATISTICAL ANALYSIS
In this step, the researcher begins the process of turning raw data into data
structures that can be used in generating meaningful and useful information
for the manager. The data analysis procedures can vary widely in
sophistication and complexity. For example, it can vary from simple frequency
distributions (which produces percentages) to sample statistical measures
(which produces mode, median, mean, range, standard deviation, and
standard error of the sample) to multivariate data analysis techniques (which
produce Anova, Regression, etc).
Selecting a Statistical Method for Testing Hypothesis

In an attempt to choose a particular test statistics to be employed, at least
three factors should be taken into consideration. The factors are: -
1. The number of groups involved in the study, whether it is one group, two
groups or more than two groups.
2. If two or more groups are involved, then the next question is: Are they
independent or dependent to each other?
3. The measurement scale employed in data collection, whether the scale is
nominal, ordinal, interval or ratio.
The above factors would determine which statistical analysis procedure is

appropriate and valid to be employed. Table 2.3 provides recommendations of
the types of statistical techniques that a researcher could use.
Table 2.3: The Recommended Statistical Techniques Based On Types of

Research
Types of Number of Variables Test Statistics

Research Involved
to use
Question
1) Univariate One variable t-test for mean,
Descriptive (e.g. exam score, salary, expenditure) F-test for variance
Question

UniSZA
2) Multivariate Two or more variables Chi-Square test for

Descriptive (e.g. gender, program of study) association, categorical
Question data analysis
3) Univariate One independent and one One-way ANOVA,

Difference dependent variable. Post-Hoc test for
Question The independent variable usually has difference between any
few levels (e.g. academic performance two groups.
1st class, 2nd class, 3rd class) LSD, Tukey, etc
4) Multivariate Three or more variables Multivariate Analysis of

Difference which usually more than two Variance (MANOVA),
Question independent variables and more than Structural Equation
one dependent variables Modelling (SEM)
5) Univariate One independent and one t-test for Pearson’s

Associational dependent variable. correlation, F-test for
Question The measurement is interval or ratio regression, SEM
6) Multivariate More than two independent and F-test for multiple

Associational dependent variables. regression, t-test for
Question The scale is interval or ratio regression coefficient,
SEM
Data analysis procedures will allow the researcher to:
1. Statistically test for the significance of difference between two or more

populations on the variable of interest in the study.
2. Statistically test for the significance of relationship between the variables of

interest in the study.
3. Statistically test the hypothesis of interdependence among the variables of

interest in the study.
4. Statistically test the significance of causal effects of independent variable on

their corresponding dependent variable.
5. Statistically test the significance of the hypothesized relationship among the

variables in the model.
UniSZA
The researcher must transform data structures into information. This step is
where the required information is created for the manager. It is in this step that
the researcher and the manager obtain the results of statistical testing and
creates a narrative description of the results. Narrative interpretation relates to
the integration of several variables into an understandable statement that the
decision maker can use to address the research question.
PROCESS 9: PREPARE THE FINAL REPORT TO THE MANAGEMENT
This section is very important and should be the main focus in any research.
The research does not have a value if the result is not properly presented and
interpreted in the findings. The report should be prepared and presented in
such a way that the manager can assess whether the objectives set earlier are
accomplished. The report presentation also should be able to address the
research questions clearly, so that no more ambiguity concerning the
underlying problem is left after the research project has been completed. You
can obtain a copy of final report of researches done previously in the library.
The most important element in the report is the highlight of information required
by the management to be used in their decision-making. The data analysis, the
interpretation of output, and the presentation of findings should be made based
on what has been agreed upon with the manager before the research project
begins. That is why the thorough understanding of the research process is
important before the researcher undertake any research project. Once the
researcher understands clearly the research process and the task involved, he
can begin writing a formal research proposal to undertake the research.
EXERCISE

The most important step in the research process is
1.
defining the problem.
Researchers should develop questions prior to formally
2.
defining the research problem.
3. Inadequate problem definition is the leading course of

UniSZA
failure in research project.

A good hypothesis should be stated in an investigative
4.
form.
The dependent variable is a variable that is to be
5.
predicted or explained.
6. The null hypothesis is always the hypothesis to be tested.
A hypothesis is a tentative statement regarding the
7.
problem.
A research proposal is a blueprint of how the researcher
8.
plans to carry out the research.
The most important step in the research process is the
9.
research design.
Researchers do not have to take a large sample if they
10.
want to carry out an exploratory research.
1. Research process includes ________________________.
a. A series of steps in identifying the variables to study, and the method for
conducting the study.
b. A step in finding conclusion to any problems.
c. Creating any databases.
2. The most important step in research process is:
a. Writing a clear and concise report

b. Specifying the sampling plan
c. Defining the problem
d. Choosing the research design
3. Defining the problem will depend on:
a. Discussions with decision makers

b. Interviews with experts
UniSZA
c. Secondary data analysis

d. All of the above factors
4. An essential step in defining a problem is:
a. Analysis of available secondary data

b. Highly structured interviews with industry experts
c. A formal questionnaire administered to decision makers
d. All of the above are correct.
5. The proper definition of a research problem begins with a ___________ to

provide perspective and then explores ___________ to provide guidelines on
how to proceed with the research effort.
a. Specific component; broad statements

b. Theoretical framework; broad statements
c. Broad statement; specific components
d. Theoretical framework; specific components
6. _____________ research is unstructured and exploratory in nature.
a. Primary
b. Secondary
c. Quantitative
d. Qualitative
7. Which is the best case to use descriptive study?
a. When causal studies are too expensive.

b. When formulating the hypothesis.
c. When little information is known about the problem.
d. When a profile of population is needed.
8. The purpose of literature review is to ________________.
a. Provide the background and critical basis for our research.

b. Provide conclusion for our findings.
UniSZA
c. Criticize the author’s credential.

9. In establishing a research approach, researchers should rely on ___________

to determine which key variable they should investigate.
a. Past results
b. The decision maker
c. Theory
d. Industry experts
10. Which of the following does not describe a hypothesis?
a. Null
b. Alternate
c. Non directional
d. Syndicated
11. Which of the following describes longitudinal study?
a. Collecting of data only

b. Collecting of data several times
c. Also known as one shot
12. A _________ is a study in which various segments of a population are

sampled and data is collected at a single moment in time.
a. Panel study
b. Cross-sectional study
c. Longitudinal study
d. Tracking study
13. The following sections are parts of a research proposal except___________
a. Introduction
b. Methodology
UniSZA
c. Literature review
d. Research finding
14. An essential step in defining a problem is:
a. Analysis of secondary data.

b. Conducting an initial interview.
c. Making an independent observation.
d. All of the above are true.
15. The formulation of research questions should be guided by:
a. Problem definitions
b. Theoretical frameworks
c. Analytical models
d. All of the above
16. An unproven statement about a factor of interest to a research is:
a. An analytical model
b. A hypothesis
c. A research question
d. Relevant characteristics
17. The formulation of hypotheses should be based on:
a. Problem definition.
b. Theoretical frameworks.
c. Analytical models.
d. Research questions.
18. When a researcher cannot find any literature to refer to as a basis for his
research, he should carry out ___________ research.
a. An exploratory
b. A conclusive
c. A descriptive
d. A causal
19. Which of the following methods is most likely to be used in an exploratory

research design?

UniSZA
a. Laboratory experiment.
b. Structured questionnaire.
c. Case study.
d. Pilot survey.
20. Which of the following is NOT the task involved in defining research
problems?
a. Discussion with decision makers

b. Developing research hypotheses
c. Secondary data analysis
d. Qualitative research
Essay question
1) A company made the following statement to define a training problem. In the

broadest sense, the problem is to determine the best ways the company
could train existing and potential users of personal computers.
a) Develop three possible research questions from the above problem

statement.
b) Develop the null and alternative hypothesis from one of the research
questions from part (a) above.
2) For each research question below, write down the possible null and research
hypotheses:
a) Do female students spend more amount of money for entertainment than

male students?
b) Do professionals who have no children and those with children have
different attitudes towards work and family?
3) A company director is concerned about the low productivity of his employees.

The literature on job performance mentioned four variables having significant
effect on job performance. The variables are the amount of training provided,
the remuneration, the motivation, and the job satisfaction. In several of the
articles, it was indicated that only if remuneration were attractive to the
employees, then their job performance would increase.
For the above situation:

UniSZA
a) Define the underlying problem.

b) Identify all the variables involved.
c) Develop the theoretical framework
d) Develop three possible research questions.
e) Develop three possible research hypotheses.
4) What are the criteria for a good hypothesis?
5) Formulate a hypothesis for the following research questions respectively.
a) What is the mean income of Proton Waja owners?

b) What age group that purchases product X?
c) What will be the result on sales if the price is increased by 20%?
6) Explain the components of a well-defined business research problem.
7) a) Define a hypothesis and explain its importance in research.

b) Briefly list all basic steps involved in testing hypothesis.
c) Differentiate between null hypothesis and alternative hypothesis. Provide
an example for each.
8) a) Briefly describe the stages involved in the research process.

b) List four advantages of doing a research proposal.
9) A medical researcher in a certain hospital carried out a study on cancer

victims to find out the probability they will successfully recover after
undergoing certain treatment. Based on literatures, he found three variables
to be important for recovery: early and correct diagnosis by the doctor, the
nurse carefully following up the doctors’ instructions and peaceful and quiet
environment. For the above situation:
a. Define the problem statement for this study

b. What are the variables involved in the study?
10) The president of “Mee Segera Sdn Bhd” suspects that half of his 500 male
and female workers are not very motivated to do their job. He also noticed
that those who are motivated are also the ones who have worked for more
than five years with the company. These workers have received salary
increase and have been rewarded accordingly over the years. Suppose you
are asked to conduct the study for this company,

UniSZA
a. State the problem statement for this study. Determine the relevant
variables and illustrate them using a theoretical framework.
b. State five research questions pertaining to the problem of the study.
c. State three research objectives of the study and describe how each
objective can be accomplished through appropriate statistical analyses.
d. Based on research questions and objectives of the study, design a short
questionnaire consisting of the following:
i. The required demographic profiles of respondents in this study

ii. The items measuring workers motivation to work at Mee Segera Sdn
Bhd

UniSZA
CHAPTER WRITING THE RESEARCH PROPOSAL
3
WHAT IS A RESEARCH PROPOSAL?
A research proposal is a detailed “blue-print” of the research process from the

beginning to the end. In every research activities, the research proposal
should be carefully prepared long before any research activity is allowed to
begin. The researcher must present the proposal to the management in order
to convince them regarding the significance of the research itself to the
company. Once the management is satisfied with the proposed research, they
would approve the research budget and anything related to the proposed
research itself.
Research proposal contains the detailed activities of how the researcher will
carry-out the proposed research project from the beginning until it is completed.
The research proposal serves as a direction for the researcher in doing his
research work. Doing research without the research proposal is like sailing the
ship without a captain! Sooner or later the ship would lose its direction and the
journey would not reach its final destination.
The general outline regarding the items that should be included when one is
writing a research proposal is presented in Table 3.1.
58
THE CHAPTERS AND SUB CHAPTERS IN ANY RESEARCH PROPOSAL
CHAPTER I- Introduction
1.1 Introductory paragraphs
1.2 Statement of problems
1.3 Objectives of the study
1.4 Research questions
1.5 Research hypotheses
1.6 Significance of the study
CHAPTER II- Literature Review

2.1 Discuss the results of previous studies in the same area
2.2 Discuss the variables involved in the previous studies
2.3 Discuss the hypothesized relationships among variables
2.4 Provide the theoretical model or framework for your study
2.5 Define the terms for every variable involved in the model
CHAPTER III- Methodology

3.1 Define properly the target Population for your study
3.2 Define the main variables involved in your study
3.3 Discuss the sampling technique to select a sample
3.4 Discuss the methods of data collection to obtain data
3.5 Explain the measuring instruments for the variables
3.6 Explain the statistical methods to analyze the data
3.7 Define the scope and limitations for the study
3.8 Draft your research schedule
59
THE STEP BY STEP GUIDE IN WRITING A RESEARCH PROPOSAL
Chapter 1 – Introduction
1.1 The Introductory Paragraph
The main intention here is to get your readers interested in your study.
Remember, you are selling your idea and you must catch the attention of your
readers and make them be with you all the way. It sets the stage for you to
proceed and elaborate on your subject. The introduction often contains
dramatic and general statements about the need for the research study. You
can use illustrations or quotes to set the tone. The objectives of the
introductory statements are:
1. To make the audience understand the urgent need for a solution to the
problem or the research to determine the right methods to remedy the
existing problem in order to bring the situation back to normal.
2. To highlight what has gone wrong and the potential threat it will bring to
the survival of an organization, and hence a research is required.
3. To highlight the gap in literature or the knowledge gap occurs. Thus, a

research work is needed in order to address that particular gap.
Example 1: An introductory paragraph

The passing of Private Higher Education and Institution Act (PHEIA 1996) by
the Malaysian Parliament has brought a competitive environment in higher
education industry to a level that no one has ever dreamt before. Today, there
are 20 public universities, 21 private universities, and almost 500 private
institutions of higher learning competing for potential candidates…..
60
1.2 The Problem Statement
The statement of problem is the centre attention of your research. It is just one
main sentence and supported by a few paragraphs of elaboration. In your
sentence, you should take an opportunity to:
1. Highlight an existing problem affecting the future survival of an

organization.
2. Highlight the present policy or method that no longer seems to be working.

3. Highlight something that need close attention such the increase in
customer complaints, declining in sales volume.
4. Highlight current trend, current need, or current technology that your

organization needs to adapt and change accordingly.
5. Highlight the changes in socio-economic variables such as life-style,

consumer income, etc that might affect the long-term survival of your
organization.
6. Highlight the literature gap. The gap can either be the specific research
has not been carried out previously, or certain important variables were
not studied in previous researches, or the population of the study has
changed significantly in term of their socio-economic status which requires
a new research, etc.
In short, you must present something critical that requires your proposed
research to be approved in order to ensure the long-term survival of the
organization and to contribute to the literature. Remember, if you fail to prove that
there is a problem, it simply means you fail to justify the need for your research.
Example 2: The Problem Statement

We live in a constantly changing environment. What happens today may be
completely different from what would happen tomorrow. Tomorrow was
essentially an extended trend-line from yesterday is no longer true nowadays!
The global economy means that your competitors are likely to come from
across the oceans, as from across towns. In the highly competitive
environment, the growth and survival of any business entity is not guaranteed
if the management is not efficient in term of time and cost.
61
Example 2a: The Problem Statement
The rate of increase in the number of higher educational institutions offering

postgraduate education has surpassed the rate of increase in the number of
students (customers) who can afford postgraduate education itself. The
situation has created stiff competition among universities for postgraduate
students.
Explain how the problem relates to business, social, or economic trends by

presenting data that demonstrate the scope and depth of a problem. Among the
evidence to support the problem are secondary data, graph showing declining
trend against time, increasing complaints, refunds by customers, and worrying
comments by the expert in the field.
1.3 The Objectives of the Study
The objective of the study is what you intend to accomplish in the study. You
need to tell your audience what you plan to achieve in solving the problem. If
you have more than one objective, you can list them specifically after your
introductory sentence.
Example 3: The Objectives of the Study
Specifically, this study aims to achieve the following objectives:

1) To identify the main factors that influence demand on Islamic Financing
2) To measure the relationship between quality and customer acceptance
3) To determine the effects of product image on customer repurchase
Remember; be practical when you state your research objectives. The

audience will have their own perception whether your objective is reasonable and
achievable within the time frame and the resources available. Please do not
fantasize your objectives.
62
1.4 The Research Questions
Research questions are „ambiguities‟ that the researcher would like to address
in his research. In the research, he intends to determine the answer to his
research questions. Research questions often came into researcher‟s mind
after he sets his research objectives. These questions are normally similar to
the questions from the audience regarding the underlying problem. A good
researcher is able to predict the questions that will come from the audience,
and will include these questions as his research questions when he prepares
his proposal. Hence, there is no more ambiguity arise after the research is
completed since all questions have been addressed by the research. If the
research is completed with many unanswered questions, then that particular
research is a failure.
Example 4: Research questions
1. Does the level of service provided to customers have significant influence

on their willingness to recommend the products to their families and
friends?
2. Does the corporate image of the organization have significant effects on
customer loyalty towards the organization?
3. Does the service recovery made by the top management manage to stop
customers from switching to other banks?
1.5 The Research Hypotheses
A hypothesis is a tentative statement made by a researcher regarding the

underlying problem in his research. It provides the direction of a research
study such as the variables of interest, the method of data analysis, and the
format of report writing. Hypothesis is a declarative statement by a researcher
and he will prove it with data analysis.
63
Example 5: (A case study)
A research study is designed to examine the relationship between the

environmental awareness and consumers‟ purchasing behaviour towards
green products. One of the objectives is to measure the influence of
environmental news on the choice of cosmetics in the market – a consumer
behaviour regarding cosmetic selection as far as environmental awareness is
concerned. A hypothesis here is simply what the researcher believes
regarding the strength of correlation between the amount of consumer
awareness towards environment and his tendency towards the cosmetic
selection. A researcher may hypothesize that if more information regarding the
increasing threat to the environment being exposed, the consumers, in turn,
will have higher tendency towards selecting the cosmetic products that is
environmental friendly. Therefore, he may hypothesize that there exist a
significant relationship between the environmental awareness and the
consumer behaviour towards green products. Remember, a research
hypothesis is something that the researcher wants to prove in his research. A
null hypothesis is a status quo, which assumes there is no significant
relationship between environmental awareness and cosmetic choice. In other
words, the choice of cosmetic products by consumers has nothing to do with
their knowledge concerning the potential environmental threat posed by that
product.
The researcher‟s hypothesis statement will be:
Ha: There is a significant relationship between customers‟ environmental

awareness and their selection of cosmetic products in the market.
Therefore, the researcher would measure two variables in his data collection
process namely, the amount of environmental news exposed to customers,
and the consumers‟ intention to purchase certain cosmetic products.
Normally there are several hypotheses in every research study. If the

researcher manages to prove his hypothesis, he can make his point to the
management that the amount of environmental awareness has a significant
influence on consumers‟ demand on green products. In other words, to
increase the demand for green products, the exposure on environmental
issues and the contribution of certain consumer products towards
environmental degradation should be made known to consumers.
64
Another hypothesis here is concerning the relationship between the level of
consumers‟ education and their level of awareness towards environmental
issues. In other words, the researcher may hypothesize that consumers with
different levels of education may have different perception on environmental
issue, and hence will act differently concerning product selection. Hence his
hypothesis would become:
Ha: There is a significant relationship between consumers‟ level of

education and their level of environmental awareness
Here the researcher will measure two variables in his data collection namely,
the level of consumers‟ education, and the level of consumers‟ awareness
concerning environmental degradation.
Hypothesis statement does not have to be long. It is enough to put the

variables involved – education and environmental awareness, hence it is
adequately understood if we put this way:
Ha: There is a significant relationship between respondents‟ education and

environmental awareness
Again, here the researcher will measure two variables in his data collection
namely, the level of consumers‟ education, and the level of environmental
awareness. The study will measure the strength of relationship between the
two variables.
A researcher may derive a hypothesis from his own observation, his previous
reading of literature, or from his own instinct and intuition regarding the issue.
A shrewd researcher manages to sense something when faced with certain
research questions. Actually, hypothesis facilitates in answering the research
question.
65
Example 6: The Research questions and their corresponding hypotheses
Research Question 1:
How significant is the influence of service quality on customer satisfaction?
Research hypothesis to address RQ 1
H1: Service quality has a significant influence on customer satisfaction.
How significant does the influence of service quality on customer loyalty?
H2: Service quality has a significant influence on customer loyalty
How significant does the influence of customer satisfaction on customer loyalty?
H3: Customer satisfaction has a significant influence on customer loyalty
Which demographic variable, if any, moderates the relationship between service quality
and customer satisfaction?
H1: The respondents‟ age group moderates the relationship between service quality and
customer satisfaction
H2: The respondents‟ gender group moderates the relationship between service quality
and customer satisfaction
H3: The respondents‟ occupation moderates the relationship between service quality
H4: The respondents‟ education level moderates the relationship between service
quality and customer satisfaction
H5: The respondents‟ income level moderates the relationship between service quality
66
Which demographic variable, if any, moderates the relationship between customer
satisfaction and customer loyalty?
Hypotheses to address RQ 5
H1: The respondents‟ age group moderates the relationship between customer
satisfaction and customer loyalty
H2: The respondents‟ gender group moderates the relationship between customer
H3: The respondents‟ occupation moderates the relationship between customer
H4: The respondents‟ education level moderates the relationship between customer
H5: The respondents‟ income level moderates the relationship between customer
*** Please note that the hypotheses must appear from research question.
From hypothesis statement, the researcher will determine the variables required
in the study, how these variables will be measured, method of data collection,
and the appropriate statistical analysis employed to test the hypothesis. That is
how the hypothesis provides a clear direction for the research.
1.6 The Significance of the Study

This section reflects the „value‟ of your proposed study. You should highlight how
your proposed study relates to larger and much critical issues in the future so that
it is justified to approve your proposal. You should take into consideration of the
following questions when presenting the significance of your study:
Why the proposed study is important?
Example 7: The significance of the study

The study would provide a clear sense of direction for the public universities,
especially UiTM, to plan the programs towards achieving the loyalty of their
outgoing undergraduates. In the stiff competition among universities for the
potential candidates into the post-graduate programs, the customer loyalty of
the outgoing undergraduates is extremely important. Thus, the
undergraduates‟ loyalty towards their university is the focus in this study since
it may bring tremendous benefits to the university in the future.
67
Chapter II - Review of Literatures Related To the Study
2.1 The Literature Review

Before starting to write your literature, you should assume the likes of
questions that will come from your audience soon after reading your literature.
As a researcher, you must obtain as much literature as possible which

relates to the topic you want to study. This is very important because you want to
revise what the previous researchers have done in the area, the variables used,
and the findings so that you will not duplicate their work.
Example 8: A Sample of Literature Review on Customer Satisfaction
The role of customer satisfaction as a performance indicator of service quality

was exposed by various studies across industries (Stauss and Neuhas, 1997;
Bloemer et al., 1998). However, the exact relationship between service quality
and customer satisfaction is still being debated and shrouded with uncertainty
(Sureshchandar et al., 2002). Many efforts have been done in measuring the
relationship between service quality and customer loyalty (Parasuraman et al.,
1994; Zeithaml et al., 1996; de Ruyter et al., 1998; Bloemer et al., 1999;
Zainudin & Zaihan, 2004). In measuring customer loyalty, Parasuraman et al.
(1994) have identified five dimensions of customer loyalty namely, loyalty to
the company, external response to a problem, propensity to switch, willingness
to pay more, and internal response to a problem.
2.2 The Definition of Terms

You should provide simple definition for all terms that may bring unclear
meaning to your reader. Do not let your audience trying to figure out what you
meant with certain terms used in your proposal. Identify a few key words and
provide definition for each of them.
Example 9: Definition of Terms

Service Quality is defined as a customer‟s evaluation of service after their
consumption.
Customer Satisfaction is defined as a general evaluation based on the result
of the product or service perceived after the consumption compared with
earlier expectations.
68
Customer Loyalty is defined as a customer who would continue to purchase
goods or services from the same company whenever possible, and who
continues to maintain positive attitudes towards goods and services from the
company
2.3 The Theorized Relationship among Variables in the Study

As a researcher, you should be able to expect how these variables relate to
each other in a logical manner. Specifically, which variable should act as
dependent variable, independent variable, mediating or intervening variable, and
moderating variable in your study?
Example 10: The theorized relationship among variables.
From a search in literature, the study discovers researches in many service

industries focused on the complex relationship among service quality,
corporate image, customer satisfaction, and customer loyalty towards the
service provider. The research on factors contributing to customer loyalty is
considered important due to its impact on the long-term growth and survival of
service providers in their respective industry. Since both service quality and
corporate image have a bearing on customer satisfaction, the study intends to
examine the two independent constructs (service quality and corporate image)
separately with customer satisfaction as a mediating construct, and customer
loyalty as an intended dependent construct. The importance of customer
loyalty has received a lot of attention by many researchers in higher education
industry (Hennig-Thurau et al., 2001; Nguyen and Leblanc, 2001). The
findings by Caruana (2002) on banking customers, and Zainudin and Zaihan
(2004) on healthcare customers, for example, prove that customer loyalty is
achievable mainly through customer satisfaction. In other words, a clearer
understanding as to the sequence of inter-relationship among service quality
provided by the university, corporate image of the university, students‟
satisfaction with the university and students‟ loyalty towards their university is
becoming paramount. The findings would help to ensure the better targeting of
limited marketing resources of the university. Hence, the above idea would
become the central theme of this study.
69
2.4 The Theoretical Framework of the Study
The theoretical framework for your study is the schematic diagram drawn to
indicate how you believe the variables involved in your study should relate to
each other. This is important since it will become the heart of your study.
Service
Quality
Customer Customer
Satisfaction Loyalty
Corporate
Image
The constructs in the model are:
1. Dependent construct (Y1)

The dependent construct is students‟ loyalty towards their university (Y1).
This is the construct of primary interest in the study. Students‟ loyalty is
measured through the extent of their willingness to continue the post-
graduate education at the university, and also their willingness to provide
positive word of mouth concerning the university. The study wants to
examine the effect of service quality provided by the university, corporate
image of the university, and students‟ satisfaction with the university on
students‟ loyalty towards their university in the future.
2. Independent constructs (X1 , X2 , X3)

The independent constructs are service quality provided by the university
(X1), corporate image of the university (X2), and students‟ satisfaction with
the university (X3). This study treats these three constructs separately to
examine the influence of each construct on students‟ loyalty towards their
university.
70
Firstly, the service quality of a university is hypothesized to influence
students‟ loyalty towards their university. Mathematically, the equation is:
Y1i = β0 + β1X1i + εi.................................................(i)
Secondly, the corporate image of a university is also hypothesized to

influence students‟ loyalty towards their university. Mathematically, the
equation is:
Y1i = β0 + β2X2i + εi.................................................(ii)
Thirdly, the study hypothesized that students‟ satisfaction with their

experience at the university would influence their loyalty towards the
university in the future. Mathematically, the equation is:
Y1i = β0 + β3X3i + εi.................................................(iii)
3 Mediating construct (Y2)

The study is also interested to examine the effect of mediating construct in the
model linking service quality (X1) and corporate image (X2) to students‟ loyalty
(Y1). The mediating construct is students‟ satisfaction with their experience at
the university. Students‟ satisfaction with the university, denoted as Y 2 is
theorized to be the positive effect of service quality provided by the university
(X1) and also the corporate image of the university (X2).
Mathematically, the equation is:
Y2i = β0 + β1X1i + εi.................................................(iv)
Y2i = β0 + β2X2i + εi.................................................(v)
The theorized effect of students‟ satisfaction on students‟ loyalty is already
stated in equation (iii).
Chapter III - The Research Methodology
In this section, you should describe clearly your research plan. The
audience wants to know your plan of work and the activities necessary for the
completion of your project. To keep the audience with you, begin with a few
attractive introductory paragraphs that restate the research problem, research
objectives, and your research questions.
71
3.1 The Target Population
Target population is the element in general population that you want to cover
in your study. For example, if your study wants to cover hardcore poor in the
community, then your population should be the people who earn less than
RM475 per month. You need to define your target population precisely and
explain the rationale for selecting that specific population for your study. Make
sure you choose the right population; otherwise, your conclusion might be
questionable since you might be shooting the wrong target.
Example 11: The population of the study
The population of this study is defined as university students studying in the

faculty of business and management in the public and private universities in
Malaysia. In this case, you should observe the specific scope such as:
1. University students in the public universities.
2. Faculty of business and management only.
Example 11a: The population of the study
The target population for this study is the undergraduate students in the
Faculty of Business and Management at Universiti Teknologi MARA (UiTM).
UiTM has branch campuses in all states in the country. Since these branch
campuses are widely scattered in term of geographical location, the study
applied the cluster sampling method whereby the peninsular Malaysia is
divided into five geographical regions namely Eastern Region, Northern
Region, Southern Region, Western Region, and East Malaysia.
The Eastern Region consists of Kelantan, Terengganu, and Pahang

campuses, the northern region consists of Perlis, Kedah, Penang, and Perak
campuses, the Southern Region consists of Negeri Sembilan, Melaka, and
Johore campuses, the Western Region consists of head-quarter in Shah Alam,
and East Malaysia consists of Sabah and Sarawak Campuses.
The regions are divided in such a way in order to ensure homogeneity

among campus within one region, hence reducing the sampling error within that
particular region. Since each region possesses almost homogeneous
72
characteristics in term of location, environment, facilities, and climate condition,
the study could select one or more than one branch campus to represent the
respective region.
3.2 The Sampling and Data Collection Procedure
Explain in detail how you plan to draw a representative sample from the
population. Since the data from this sample will be used to draw conclusion on
the overall population, tell your audience how you ensure the selection process is
representative and unbiased. What are your sampling method, and the rationale
for using that method? Convince the audience of your knowledge regarding
statistical validity and reliability. Tell specifically the sampling process you plan to
execute. The process include sampling frame, how you obtain the sample, how
you ensure the sample is randomly selected, how you contact the respondents,
the instruments to use for data collection, who does the data collection for you,
are they qualified for the job, how do you ensure the data collection will take
place accordingly without prejudice, bias, or cheatings etc.
Example 12: The sampling and data collection procedure
Once the branch campus is selected, the list of students in the faculty of business
and management is obtained from the office of academic affairs at the respective
branch. The study obtained only the list of students in semester 3 and above
since the students who have been in the university longer are in a better position
to provide their opinion for items under the respective construct of the study.
In other words, these students have been experiencing multiple service

encounters at the university, and they have “known” the university very well.
Hence, the list of students from semester three and above would be the sampling
frame for that particular campus in which the respondents are randomly selected
using certain sampling procedure.
Below are the steps taken in the sampling and data collection processes:
1. The permission to collect data is obtained from the faculty of business and
management at the respective campus. At the same time, the list of students
to be used as a sampling frame for the study is also obtained from the faculty
together with the class schedule where the students are having their lecture.
73
2. The respondents are selected from the sampling frame using the random
numbers generated through the computer. For example, if the sampling frame
consists of 100 students and the study decided to take only 20 students as
respondents, then the 20 numbers are generated using the following
computer statement (Rand 20 1-100). The computer would generate
randomly 20 numbers in the range of 1 to 100. Let‟s say the 20 generated
numbers are 5, 9, 11, 17, 23, 25, 36, 39, 44, 47, 51, 58, 66, 68, 73, 78, 88, 89,
96, and 98. Then these numbers are matched with the list in the sampling
frame. The students whose number on the list matched the numbers
generated by the computer would become the respondents for the study.
3. Once the respondents are identified, the research assistants would proceed
to the respective lecture room during the time where the students are having
their lecture session. The research assistants would obtain the permission
from the respective lecturer to distribute the questionnaires to the respective
respondents.
4. If the research assistants failed to find the respondents in that particular time,
they would proceed to the next lecture session or with the help of their friends,
the questionnaire is delivered to the respective respondent.
5. The respondent can respond to the questionnaire at their convenient time and
mail the completed questionnaire using the self-addressed envelope. In order
to ensure confidentiality of the responses, each questionnaire is enclosed with
a cover letter that explains clearly the purpose of the study. The respondents
do not need to provide any identification on the questionnaire. They could
also submit the completed questionnaire using the sealed envelope to the
representative at the office of academic affairs the following day.
The study employed the above method in order to allow respondents to attend
the questionnaires without any pressure so that whatever response provided
reflects the true opinion from the respondents themselves. Therefore, the
response bias due to time constraint and the presence of researcher would not
arise.
74
3.3 The Measuring Instruments employed in the Study
Normally, researchers employed the survey instruments designed by

previous researcher who are prominent in the respective field. For example, the
prominent researchers have designed instruments to measure service quality,
customer satisfaction, customer loyalty, etc. If you want to carry out researches
on those topics, you need to adopt and customize the items in the questionnaire
to suit your research area. You need to mention the source of the instrument and
describe the previous validity and reliability assessments. After data is obtained,
you need to run the confirmatory factor analysis (CFA) to examine whether the
items conform to the respective construct. In CFA, the researcher could check
the unidimensionality, validity, and reliability of the instruments prior to doing the
statistical analysis.
Example 13: The measuring instruments to be used in this study
The study adopts the questionnaires developed by Zainudin (2007) to

measure service quality, corporate image, customer satisfaction, and
customer loyalty in banking industry. Since Zainudin (2007) developed the
instrument for higher education industry, this study would customize the items
accordingly in order to suit customers in the banking industry.
3.4 The Assumptions Required
All processes and procedures must follow certain assumptions to avoid the
threat of invalid methodology. For example, if the sampling distribution of your
variables does not meet the required assumptions for statistical distribution, then
the statistical methods used in the data analysis is invalid. Hence, all conclusion
based on statistical test is considered to be invalid.
Some of the required assumptions:

1. -The population from which the sample is selected is normally distributed
2. -Respondent is randomly selected, no bias in the selection process, and
hence it is a representative of the population under study
3. -Respondents are given a wide variety of response choice. The choice of
response is from 1 (for strongly disagree with the statement) to 10 (for
strongly agree with the statement)
75
4. -Respondents answer the questionnaire at their own convenience without
bias, prejudice, threat, influence or pressure from any party.
3.5 The Procedure for Data Analysis
Describe the procedure you plan for data analysis. Explain how you will
treat the questionnaires – the checking, editing, coding, and data transcribing.
Describe how you will address each individual research question. What
variables are you looking for? How these variables provide answers to your
research questions? What statistical computations are required? State the types
of statistical test you plan to perform. This is important when you have certain
hypotheses to prove.
Example 14: The procedures for analysis of data
The researcher analyzed the data using the Statistical Package for Social
Science version 18 (SPSS 18.0) and AMOS program version 18.0 (AMOS 18).
This SPSS helps to facilitate data clearing, and checking for logical
inconsistencies. The researcher carried out data mining procedure before the
actual data analysis is performed. In order to make your proposal more
convincing, particularly your proposed hypothesis; you should include the
relevant statistical analysis procedure that you want to employ in testing your
proposed hypotheses.
76
Example 14a: The Hypothesis and its Statistical Analysis
THE RESEARCH HYPOTHESIS STATISTICAL

ANALYSIS
Ha: Service quality has a significant and direct effect on customer satisfaction Regression or
Path analysis in SEM
Ha: Service quality has a significant and direct effect on customer loyalty Regression or
Ha: Corporate image has a significant and direct effect on customer satisfaction Regression or
Ha: Corporate image has a significant and direct effect on customer loyalty Regression or
Ha: Customer satisfaction has a significant and direct effect on customer loyalty Regression or
Ha: The respondents‟ age moderates the relationship between customer Multi-group CFA
satisfaction and customer loyalty In SEM
Ha: The respondents‟ gender moderates the relationship between customer Multi-group CFA
Ha: The respondents‟ education moderates the relationship between customer Multi-group CFA
3.6 The Scope and Limitations of the Study
All research studies have limitation and a finite scope. Since the results of your
study are not valid for the whole world, not applicable across all industries, not
apply to all population, and might not be true under all circumstances, you
should think carefully about the scope and limitations of your study. It is
irritating to you when someone tells you that your research is not valid under
certain conditions, which is within your limitation!
77
Example 15: The scope of the study
Like in other studies, this particular study has its own limitations, both in scope
and methodology. First of all, the study was carried out in the Faculty of
Business and Management. Thus, the result might only be generalizable to the
above population. In other words, the findings might be different if the scope is
increased to include more faculties since different faculties might pose
different characteristics.
Methodologically, this study only measures the perception of respondents for

each item on a questionnaire. Should the expectation score also be taken into
account, the results would be more informative when the gap score between
perception and expectation is computed. Thus, the management of a
university would become more concerned when presented with the extent of a
gap between their actual performance and the expectation of their customers.
Making the gaps score handy for all components, the university could plan the
appropriate program to close the respective gaps.
Another limitation is this study employed the cross-sectional design whereby

the students‟ perception score is only measured once throughout the study.
Since higher education is a long term commitment, and this study found the
length of time in a university has a significant effect on the perception score,
the result would be more informative if the longitudinal design is employed.
In longitudinal design, the perception score of the same respondents is

assessed at different time interval throughout their study. Using this method,
the management of a university could monitor how their students‟ perception
changes against time when they are consuming more service and getting
more information concerning the university.
3.7 The Research Schedule

Research schedule is a tentative time schedule estimated by a researcher
for every activity involved from the beginning until the end. As a researcher, you
must be able to estimate in detail how much time you require to complete every
activity in every step in your research. This is important to convince the audience
that you can complete your study on time. At the same time, it shows that you
know every activity very well, the time required to complete that particular activity,
and the overall time you need to submit the final report. Remember, the time
estimate must be reasonable because your audiences might know the
exaggerated figures! This will erase their confidence in you.
78
EXERCISES
1. You have been approached by the chief editor of “People Today” to carry
out a research study. The magazine has been unsuccessful in attracting
corporate officers to purchase the magazine. As a research consultant, you
need to understand and gain insight on the problem of the study. You
believe that an exploratory design will help them to identify the problem that
is causing the resistance among the corporate customers. You are also
required to prepare a research proposal for the marketing manager.
a) Discuss the steps in preparing the research proposal.
b) Describe the exploratory technique that you would use in defining the
problem of the study and explain how it is done. Is it appropriate for you
to proceed using the descriptive design? Why?
c) State two research questions that could help you to identify and solve
the chief editor‟s problem.
d) Construct three questions that would help you to obtain information

from the customers with regard to their desire to purchase the
magazine.
2. Describe how you would develop the theoretical framework for your study.
3. State and describe three important variables that are needed in developing a
theoretical framework. Illustrate using a schematic diagram.
79
Chapter 4: Sampling Design
CHAPTER THE SAMPLING DESIGN

4
BASIC SAMPLING TERMINOLOGY
Population and Sample
Population is a total set of people or items of interest in a research project.

Population is the entirety of the same group. The subset of population is
sample. In research, sometimes it is impossible for the researcher to study all
elements in the population. This is because the number of elements in the
population is large and the researcher has constraints of time and resources to
study all of them. Hence, the researcher needs a sample. A sample is a
portion of elements in the population selected to represent the population
under study. The finding from a sample is generalized on the whole
population. That is why the researcher needs to apply the proper sampling
technique to obtain sample so that it is a representative of a population. There
are many sampling techniques available to help researchers obtain sample
from a population.
The main reasons for studying sample instead of the whole population:
1. Cost constraint: Working with a sample does not incur much expenses
compared to working with the whole population.
2. Time constraint: Working with sample does not require longer time
compared to working with the whole population.
3. Manpower constraint: Working with sample does not require as many

manpower compared to working with the whole population.
4. The precision of measurement: The degree of precision in measuring a

sample increased because it involves dealing with a small number of
elements compared to elements of the whole population.
5. Process of analysis will be more systematic since the number of sample

to deal with is small.
UniSZA
6. Saving the elements in the population if the study requires the

destruction of sample in measuring the performance.
Elements
Elements are the selected respondents or objects from the defined target
population from which the researcher will study by collecting the required data
and information.
Sampling Units
Sampling units are the elements in the target population available to be

selected randomly during the sampling process. The researcher selects the
sample for the study from the sampling units by using an appropriate sampling
technique. The sampling technique employed would ensure representativeness
of the sample towards its population.
Sampling Frame
Sampling frame is the list of all eligible sampling units and elements in the
population from where the sample can be selected. Examples of sampling
frame are a phone directory, a list of companies selling health products, a list of
diabetic patients, a list of voters in the area, a list of houses in the district, a list
of credit card holders who apply for additional credit limit, etc. Using the
appropriate sampling technique the researcher can select the sample from this
sampling frame for the study.
CHOOSING THE APPROPRIATE SAMPLING DESIGN
The choice of sampling design to be employed in the study depends very much
on certain factors deemed to be crucial. Among the factors that should be
considered are:
1. Research Objectives and Research Questions
A full understanding of the research objectives and the research questions

provides the initial guidelines for determining the appropriate sampling
design to be employed in the study. If the research objectives include the
intention to generalize the findings to the defined target population, then the
UniSZA
researcher must seriously consider using some type of probability sampling

method rather than a non-probability sampling method. This is because the
selection of sample using the probability sampling method occurs at random
and the representativeness of the sample on the population cannot be
questioned. However, if the researcher uses the non-probability sampling
method, the selection of sample is not at random (sometimes biased), and
hence the findings cannot be generalized on the whole population since the
sample itself is not a representative of a population in the first place.
2. Degree of Accuracy Required
Accuracy is the degree to which bias is absent from the sample. There is no
variation in measures due to any influences that causes skewness in the
distribution with an accurate sample. The degree of accuracy required or the
researcher’s tolerance for error may vary from one project to another. This is
true especially when cost constraint or other considerations may be traded
off for a reduction in accuracy.
3. Availability of Resources
If the researcher’s financial and human resources are restricted, these
limitations will most certainly eliminate the more sophisticated time-
consuming and complex probability sampling methods to be employed.
Marketing practitioners who are concerned by the cost versus the precision
of the information often choose the non-probability sampling method for their
study rather than do not conduct any study at all. This is because most of
marketing research projects is concerning the customers’ opinion, customers’
preference, and customers’ perception about certain product or service in the
market; and these types of researches do not require the measurement
which is too precise like the scientific research.
4. Time Constraint
Researchers who need to complete a project in a short time will be more
likely to select a simple, less time-consuming sampling method rather than a
more complex and accurate method. For instance, a telecommunication
company would like to introduce a new plan, which requires them to do a
research on that matter. If the study is too time-consuming, several
prospective customers might move away from the company to their
competitors. So, doing the research within the time frame, the company will

UniSZA
not only keep current customers coming back but also encourage
prospective customers to buy the new plan.
5. Advanced Knowledge of the Target Population

In many cases, a complete list of population elements or a sampling frame is
not available to the researcher. A lack of sampling frame may automatically
rule out the use of systematic random sampling, stratified random sampling,
or any other type of probability sampling method which requires the complete
list of population elements before the sample can be selected.
6. Scope of the Research

The researcher needs to discuss with the manager and agree on the scope
of a research project. This is important since the scope of the research
project will influence the choice of the sampling method. Be sure of the
geographical area to be covered, the demographic characteristics of the
population, and the specific focus for the study. When the target population
elements are scattered in the unequally distributed geographical area, a
cluster sampling method may become much more suitable than other
sampling methods. Generally speaking, the broader the geographical scope
of the research project, the more extensive and complex the sampling
method becomes to ensure proper representation of the target population.
7. The Required Statistical Analysis

In deciding the sampling method to be used in the study, the researcher
should consider the statistical analysis techniques that will be employed to
analyze the data. The choice of sampling method that does not fulfill the
requirement for the method of analysis may land the researcher into trouble
since the analysis might be invalid. This is because certain data analysis
technique is only valid under certain condition concerning the distribution of
sample. Sometimes the researchers are using the non-probability sampling
methods to collect the data, but instead, using the statistical tests which
require the distribution of sample to be randomly and normally distributed. Of
course the results will be statistically invalid since it pre-requisite is not met.
UniSZA
THE THEORETICAL TOOLS TO ASSESS THE QUALITY OF SAMPLE
Errors can be classified as either sampling error or non-sampling error. Random

sampling error can be detected by observing the difference between the sample
results and the census conducted using identical procedures. Two of the
difficulties associated with the detection of sampling errors are:
1. A census (study on the whole population) is seldom carried out in survey

research. Hence, there will be no basis to detect the sampling error.
2. Sampling error can be determined only after the sample is drawn and data
collection has been completed.
What is sampling error? Sampling error is any type of bias attributed to

mistakes made by the researcher either in the sampling process or the size of
sample used. Random sampling error occurs due to some random variations in
the scientific selection of sampling units.
From the principles of the Central Limit Theorem (CLT), the degree of
sampling error and its impact can be reduced by increasing the size of sample to
be taken from its population. CLT is the theorem which states that for any target
population, the sampling distribution of the mean and proportion derived from a
random sample will be approximately normally distributed provided that the
sample size is sufficiently large. The CLT plays an important role in
understanding the concepts of sampling error, the statistical significance, and the
determination of sample size.
Of course, it is impossible to select a sample that represents the

populations perfectly, that shows why the sampling error occurs. The lack of fit
between sample and the population is expressed as sampling error. The
sampling error is the magnitude of difference between the characteristics of a
sample and the characteristics of the population. The researcher can minimize
the amount of sampling error through the following procedure:
1. By using the appropriate sampling technique when drawing the sample from
its population.
2. By increasing the size of sample or respondents to be obtained from its
population.
Sample size affects data quality and generalizability:

UniSZA
1. If the researcher is using the probability-sampling plan, the sample size is

determined using either appropriate formulas or using the experience from
previous researchers.
2. If the researcher is using the non-probability sampling, it is up to his own

judgment since there is no specific formula.
Determining the appropriate sample size is not an easy task. The

researcher must consider how precise the estimates must be and how much time
and money are available to collect the required data, since data collection is
generally one of the most expensive components in the study.
THE IMPORTANT FACTORS TO DETERMINE THE APPROPRIATE SAMPLE

SIZE:
1. The variability of the population characteristic under investigation. The greater

the variability of the characteristic, the larger the size of the sample necessary
for that research.
2. The level of confidence desired in the estimate (CL). The higher the level of
confidence desired in the findings, the larger the sample size needed.
Confidence interval represents a statistical range of values within which the
true value of the target population parameter is expected to lie.
3. The degree of precision desired in estimating the population characteristic.

The more precise the result from the study required by the manager, the
larger the sample size need to be drawn from its population.
The non-sampling errors are biases that may occur in any research study
regardless of whether a sample or a census is used. These errors may occur at
any stage of the research process.
THE TYPES OF SAMPLING DESIGN
Sampling techniques consist of two types that can be classified as probability

and non-probability sampling (See Figure 4.1). The major difference between
these two techniques is that - in the probability sampling, the sample is selected
by chance; while, in the non-probability sampling, the sample is selected based
on the personal judgment of the researcher rather than chance.
UniSZA
Sampling Design
Probability Non-probability
Figure 4.1: The Classification of Sampling Design
TYPES OF PROBABILITY SAMPLING DESIGN
Probability sampling methods are the most commonly used method because the
selection of participants is determined at random. In the probability sampling, any
element of the population has an equal chance or equal probability of being
selected as a sample. Thus the representativeness of a sample cannot be
questioned.
Probability
Simple Random Systematic Stratified Cluster Multistage

Sampling Sampling Sampling Sampling Sampling
Proportionate Disproportionate
Figure 4.2: Types of Probability Sampling Techniques

UniSZA
The types of probability sampling techniques are:
1. The Simple Random Sampling
This sampling technique is applicable only if the target population is homogenous in

term of the characteristics of interest for the study. The characteristics of the study
can be in term of gender, socio-economic status, ethnic group, the culture, and the
religion of the potential respondents. This sampling procedure allows for the equal
chance for any element in the population to be selected as a sample for the study. In
other words, a probability sampling procedure ensures every sampling unit in the
target population has a known, equal, nonzero chance of being selected.
The process of simple random sampling consists of the following four steps:
a. The definition of target population from which the researcher wants to

select a sample for research.
b. The list of all elements in the population. This list is sometimes called a
sampling frame form which the sample will be selected. For example, a
complete list of marketing students in UiTM is a sampling frame.
c. The assignment of number to each element of the population. In this

example, the numerical list represents the member of the population
(list of marketing students).
d. The use of a criterion to select the sample:
a) Obtain the table of random numbers (manual procedure).
b) Obtain random numbers from a computer (automatic procedure).
c) Match the random numbers obtained to the numerical list of

students and select these students as the respondents for data
collection or further research activities.
d) The respondents selected this way are totally random; hence the
problem of biasness does not arise.

UniSZA
The simple random sampling might not be practical if: -
a. It requires a sampling frame that is often not available.
b. The sampling fails to use all the information of a population that might
result a waste of the design.
c. It may take lots of time and not cost-effective.
2. The Systematic Sampling
Like simple random sampling technique, the systematic sampling technique is also
applicable for the homogenous population. In this sampling technique, the element
which falls on the kth number on the list is chosen as a respondent. This sampling
procedure fixed the increment in order to determine the following respondent, for
example, every 5th person on the list is selected as respondents.
The process of systematic sampling consists of the following four steps:
a. Selection process
i Determine the size of population (N).
ii Determine the required sample size (n).
iii Determine the interval (k) when k is the ratio between the size of
population and size of sample (k = N/n). For example if N = 500 and
n = 25. In this case, K = 500/25 = 20.
b. Select one random number between 1 to k, in this

example from 1 to 20. Lets say, number 5 happens to be randomly
selected here. Then the first respondent for the study is person number 5
on the sampling frame.
c. Then, the following respondents are K + 5, K + (K + 5),

and so on until the 20th respondent is selected since the sample size is 20.
d. In this example, the 20 respondents would be obtained based on the

following numbers on the sampling frame:

UniSZA
5 25 45 65 85
105 125 145 165 185
205 225 245 265 285
305 325 345 365 385
405 425 445 465 485
3. The Stratified Sampling
When the characteristic of the population under study is not homogenous, then the
researcher has to stratify them into homogenous groups. When the characteristic in
each group is homogenous, then the researcher can employ either the simple
random sampling or systematic sampling technique to randomly select the
respondents. The stratified sampling is the process of selecting a sample that
represents each stratum in a population. The researcher can stratify the
heterogeneous population into homogeneous population within a stratum in terms of
socio-economic status, ethnicity, religion, gender, marital status, type of houses, etc.
The process of stratified sampling is:
a. Specify the strata (elements within the same strata is homogenous, and
elements between strata is heterogeneous).
b. Assign the elements according to the strata having the similar

characteristics of interest in the study.
c. Now we can select the sample within each strata using simple random
sampling since the population is already uniform.
d. The number of samples from each stratum must be proportionate to the

number of units in that stratum. For example, a study is to be conducted
on the investors who made their investment in the Amanah Saham
Bumiputera (ASB) in January 2008. The study wants to look specifically on
how the marital status of investors influences their investment decision in
ASB.

UniSZA
e. In this case, the population of investors is heterogeneous in terms of their

marital status (namely single, married, divorced, and widowed). Thus, the
researcher needs to stratify them according to their marital status before
the sample from each stratum can be selected. Now the investors in the
same stratum have similar characteristics. Suppose the researcher
decided to obtain 100 investors or 10% of the population of 1000 investors
as the respondents for his study.
The stratification process is shown below:
Sample from
Whole Population in each stratum
Population each stratum
Single (10)
Total Single (100)
investors
who bought Married (400) Married (40)
the ASB unit
trust in Jan Divorced (300)
2008 is 1000 Divorced (30)
Widowed (200)
Total = 1000
Widowed (20)
Total = 1000
Total = 100
Figure 4.3: The Stratified Sampling Process
The advantage of stratified sampling is the estimate obtained from the sample
is more accurate since it ensures all strata are represented in the sample.
The researcher also can compare whether different strata perform differently
regarding certain variable of interest.
4. The Cluster Sampling
Cluster sampling is the probability sampling technique applicable when the study
covers a large geographical area, the population is not homogeneous, and the
sampling frame is not available. The clusters can be the area in a residential district
(section 1, section 2, etc), the growth corridor in the country (WPI, NCER, ECER,
UniSZA
etc), or the developmental region such as eastern region, northern region, central
region, and southern region.
The process of cluster sampling is:
a. First of all, the researcher needs to specify the clusters of interest within
the population. The researcher can classify certain groups into a cluster in
one geographical area.
b. For example, UiTM has campuses scattered across the country. The spread of
UiTM campuses can be classified based on the respective region namely eastern
region (Kelantan, Terengganu, Pahang), northern region (Perlis, Kedah, Penang,
Perak), central region (Shah Alam, Petaling Jaya, Meru), southern region (Johor,
Melaka, Negeri Sembilan), and east Malaysia (Sabah, Sarawak). In this case,
each region is a cluster and under each cluster there are a few campuses
available.
c. In the study, firstly the researcher needs to select randomly a few clusters
from the available clusters (regions). Secondly, the researcher needs to
select randomly a few campuses among the campuses available in the
selected clusters.
d. All students in the selected campuses are taken as respondents in the
study. In this case the sampling technique is applied in the selection of
clusters and in the selection of campuses within the selected clusters.
Thus, no sampling technique is applied on the respondents in the selected
campuses since all of them will be covered in the study.
TYPES OF NON-PROBABILITY SAMPLING DESIGN
In the non-probability sampling, the element in the population does not have equal
chance of being selected as a sample. This technique is used in research where no
inference concerning the population needs to be made. This means the result of the
study is only applicable to the particular sample and does not necessarily reflect the
whole population. However, this type of sampling is very economical and easily
performed.

UniSZA
Non-probability
Convenience Judgmental Quota Sampling Snowball

Sampling Sampling Sampling
Figure 4.4: Types of Non-probability Sampling Design
1. The Convenience Sampling
A simple definition of convenience sampling is to obtain the sample at convenience.

Convenience sampling is a non-probability sampling procedure where the selected
sample represents the captive audience. The captive audience is a very convenient
way to generate a sample. The procedure is not random whereby the respondents
are selected because they happen to be at the right place and at the right time. For
example, the researcher goes to the supermarket at noon and obtains a few
shoppers as respondents to complete the structured questionnaire.
Convenience sampling offers the least expensive and least time consuming
compared to all sampling techniques. In addition, the sampling units are accessible,
easy to measure, and cooperative. Often, it is used in exploratory research for
generating ideas on certain phenomena, insights into certain problem, to develop
hypotheses, to pre-test questionnaires, and to conduct pilot studies.
However, in spite of these advantages, convenience sampling also has serious

limitations. The limitations of convenience sampling can be stated as follows:
a. It may pose the potential sources of selection bias including self-selection

respondents. Hence, the data obtained is not randomly distributed.
b. The sample is not representative of a population and theoretically cannot be

generalized to any population of interest.
c. The sampling process is not appropriate for population inferences, and the
technique is not recommended for descriptive and causal research.
UniSZA
2. The Judgmental Sampling
Judgmental sampling is another non-probability sampling technique in which the

respondents are selected based on the judgment of the researcher that they have
the required characteristics to be included in the study. In this technique, the
researcher selects respondents that meet the specified criteria for the study. For
example, a cosmetic manufacturer selects respondents from the female artists since
they believe that all female artists are consuming cosmetic products. Another
example, ASTRO selects teenagers who posses certain characteristics as
respondents in order to obtain their feedback concerning the Akademi Fantasia (AF)
program. In this case, the management of ASTRO believes that these particular
teenagers are the most likely viewers of AF programs.
3. The Quota Sampling
Quota sampling is also a non-probability sampling technique which is similar to the

stratified sampling in that a particular stratum is the focus of the study. The purpose
of quota is to ensure that various subgroups in a population possessing different
characteristics are represented in the sample. This technique selects respondents
possessing certain characteristics of interest by the researcher. The characteristics
of respondents to be selected are pre-determined by the researcher himself such as
students from urban areas and the students from rural areas. Another example of
characteristics of interest is employment status such as government staff, staff of
private firms, professionals, self-employed, etc.
Example 4.1:
Suppose a firm wishes to investigate consumers who own digital versatile disc
(DVD) players in their home. Since different brands of DVD pose different
characteristics, the researcher may wish to ensure that each brand of DVD is
included in the sample. In a quota sampling, the researcher determines the number
of respondents (quota) to obtain for each DVD brands. For example, a researcher
may choose 100 respondents to interview. The respondents must consist of
customers using the following DVD brand, 40 Sony, 30 Samsung, 20 Toshiba, and
10 of other brands.

UniSZA
4. The Snowball Sampling

In the snowball sampling, firstly the researcher selects one respondent who
fulfills certain characteristics for the study. After obtaining the required data
from the first respondent, the researcher asks this particular respondent to
locate his friends who possess the same characteristics. The second
respondent will also help the researcher to locate the next respondent.
Snowball sampling is typically used in research situation where:
a) The defined target population is very small and the characteristic of the
population is unique.
b) Compiling the complete list of sampling units is nearly an impossible task.
For example, the research project is to study the success story of single
mother in the direct-selling business. First of all, the researcher needs to find
the first respondent, the single mother doing direct-selling business. After
obtaining the required data (through personal interview or structured
questionnaire) from this respondent, the researcher would ask that
respondent to locate her friends who possess the same characteristics. The
researcher would stop once he has obtained enough number of respondents
required for the study.

UniSZA
Exercises
A list containing elements in a target population is known as a
1.
sampling frame.
A sample is any subgroup of a population selected for
2.
participation in a study.
A stratified sampling is usually the most efficient sampling
3.
technique in a statistical sense.
A sampling unit is the object in the population from which
4.
information is desired.
Resource constraints should not be used in determining sample
5.
sizes for research project.
6. Systematic sampling does not require a sampling frame.
7. Quota sampling does not guarantee representativeness.
A census involve the complete enumeration of all of the
8.
elements of a population.
The sampling system consists of a list of directions for
9.
identifying the target population.
A census can increase non-sampling error to the point that
10.
these errors exceed the sampling errors in a sample.
1. The first step in the sampling design is:
a. Determine the sampling frame

b. Define the population of interest
c. Define the problem statement
d. Determine the sample size
2. Elements that share a common set of characteristics are called:
a. Population
b. Census
c. Sample
d. Sampling unit

UniSZA
3. A subset of the elements of a population selected for study is known as:
a. Quota
b. Sample
c. Census
d. Panel
4. A complete enumeration of the elements of a population is called
a. Sample
b. Quota
c. Population
d. Census
5._____________ consists of a list the target population.
a. The sampling frame

b. The sampling element
c. The target population
d. The sampling unit
6. A census should be chosen in favour of a sample when the population size is

__________, the sampling error is ___________, and the research budget is
__________.
a. Small; high; large

b. Small; low; large
c. Large; high; large
d. Large; low; small
7. Among the factors that the researcher should consider when determining the
sample size for his research is:
a. The number of variables

b. Random error
c. Sampling error
d. All of the above are true.

UniSZA
8. A good sample is the one in which there is no bias from the sampling process.
This is defined as _____________.
a. Precision
b. Consistency
c. Accuracy
d. Random
9. When a researcher chooses a segment of the population that represents the

population as a whole, he has chosen a ______________.
a. Group
b. Bi-variant population
c. Sample
d. Market
10. The number of elements to be included in a research study is called:
a. Sampling frame
b. Population
c. Sample size
d. Sampling unit
11. When the sampling error cannot be calculated due to the method used to
select the sample, the researcher has just used a (n) _______________.
a. Sampling unit sample

b. Inverted sample
c. Probability sample
d. Non-probability sample
12. A researcher begins by selecting one customer from the list of 1300
customers, and then continues the process by selecting every 30th customer
on the list is using a ______________.
a. Simple random sampling

b. Systematic sampling
c. Stratified random sampling
d. Judgement sampling

UniSZA
13. A researcher gives instructions to select a sample consisting of 90 people

under the age of 30 years, and 60 people over the age of 30 years because
that is the proportion of people with that characteristics in the population.
This is an example of a ________ sample.
a. Judgement sampling
b. Quota sampling
c. Proportional stratified sampling
d. Snowball sampling
14. Using internet sample for the survey is an example of:
a. Quota sample
b. Convenience sample
c. Judgement sample
d. Snowball sample
15. What type of sampling is suitable when there are great variations among the
sub-groups in the population?
a. Stratified sampling
b. Cluster sampling
c. Systematic sampling
d. Simple random sampling
16. Which of the following techniques is typically the least expensive and least
time consuming?
a. Simple random
b. Convenience
c. Stratified
d. Cluster
17. Which of the following is not a probability sample?
a. Systematic sampling
b. Proportional stratified sampling
c. Disproportionate stratified sampling
d. Quota sampling

UniSZA
Essay Question
1. A university is planning to subsidize the expansion of its students’ recreational

centre through a special RM 24.00 annual fee to be paid by each student for
the next 4 years. Since the project will take 2 years to complete, the students
who are currently in the third and fourth year will not benefit from the
expansion. The survey committee believes that the opinions of students are
likely to depend on their current class status. There are currently 4000 first
year students, 3200 second year students, 2800 third year students and 2000
fourth year students enrolled in the university. The sample will contain 300
students and the size of the sample from each group is to be proportional to
the size of the particular sub-population.
a. How many first year students should be included in the sample?
b. What is the probability sampling method that is appropriate for the above
opinion poll? State your reason(s).
c. Assuming that the list of 2800 third year students is obtained from the
registrar office, the numerical list of names is between 1 and 2800. Using
the table of random numbers below, explain the procedure you would
follow to select six samples from the list.
9156 7425 6535 2478 2810 1795

1795 5563 9527 5830 1290 4650
4650 3185 4990 9949 2533 2020
9215 7986 8418 4596 1457 7547
1457 7627 3494 0008 0739 6679
d. The committee wants 70 students in the sample. Instead of using a table

of random numbers, state another method of drawing the random sample.
e. What method of data collection should be used? State your reason(s).
2. a. Give two main reasons for using cluster sampling over simple random
sampling.
b. What are the main reasons for using a sample rather than a census?

UniSZA
c. The 5 year old Shah Alam housing directory lists the residents’ names
alphabetically together with their addresses. TNB wishes to conduct a
research on the perception of Shah Alam’s residents towards the
campaign on using Solar Panel for house use. For the above situation:
i. State the required sampling frame.

ii. What problems could arise by using this frame?
iii. How can these problems be solved during the fieldwork?
iv. In using the directory, which probability sampling is appropriate for
the above study? State your reason(s).
3. Explain the differences between stratified random sampling and cluster

sampling by discussing the sample selection methods for each technique.
Give one example for each type of sampling.
4. Describe the differences between a probability sample and a non-probability

sample. Give two examples of each type of sample.
5. What are the advantages of:
a. Convenience sampling
b. Judgmental sampling
c. Snowball sampling
6. a. Distinguish between primary and secondary data.

b. List the advantages and disadvantages of using secondary data
7. a) Suppose that a survey is to be taken on shoppers at a large shopping mall.

For each of the following, indicate the type of sampling technique that can
be employed. A questionnaire is administered to:
i) A random sample of customers at each entrance to the mall.

ii) Every sixteen customer who enters the mall.
iii) All customers that enter a set of randomly selected stores in a 15-
minute period during the day.
iv) Customer who enter any bookstore outlet at the mall.
b) State the differences between Quota sampling and Stratified sampling

UniSZA
8. You are requested to conduct a survey and choose an appropriate

communication method to find out students’ feelings and opinions about the
various food services available on the campus. A survey instrument was
developed for the purpose of gathering the information.
a) Define survey in the above context. State THREE (3) advantages of a

survey method.
b) What would be a suitable method of communication to capture large

amount of responses? Justify your answers.
c) What are the classifications of non-sampling errors that could arise from
the study?
9. What is a cluster sampling? By giving example, describe briefly how would

you draw a sample from a known population using a cluster sampling?
10. A prominent bank located in Kuala Lumpur has some 400,000 users of its
credit cards scattered throughout Malaysia. The application forms for the
credit card asked for the usual information about name, address, income,
level of education and so on that is typical of such applications. The bank is
currently interested in determining if there is any directional relationship
between the frequency of card usage and the amount of credit charged on the
card for various purchases made within a month, i.e. the more frequent the
card is used, the higher is the credit amount charged on the card.
a) Identify the sampling frame that would be used by the bank to conduct the
investigation. How is the sampling frame acquired?
b) Choose a suitable sampling technique and indicate how you would draw a
sample from the target population using the sampling technique chosen.
c) Determine and describe the appropriate data collecting method for the
study. State TWO (2) advantages and disadvantages for choosing the
method.
d) What is a suitable instrument used to collect the data in this study?

Describe the purpose of the instrument.

UniSZA
Chapter 5: Types of Data, Measurement, and Scales
CHAPTER DATA, MEASUREMENT, AND SCALES

5
TYPES OF DATA SOURCES
Secondary and Primary Data Sources
The sources of data and information in research can be classified into

primary data and secondary data. The primary data is basically the data that
is collected and assembled by the researchers of a current research problem.
Some experts refer to primary data as “first-hand” data structures.
The secondary data includes historical information or data structure

which was previously collected and assembled for solving other problems and
not the current problem. These data can be in the form of newspaper report,
annual report of certain department, or the published report by certain
organization.
The Primary Data Sources
Primary data source is the source that presents first-hand information. The
sources can be the original records, original documents, personal
interviews, and records of eyewitnesses that are original in nature. In other
words, primary sources are the original data collected by the researcher
himself for his study. Primary data sources can come from direct
outcomes of an event the actual experience that is recorded by the person
involved. Among the examples of primary data collection activities are:
1. Conducting surveys to obtain demographic information on customers of

certain products using research instrument such as structured
questionnaire, personal interviews, focus groups discussions, etc.
2. Recording the observations on customers‟ behaviour when choosing

cosmetic products to purchase in the market.
3. Conducting laboratory experiments on sampling units to determine how

they respond to certain stimulus.
UniSZA
The Secondary Data Sources
Secondary data is the data obtained from historical information such as

company‟s annual report, financial statement, Bank Negara Report,
newspaper, summary statistics, etc. In other words, it is a second-hand
data or at least once removed from the original event. The most important
consideration when using secondary data is how much you can trust the
original source of the data.
The secondary research activities are:
1. The review and analysis of the existing data on the target markets. These
data are available in the business magazines, research studies,
government publications, annual report of a company.
2. The evaluation of weekly demand trend on competitors‟ product for the

last 5 years in comparison with the company‟s product.
3. The assessment of an impact on the economy due to the increase in fuel

prices, food prices, and also political unrest.
4. The analysis of voting trend among the Malay voters towards the non
Malay candidates in the 2008 General Election.
The secondary research does not require any method to get the data
because it already exists. Thus, no specific method or technique is needed to
collect information. Instead, the time and effort are spent to locate and gather
the information from reliable sources. Some resources for secondary
research information include:
1. Libraries and other public information centres.
2. Books, business publications.
3. Magazines and daily newspaper.
4. Trade associations - most associations have reports on the industries they

serve, the standards they operate under and leaders in the field.
5. Local agencies such as district office, post-office, land-office, etc.

UniSZA
6. Consumer association, non-governmental organizations, etc.
7. Banks, real estate agencies, insurance companies.
8. Wholesalers and manufacturers.
9. Federal government resources can provide extensive demographic data

on population, markets and the economy.
10. Regional planning organizations, local governments have historical as well

as current data on community growth trends. Many offices also have
forecasted future demographic statistics for the area.
11. Media representatives - advertising and sales personnel for television

station, radio and print media outlets keep information on their viewers,
listeners, and readers to help influence potential advertisers.
TYPES OF DATA: QUALITATIVE AND QUANTITATIVE DATA
1. The Qualitative Data
Sometime the researcher does not have to work with numerical data in his
research project. For example, the study on public‟s perceptions concerning
certain government policy, the study on customers‟ acceptance towards
Islamic Insurances, or the public preference concerning new Proton models
just launched in the market. The above research project is called qualitative
research. Qualitative research provides definite market information regarding
the opinions, perceptions, and behaviours of the subjects in the market
research study. The information obtained is in terms of non-numerical.
For example: The customer‟s perception on the new proton model:

a. Excellent
b. Good
c. Average
d. As expected
e. Below average

UniSZA
Qualitative data is less accurate and limited statistical methods available for
data analysis. Qualitative are measured according to their specific categories.
For example, gender is measured as „male‟ or „female‟. Marital status is
measured as „single‟ or „married‟, opinion is measured as „strongly disagree‟,
„disagree‟, „neutral‟, „agree‟, or „strongly agree‟ and so on. Qualitative
research is employed to achieve a variety of objectives. Among the objectives
of qualitative research are:
a. To obtain certain information on a market segmentation.

b. To explore the concepts and positioning of a product or service.
c. To identify attitudes and opinions shared by the target customers.
d. To prioritize the variables of interest for further study.
e. To clarify certain issues that might arise when defining the problem.
f. To provide proper direction for the development of questionnaires.
Personal interviews and focus group discussions are the most common
qualitative research techniques.
a. Personal Interviews
i Conducted on a one-to-one basis

ii Uses an unstructured questionnaires
iii Uses open-ended questions
iv Extensive probing questions
v Usually recorded on audio visual tape-
vi Interview typically lasts one hour
vii Usually no more than 50 persons are interviewed
b. Focus Groups Discussion
i Conducted in a group of 8-12 people in a session

ii Led by a moderator who follows the pre-designed script
iii The sessions are audio or video recorded
iv The whole session normally lasts between 1 to 2 hours
v Minimum of two groups required per target market
vi Discussion is lively and proactive. Generates required information.

UniSZA
2. The Quantitative Data
When the numerical data is required in the study, the research is called
quantitative research. Data on demand, supply, interest rate, annual income,
total export, total import, and the volume of daily sales are numerical. The
quantitative research creates statistically valid market information. The
information is in terms of number. For example: The weight of new Nokia
hand phone is:
a. Below 300 grams

b. Between 300-400 grams
c. Between 400-500 grams
d. Above 500 grams
Quantitative data is more accurate compared to qualitative data and more

statistical methods can be applied in data analysis. Some common
application of quantitative research includes:
a. Testing to prove certain hypothesis

b. Estimate the demand pattern of a product or service
c. Projecting the economic and business growth
d. Modelling the Gross National Product (GNP) growth
Quantitative variable can further be classified as either continuous or

discrete. The continuous variable is one with an unlimited number of values
that may take place. The examples of continuous variables are the distance
of travel, the amount of petrol consumption, the weights of individuals, the
monthly expenditure, the amount of loan, and the volume of export recorded.
The discrete variables are whole numbers obtained through counting.
Examples are number of students in a class, number of employees in a firm,
number of American soldier died in plane crashes, number of people
attending the Akademi Fantasia (AF) concert, and the number of shoes
available in stock. Personal interviews, telephone interviews and mailed
questionnaires are the most common techniques to obtain quantitative data.

UniSZA
METHODS OF DATA COLLECTION
1. Personal Interview
Advantages of personal interview:
a. Interviewer can observe reactions of respondent, can probe, and can

clarify ambiguities, able to note facial expression of respondents.
b. Technique usually yields in a high percentage of responses.
c. Flexibility with location and time for gathering information.
d. Interviewer can use visual displays to assist respondents who have

difficulty in understanding the question.
e. Interviewer can convince respondents to corporate and give honest

response.
f. Allows for good sampling control.
Disadvantages of personal interview:
a. Too costly, hence cannot collect many respondents.
b. Too time consuming, not suitable for big project with small budget.
c. May contain interviewer biases.
d. Facial expressions and interviewer statement can influence responses

given by the respondent.
e. May contain error in recording made by the interviewers.
2. Telephone Interview
Advantages of telephone interview:
a. Fast and easier method to obtain data.
b. Lower cost compared to personal surveys.

UniSZA
c. No travel and lodging expenses required for interviewers.
d. Cover wider geographical area compared to personal surveys.
e. The researcher can monitor the process.
f. Disadvantages of telephone interview.
g. The length of each question is limited.
h. Difficult to reach respondents who are always busy.
i. Difficult to convince the respondents in order to discuss certain issues.
j. Some people are not keen to talk long over the phone.
k. The cost is more expensive compared to mailed questionnaires,

especially for long distance respondents.
3. Mailed Questionnaire
Advantages of mailed questionnaires:
a. Can cover wider geographical area.
b. Lower budget required compared to personal and telephone interview.
c. No response bias from interviewer.
d. Anonymity of respondents assured.
e. Respondent completes questionnaire at their own convenient time.
Disadvantages of mailed questionnaires:
a. Accurate address is not always available.
b. Response received is not necessarily representative of the target

population.
c. Delayed response does occur.

UniSZA
d. Clarification of response cannot be made.
e. Low response rate since respondents do not feel obliged to complete.
4. Self-Administered Questionnaire
Advantages of self-administered questionnaires:
a. Lower budget required since the questionnaire is self explained and can
be distributed to many respondents at the same time.
b. Busy respondents can complete the questionnaire at their convenient

time and the researcher can collect later.
c. Response rate is high since the researcher manages to talk and

convince respondents personally prior to giving the questionnaires to
them.
5. Direct Observation
Advantages of direct observation:
a. Researcher can record the real scenario since the respondents do not
know that their attitudes are being observed.
b. No response bias and also no influence from the researcher involved

here.
c. More suitable for the market research project especially on customer

behaviour.
Disadvantages of direct observation
a. Errors due to different interpretation might be recorded by different

observers on the same respondents.
b. Expensive and time consuming to complete data collection stage.
c. Not suitable for serious type of research project where the actual
responses from the respondents are required.
UniSZA
TYPES AND LEVELS OF MEASUREMENT
To measure a variable is to assign a value to objects, events, or outcomes of an

event according to specified rules. In a research project, the researcher does not
measure the respondent but measures the characteristics of the respondents.
The characteristics of interest for the study are called variables. For example, in
a study on consumer preferences towards certain food outlets; the researcher
does not measure consumers but the variables such as their age, gender,
preferences, perception, attitudes, amount of purchase, intention to revisit, or
other relevant characteristics required in the study.
The researcher assigns numbers to objects of the study according to certain

pre-specified rules. The specification rules for assigning numbers to the
characteristics of the objects are the most important aspect of measurement.
That is, the assignment of values to the object must be isomorphic. Isomorphic
means that there must be a one-to-one correspondence between the figure and
the characteristic being measured. For example, the same dollar figures are
assigned to households with identical annual incomes to ensure that the
numbers can be associated with specific characteristics of the measured object,
and vice-versa. The purpose of assigning figures to the characteristics of the
objects is to facilitate the communication of measurement rules and results; and
to permit statistical analysis of the resulting data.
Level of measurement is the scale representing a hierarchy of precision in

which a variable might be assessed. There are four types of measurement
namely nominal, ordinal, interval, and ratio.
1. The Nominal Scale
Nominal scale data describes the variable in terms of its category and the
measure differs in terms of quality rather than quantity. The nominal scales
are assigned to represent the categories of which the characteristic of data
belongs. Nominal scale measure is the most basic measure among the four
levels of scale designs and used mainly for the purpose of identifying the
respondents in terms of their gender, ethnic groups, socio-economic status,
preferences, etc. In this scale, the question requires respondents to select
only one descriptor as their response. The responses provided do not contain
any level of intensity; therefore, it is impossible to establish any form of rank
order or to calculate any meaning values to describe the respondent.
UniSZA
Examples 5.1: Data to obtain using the nominal scale

a) Marital status of respondent
 Married
 Single
 Widowed
b) Gender of respondent
 Male
 Female
c) Do you like chocolate ice cream?

 Yes
 No
2. The Ordinal Scale
The ordinal level of measurement describes variables of interest along some

type of continuum. This measurement refers to the variables in terms of its
rankings of various outcomes such as big and small. Thus the scale provided
indicates the relative positions of the objects in its group but not the
magnitude of differences between them. This scale allows respondents to
express the relative magnitude between the answers to a question as
illustrated in following example. The researcher can rank-order the raw
responses into a hierarchical pattern.
Example 5.2
Instruction
Rank the following brands of shampoo based on your preference.
Brand Your Rank

1. Follow Me
2. Vidal Sasson
3. Rejoice
4. Head & Shoulders
5. Sunsilk
6. Pantene
UniSZA
Example 5.2a
Instruction
Please indicate how much you agree or disagree with the following
statements concerning the entertainment program on TV. Select the
most appropriate number to represent your opinion.
1. Strongly Disagree
2. Disagree
3. Not Sure
4. Agree
5. Strongly Agree
3. The Interval Scale
The interval scale measurement describes the variables in terms of some

equal intervals between them. The scale allows respondents to select an
appropriate response in which the value falls in a given interval.
Example 5.3: Data to obtain using the interval scale

a) Select your age category:
 Below 20 years
 20-39 years
 40-59 years
 Above 60 years
b) Select your income category:

 Below RM 1000
 RM 1001- RM1999
 RM 2000- RM3999
 Above RM4000
c) Please indicate how often you access the internet in a month.

 Do not access at all
 1-5 times
 6-10 times
 11-15 times
 More than 15 times

UniSZA
Example 5.4: The Questionnaire which produce interval scale data
This method is the most commonly used in business and social sciences to
obtain data concerning respondent‟s attitude, opinion, perception, or intention
on something. The method produces an interval-scaled data since the data
fall anywhere between two extreme intervals namely “strongly Disagree” and
“Strongly Agree”
Please indicate how strongly you agree or disagree with the following
statement concerning the corporate image of this university. You can mark
your sincere response anywhere between 1 to 10. Please remember, there is
no right or wrong answers!
Corporate Image of a university

This university......... Strongly Strongly
Disagree Agree
1. has a quality of academic staff (lecturers) 1 2 3 4 5 6 7 8 9 10
2. maintains a tradition of academic excellence 1 2 3 4 5 6 7 8 9 10
3. has lecturers with good academic reputation 1 2 3 4 5 6 7 8 9 10
4. maintains good industrial linkages 1 2 3 4 5 6 7 8 9 10
5. practises an ISO 9001:2000 in its system 1 2 3 4 5 6 7 8 9 10
4. The Ratio Scale
The ratio scale describes variables that have equal intervals between them
but also have an absolute zero. In the simplest terms, this means they are
variables for which one possible value is zero or the actual absence of the
variable or trait is possible. Ratio scales assign absolute rather those relative
quantities. For example, both monetary values and weight are ratio scales
because they possess absolute zeros and interval properties. The absolute
zero represents a point on the scale at which there is an absence of the given
attribute. Ratio scale is the highest level of measurement among the four
scales and the statistics that can be used in the analysis are the most flexible.
However, in behavioural research interval scale typically is the appropriate
measurements.

UniSZA
Example 5.5: Data to obtain using the ratio scale

1) How much do you pay for your monthly rent?
2) How long does it take for you to drive to the office every day?
3) How many school-going children are there in your family?
THE COMPARISON AMONG FOUR TYPES OF SCALES
Table 5.1: The Comparison among the Types of Measurement Level

STATISTICS
Basic
Scale Examples Descriptive Inferential
characteristics
Number assigned to id number, gender, frequency,
Chi-square,
Nominal classify objects. program, types of percentages,
Binomial Test
house, ethnic group mode
Number assigned to social class,
median, Spearman Rank
indicate the relative qualification, Job
Ordinal percentile, Correlation,
positions of the ordered position, rank of
ranking Friedman ANOVA
objects. opinion, perception
Number assigned to Pearson
age, income, range, mean,
indicate the magnitude Correlations,
attitudes, opinions, variance,
of differences between t-tests, ANOVA,
Interval objects. Normally in the
the strength of standard
Factor Analysis,
agreement or deviation,
multiple-choice Regression
disagreement skewness
response.
range, mean, Coefficient of
Zero of point is fixed; length, weight,
variance, Variation, almost all
number assigned to income, cost, sales
Ratio indicate the actual value quantity, amount of
standard statistical analysis
deviation, methods can be
or amount of variable. expenditure
skewness used
The comparison among four types of measurement scales can be explained

through the following example:
Table 5.2: The Relationship between Nominal, Ordinal, Interval, and Ratio
Nominal Ordinal Interval Ratio
scale Scale Scale Scale
Preference
Preference Money spent in last
Name of ratings of food
No. rankings among two months at the
restaurant quality in the
the restaurant respective restaurant
scale of (1 to 10)
1. Delima Café 2 5.5 $180.00
2. Sigai Café 1 7.1 $300.00

UniSZA
3. Sel Kampung 4 3.4 $ 10.00

4. Cenderawasih 5 1.0 $ 0.00
5. MakDara 3 4.0 $100.00
Example 5.6: The illustration of real-life application for four different scales.
The marketing graduate plans to open a fast-food restaurant near the campus.
She conducted a study on the students‟ preference towards the existing
restaurants in the area. Her sample consists of five main restaurants in the
area.
Using the nominal scale, she assigns the five restaurants as follows:
1 = Delima Café, 2 = Sigai Cafe, 3 = Selera Kampung, 4 = Cenderawasih, and
5 = Restoran MakDara. In this case, she uses the numbers to nominate the
restaurants. Thus, in the above numeric, it does not imply that Delima Café,
which was assigned 1, was in any way the most preferred restaurant as
compared to Restoran MakDara, which was assigned the number 5. The
number only represents the name of the respective restaurant, and not the
performance of that particular restaurant compared to other restaurants.
Nominal scale allows the researcher to run some descriptive statistical analysis
such as percentages and mode. However, the statistical analysis such as mean
is not meaningful in this level of measurement.
Using the ordinal scale, the respondents were asked to rank the five
restaurants based on their order of preference. Rank 1 indicates certain
restaurant is the most preferred restaurant. Based on the result in Table 5.2,
restaurant Sigai is the most preferred while restaurant Cenderawasih is the
least preferred restaurant in the area.
By using the interval scale, the respondent‟s preferences for each restaurant
can be expressed on a 10 point rating scale. The respondents can provide their
score from 1 to 10 to reflect how strong their preference towards each particular
restaurant. From the table, Cenderawasih restaurant obtains rating 1.0 while
MakDara restaurant obtains rating 4.0. However, the rating for MakDara 4.0
compared to Cenderawasih 1.0 does not, in any way, imply that the customers
prefer MakDara restaurant 4 times when compared to Cenderawasih.

UniSZA
The ratio scale in Table 5.2 shows that a respondent was asked to indicate the
amount of money he spent at each of the four restaurants in the last two
months. Based on the table, the respondent spent $100 in MakDara and only
$10 in Selera Kampung. Here we can say this particular respondent spent 10
times as much in MakDara as Selera Kampung. The $0 score indicates that the
particular respondent did not spend his money at that restaurant.
THE CLASSIFICATION OF SCALING
Scaling can be defined as a procedure for the assignment of symbols (numbers)

to a property of objects in order to impart some of the characteristics of numbers
to the properties in questions.
There are many ways to classify the scales of an object in the study.
However, most of the researcher employs three types of scale to measure the
response namely categorical scale (sometimes-called voting scale), comparative
scale (sometimes-called ranking scale), and rating scale.
The rationale for this classification is to separate these two methods

according to the level of measurement. Nominal scale data is often used for
voting (category scale), while ordinal scale data is often used for ranking
(comparative scale). Interval data is obtained using rating scale since the
respondents‟ are asked to rate how strongly they agree or disagree with certain
statement. Rating scale is used to measure most behavioural concepts such as
respondents‟ opinion and perception. Meanwhile, ranking scales is used to make
comparison or rank the variables that have been tapped on a nominal scale.
The Categorical Scale (Voting Scale)
Voting scale has several response categories and used to elicit responses
concerning the objects, events, or respondents. Basically, voting scale is used to
judge properties of objects without reference to other similar objects. The voting
scale may be in the forms of two categories “like or dislike” or other
classifications using more categories such as three-point, five-point, seven-point
or more points. One argument is that more points on a scale provide greater
sensitivity of measurement.

UniSZA
1. Dichotomous Scale
The dichotomous scale is used to elicit one of two responses (Yes or No)
and sometimes three responses as in the example below. Basically, a
nominal scale is used to vote the response. This method produces a
nominal scaled data.
Example 5.7
Do you like pizza?
Yes ____
No ____
Not Sure ____
2. Category Scale
The category scale uses multiple items of object characteristics to elicit a
single response. It is often used to separate the objects according to their
categories (gender, age, etc.) This method also produces the nominal
scaled data.
Example 5.8
Please state your gender?
Male ____
Female ____
3. Likert Scale
Likert scale was developed by Rensis Likert and designed to examine how
strongly the subjects (respondents) agree or disagree with the given
statements about the stimulus object. Typically, each scale item has five
specific responses ranging from “Strongly Disagree” to “Strongly Agree”.
The example below illustrates how likert scale method is used in
measuring the image of hospital.
The respondents are asked to indicate their attitude about the object being
rated. The likert scale applied below would produce the ordinal scaled
data since each number correspond exactly to one response statement.

UniSZA
Hence, the data can only be analyzed using the non-parametric methods,
not the parametric methods.
Example 5.9
Instructions
The following statements are about the image feature of this hospital. Please indicate the
extent to which you think this hospital possess the feature described by each statement by
ticking at the appropriate number of the scale provided on the top.
Strongly Disagree Not sure Agree Strongly

disagree agree
1 2 3 4 5
1. The service by staffs is within my
expectation.
2. The staffs of this hospital treat their
patients respectfully.
3. The service time at this hospital is
convenience to customers.
4. Semantic Differential Scale

The semantic differential scale is a seven point rating scale associated with
bipolar attributes that have semantic meaning. Several bipolar attributes are
identified at the extremes of each scale. The respondents can mark
anywhere between the two extremes depending on where they perceive the
given statement is more appropriate to describe the subject. This scale
produces the interval-scaled data. The researcher can use the parametric
methods to analyze the data.

UniSZA
Example 5.10
Instructions
Please marked (X) in the blank which best indicates what you think about your
organization. This university is:
Modern __ :__:__:__:__:__:__: Old Fashioned
Powerful __ :__:__:__:__:__:__: Powerless
Excellent __ :__:__:__:__:__:__: Poor
Organized __ :__:__:__:__:__:__: Disorganized
5. Numerical Scale
The numerical scale is quite similar to the semantic differential scale with
bipolar labels placed at the endpoints. The difference is the five points or
seven-point numerical scale in between the two endpoints as shown in the
example below. This method produces an interval scaled data. Thus, the
researcher could employ the parametric as well as the non-parametric
methods to analyze this data.
Example 5.11
How comfortable are you with the new arrangement in the classroom?
Very Uncomfortable 1 2 3 4 5 Very Comfortable
6. Stapel Scale
Stapel scale is developed by Jan Stapel. This scale is designed to
simultaneously measure the attitude of respondents at both direction
(positive and negative), using 6 (+3 to -3) or 10 categories (+5 to -5), without
a neutral point (zero). This scale obtains an interval data since it does not
have an absolute zero point.

UniSZA
Example 5.12
Instructions
Please circle the number to evaluate how accurately each word or phrase below
describes the healthcare services in this hospital. The larger the plus number
you mark the more accurate the phrase describes the services. The minus
number shows the less accurate that you think the phrase describe the service.
Healthcare Services in this Hospital

+3 +3 +3 +3
+2 +2 +2 +2
+1 +1 +1 +1
Convenience Respectful Equipment Advice
-1 -1 -1 -1
-2 -2 -2 -2
-3 -3 -3 -3
7. Graphical Rating Scale

In graphical rating scale, the respondents are presented with graphical
representations to show the response to a particular question. This scale is
easy to respond and obtains an ordinal data. The example shows the scale
which depicts faces ranging from sour face to sweet smile to express the
respondent‟s preference about a movie.
Example 5.13
Instructions
Please tell us how much you like this class by pointing to the face that
best describes your feeling.
1 2 3 4 5

UniSZA
The Comparison Scale (Ranking Scale)
In ranking methods, the subject directly compares two or more objects and
makes a choice among them. It is often used in the situations where the
respondent is asked to select one as the “best” or the “most”. This method is not
satisfactory when more than two choices are given. For example, when the
respondent is asked to select the most preferred among three brands A, B, and
C. Assuming that 40 percent choose brand A, 30 percent choose brand B, and
30 percent choose brand C. The question is, “Which is the preferred brand?” The
answer suggests that A is the most preferred is correct, but 60 percent of the
respondents choose the same brands other than A. The researcher can avoid
the confusion by using the following types of comparative or ranking scale.
1. Paired Comparison Scale

Paired comparison scaling is used when a respondent is presented with two
objects (brands) and asked to select only one object according to some
criterion. The data obtained is ordinal. This scale helps to assess preferences
in which customers consistently show a preference for certain brand of
product instead of other brand. For example, in assessing customer
preference towards notebook brands, customer can make 6 paired choices to
evaluate 4 brands, Acer, Dell, Hewlett-Packard (HP), and Apple. The paired
choices can be calculated using the formula [n (n – 1)/2]. In analyzing the
data, the researcher can calculate the percentage of each brand in the
column by summing the matrices for all respondents, dividing the sum by the
number of respondents and multiplying by 100.

UniSZA
Example 5.14
Instructions
We present you with 8 pairs of notebook brands. Please indicate which of

the two you would prefer the most. Please put (1) if you prefer the brand in
the column over the row. However, if you prefer the brand in the row over
the column, you can put (0).
Acer Dell HP Apple

Acer 0 1 0
Dell 1 1 0
HP 0 0 0
Apple 1 1 1
Number of times 2 1 3 0
preferred
2. Forced Choice or Ranked Order

Forced choice or ranked order enables respondents to rank objects according
to some criterion. For example, the respondents are asked to rank the brands
of digital camera according to their overall preference.
Example 5.15
Instructions
Rank the following brands of digital camera that you would like to buy in
order of preference. Begin by assigning 1 for the most preferred and 5 for
the least preferred.
Brand Rank Order

1. Kyocera
2. Canon
3. Ricoh
4. Sony
5. Fujitsu

UniSZA
3. Fixed or Constant Sum Scaling

Fixed or constant sum scaling asks the respondents to allocate a constant
sum of units such as points, dollars, and percentages across items or criterion
of the object. The data obtained through this scale is considered as interval
data. The example shows how the respondents are asked to allocate 100
points to five attributes of mobile phone.
Example 5.16
Instructions
You are given 100 points to allocate to five attributes of mobile phone.
Please allocate your points to indicate the importance of the attribute in
a mobile phone.
1. Design
2. Colour
3. Shape
4. Size
5. Price
Total points 100
4. Comparative Scale
The comparative scale provides a benchmark or a point of reference to
assess attitudes toward the object, or situation under study. The example
below illustrates the use of comparative scale in assessing service quality in
the university.
Example 5.17
Instructions
Please circle the appropriate response.
Over the last two years, the overall quality of my university service has:
1 2 3 4 5 6 7
Decline Remained Improved
Significantly the same significantly
UniSZA
Exercises
The brands or types of stores are classified or measured using
1.
a nominal scale.
The numbers in nominal scale reflect the amount of the
2.
characteristic each object possesses.
An ordinal scale reflects the magnitude of differences between
3.
objects.
Data analysis using Regression methods requires at least an
4.
interval-scaled data.
Ratio-scaled measurements assign actual quantity rather than
5.
those relative quantities.
6. Zero points are fixed with ratio scales.
7. Preference rankings are example of an interval level of scaling.
8. Comparative scaling is considered a form of non-metric scaling.
In constant sum scaling, if an attribute is perceived as not

9.
important, the respondent could assign a zero score.
The major benefit of comparative scaling is that small
10.
differences between stimulus object can be detected.
1. ____________ indicates only relative size differences between objects, and

permits researcher to rank order the respondents responses such as first,
second, third choices of brand.
a. Ratio scale
b. Ordinal scale
c. Interval scale
d. Nominal scale
2. The level of measurement can be classified as ______________.

a. Nominal, ordinal, interval, ratio
b. Binomial, multinomial, normal
c. t-test, Z-test, Chi Square, F-test
d. none of the above

UniSZA
3. Coding males as “1” and females as “2” is a (n) ________ scale.

a. Nominal
b. Ordinal
c. Interval
d. Ratio
4. Which of the following is a possible measure of a salesperson‟s performance?

a. Number of customer accounts created
b. Ringgit Malaysia sales
c. Volume of unit sales
d. All of the above
5. The simplest type of measurement scale is:

a. A nominal scale
b. An ordinal scale
c. An interval scale
d. A ratio scale
6. Whenever you divide people into groups you are probably using a
___________ scale of measurement.
a. Ratio
b. Ordinal
c. Interval
d. Nominal
7. What is the mode of the following set of scores? 4,3,5,6,8,4,5,6,4,8,9,3,4,11

a. 3
b. 4
c. 5
d. 6
8. If we use a scale using 5 categories of races, what type of scale would this
represent?
a. Nominal
b. Ratio
c. Interval
d. Ordinal

UniSZA
9. For which of the following types of scales is the mode the most appropriate
measure of central tendency?
a. Ordinal
b. Ratio
c. Interval
d. Nominal
10. Which of the following is an incorrect classification of “scales”?

a. Attitude measured on a semantic differential scale. (Interval)
b. Weight measured in kilograms. (Ratio)
c. Gender with males measured as 1 and female measured as 2. (Ordinal)
d. Rank Position in an examination (Ordinal)
Essay Questions
1. a) Give a reason why the knowledge of level of measurements is important in

any research.
b) Classify the variables “service quality”, “class status”, and “parents income”
asked in the questionnaire below into quantitative or qualitative variables.
Then state whether they are nominal, ordinal or ratio level of measurement
i. Quality of service at this recreational centre:

Poor Excellent Good Fair
ii. Your student status:

First year Second year Third year Fourth year
iii. Your parents‟ annual income:

Under RM 1000
RM 1001-2000
RM 2001-3000
RM 3001 and above

UniSZA
2. A group of researchers plan to carry out a survey on the repair workshops in

Kota Bharu, Kelantan. These workshops can be categorized according to the
types of vehicles of being repaired. They are 35 bicycle workshops, 50
motorcycle workshops, 85 car workshops, and 20 heavy vehicle workshops.
In order to save time and cost, the researchers plan to survey only 57 of
these workshops.
a) Is the data primary or secondary?
b) State the population for the study.
c) State the required sampling frame.
d) Give an example of questionnaire for each of the following situations:
i. A question that will derive a quantitative data and

ii. A question that will derive a qualitative data.
(The question must be related to the above study)
e) Which is the most suitable method of data collection to be used? Give

one advantage of this method.
3. State the levels of measurement (nominal, ordinal, interval or ratio) for each
of the following variables of interest:
i. Working experience in years.

ii. Method of payment used for textbooks and supplies.
iii. The importance of eating dinner together with family.
iv. The education fees paid by students.
4. In the following situation, decide whether you would use a personal interview,
telephone survey, or self-administered questionnaire. Give your reasons.
a) To get opinions from students at UiTM on their preferences among three

candidates who are running for students‟ leadership.
b) A survey on the residents in Shah Alam on why they choose the area.
You also wish to get some information about what they like and do not
like about this area.
c) A survey of 58 wholesale companies scattered all over West Malaysia

on their personnel management policies for warehouse personnel.
UniSZA
5. You plan to conduct a mail survey of the lecturers in the whole of UiTM
across the country. The study concerns the Universities policies regarding the
payment of moving expenses of employees who are transferred. Suggest 4
ways to improve the response rate of such a survey.
6. a) The following questions are not written clearly. Improve on the questions.
i. What is your income?
 RM 900 and less
 RM 901 to RM 1000
 RM 1001 to RM 2000
 RM 2001 to RM 3000
 RM 3001 to RM 4000
 RM 4000 and more.
ii. How often do you go to a cinema?

Never Occasionally Always
iii. With whom do you normally go to the cinema?
Family Partner Friends Others
b. Develop ONE question that will derive

i. quantitative data
ii. qualitative data
7. a. Suppose that the following information is obtained from Rahman Salleh on

his application for a housing loan at UMB Finance.
i. Age : 34 years
ii. Marital Status : Married
iii. Annual Family Income : RM 100,000
Identify the levels of measurement for each of the 3 information given

above and state your reasons.
b. Identify the highest and the lowest levels of measurement and state their
differences.

UniSZA
c. State the characteristics of an ordinal and an interval level of

measurement.
8. Design a question and name the level of measurement you would use for the
following:
a) To find out the popular brands of soda consumed by your customer.
b) To find out students‟ preference among the three types of exams namely multiple
choices, essay type and a combination of both.
9. Over the past five years, a manager has noted a steady growth of sales and profits
for her division‟s product line. Is there a need to utilize any secondary date for further
evaluate the division‟s condition. Give your reasons.
10. What would be the most appropriate method of data collection under the following
situations? Give reasons for your answers.
a. Survey of the achievement motives of industrial engineers

b. Survey of the satisfactions of rent-a-car users
c. Survey of television commercial advertising awareness
d. Survey of top corporate executives.
11. Indicate whether the following measures are nominal, ordinal, interval, or ratio
scales:
a. Prices on the stock market

b. Gender
c. Temperature
d. Educational level: secondary school, diploma, degree, master, PhD.
12. Comment on the weaknesses of the following questions:
a) A student survey:
How frequent do you get to campus using each of the form of
transportation available per week? Select your answer based on the
following scale:
1. Most frequent
2. Frequent
3. Never
UniSZA
4. Seldom
5. Very seldom
i. Walk …………….
ii. Bicycle …………..
iii. Public transportation …………..
iv. Motor vehicle …………..
b) A survey of employers:
Should the company continue its generous medical insurance
program?
Yes No
c) A personnel manager‟s survey of employees:

In your opinion, are women discriminated against, treated equally or given
preference in promotion practices?
1. Against 2. Treated equally 3. Treated preferentially
13. Indicate whether the following examples are nominal, ordinal, interval or ratio
scales. Explain briefly the meaning of these scales.
i. Numbers assigned to swimmers who participated in the 100 meter

butterfly stroke.
ii. Ranking of teams in a tournament (First, Second, and Third place)
iii. Time of finish in the 100-meter run in seconds
iv. Performance rating of runners on a 0 to 10 scale.
14. A research can use both primary and secondary data.
a) What is secondary data? State four advantages of using secondary data

b) Provide three reasons why some organizations choose to hire external
researchers or consultants to assist in obtaining primary data.
c) List and briefly explain two methods of collecting primary data.

UniSZA
15. For each of the following, indicate the type of measurement scale it
represents, that is nominal, ordinal, interval or ratio scale.
i. Weight of an automobile
ii. Exterior colour of an automobile
iii. Size classification of an automobile
iv. Original cost of an automobile
v. Rankings of automobile models by a consumer organization
vi. Trunk size of an automobile
16. a) A researcher wishes to compare between two counters (HEA and HEP) on
the following attributes: friendly personnel and convenience of location.
i. Design a Likert Scale to accomplish this task.

ii. Design a semantic differential scale to accomplish this task.
iii. Design a graphical or pictorial scale to accomplish this task.
b) Briefly discuss the THREE (3) criteria of a good measurement.

c) Discuss with example FOUR (4) basic scales in research.
17. a) Distinguish between rating and ranking scale. Which is a better attitude
measurement?
b) A researcher wishes to compare two hotels on the following attributes:

friendly personnel, convenience of location and value for money. Design
a Likert scale and semantic differential scale to accomplish this task.
18. Describe four levels of measurement in research. Give examples of each.
19. A researcher is interested to determine the extent of smoking among

university students. A random sample of 100 students was selected to
complete a questionnaire. Part of the questionnaire is provided below.
a) Identify the scale of measurement for each of the following questions

posed in the questionnaire provided below.
b) Identify the appropriate charts or graphs to present the distribution of

each variable obtained. Justify your choice.

UniSZA
SMOKING HABIT QUESTIONNAIRE
A: Profile
1. Gender □ Male □ Female
2. How old are you? Years old
3. Where do you stay? □ On campus □ Off campus with friends

□ Off campus with family
4. Do you smoke cigarettes now?
□ Yes, regularly Go to Section B

□ No Go to Section C
□ Occasionally Go to Section D
B: Smokers
1. On the average, how many cigarettes do you smoke a day? _________
2. How long have you been smoking? _________Years___________

UniSZA
Chapter 6: Primary Data Collection Methods
CHAPTER THE PRIMARY DATA SOURCES

6
NATURE OF PRIMARY DATA
Primary data is the data obtained by the researcher or the agent hired by the
researcher. The persons can be the research assistant, the staff of consultant
firms, or the individuals who are doing the data collection work on behalf of the
researcher.
On the other hand, secondary data is the data already obtained and compiled by
other parties such as government departments, company‟s annual report,
newspapers, business magazines, or the internet web pages. This type of data is
already published by relevant parties for their own purpose such as the annual
economic reports prepared by Bank Negara to the government, the company‟s
annual report prepared by the company to its shareholders, the productivity
report produced by the department of statistics, and the annual report prepared
by each ministry.
A researcher can obtain data from secondary sources for his research but
these data might not be suitable and accurate for his research since these data
might be outdated, or the data collected by the previous parties were for different
objectives and had used different scales. Hence, whenever possible, the
researcher should collect his own data unless the required data is not possible to
obtain from primary sources.
For example, the data regarding the monthly volume of palm oil production
in Malaysia, the daily price of gold and petroleum traded on The World Market,
the dividend payments made by ASB to the investors, and many other data are
called historical data and not available through primary sources. These data are
only accessible from the Department of Statistics, the Bank Negara report, the
Ministry of Finance, the Ministry of Primary Industries, the Malaysian Commodity
Exchange etc. In other words, the secondary data are data obtained from the
secondary sources

UniSZA
PRIMARY DATA COLLECTION METHOD
Once the researcher has determined his research objective, research question,
and the corresponding hypothesis for his research project, what he needs to do
now is to collect the required data. Data is information from sample that the
researcher would analyse in order to meet his research objective, facilitate his
research question, and prove his research hypothesis. For example, the data or
information about customers are gender, age, qualification, marital status,
number of kids, monthly income, brand of car, type of house, religion, hobby,
sports, leisure activities, credit cards, golf membership etc.
Method for Collecting Primary Data
Survey Method Observation Method
Direct
Personal Self-Administered Observation
Interview Questionnaires
Mechanical
Face-to-Face Mail Survey Observation
Interview
Computerised Content
Telephone Questionnaires Analysis
Interview
Figure 6.1: Primary Data Collection Methods
Actually, what variable to measure will depend on your research objectives,

research questions, and research hypotheses. The researcher should always
refer to the three elements above when determining what data to collect in order
to avoid collecting the unnecessary data. Based on Figure 6.1, the method of
data collection can be classified into survey and observation method.

UniSZA
1. Survey Method
In survey method, questionnaires are given to respondents to elicit

information for the study. Respondents are asked the questions based on the
information needed by the study. The questions may be asked in the verbal
forms (interview), writing (mail questionnaire), or through computer (internet
or e-mail).
2. Structured Questionnaires
One of the most popular methods of collecting research data is through the
structured questionnaire. These questionnaires are self-explained and self-
administered. In using this method, the researcher designs carefully a series
of questions that cover the variables of interest in the study such as the
respondents‟ demographic background, their opinions concerning certain
issues, their perception concerning certain service performance, and their
intentions to do in the future etc. Structured refers to the degree of
standardization imposed in the process of data collection (questionnaires). In
other words, the researcher arranges the questions properly on a paper
together with the cover letter to explain the purpose of data collection, the
instruction to the respondents on how to respond to the questions, and the
assurance of confidentiality of information provided.
Advantages of a Questionnaire
a. Lowest cost incurred. The cost is low due to no interview, no training of

interviewers, no travelling, no lodging allowance involved.
b. No monitoring cost incurred. Usually the sending of questionnaires and

receipt of responses from respondents are done by the researcher
himself.
c. More respondents and wider area can be covered. The researchers can
send his questionnaires to as many respondents as he likes since the cost
for each respondent is very small.
d. Response more accurate. Since there is no influence, no gesture, no

facial expression, and no interruption from the interviewer, the
respondents can answer the questions at his own convenience. This
manner will assure the accuracy of responses.

UniSZA
Disadvantages of a Questionnaire
a. The poor rate of response from respondents. Since the researcher does
not have gestures or face-to-face contact with respondents, the
respondents can just throw away the questionnaires. In this method, the
researcher should make a regular follow up contact with respondents
using letter or telephone call. Usually the researcher would call the
respondents earlier to inform them that he is sending a questionnaire to
obtain data for his specific objective. The call is made as a gesture in
order to obtain good cooperation from respondents.
b. Once the respondent has difficulty regarding certain terms or questions in

the questionnaire, he will not bother completing the remaining questions
since nobody is available to explain. Sometimes the respondents simply
send back the incomplete response.
c. There is no assurance that the person who responds to the questionnaire

is the intended respondent or the legitimate respondent. This matter will
result in sample not being representative of the population.
Private agents who conduct surveys found that people are more likely to
respond to mail questionnaire that has professional appearance and come
together with attractive prizes such as lucky draws for the respondents who
send the completed questionnaire in time. Some questionnaires obtain good
responses from respondents especially the questionnaire that comes together
with the warranty cards when customers purchase certain products.
Customers will have to answer series of questions on the warranty cards
before sending back to the manufacturer for product warranty.
TYPES OF SURVEY METHOD
Personal Interview
1. Face to Face Interview
In the face-to-face interview (sometimes called personal interview), the
researcher will prepare the questions to be asked during the interview with
respondents. Each question represents the variable that the researcher
wants to obtain its data. The questionnaires cover all variables required from
a respondent. Before the interview begins, the researcher would explain the
objective of research, ask for cooperation, and give assurance that the
UniSZA
response given is only for research purpose and the information is treated
with confidential. This is important since personal data is confidential, and no
one will reveal his personal information if the confidentiality is not assured.
The interviewer should posses good personality so that the interview session
proceeds smoothly and in a friendly atmosphere. During the interview, the
interviewer will read the question and record the response. Personal
interviews may be conducted at the respondents‟ home, offices, or
anywhere. Below are the common examples of places which are basically
used to conduct the interview.
a. Door-to-door Interview
Door-to-door interview refers to the interview in which the

respondents are interviewed face-to-face in their homes. The major
advantage of this interview is high participation rate, but the
disadvantage is high cost.
b. Mall Intercept Interview
Mall intercept interview refers to the personal interviews conducted in

shopping malls. Interviewers typically intercept shoppers at a central
point normally at the entrance to the mall. The advantage of this
method is low cost since no travel required to the respondent‟s home.
Mall intercept interview is appropriate when the respondents need to
see, touch, or taste the product before they can provide meaningful
information.
c. Computer-Assisted Personal Interview (CAPI)
This interview uses a computer to get the information from the

respondents using several user-friendly electronic packages to
design questions easier for the respondent to understand. However,
this method is classified as personal interview technique because an
interviewer is usually present to serve as a host and to guide the
respondents. CAPI is normally used to collect data at shopping malls,
product clinics, conferences, and trade shows.

UniSZA
Advantages of Face-to-Face Interview
a. Higher response rate. With proper plan and approach, the respondents
will not turn down the request for an interview. The interviewer must be
friendly and creative in getting cooperation from respondents. The
interview method normally achieves a response rate of more than 70%.
b. Data more accurate. The face-to-face meeting allows the interviewer to

clarify terms or anything which might confuse the respondents. Once the
respondent understands the question, he will provide an accurate
response.
c. The interviewer can note specific reactions by respondents during

interview. The physical reaction and facial expression by respondent can
tell whether he is providing accurate responses or not. The interviewer can
also note the physical environment surrounding the interview such as the
respondents‟ office, the house, the dress etc that should tally with his
response.
d. People will usually respond with good gesture and provide accurate
responses when approached in person. The Malay culture of not saying
“no” still holds true when someone comes to the door with polite and
peaceful manner.
e. An experienced interviewer can sense out if the respondent is trying to

hide some information. In this case, he will use his creativity to clarify the
intention or terminate the interview.
Disadvantages of Face-to-Face Interview
a. High cost. Interviewers must be given specific training on methods such

as the art of making personal approach, the art of asking questions, the art
of requesting cooperation etc, which is expensive and time consuming.
The interviewer must be confident enough to work on his own. At the
same time, the daily allowance for travelling, food, lodging etc is very high.
b. Incomplete response. Any small mistake by interviewer can cause error in

the response. The mistake such as using a wrong approach, bad facial
expression, coming at the wrong time, not sensitive to certain issue in the
conversation sometimes could hinder respondents from giving truthful
response.
UniSZA
c. Error in recording. This happens especially when the interview session is

in a hurry due to time constrain on part of respondents. The interviewer
who needs to read the question and record the response quickly and
simultaneously is prone to making mistakes.
d. Require close supervision. The interviewers‟ work should be supervised

closely to avoid interviewer cheating. Sometimes, due to many reasons
such as time constraint, respondent is difficult to contact or the respondent
is not available during appointment time, the interviewer will fill the
response on his own.
2. Telephone Interview
Sometime it is possible to collect data through telephone conversation. This

method is possible if the researcher has complete directory of telephone
numbers of population under study. If the respondents are government
officers, employees of private firm, professional people such as doctors,
lawyers, accountants etc, then the researcher have the option of using this
method.
a. Traditional Telephone Interview
In traditional telephone interview, respondents are called through the

telephone and the interviewer will ask a series of questions and record
the responses. Respondents are more willing to provide detailed and
reliable information on a variety of personal topics over the telephone
than with personal interviews.
b. Computer Assisted Telephone Interview (CATI)
CATI uses a computerized questionnaire administered to respondents

over the telephone. The interviewer would contact respondents over the
telephone, read questions posed on the computer screen, and record the
respondent‟s answers directly into the computer memory bank. The
computer systematically guides the interviewer and checks the
responses for appropriateness and consistency.

UniSZA
Advantages of Telephone Interview
a. Less expensive compared to the face-to-face interview. The financial cost

for travelling, lodging, and outstation allowance is not involved.
b. Less time consuming. The number of respondents interviewed through

telephone in one day is much higher than the number interviewed through
face to face.
c. Easy monitoring. The researcher can monitor the interviewers‟ work more
easily since he can check the telephone numbers of respondents and the
time called.
Disadvantages of Telephone Interview
a. Lower response rate. The rate of response is lower since the respondents
can just “hang-up” the call when he realised it is time consuming, or
unsure of the confidentiality of the conversation.
b. Fewer questions could be asked. Usually the conversation through the

telephone cannot take long especially when the topic of discussion is not
interesting, especially for the respondent since he has no particular
interest in it.
c. Difficult to get good cooperation. Conversation through the phone is not

convincing enough especially when trying to get cooperation from the
respondent. It is difficult to convince someone when they cannot see you
in person, your facial expression, you body gesture etc. Imagine the
accuracy of data when you are obtaining personal information such as
address, income, marital status, occupation, etc. from someone who do
not know you and still doubtful of your real intention.
3. Self-Administered Questionnaire
In this section, we will discuss another survey method that does not require
the presence of the interviewer. In this technique, the researcher distributes
questionnaires to respondents personally, through mail service, by inserting
in the newspapers, or by sending to the email address. The difference
between self-administered questionnaires compared to personal interview is
in the self-administered questionnaire, the respondents make an effort to
read and respond to the questions. However, the effectiveness of self-
UniSZA
administered questionnaire will depend on the efficiency of the written words

rather than the soft skill of interviewers.
4. Mail Survey
A mail survey is a self-administered questionnaire sent to pre-selected

respondents through the mail. Basically, a mail interview package consists of
the outgoing envelope, cover letter, questionnaire, return envelope, and
possibly an incentive.
5. Mail Panel
A mail panel consists of a large, nationally representative sample of

households that have agreed to participate in periodic mail questionnaires
and product tests. The data on the panel members is updated every year and
households are compensated with various incentives. Mail panel is
appropriate for longitudinal design studies which allow obtaining information
from the same respondents repeatedly.
Advantages of Mail Survey

1. Low cost.
2. No interviewer bias
Disadvantages of Mailed Survey

1. Low response rate
2. Slow data collection speed
6. Computerised Questionnaire
a. Internet Survey
Lately, the internet poll has become one of the popular methods to obtain
information from the public especially their opinion concerning certain
issue of public interest. In internet survey, the researcher brings an issue
to attention and requests the opinion from the public. The public can
respond by voting the statement that closely resembles their opinion. At

UniSZA
the same time they can view the current standing in term of the most
popular opinion and the corresponding vote obtained.
Internet survey appears when a computer user is asked to go to a
particular Web site location and answer the series of questions displayed
in the website. In this technique, the respondents are not selected using
specific technique but those who visit the Web site where the survey is
posted are invited to participate in the survey.
b. E-mail Survey
In the e-mail survey, questionnaires are sent out to the respondents

directly through their e-mail address. The respondents would reply the
email by providing their response to each item on the questionnaire. The
professional market research groups are using the internet to send their
questionnaires to the respondents‟ email. The respondents would
complete their respective questionnaires and also return through the
internet to the researcher‟s e-mail address. Normally, these research
agencies would provide certain rewards in terms of discount coupons etc
in order to encourage the respondents to participate in the study.
Advantages of Computerised Questionnaire
1. Low cost.
2. Very high data collection speed.
3. Non interviewer bias
Disadvantages of Computerised Questionnaire
1. Very low response rate

2. Low control of data collection method

UniSZA
The Observation Methods
1. Direct Observation
In this method, the researcher will identify his respondents and records the
required data based on what he observes. This method is suitable for a
research to study the behaviour of respondents. For example, the research is
carried out to identify how car drivers behave on the road during traffic jam.
In his observation, the researcher will record the type of drivers who follow
the traffic rules properly, and the type of drivers who choose to ignore rules
such as queue jumping or overtaking using emergency lanes. The
researcher may also be interested to know the types of vehicles
(motorcycles, cars, buses, and lorries) which always ignore traffic rules
during traffic jams. Another area where this method is suitable is to observe
customers‟ behaviour in the supermarket. In the market research study for
example, the researcher is interested to know how customers make
decisions in choosing which shampoo to buy. The researcher will stand in the
area where hundreds of shampoos of different brands are placed on the
rack. He will record how the customers choose the shampoo. Most probably
some customers have decided earlier which brand to buy; some customers
will compare prices, packaging, and even the smells before buying the
shampoo. The researcher will record specific characteristics of customers
who prefer certain brands of shampoo etc.
Advantages of Direct Observation
1. The data obtained reflect the actual behaviour of respondents. The

respondents will tend to hide their actual behaviour when approached in
person or when answering the questionnaires.
2. The researcher gets a clearer picture and a better feeling of the

situations for his study. Hence, he will be in a better position to make a
proper recommendation regarding the underlying phenomena in the
study.
Disadvantages of Direct Observation
1. The respondent will not act his normal behaviour if he knows that he is
being observed.
2. The data collection process is cumbersome and tedious.

UniSZA
2. Mechanical Observation
Sometimes certain mechanical devices such as video cameras, rather than

human observers to observe are used and record customer behaviours. The
devices do not require the respondents‟ direct participation in the study but
they will capture the respondents‟ behaviour for analysis. The early
application of this technique is in a study to determine the level of comfort
among train passengers by taping or recording how they sat and moved in
their seats.
Advantages of Mechanical Observation
1. It offers high disguise by using the hidden camera. However, other

mechanical device such as the use of psycho galvanometers is very
difficult to disguise.
2. Low observation bias since mechanical observation involves the non-

human observer.
Disadvantages of Mechanical Observation
1. Low ability to observe in a natural setting. However, it depends on the

mechanical tools used in the observation. The degree is low when using
psycho galvanometer but high if using turnstiles.
2. Less flexible.
3. Content Analysis
Content analysis is usually used to study communication rather than

behaviour, or physical objects. It is defined as the objective, systematic, and
quantitative description of the manifest content of a communication. Content
analysis obtains data by observing and analyzing the contents or messages
of advertisements, newspaper articles, television programs, etc. It involves
analysis as well as observation which systematically analyze people‟s
communication to identify the specific information contents and other
characteristics such as words, characters (individuals or objects), themes
(propositions), space and time measures (length or duration of the message),
or topics (subject of the message).

UniSZA
Advantages of Content Analysis
a. High degree of disguise since the data is collected over the situation to
be observed.
b. High degree of observation specification and measurement.
Disadvantages of Content Analysis
a. Low ability to observe in natural setting because observation takes place

after the behaviour has occurred.
b. Potential of observation bias because human observer is involved in data

collection process.

UniSZA
Exercises
Primary data is the data obtained by the researcher or a person hired
1.
by the researcher to collect data.
Method for collecting primary data consists of survey method,
2.
observation method and quantitative data.
In survey method, questionnaires are given to respondents to elicit
3.
information for the study.
Structured questionnaire refers to the process of drafting questions
4.
such as question content, question wording, and question ordering.
5. Telephone interview is expensive than face-to-face interview.
The effectiveness of self-administered questionnaire will depend on
6.
the efficiency of the written words rather than skill of interviewer.
A mail panel consists of a large, nationally representative sample of
7.
households that have agreed to participate in mail questionnaires.
The computerised questionnaire consists of internet survey, email
8.
survey and CD-ROM databases.
Content analysis obtains data by observing and analyzing the
9. contents or messages of advertisements, newspaper articles,
television programs, etc.
Mechanical observation requires the respondents‟ direct participation
10.
in the study.

1. A method of primary data collection based on communication with a
representative individual is called __________.
a. Conclusive research
b. Scientific observation
c. An experiment
d. A survey
2. Which of the following is an advantage of depth interview?
a. High degree of structure

b. Valuable insights into motivation.
c. Need for a skilled interview
d. Quickly and easily completed.

UniSZA
3. Which of the following is an advantage of secondary data as compared to

primary data?
a. Obtainable from reliable suppliers

b. More up-to-date
c. Usually available in less time
d. Available in usable form that meets a manager‟s specific needs.
4. Obtaining accurate, sensitive information, such as consumption of alcoholic

beverages, is best obtained through a:
a. Telephone survey
b. Personal interview
c. Mall intercept interview
d. Mail survey
5. Which of the following is NOT an advantage of the personal interview?
a. High participation rate

b. Anonymity of respondents
c. Good opportunity to probe
d. Ability to use visual aids
6. Observation method is used rather than mail questionnaires in which of the

following situation?
a. To collect data as the event occurs

b. When the researcher does not want to disturb the respondent
c. To get rid of the problem of recall (remembering certain event) in
questionnaire responses
d. All of the above
7. ________________ provides data at the lowest cost.
a. Personal interview
b. Telephone interview
c. Mail survey
d. Internet survey

UniSZA
8. Which of the following is the best reason for using telephone interviews?
a. Low cost compared to field interviews.

b. Use of random number dialling.
c. Easy availability of directories.
9. Which of the following is the disadvantage of using personal interview in

collecting data?
a. Accuracy of information
b. A two way relationship exists
c. A high response rate
d. The cost is higher compared to other methods.
10. Which of the following is a good practice in becoming a good interviewer?
a. Be patient
b. Pay attention to accuracy and details
c. Keep responses confidential
d. All of the above
Essay Question
1. a) Why do researchers prefer the personal interview to mall interview?
b) What type of survey method would you choose when you are dealing in
the following situations?
i. Survey the achievement of industrial engineers
ii. Survey the satisfaction of rent-a-car users.
iii. Survey the awareness of television advertisements
iv. Survey the mission of corporate executives.

UniSZA
2. a) A telephone interviewer asks: „I would like to ask you about ethnic groups
in East Malaysia. Are you an Iban, Bajau or Dayak?‟ After the respondent
replies, the interviewer says, „We have conducted a large number of
surveys with people of your background and we do not need to question
you further. Thank you for your cooperation‟. What type of sampling
technique was utilized?
b) A motorcycle manufacturer, Modenas, decided to research consumer

characteristics by sending 100 questionnaires to each of its dealers. The
dealers would then use their sales record to trace buyers of this brand of
motorcycle and distribute the questionnaires to them. Identify the sampling
technique used in the study. Discuss your agreement or disagreement on
the method of data collection and sampling technique used in the study.
3) Compare and contrast cross-sectional and longitudinal studies
4) What factors would you consider when choosing between observation and
survey methods?
5) Describe how you would overcome respondents‟ inability and unwillingness to

answer a questionnaire.
6) What is an internet survey? Describe a situation where internet survey is

preferred over mail survey?

UniSZA
Chapter 7: Secondary Data Sources
CHAPTER
7 THE SECONDARY DATA SOURCES
NATURE OF SECONDARY DATA
Secondary data is the data collected by other parties. This data might be
obtained for a different purpose. The current researcher who uses secondary
data has little knowledge about the original respondents, the sampling
methods, the method of data collection, and the accuracy of the measuring
instruments employed by the previous researcher to obtain data. Thus, the
researcher needs to be careful concerning its validity before using the
secondary data.
The Usage of Secondary Data in Research
Secondary data utilized to serve three purposes for the existing research project.
First, it serves as a specific reference of research in some point in time. For
example, the total population of Malaysia in 1990 was 25 million; the volume of
Malaysian cars on the road in 2000 was 500 thousand units etc. Second, it is
used to support a larger research study. Research procedures typically called for
at least a minimum amount of exploratory study to learn if the past trend can
contribute to the present study. Secondary data will help to avoid the researcher
from “reinventing the wheel” and provide the relevant reference for developing
hypothesis. Finally, secondary data may be used as the sole basis for a research
study, especially research that requires the use of historical data such as time
series analysis etc
Advantages and Disadvantages of Secondary Data
The main advantage of secondary data comes from the availability and easily
accessibility. This is particularly true when electronic retrieval such as internet is
used to access digitally stored data. However, the researcher should be cautious
when using the secondary data because these data have some limitations and
disadvantages. The limitation of secondary data is that they were not meant for

UniSZA
the research at hand. The usage of secondary data to the current research might
not be relevant and accurate because of the following reasons:
1. Data already outdated,

2. Terms used differently,
3. Different units of measurement, and
4. Lack of evidence to verify its accuracy
Therefore, it is important for the researcher to evaluate secondary data on

those factors before using the data for current study.
THE SOURCES OF SECONDARY DATA
Data are classified as either statistical or non-statistical in nature. For example,

government documents and company publications are often statistical in nature,
while periodicals and books tend to be non-statistical. Before going on to the
general problem of good data search procedure, let us look into the major
sources of secondary data which was classified as internal and external sources.
1. Internal Sources
Most organizations routinely gather, record, and store internal data as a

reference for the future. However, internal organization sources vary among
them concerning the usage. For example, the departmental reports,
production summary, financial and accounting reports, marketing and sales
reports differ greatly. For the research purposes, the collection methods are
unique to the specific situation, and the success depends on knowing where
and how to look for these data. However, the usage of data mining
technique might be useful in helping the researchers or managers in their
studies.
Data Mining Analysis
The decision system of large corporations often contains millions or even

hundreds of millions of recorded data. These complex data volumes are too
large for a manager to understand. Data mining analysis helps the
managers to understand the underlying meaning of the data. It refers to the

UniSZA
use of powerful computers to dig through the volumes of data to discover

patterns about the customers’ demand and the products’ life cycle.
Example of Data Mining Analysis
Market basket-analysis is one form of data mining procedures that analyze

anonymous point-of-sale transaction databases to identify the relationships
between trend of product purchase and specific socio-economic variables of
customers.
2. External Sources
External sources are more varied than the internal source and the
researcher should always find better methods for the search. External data
are created, recorded, or generated by an entity other than researcher’s
organization. Published data can be classified into five categories. First
category is computerized database that includes of both online and CD-
ROM databases. Access to these databases is available through the online
service system.
Second category is books in which thousands of books are published each

year and are easily accessible at bookstores or libraries. The third major
source of secondary data is periodicals. This source can be accessed
through online services, which are capable of reaching thousands of
periodicals throughout the world. The fourth category of secondary data is
from the government sources. From time to time, the government produces
data regarding the policies, economics, and industries through their
respective departments. The last category consists of data from the
miscellaneous sources that include university publications such as master’s
theses, doctoral dissertations and research records; company’s publications
such as financial reports, policy statements, speeches by prominent
executives, sales record, product specifications and many others.
Furthermore, there are miscellaneous secondary data sources consisting of
the production of various trade, professional, and other associations. These
organizations often published statistical compilations, research reports, and
proceedings of seminars, etc.

UniSZA
PROCEDURE FOR COLLECTION OF SECONDARY DATA
In searching for secondary data, the researcher normally faces two difficulties,
namely:
1. To ensure the appropriate data source;

2. To organize data into appropriate form;
Task Involved in Searching Secondary Data
To begin the data search, the researcher must have a complete

understanding on the purpose of the study. The researcher should acquire some
background information about the study by looking into general sources such as
appropriate encyclopedias, elementary textbooks, etc. A good search technique
moves from the general to the specific and from broad area search to a narrow
area.
1. The Bibliography Search
After the researcher has made some preliminary search into encyclopedias
and textbooks, he might have some knowledge about the topic. This
knowledge will help to develop a bibliography for the study which is the
essential step for secondary data search and much important to the study.
As explained in earlier, bibliography provides information about citations such

as the name of the author, the title of the article or book, source of
publication, year, volume, and the page numbers. By inspecting through this
information, the researcher can select the priority sources for further studies.
The researcher might also see the development of the subject over time,
discover the authority sources, and is assured of covering the full range of
materials on the subject.
2. The Computerized Search

UniSZA
Computerized database consists of information that has been made available

for electronic distribution. Computerized information retrieval systems, using
databases enable the researcher to have access to an almost unlimited
variety of current material and resources. A database is a large collection of
data in a computer which can be expanded, updated, and retrieved rapidly
for various purposes. Databases are classified into two broad categories
namely, reference and sources. A reference database contains citations that
require further investigation by the researcher. Examples of references
database include bibliographic, abstract, and indexing services such as
ABI/INFORM (Abstract Business Information), UMI (University Microfilms
International’s Dissertation Abstracts) etc. This database basically contains
indexes and detailed abstracts on articles gathered from over hundreds of
journals related to business, management, administration, sales, marketing,
etc.
Meanwhile, source databases contain complete information such as census

data, complete full-text databases in which every word of the original sources
can be used for future reference. The examples of full-text databases are
EMERALD, PROQUEST, Harvard Business Review’s, etc. Access to
information on the databases is accomplished in one of two ways. The
computerized information retrieval system will be either through online
database or through a CD-ROM (off-line) database.
a) The Online databases
The term online refers to the process by which a researcher gains

access to the computerized databases through computer systems.
Online database consists of a central data bank which has access
using a computer, a direct communication line to the system or a
telephone line with a modem.
b) The CD-ROM or Off-line databases
The term “CD-ROM” stands for Computer Disk Read Only Memory.
Data is stored on CD-ROM disks and read from the CD-ROM through
computer screen or printed on a printer. The disks are purchased on a
subscription basis from a publisher or vendor. Basically, the libraries
UniSZA
subscribe to this database for the search process and update the
information. The example of this database is Infotrac (Information
Access Company) which is widely used and contained three indexes;
General Periodicals, Legaltrac and National Newspaper Index.
Planning for a Computerized Search
Doing a database search requires the researcher to think through the topic
very carefully and to plan the research strategy. Understanding the specific area
can help to clarify the topic and highlight possible problem areas. However,
several steps might be useful in conducting computerized search as the
following:
1. Select the appropriate online database.

2. Type in the search keyword and submit.
3. Examine the number of possible citations available.
4. Select the citation and retrieve.
5. Review the citation and decide if the search needs to be modified to get
more precise citations.
6. Examine the citations.
Advantages of Computer Search
a. Speed; the computer can scan many citations in a short amount of time.
b. Flexibility; the computer can combine subject items in which every

meaningful word can be searched and the researcher can focus the result
only on citations that fit the research.
c. Up-to-date; computerized databases are frequently updated with more

current materials.
Disadvantages of Computer Search
a. Imprecision; the computerized search allows only one or more key words
typed into the system. As a result, synonyms and alternate spellings are
not automatically searched.

UniSZA
b. Wide capability; the computerized search will often find many sources that
contain the key words that the researcher has used in the system, but the
citation is not related to the subject of research.
c. Limited sources; the computerized databases are limited in the numbers

of years the information stored in their collection. For example, it is usually
necessary to revert to a traditional library search for journal articles written
prior to the mid 1970s.
3. The Library Search
Library search is more time consuming compared to computerized search.

However, the researcher will know all possible sources of secondary data
only by doing a thorough search. Furthermore, not all databases are
available for everyone due to some constraints such as cost and the internet
access.
a. The Books Sources
In searching the information from this source, the researcher must obtain
information about the title of the books. This information can be obtained
from the general bibliography of books in the library. The researcher
should make sure that the key words about the topic used is not too
narrow and if needed should expand into different categories to ensure
that searching process is smooth. Basically, the researcher obtains
secondary data from books for the purpose of supporting his ideas and to
illustrate something such as the underlying theory of the research project
at hand.
b. The Periodicals Sources
Actually, periodicals are the best single source of information for the
researcher because there are thousands of publications covering almost
all subjects. Periodicals refer to professional journals, public opinion
surveys, business periodicals, etc., which contain important information
to support research. Professional journals comprised of Journal of
Management, Journal of Marketing, Journal of Education, Journal of

UniSZA
Medicine, and so forth. Periodicals are useful in providing the most

current information to support your research form its data source.
c. The Government Sources
Government publications are good sources of information in many areas.

There are important documents concerning many scientific fields,
government relations to business, and business operations. Government
agencies produce data every year. Materials from all branches of the
government are listed by the name of the issuing department, for
example Ministry of Health, Ministry of Higher Education, Kuala Lumpur
City Hall, and many more. Most of the data published by the federal
government are assured of the accuracy and quality of investigation.
d. The Miscellaneous Sources
After the researcher has gone through all the three sources, several
sources might be useful in searching secondary data for a research
project. There are many other sources such as data published by
business organization, non-governmental organization, local
governments, etc. One of the major miscellaneous sources is the
publications from higher education such as university’s bulletin,
academic researches, and doctoral dissertations. For example, UiTM
publishes its bulletin for internal circulation called Buletin UiTM every
three months; UiTM also published the research papers produced by its
academic staff every year after the annual Conference on Scientific and
Social Research (CSSR).
EVALUATING THE QUALITY OF SECONDARY DATA
The researcher should evaluate the quality of secondary data according to

some limitations and disadvantages we have already discussed. The
evaluation of secondary data comprises of two forms; first, how well does the
data fit the research needs; and second, how confident is the researcher with
the accuracy and legitimacy of the data. Before using secondary data, the
researcher must ask how relevant the data is to his research project.

UniSZA
The researcher should ask the following questions to evaluate secondary data:
1. Is the data obtained relevant with the problem definition?

2. Is the data obtained relevant to the population of the study?
3. Does the data cover the period of the proposed study?
4. Does the data appear in the correct units of measurement?
5. Does the data adequately cover the subject of interests?
APPLICABILITY USE THE

OF DATA DATA
Does the data help to No Yes

define the problem? Can the accuracy of
No data collection be
Yes
verified?
Is the data relevant to No
Yes
Can the data be reworked? If yes, Continue
the period of interest?

No Is there a possibility of
Yes data bias?
Is the data relevant to No
the population? Yes
No
STOP No Is the cost of data
Yes acquisition worthy?
Do the terms and No
variable classifications Yes
apply? No Is it possible to get from
Yes the original source?
No
Are the units of
measurement ACCURACY OF
comparable? DATA

UniSZA
Figure 7.1: Secondary data evaluation process
However, the researcher may also look at the following criteria, issues, and
remarks for the purpose of obtaining suitable secondary data for the study.
Criteria Issues Remarks

What is the Method of data collection, Data should be reliable, valid,
methodology used Response rate, Quality of and generalizable to the problem
by the previous data, Sampling technique at hand.
researcher in data used, Sample size used,
collection? Questionnaire design,
Field work, Data analysis
How accurate is the Errors in research design, Assess accuracy by comparing
data? Sampling, Data collection, data from different sources.
Data analysis, Reporting
How recent is the Time lag between data Census data are periodically
data? collection and its publication, updated by syndicated firms
Frequency of data updates.
What is the original Why were the data The original objective of research
research objective? collected? will determine the relevance of
the data to the current study.
What is the nature of Definition of terms, Reconfigure the data to increase
the original data Unit of measurements used, their usefulness if possible.
Categories used,
Relationship examined
How reliable is the Expertise, credibility, Data should be obtained from the
data? reputation, and original source rather than from
trustworthiness of data the acquired sources. What is the
sources. measure of reliability?

UniSZA
Exercises
True-False Question
Primary data are data which are collected for purpose other than
1.
the current study at hand.
The question of its relevance and accuracy are common
2.
disadvantages of using secondary data.
Secondary data should be carefully examined in order to
3.
determine how key variables are defined and measured.
Relevance and accuracy are common advantages of using
4.
primary data.
Before collecting external secondary data, internal secondary
5.
data sources should be carefully examined.
Published external sources of secondary data are broadly
6.
classified as general business data or government data.
Full-text databases are composed of citations to articles in
7.
journals, magazines, and newspapers.
Dependability of data is typically determined through examining
8.
the credibility or reputation of the source.
The accuracy of secondary data is about the same for most
9.
countries in the world.
The researcher is ethically obliged to ensure the relevance and
10.
usefulness of secondary data to the problem at hand.
Multiple Choices Question

UniSZA
1. Data collected for some purposes other than the problem at hand is known as:
a. Primary data
b. Secondary data
c. Qualitative data
d. Quantitative data
2. Secondary data refers to data that ________________

a. Were purchased from an organization hired to collect data.
b. Were collected by intermediaries in the channel of distribution.
c. Were collected without direct contact with respondents.
d. Were collected for a purpose other than the problem at hand.
3. The difference between secondary data to primary data is:

a. More easily accessible
b. More expensive
c. Require more time to obtain
d. More accurate
4. The data originated by a researcher for the specific purpose of addressing the
problem at hand is called ___________.
a. External data
b. Internal data
c. Primary data
d. Secondary data
5. Factors used to evaluate secondary data should include:

a. The methodology used to collect data.
b. The accuracy of the data.
c. The dependability of the data
d. All of the above factors
6. If you want to examine the income statements and balance sheets of one of
your company’s biggest competitors, you would use a ____________
database.
a. Geographic
b. Citizen
c. Financial
d. Social
UniSZA
7. The starting point for any search for secondary data should be __________
a. Internal sources
b. Original sources
c. Acquired sources
d. External sources
8. Citation references to articles in journals, magazines, and newspapers are

example of a _________ database.
a. Full-text
b. Numeric
c. Directory
d. Bibliography

UniSZA
Chapter 8: Designing a Questionnaire
CHAPTER DESIGNING A QUESTIONNAIRE

8
WHAT IS QUESTIONNAIRE?
Questionnaire is defined as a set of structured questions designed to collect

the data required for research. It is also defined as a set of focused
questions to obtain information from the targeted respondents. Collecting
data using questionnaire helps to reduce time and cost since it allows
respondents to complete it at their own convenient time without any
intervention from the researcher. Hence, the response obtained reflects the
true opinion from the respondents.
Advantages of using questionnaires (besides their self-administered):-
1. Can send to respondents through ordinary mail as well as e-mail, hence

it is faster, cheaper, and can cover a broader geographical area.
2. Much cheaper than the face-to-face interview or any other methods of

data collection.
3. Respondents are more willing to be truthful in their opinion since their

anonymity is guaranteed and they are not influenced by the researcher.
THE PURPOSE OF QUESTIONNAIRE
1. Obtains accurate information from the respondent. The researcher is trying

to obtain the accurate information concerning the problem in the study.
Accurate information is obtained by asking the right question to the right
person, and allows him to respond at his own convenient time.
2. Provides a structure to the interview so that it flows smoothly and orderly. It
is important in any survey that all respondents are asked the same set of
questions in the same way. Without this structure, the responses might be
inconsistent and would be impossible to build the clearer picture regarding
the problem at hand.
167
3. Provides a standard format on which facts, comments and attitudes can

be recorded. This is important since different format might produce
different information even if it is given to the same respondent.
4. Facilitates data processing. Various responses are recorded in a
specified place in a questionnaire and it helps during data input.
THE CRITERIA OF GOOD QUESTIONNAIRE:

1. It must translate the information needed into a set of specific questions to
the respondents, and the questions will yield the desired information.
2. A questionnaire must uplift, motivate, and encourage the respondents to

become involved in the study and become willing to complete it.
3. A good questionnaire should have no biased questions which would

discourage respondents from giving the true response.
Questionnaire is a good instrument for data collection but its development

requires time and effort. Questionnaire is used to obtain three types of
information. First, the most important information is target data such as facts,
attitudes, preferences, expectation about the central topic of the study, etc. The
second type of information is classification data which concerns with the
respondents‟ characteristics needed for classification and analysis such as
gender, age, family income, social classes etc. The third type of information
called administrative data includes the respondent identification, interviewer
identification, date, place of interview, etc.
THE PROCESS OF DESIGNING A QUESTIONNAIRE
The steps involve in designing questionnaire can be divided into 4 stages. The
first and second steps are included in the first stage called the identifying
process. The second stage consists of drafting questions. It consists of three
steps namely,
(1) Determine the content of question

(2) Determine the wordings of question
(3) Determine the structure of response
168
The third stage is organizing the questions, deciding the form and its
layout. Finally, after the questionnaire has been developed, a series of test
should be conducted to eliminate sources of error and assuring the accuracy of
data (see Figure 8.1).
1. Identify the information required

Identifying Process
2. Determine method of data collection
3. Finalise the content of question
4. Determine the wordings of question

Drafting Questions
5. Determine the structure of responses
6. Decide on the order of question
7. Decide on the form and layout Organizing Questions
8. Pre-test the questionnaire

Questionnaire Testing
Figure 8.1: Steps in Questionnaire Development
1. First stage: Identifying Process
Identify what information is required for this research. There are few steps
involved:
1. Identify the research questions concerning the problem which the top
management must address.
2. Develop the questions that the researcher must address to contribute to

the solution of the management problem.
3. Develop specific questions in order to answer the research question. In

this stage, the researcher should go back to each research questions.
4. Develop questions that the respondents must answer to provide the

information to accomplish the objective of the study.
169
Identify method of data collection
The researcher should identify and determine the data collection method to be
used in the study before designing the questionnaires. This step involves the
following tasks:
a. Identify the mode of communication
The decision on mode of communication will affect the design of the

questionnaire in terms of length of questionnaire including the complexity
and variety of questions. For example, it is possible to use graphic in
personal interview since the respondents interact face to face with
interviewer. However, the graphic design is not appropriate for telephone
interviews.
b. Decide the structure of questionnaire
The structure of question varies according to the method of data collection

used. Different data collection method requires different structure of
questionnaire. For example, the unstructured questionnaire that consists of
open-ended question is basically used in in-depth interview in which the
interviewer‟s task is to encourage the respondent to participate in the study.
Open-Ended versus Close-Ended Question
a) Open-ended question
i Respondents have the freedom to provide their views or opinions

concerning certain issues.
ii Respondents answer the questions in their own words. Completely

unstructured questions allow respondents to answer in any way they
like. Types of open-ended questions include:
 Questions that ask respondents to state the first word that

comes to their mind when a particular word is mentioned.
For example, when the particular word is “BMW”, the response
from a respondent would be “exclusive”, “expensive”, and
“stylish”. When the first word is “Proton”, the response might be
“economical”, “affordable”, and “home-made”.
170
 Sentence, story or picture completion questions ask respondents

to complete partial sentences, stories or pictures in their own
words.
iii Number of question is limited since this type of response is open-

ended which is very cumbersome to analyze.
b) Close-ended question
 The responses are already provided in the questionnaire. The

respondent only need to tick one response most related to him.
 The respondents have the option to choose “others” if none of the

choices are related to them.
 Types of close-ended questions include:
i. Multiple-choice questions that require respondents to answer

“yes” or “no” or choose from a list of several answer choices.
ii. Questions that ask respondents to rank their response based on

their preference on a given item.
iii. Questions that measure the respondents‟ perception of how

strongly they agree or disagree with the given statement.
171
Example 8.1
Instruction:
I would like to assess how strongly you agree or disagree with every statement
provided here concerning your university. You are free to indicate your
response between one (1) to ten (10). Your score of (1) would indicate you
strongly disagree with the statement and score (10) would indicate you strongly
agree with the respective statement.
Remember, you can tick your response anywhere between 1 to 10.
The statement provided How strongly you agree or disagree with the statement
This university ........... Strongly Strongly

Disagree Agree
1. has adequate car parks. 1 2 3 4 5 6 7 8 9 10
2. has a beautiful campus environment. 1 2 3 4 5 6 7 8 9 10
2. Second Stage: Drafting the Question
a. The Content in a Questionnaire
The questionnaire contains questions for the respondent. Sometimes the

respondents are asked questions which they have no idea to respond.
For example, the respondent is asked about the amount spent on petrol
per week in his family. Since the family consist of many individuals, the
respondent has no information about the petrol expenditure in a week by
his spouse, and also his sons or daughters.
Example 8.2: An incorrect question

How much is the weekly expenditure for petrol in your family?
172
Example 8.2a: The correct question
How much is your weekly expenditure for petrol?

The questions should centre on the topic that the respondents can
remember. The incorrect question attempts to discover the topic that is
beyond the ability of respondents to remember. The inability to remember
will lead to error of omission, telescoping, and creation. Omission is the
inability to recall an event that actually took place. Telescoping happens
when the respondents remember an event that occurs more recently than
the actual time. Finally, creation is an error that occurs when a respondent
tells something concerning the fake event, or the event that does not take
place.

What are the items you purchase from this store last year?
What are the items you purchased from this store in the last two days?
In designing a question, the researcher should be careful with the question

that the respondent can articulate the response. Based on the example
below, the respondent is asked to describe the best university he wishes to
further his studies. This question will lead the respondents to ignore
because of difficulties to phrase their answer. The researcher should
provide a list of possible answers that the respondents could choose from.
Please describe the characteristics of an excellent university you wish to

further your studies.
173
Please choose at the following statements that coincides your feelings

about the excellent university you wish to further your study.
1. The university should have nationally known professors.

2. The university should have excellent academic facilities.
3. The university should have excellent sport facilities.
b. The questionnaire contains questions that are straight forward

and easily understood by the respondents.
c. The question content should perform the function to obtain

information needed by the study. Each question in the
questionnaire should contribute to the information to be used in
data analysis.
d. The question content should cover the issues and obtain sufficient
data for specific problem. The researcher should avoid double-
barrelled questions that attempt to cover two issues in a single
question. In the following example, the researcher wants to cover
two issues (variety and taste) in a single question.
Example 8.5: The incorrect question
Indicate how far you agree or disagree with the following statements.
The foods provided at the dining hall are varied and tasty.
The response from the respondents will not be accurate since there are two
issues in one question. Furthermore, the respondents might refuse to
answer if they agree with one issue but not with the other issue. It is
advisable to break this question into two questions as shown in the
following example.
Indicate how far do you agree or disagree with the following statements.
The food provided at dining hall has variety.
The food provided at dining hall is tasty.
174
b. The Wordings of a Question
Question wording involves the process of converting the content into

words which are clear and easily understood by respondents. Question
wording is the most difficult task in developing a questionnaire because if
the question is poorly worded, it will sway the respondent to respond or
the response provided is incorrect. To ensure that the question is
properly worded the researcher must make much revision on each
question to satisfy the six checklists below:
i Is the question stated in normal words of conversation?

ii Is the question clearly stated?
iii Are there any misleading assumptions?
iv Is there a biased wording in a question?
v Is there the right degree of personalization?
vi Are adequate alternatives presented?
However, the following guidelines might help the researcher in the process
of wording question.
a. The question must be stated in a clearly defined term.
Example 8.6: The incorrect question
How much water do you take every day?

The question is not clear concerning the measurement of water.
How many litres of water do you take every day?

Another example: “How do you pay your monthly bills?” The question
does not state exactly what bill (water, electricity, phone, maintenance,
or rental)
175
b. Use ordinary words
Take note that a personal interview is a correspondence between an

interviewer and the respondents. The challenge to the researcher is to
ensure that the respondent understands what he says. In doing this, the
words must be simple enough to ensure the respondents understand the
exact intention of the researcher. The task involved is to reduce the level
of language difficulty.
Another problem is the use of technical jargon. The jargons will confuse
respondents and they might refuse to respond since the terms used are
quite alien to them.
Do you think that the market of shoes in this area is already saturated?
Do you have plenty of choices when buying your shoes?
c. Avoid ambiguous words
The ambiguous words such as “usually”, “normally”, “frequently”,

“sometimes” should not be used in the questions since they have
different meanings among the respondents. The use of ambiguous
words will lead to inaccurate response since the alternative answers
chosen by the respondents might be overlapped.
How often do you go to cinema in a month?

1. Never
2. Occasionally
3. Sometimes
4. Often
5. Regularly
176
How often do you go to cinema in a month?

1. Less than once
2. 1 or 2 times
3. 3 or 4 times
4. More than 4 times
d. Use personalised words
Using personalised words is useful when designing a sensitive question.

The personalised words will encourage the respondents to answer the
question by expressing their attitudes or opinions on behalf of other
people.
What do you think about wearing sexy clothes to class?

What do you think about smoking in public places?
What is your opinion on overweight?
What do people think about wearing sexy clothes to class?

What do people think of smoking in public places?
What do people think of overweight?
e. Making explicit assumption
If you want to make certain assumption, you should ensure that the
assumption is worded explicitly in the question to prevent the respondent
from overestimating the main subject or topic of question.
177
Do you like swimming? (The response might be „yes, but there is hardly a
swimming pool in this area‟ or „no since the swimming pool is not available
in this area)
Do you like swimming if the swimming pool is available in your area?

(The response might be straight „yes‟ or „no‟)
f. Avoid bias wordings or leading question.
Bias wording or leading question occurs when the respondent is given

clues in a question which will lead the respondents to respond in favour
of the clues. The clues for bias wordings or leading question come from
positive or negative statement about the subject.
i. Do you think Proton Savvy is the best car in its class since it is proven in
the test conducted by an expert?
ii. Cigarette smoking has high probability of male impotence and many
cases of sexual dysfunction; my question to you is “Do you smoke?”
i. Do you think Proton Savvy is the best car in its class?

ii. Do you smoke a cigarette?
178
g. Use adequate alternatives
The researcher needs to provide adequate alternatives with respect to

the subject in question. These alternatives should be explicitly expressed
to avoid bias.
Do you think the government should reduce the tax in order to reduce the
smuggling activities of products into the country?
Do you think the government should reduce the tax in order to reduce
smuggling activities of products into the country, or you think that
government should determine other mechanism to reduce the smuggling
activities?
3. The Response Options in a Question
After the researcher has properly worded the questions, the next process is
to determine the response options for each question. There are two options
that range from the open-ended response (free choice of words) to the
closed-ended response (specific alternatives provided). Open responses
refer to the freedom given to the respondents to phrase their answer. Close
responses typically are categorized as dichotomous or multiple choice. The
best decision in designing the response is by giving the respondents some
freedom to express their opinions. This can be achieved by providing more
choices of responses and ask the respondents to indicate the most
appropriate response.
179
4. The Ordering of a Question
a. Ordering by section
Normally, the questionnaire is employed to obtain three types of

information required in research. The questions are classified as (1)
basic or target information, (2) classification information, (3)
identification or administrative information. The basic or target
information is the most important data for the research since it directly
relates to answering research question. This information consists of
facts, attitudes, preferences, and expectations about the research
problem. The second type of information is classification information
that is used to classify respondents based on socio-economic status
(education, occupation, income, etc) and demographic variables
(gender, age, religion, race, etc). However, the use of classification
question in the questionnaire would depend on the research objectives.
Finally, the third type of information is called identification or
administrative data. Among the information required are respondent‟s
names, place of birth, mail address, and telephone number or the
interviewer identification such as date, time, and place.
In a general guideline of questionnaire design, the basic or target

information must be obtained first, and then followed by the
classification information which is normally placed in the last section of
questionnaire. Identification or administrative information is not in the
questionnaire, but is obtained by the researcher for the record.
b. Opening question
The opening question is extremely important since it will establish
rapport, build confidence, and encourage respondents to complete the
questionnaire. This question should be simple, interesting and not
threatening.
c. General question followed with specific question

The researcher should ask general questions in the early stage of
questionnaire followed by the more specific questions. This technique
is called funnel approach that refers to a strategy for ordering question
beginning with the general questions then followed by more specific
questions.
180
Example 8.13: A general question
Do you think sales promotion is important for purchasing a new car?
Example 8.13a: The specific question
How important is free gift when purchasing a new car?
d. Difficult or sensitive question
The respondent is less likely to ignore the question after rapport has been
established and the respondent has been involved in answering general
and specific question. Therefore, a difficult or sensitive question must be
placed last in each section in the questionnaire. For example, the question
on income and telephone number must be the final item in the section of
classification question.
e. The order of the possible responses to a particular question is also

important; examples strongly agree, agree, not sure, disagree, and
strongly disagree.
Example 8.14:
Strongly Strongly
Disagree Agree
1 2 3 4 5 6 7 8 9 10
3. Stage three: Organizing the Questions
Once the researcher has drafted the questions, he would organize them in
a form or layout called questionnaires. The forms of a layout such as the
format, instructions, spacing, and positioning of questions in the self-
181
administered questionnaires would likely influence the respondents to

respond.
a. The Format of a Questionnaire

i The items and the questionnaire should be presented in an
attractive, professional and easy to understand format.
ii All questions and pages should be clearly numbered. It is important

particularly when branching is used. Meanwhile, the numbering of
question assists the process of coding responses.
iii The questionnaire should contain clear and explicit directions as to

how it should be completed and returned. Basically, the cover letter
and instructions would explain the importance of the study, the
contribution of respondents in the study, and the secrecy of
responses.
iv The questions should be objective.
v The questions should be ordered from light to heavy tone and also
from general to specific.
vi Transitions should be used from one topic to the next. A good

questionnaire must consist of several parts based on the types of
information to obtain. More importantly it would prevent from
confusing respondents while responding to the questions.
vii Examples should be given when necessary. Basically, an example

is given to complement the instructions to explain how to respond
to the questions.
b. The Form and Layout
i A questionnaire must be produced on a quality paper with

professional appearance. It is important to build confidence and
positive attitudes of the respondents concerning the study.
ii It is advisable to produce a questionnaire into a booklet if the

questions consist of several pages. The booklet format is easier to
handle and provide convenience to the respondents.
182
iii The questions should not be crowded into a page in order to show
the questionnaire look simple. It will cause the questionnaire to look
complex which will result in lower cooperation and completion
rates.
iv The instructions on how the question should be answered must be

placed before the questions and as close to the questions as
possible.
v The use of colours also will give an advantage to the questionnaire.

Colour coding in the questionnaire might be used for branching
questions and isolate the different groups of respondents.
c. The Cover Letter

An essential part of any questionnaire is the cover letter that
accompanies it. The message in a cover letter is important since it
helps set the scene for what is to come. Cover letters can make or
break the success of a project. A good cover letter is especially
important for questionnaires that are mailed to respondents so that
the sense of authority is established and importance of the project is
conveyed. What the cover letters should have:
i It is done on official letterhead, which helps to impress

respondents and increases the likelihood of responding.
ii It is dated recently, indicating there is some urgency to the

request for cooperation.
iii The letter is personalized. It does not open by stating “Dear

participant” but Dear Mr., Mrs., Miss etc.
iv The purpose of the questionnaire and the importance of the study

are clearly stated.
v A time estimate is given so they know when to return it. For face-
to-face interview this is not needed.
vi It promises the confidentiality of respondents and their responses

will not be used for other purposes except for this particular
research.
183
vii It begins with the request for a sincere participation and ends with
sincere thanks for cooperation.
Example 8.15: The cover letter
Dear respondent:
I am doing an academic research entitled:
The Effect of Service Quality and Corporate Image on Students’ Loyalty in

Higher Education: The Case of a Local University
Congratulations, you have been selected as a respondent for this study. Your
sincere cooperation is crucial in determining the success of this study. In this
study, I would like to assess your perception concerning certain factors of your
university. Please read carefully all instructions pertaining to every section and
answer every question sincerely. There is no right or wrong answer. All
responses provided will be treated with strict confidential and will be used for
this academic research only. Thank you very much for your cooperation.
Sincerely,
………………......................………
Assoc Prof Dr Zainudin Hj Awang
Centre for Graduate Studies (CGS)
Universiti Teknologi MARA (UiTM)
15100 Kota Bharu Kelantan
e-mail: zainudin888@kelantan.edu.my
4. Final Stage: Pre-testing the Questionnaires
The final stage in questionnaire designing process is pre-testing. Pre-testing is

used to improve the questionnaire by identifying and eliminating its weaknesses
based on the responses from the respondents. Pre-testing the questionnaire
will examine all aspects in the questionnaire including the question content, the
wordings, the ordering of questions, the instructions, and the layout. The
respondents for pre-testing should be drawn from the target population of the
184
study. Pre-test may be conducted through personal interview, mailed

questionnaire, telephone interview, and also through email. However, personal
interview is the best method for pre-testing since the interviewer can observe
the reactions and attitudes of respondents for every item in the questionnaire.
There are two common procedures of pre-testing the questionnaire namely;

protocol analysis and debriefing. In the protocol analysis procedure, the
respondents are made known of the pre-test activity. The respondents are
treated as collaborator in the refinement process. This technique allows
detailed probing in each part of questionnaire. However, this technique will
introduce bias of friends or associates to be used in pre-testing process.
In debriefing technique, the researcher does not inform the respondents about
the pre-testing. The respondents are told about the pre-testing process only
after they have completed the questionnaire. The probing of questions is done
through asking respondents to explain their response for each question in the
questionnaire. The respondents are also encouraged to state any problem they
encountered while answering the questionnaire. The weakness of this
technique is the difficulty to get cooperation from respondents. Hence, the
number of successful probing is limited.
185
A SAMPLE OF A COMPLETE QUESTIONNAIRE 1
This Questionnaire is obtained from Zainudin (2007)
Dear respondent:
I am doing an academic research entitled:
The Effect of Service Quality and Corporate Image on Students’ Loyalty

in Higher Education: The Case of a Local University
Congratulations, you have been selected as a respondent for this study. Your
sincere cooperation is crucial in determining the success of this study. In this
study, I would like to assess your perception concerning certain factors of your
university. Please read carefully all instructions pertaining to every section and
answer every question sincerely. There is no right or wrong answer.
All responses provided will be treated with strict confidential and will be used for
this academic research only.
Thank you very much for your cooperation.
Sincerely,
………………………………….........
Assoc Prof Dr Zainudin Hj Awang
Centre for Graduate Studies (CGS)
Universiti Teknologi MARA (UiTM)
15100 Kota Bharu Kelantan
e-mail: zainudin888@kelantan.edu.my
186
Instruction:
I would like to assess how strongly you agree or disagree with every
statement provided here concerning your university. You are free to indicate
your response between one (1) to ten (10). Your score of (1) would indicate you
strongly disagree with the statement and score (10) would indicate you strongly
agree with the respective statement.
Remember, you can tick your response anywhere between 1 to 10.
Example:
The statement provided How strongly you agree or disagree with the
statement
This university ........... Strongly Strongly

Disagree Agree
1. has an adequate parking lot on campus. 1 2 3 4 5 6 7 8 9 10
2. has a beautiful campus atmosphere. 1 2 3 4 5 6 7 8 9 10
187
SECTION 1: Corporate Image Items
statement concerning the corporate image of this university.
You can mark your sincere response anywhere between 1 to 10. Please
remember, there is no right or wrong answers.
1. Corporate Image of this university

Disagree Agree
1. has a quality of academic staff (lecturers) 1 2 3 4 5 6 7 8 9 10
2. maintains a tradition of academic excellence 1 2 3 4 5 6 7 8 9 10
3. has lecturers with good academic reputation 1 2 3 4 5 6 7 8 9 10
4. maintains good industrial linkages 1 2 3 4 5 6 7 8 9 10
5. practices an ISO 9001:2000 in its system 1 2 3 4 5 6 7 8 9 10
6. contributes significantly to the society 1 2 3 4 5 6 7 8 9 10
7. has a good relationship with its alumni 1 2 3 4 5 6 7 8 9 10
8. enjoys good cooperation with other universities 1 2 3 4 5 6 7 8 9 10
9. achieves excellent in its research programs 1 2 3 4 5 6 7 8 9 10
10. produces excellent publications 1 2 3 4 5 6 7 8 9 10
11. wins awards in its research activities 1 2 3 4 5 6 7 8 9 10
12. always organizes seminars or conferences 1 2 3 4 5 6 7 8 9 10
13. enjoys close cooperation with the government 1 2 3 4 5 6 7 8 9 10
14. offers a variety of courses to choose from 1 2 3 4 5 6 7 8 9 10
15. offers a flexible learning system 1 2 3 4 5 6 7 8 9 10
16. always makes quality as the main priority 1 2 3 4 5 6 7 8 9 10
188
SECTION 2: Service Quality Items
statement concerning the quality of services at this university.
2. Service Quality of this University

Disagree Agree
1. has an adequate infrastructure facilities 1 2 3 4 5 6 7 8 9 10
2. has a beautiful campus environment 1 2 3 4 5 6 7 8 9 10
3. has a good campus transportation system 1 2 3 4 5 6 7 8 9 10
4. has adequate academic facilities 1 2 3 4 5 6 7 8 9 10
5. has a comfortable cafeteria or food courts 1 2 3 4 5 6 7 8 9 10
6. has good sports facilities 1 2 3 4 5 6 7 8 9 10
7. has conducive rooms in hostels 1 2 3 4 5 6 7 8 9 10
8. has a library with adequate materials 1 2 3 4 5 6 7 8 9 10
9. offers a flexible class schedule 1 2 3 4 5 6 7 8 9 10
10. has an effective information delivery system 1 2 3 4 5 6 7 8 9 10
11. has an up to date syllabus in its programs 1 2 3 4 5 6 7 8 9 10
12. has an adequate laboratory facilities 1 2 3 4 5 6 7 8 9 10
13. has lecture rooms equipped with technology 1 2 3 4 5 6 7 8 9 10
14. offers courses suitable for the future career 1 2 3 4 5 6 7 8 9 10
15. has lecturers with the experience in the field 1 2 3 4 5 6 7 8 9 10
16. provides courses with the required skills 1 2 3 4 5 6 7 8 9 10
17. has curriculum which is relevant to job prospects 1 2 3 4 5 6 7 8 9 10
18. has conducive lecture rooms 1 2 3 4 5 6 7 8 9 10
19. has the multi-media and ICT facilities 1 2 3 4 5 6 7 8 9 10
20. has counter personnel who are student-focused 1 2 3 4 5 6 7 8 9 10
21. has non-academic staff who are helpful 1 2 3 4 5 6 7 8 9 10
22. has counselling staff who provide useful advice 1 2 3 4 5 6 7 8 9 10
23. has lecturers who are knowledgeable 1 2 3 4 5 6 7 8 9 10
24 has support staff who are always available 1 2 3 4 5 6 7 8 9 10
25 has lecturers who create good atmosphere 1 2 3 4 5 6 7 8 9 10
26. has lecturers who use suitable teaching materials 1 2 3 4 5 6 7 8 9 10
189
27. has staff whose appearance is always neat 1 2 3 4 5 6 7 8 9 10

28. has lecturers who are easy to meet during office 1 2 3 4 5 6 7 8 9 10
29. has lecturers with good relationship with students 1 2 3 4 5 6 7 8 9 10
30. has lecturers who motivate students to perform 1 2 3 4 5 6 7 8 9 10
31. has lecturers who make learning interesting 1 2 3 4 5 6 7 8 9 10
32. has lecturers who give explanation 1 2 3 4 5 6 7 8 9 10
33. has lecturers who evaluate assignment fairly 1 2 3 4 5 6 7 8 9 10
34. has lecturers who are concern with performance 1 2 3 4 5 6 7 8 9 10
190
SECTION 3: Customer Satisfaction Items
statement concerning your satisfaction being a student of this university.
You can mark your sincere response

anywhere between 1 to 10. Please remember, there is no right or wrong
answers.
3. Student’s satisfaction towards this university

As a student of this university, I….. Strongly Strongly
Disagree Agree
1 can follow my study programs successfully 1 2 3 4 5 6 7 8 9 10
2 can complete my academic work with ease 1 2 3 4 5 6 7 8 9 10
3 have a self-confidence to succeed in my study 1 2 3 4 5 6 7 8 9 10
4 have obtained the required knowledge 1 2 3 4 5 6 7 8 9 10
5 am satisfied with the development of human capital 1 2 3 4 5 6 7 8 9 10
6 am satisfied with the process of course registration 1 2 3 4 5 6 7 8 9 10
7 am satisfied with the process of knowledge transfer 1 2 3 4 5 6 7 8 9 10
8 am satisfied with the system at this university 1 2 3 4 5 6 7 8 9 10
9 am satisfied with the continuous evaluation system 1 2 3 4 5 6 7 8 9 10
10 am satisfied with academic rules and regulations 1 2 3 4 5 6 7 8 9 10
11 am satisfied with the process in obtaining references 1 2 3 4 5 6 7 8 9 10
12 am satisfied with academic facilities provided 1 2 3 4 5 6 7 8 9 10
13 am satisfied with the security system on campus 1 2 3 4 5 6 7 8 9 10
14 have made the right decision to study here 1 2 3 4 5 6 7 8 9 10
191
SECTION 4: Customer Loyalty Items
Please indicate how strongly you agree or disagree with the following statement
concerning your loyalty towards this university.
4. Student’s loyalty towards this university

Your statement...
Strongly Strongly
Disagree Agree
1 I will choose this university again for my post-graduate studies 1 2 3 4 5 6 7 8 9 10
2 I will study at this university even if the cost has increased 1 2 3 4 5 6 7 8 9 10
3 I am willing to pay more due to many benefits at this university 1 2 3 4 5 6 7 8 9 10
4 I will participate in programs organized by this university 1 2 3 4 5 6 7 8 9 10
5 I will tell positive things about this university to others 1 2 3 4 5 6 7 8 9 10
6 I will recommend this university as a suitable place to study 1 2 3 4 5 6 7 8 9 10
7 I will encourage my friends to study at this university 1 2 3 4 5 6 7 8 9 10
8 I will propose the firms to have linkages with this university 1 2 3 4 5 6 7 8 9 10
9 I will recommend this university to employers for recruitment 1 2 3 4 5 6 7 8 9 10
192
Chapter 8: Questionnaire Design
SECTION 5: DEMOGRAPHIC PROFILES
Provide your response by circling the appropriate number
1 Your Gender:
Male ………................................ 1
Female …………............................. 2
2 Your Age
Below 20 years …………............................. 1
20 - 22 years …………............................. 2
23 - 25 years …………............................. 3
Above 25 years …………............................. 4
3 Length of time in this university:

Less than 2 semesters …………………….............. 1
2 but less than 4 sem …………………….............. 2
4 but less than 6 Sem ........................................... 3
More than 6 Semesters ........................................... 4
4 Parent‟s monthly income:

Less than RM 2000 ………..........………........... 1
RM 2001 – RM 3000 ..…………………................ 2
RM 3001 – RM 4000 ........................................... 3
RM 4001 – RM 5000 ………................................ 4
RM 5000 and above ........................................... 5
5 The highest academic qualification for your parent:

No Schooling …………………................ 1
SPM/STPM ......................................... 2
Diploma ......................................... 3
Ijazah ………….………............... 4
Sarjana ......................................... 5
PhD or Profesional ......................................... 6
Assoc Prof Dr Zainudin Awang 191

UiTMKelantan
6 Your parent‟s occupation: Father Mother

Profesional field .................... 1 1
Government servant .................... 2 2
Teachers .................... 3 3
Clerical works .................... 4 4
Self employed/business .................... 5 5
Pensioners .................... 6 6
Others (please specify) .................... 7 7
7 Is this university your first choice in your application for

admission? (in UPU application)
Yes ......................................... 1
No ......................................... 2
Not Sure ......................................... 3
Thank you very much for your sincere cooperation in this study.

UiTMKelantan
A Sample of a Complete Questionnaire
SEasonS
HOTEL
K U A L A L U M P U R
Dear Guest,
We are honoured that you have selected our Hotel for your recent business trip/leisure
travel. Thank you for your patronage.
The objective of the Legend Group of Hotels and Resorts is to provide our guests with
gracious service and comfort. We would very much appreciate your help in achieving this
objective.
We randomly select a number of guests and invite them to let us know their feelings
about their current hotel stay. As one of our valued guests, we seek your co-operation in
completing this questionnaire and returning it by folding this form and securing the
adhesive flaps.
All information provided will be used in our planning to serve you better in the future.
Thank you for your kind assistance. We look forward to your next visit with us.
Yours sincerely,
Stacy Woo Wee Wren

President
1. Taking all things into consideration, how would you rate our overall services and
facilities? From a scale of 1 being extremely dissatisfied to 10 being outstanding,
how best would you describe your experience?
1 2 3 4 5 6 7 8 9 10
□ □ □ □ □ □ □ □ □ □
2. Please tick the boxes that best describe your feelings about the items listed below
with which you have had experience this visit.

UiTMKelantan
1 2 3 4 5 6 7 8 9 10
Overall Staff Courtesy
Overall Staff Helpfulness
Overall Staff Efficiency
Door Service
Luggage Service
Front Desk Service
Housekeeping/Maid Service
Room Comfort
Efficiency of Bathroom Equipment
Value for Money
Please comment
________________________________________________________________________
________________________________________________________________________
________________________________________________________________________
3. Below are some of the other services which you may or may not have used. Please
tick the boxes that best describe your feelings about these services which you may
have used on this visit.
1 2 3 4 5 6 7 8 9 10
Laundry/Valet Service
Telephone Service
Business Centre Service
Message Service
Recreation Facilities
Food & Beverage
Room Service
Overall Restaurant Service
Overall Lounge and Bar Service
Room Maintenance
Please comment
________________________________________________________________________
________________________________________________________________________
________________________________________________________________________

UiTMKelantan
4. Below are the individual Food and Beverage outlets, Banqueting and other facilities
which you may or may not have used. Please share your experience with us by rating
their service and facilities
Coffee House
Seasons Café 1 2 3 4 5 6 7 8 9 10
Food Quality
Service
Friendliness
Lounge
Lobby Lounge 1 2 3 4 5 6 7 8 9 10
Food Quality
Service
Friendliness
O’las Restaurant 1 2 3 4 5 6 7 8 9 10
Food Quality
Service
Friendliness
Room Service 1 2 3 4 5 6 7 8 9 10
Food Quality
Service
Friendliness
Timeliness
Bar
Kasbah 1 2 3 4 5 6 7 8 9 10
Food Quality
Service
Friendliness
Entertainment

UiTMKelantan
Seasons Crest Lounge 1 2 3 4 5 6 7 8 9 10
Food Quality
Service
Friendliness
Banquet / Conference 1 2 3 4 5 6 7 8 9 10
Set-Up
Food Quality
Service
Friendliness
Helpfulness
Light and Sound
Please comment
________________________________________________________________________
________________________________________________________________________
________________________________________________________________________
5. Is this your first visit to this Hotel? □ Yes □ No
6. Thinking of your most recent previous stay, how would you rate us in terms of
overall services and facilities on that particular experience?
1 2 3 4 5 6 7 8 9 10
□ □ □ □ □ □ □ □ □ □
7. What influenced you to stay at Grand Seasons Hotel?
□ Standard of Maintenance □ Prestigious Image

□ Quality of Service □ Conference Facilities
□ Quality of Food and Drink □ Location
□ Business Facilities □ Comfort of Guest Room
□ Value for money □ Lobby Ambience

UiTMKelantan
8. Based on what you have had experienced, would you recommend this Hotel to your
friends and business associates?
□ Yes □ No
9. If you were to return to this city, would you stay in this Hotel again?
□ Yes □ No
10. Did you experience any problem which is of concern to you during your stay with
us?
□ Yes □ No
11. If so, in which of the following areas and what was the problem?
□ Unfriendly staff □ Inefficient staff

□ Unhelpful staff □ Inadequate supplies
□ Faulty equipment □ Inefficient system
□ other
Please specify
______________________________________________________________
Comment
______________________________________________________________
12. Who did you report the problem to?
□ To Management Staff □ To Non-Management Staff

□ Did not report
13. Was your problem resolved to your satisfaction?
□ Yes □ No

UiTMKelantan
14. Compared to other hotels which you have had experienced, how would you rate our
Hotel?
□ Exceptionally Good □ Average

□ Good □ Below Average
15. Which Hotel would be your next preferred choice to stay in aside from this Hotel?
Please fill in the name of the hotel

___________________________________________
16. What is the main purpose of this trip?
□ Business □ Leisure
□ Conference/Corporate Meeting □ Airline Stopover
17. What age group are you in?
□ Below 26 □ 46-50 years

□ 26-30 years □ 51-55 years
□ 31-35 years □ 56-60 years
□ 36-40 years □ 61-65 years
□ 41-45 years □ 66 years and above
18. Please tick below your country/state of residence and write down your country/state
of origin on the space provided.
Country/State of origin
__________________________________________________
Malaysian States
□ Africa □ Sabah
□ Australia □ Sarawak
□ Brunei □ Selangor
□ Canada □ Wilayah Persekutuan – Kuala Lumpur
□ China □ Wilayah Persekutuan – Labuan
□ Europe □ Wilayah Persekutuan – Putrajaya
□ Hong Kong □ Penang

UiTMKelantan
□ Indian Sub-Continent □ Perak

□ Indochina □ Pahang
□ Indonesia □ Terengganu
□ Japan □ Kelantan
□ Korea □ Malacca
□ Middle East □ Perils
□ New Zealand □ Kedah
□ Philippines □ Johore
□ Singapore □ Negeri Sembilan
□ South America
□ Taiwan
□ Thailand
□ United States
□ Others
Once again, thank you for returning this questionnaire. We look forward to
welcoming you back to Grand Seasons Hotel.
Name: __________________________________________
Room No: _________
Address: ________________________________________
E-mail: _________________________________________
Today’s Date: ________

UiTMKelantan
Exercises

1. Open-ended questions are considered to be superior
when you wish to discover opinions.
2. A questionnaire must uplift and motivate respondents to
participate.
3. The first step in the questionnaire design process is to
determine the content of individual questions.
4. Sensitive topics, such as politics and religious life, should
be placed early in the questionnaire to improve response.
5. A disadvantage of open-ended questions is that coding of
response is expensive and time consuming.
6. Question content should not be influenced by the
interviewing method.
7. As a rule, general questions should precede specific
questions.
8. Questionnaire design is influenced by education of
respondents.
9. General questions should be placed first before specific
question.
10. A questionnaire should be produced in booklet form if it
consists of several pages.
Multiple Choices Questions
1. The purpose of questionnaire is to:
a. Translate the information needed into a set of specific questions that the
respondents can and will answer.
b. Uplift, motivate, and encourage the respondent to become involved and
cooperate with the interview.
c. Minimize response error.
d. Accomplish all of the above.

UiTMKelantan
2. The formulation of questions in a questionnaire should be guided by

____________.
a. Hypotheses.
b. Research questions.
c. Problem definition.
d. Both a and b are true.
3. To ensure that the information obtained from a questionnaire fully addresses

all research problem components, the researcher should always:
a. Make certain that the form and layout of the questionnaire is appropriate.
b. Prepare a set of dummy tables to show how the data analysis will be
structured.
c. Carefully decide on question order.
d. Reproduce the questionnaire to make sure it is readable and useful.
4. The first step in designing any questionnaire is to:
a. Specify the information needed.

b. Specify the appropriate type of interviewing method.
c. Determine the content of individual questions.
d. Decide on the question structure.
5. Which of the following is a demographic variable?
a. Choice of place to visit during a holiday.

b. Type of account maintained at a certain bank.
c. Monthly expenditure on fast food.
d. Educational background of respondents.
6. Questionnaire design is influenced by the respondent‟s ____________.
a. Age group
b. Education
c. Occupation
d. All of the above characteristics.

UiTMKelantan
7. Qualitative information is usually obtained using:
a. Structured questions
b. Unstructured questions
c. Disguised questions
d. Undisguised questions.
8. A structured question with two response alternatives, such as yes and no, is
called a:
a. Double-barrelled question.
b. Dichotomous question.
c. Telescoping question.
d. Leading question.
9. In drafting questions for a questionnaire, a researcher should ensure that:
a. The words do not have other meanings

b. The questions can be double-barrelled.
c. The questions are leading questions.
d. The questionnaire starts with difficult questions.
10. Which of the following is not an advantage of open-ended questions?
a. Respondents are free to answer whatever they want

b. Discovering unanticipated responses
c. Minimize interviewer bias
d. All of the above are advantages.
11. Open ended questions that respondents answer in their own words are
known as:
a. Unstructured questions.
b. Structured questions.
c. Dichotomous questions.
d. Telescoping questions.

UiTMKelantan
12. An „M‟ (Male) or „F‟, (Female) in a demographic section of a questionnaire

represents a:
a. Categorical variable
b. Unknown variable
c. Dependent variable
d. Continuous variable
13. What type of question is the following: “I‟m going to show you a sample of
magazine advertisement. What does this advertisement mean to you”?
a. Open-ended
b. Closed-ended
c. Dichotomous
d. Multiple choice
14. “Smoking is not just a bad habit but it is the main cause of cancer. I would
like to ask you one question, Do you smoke?” is an example of:
a. a generalization question
b. an implicit alternative question
c. a position bias question
d. a leading question
15. A disadvantage of open-ended questions is that:
a. High potential for interviewer bias.

b. Open-ended questions are not suitable for self-administered
questionnaires.
c. Coding of responses is expensive and time consuming.
d. All of the above are disadvantages of open-ended questions.
16. Compared to open-ended questions, multiple-choice questions:
a. Reduce interviewer bias.

b. Are easier to design
c. Are less costly for coding and data processing.
d. Both a and c are correct.

UiTMKelantan
17. Information should be obtained in a questionnaire in the following order:
a. Identification; classification; basic

b. Basic; identification; classification
c. Classification; basic; identification
d. Basic; classification; identification
18. Difficult questions should be placed ____________ in the questionnaire.
a. Early
b. In the middle
c. Late
d. Anywhere
19. Which form of response is considered superior when you wish to discover
opinions and degrees of knowledge?
a. Open-ended questions
b. Dichotomous questions
c. Multiple-choice questions
20. Which of the following are reasons for NOT including a question in
questionnaire?
a. Though interesting, the information cannot be used.

b. The question is relevant and focused.
c. Different people can interpret the question differently.
d. Both (a) and (c)

UiTMKelantan
Essay Questions
1 The President of UiTM Alumni Association wishes to carry out a survey from
the members who were from classes of 2010 onwards to determine their
achievements, current activities and future aspirations. Information pertaining
to the following areas is desired: gender of members, major area of study,
grade point average, further educational pursuits (that is master‟s degree or
doctorate), current annual salary, number of full time position held since
graduation, annual salary anticipated in five years, political party affiliation
and marital status. You are asked to conduct the survey.
a. Write a draft of the cover letter needed for this survey.

b. In developing a questionnaire for the above survey, what are kind of
questions to be avoided?
2. The management of UiTM is wondering why there are many students who
refuse to consume the food that is provided at dining halls at every college.
Every day, a large amount of food is not taken and wasted, whereas the
government is paying the contractor certain amount of money per head for
the food. This has resulted in a waste of millions of ringgit every year. In
addition, every day a large amount of food is sold cheaply to local farmers to
feed their livestock. Instead of eating the food provided at the dining halls,
these students buy food at the cafeteria or from the stalls nearby.
The management of UiTM wants you to look into this matter and carry out a
research. The findings of your research should determine the main factors
that contribute to the above problem so that appropriate actions can be taken
to rectify the situation.
At the same time, you should come out with the recommendations (based on
your findings) so that the government fund is not wasted, and the students
can use their monetary loan to buy books or save for their future.
a. Draft 5 questions to obtain the information regarding the quality of food

being served at the dining halls.
b. Draft 5 questions to obtain data of the following types.
i. Nominal (1 question)
ii. Ordinal (1 question)
iii. Interval (1 question)
iv. Ratio (2 question)
UiTMKelantan
3. Briefly explain the questionnaire design process.

Give an example in the form of a question to reflect each of the following
errors.
a. Double-barrelled question
b. Loaded question
c. Leading question
d. Ambiguous question
4. a) The following questions are not written clearly. Rewrite the questions.
i. What is your income?
o RM 900 and less

o RM 901 to RM 1000
o RM 1001 to RM 2000
o RM 2001 and RM 3000
o RM 3001 to RM 4000
o RM 5000 and more
ii. How often do you go to a cinema?
o Never
o Occasionally
o Always
iii. With whom do you normally go to a cinema?
o Family
o Partner
o Friends
o Others
b) Develop ONE question that will derive:
i. Quantitative data
ii. Qualitative data

UiTMKelantan
5. State five most common mistakes that should be avoided when developing a
questionnaire. Give examples of each.
6. What are the factors that constitute bad questions in questionnaire design?
Develop TWO (2) examples each of bad and good questions
7. Rephrase the following open-ended questions and use appropriate category of

response scales for each question.
a. How much do you spend per week on groceries?

b. How satisfied are you with the car rental service?
c. Which brand of hand phone would you prefer to buy?
8. What are the advantages and disadvantages of structured and unstructured

questions?
9. State SIX (6) steps involved in a questionnaire construction

UiTMKelantan
Chapter 9: Data Management
CHAPTER
9 DATA MANAGEMENT
Data Management Process
The goal of any research is to obtain the required data, analyze these data
accordingly, and write a report based on the output of data analysis. Raw data
is simply a recorded measure of certain information collected to accomplish the
objectives of the study. The conversion of raw data into a piece of information
using appropriate statistical analysis for answering the stated research
questions is called data analysis.
After data have been collected from the respondent, the next step is to analyze
them using various software programs such as SPSS, SAS, Excel, etc. However,
before we begin to analyze the data, we need to conduct some preliminary steps
to ensure that the data are ready for further analysis (see Figure 9.1).
The process involves the following tasks;

1. Data editing
2. Data coding
3. Data entry

UniSZA
Data Editing
Data Coding
Error checking
and verification
Data Entry
Data Analysis
Descriptive Univariate Bivariate Mulitvariate

Analysis Analysis Analysis Analysis
Interpretation of output
Figure 9.1: Steps in Data Analysis
1. Data Editing
Occasionally, the researcher makes mistake when recording certain

responses. Some responses may be contradictory with the other, and must be
verified before data can be coded. Data editing is the process in which raw
data are screened for errors that might occur during sampling or data
collection process. Data editing can be either field editing or in-house editing.
Field editing refers to the editing made by the interviewer on the field during
the interview; while in-house editing refers to the editing process made by the
researcher after the interview.

UniSZA
The purpose of data editing is to increase the accuracy and precision of the
data by checking certain omissions, such as blank response on the
questionnaire, legibility of handwriting, inconsistent responses, etc. The
researcher can check several areas of concern namely:
a) Proper asking of questions

b) Proper recording of responses
c) Proper screening of respondents
d) Proper recording of open-ended questions
a) Proper asking of a question

Asking the question is one aspect of editing process to ensure that the
question is properly asked to the respondent. Sometimes the interviewer
forgets to ask certain questions, and as a result, the corresponding
responses are missing.
b) Proper recording of a response

At this stage, the researcher should check for the consistency of the
responses. There are certain obvious inconsistencies that can be easily
detected. For example; in the first question, a respondent indicated his
occupation is student, but in another question the monthly income was
recorded as RM 20,000. These inconsistent responses must be rectified
and edited.
c) Proper screening of respondents

The researcher should make a proper screening of respondents to ensure
that they are the targeted respondents and eligible to participate in the
interview.
d) Proper recording of open-ended responses

Responses to open-ended questions provide a highly meaningful data to the
research. However, the open-ended questions are difficult to interpret. For
example, one response to the question “Why do you choose KFC restaurant
instead of other fast food restaurant?” The response is “The restaurant
provides good service.” This response is ambiguous since it does not
provide adequate information as to what the respondent means by “good
service.” The meaning of “good service” is too broad. The interviewer should
probe deeper for a clearer response. After probing, the response is “the
employees are friendly and helpful, they serve customers with smile, and
the waiting time is short”. From the responses, the researcher must interpret
UniSZA
the “good service” into a more specific response such as “provide fast,
efficient, and friendly service”.
Editing of a Missing Data
There are a few options which the researcher can take when facing with the
problem of “no response” or “missing data”. Among those options are:
a) Go back to the respondent

The researcher should go back to the respondent if the number of missing
response on that particular respondent is large and the number of sample in
the study is small. This method is the most effective in data editing process
such as to rectify the missing data, verify ambiguous responses, confirm the
inconsistent response, etc. However, the data obtained might be different from
the original data due to differences in the method of approach.
b) Assigning missing values

The researcher may assign values to the missing data if the previous method
is not applicable due to the difficulties in contacting the respondents. However,
the following elements must be identified before performing this approach.
i. The number of respondents involved in missing data is small

ii. The proportion of missing data for each respondent is small.
iii. The missing data are not on the main variables in the study.
The researcher can assign a missing value using a dot (.) in a data entry. The
computer recognizes a dot as no response for that particular item.
c) Excluding missing responses

If the researcher fails to rectify the missing response, he may exclude that
particular respondent from the study. However, this method is recommended
when the following requirements are met:
i. The proportion of respondents with missing data is small (less than

10%).
ii. The sample size taken in the study is large.
iii. The respondent with missing data has similar characteristic with other
respondents in the study (e.g. demographic, gender, social-class etc).

UniSZA
iv. The proportion of missing data for each respondent is large and it is
impossible to edit.
v. The missing data involved is the main variables in the study.
Nevertheless, the decision to exclude the respondents from the study may be
subjective and might lead to bias results. If the researcher has to resort to this
procedure, he should record the procedure adopted to identify these
respondents and their reference. A major problem in editing is subjectivity.
There are many procedures to be adopted by the researcher in doing editing
process. A good researcher should have a systematic procedure for assessing
the questionnaires.
2. Data Coding
After the editing and cleaning process of raw data is completed, these data
have to be coded in order to facilitate data entry. If the data is not coded
systematically, it will cause the lack of clarity of data, which will result later in
serious confusion in the analysis. Data coding refers to the process of
identifying and classifying each response with a numerical score or symbol. In
other words, the researcher needs to assign a code (usually number) to each
possible response in a question. Generally, codes are numbers or symbols.
However, codes are broadly defined as rules for interpreting, classifying, and
recording the data. In this process, the researcher must specify the column
position, number of variable, name of variable and coding instructions.
Coding will facilitate the process of transcribing the data from the
questionnaire into the computer. In the coding process, the researcher will
look through every question and each response to prevent confusions
especially when there are many questions in a questionnaire and a large
number of respondents in the study. The codes are numerical because
numbers are easy to input into a data file. For example, using a scale from 1 to
10, how would you rate the following statement concerning your university?

UniSZA
Your statement... Strongly Strongly

Disagree Agree
1. I will choose this university again for my post-graduate studies 1 2 3 4 5 6 7 8 9 10
2. I will study at this university even if the cost has increased 1 2 3 4 5 6 7 8 9 10
3. I am willing to pay more due to benefits obtained at this university 1 2 3 4 5 6 7 8 9 10
4. I will participate in programs organized by this university 1 2 3 4 5 6 7 8 9 10
5. I will tell positive things about this university to others 1 2 3 4 5 6 7 8 9 10
6. I will recommend this university as a suitable place to study 1 2 3 4 5 6 7 8 9 10
7. I will encourage my friends and families to study at this university 1 2 3 4 5 6 7 8 9 10
8. I will propose to the firms for industrial linkages with this university 1 2 3 4 5 6 7 8 9 10
9. I will recommend this university to employers for recruitments 1 2 3 4 5 6 7 8 9 10
In the above example, the code of response is readily available for each item.
Another example of a coded response:
1) Which of the following restaurant you have visited most frequently?
KFC (1) Pizza Hut (4)

McDonald (2) Burger King (5)
A&W (3)
2) What is the main reason you prefer this fast food restaurant?
Tastes (1)
Cleanliness (2)
Promotion of the restaurant is interesting (3)
Others, please specify_________________________ (4)
For an open-ended question, it is difficult to code because the responses may vary
largely among respondents. The researcher should develop codes for the
anticipated responses. It begins by generating a list of as many potential responses
as possible. These responses can be given values within a range determined by the
actual number of separate responses identified.

UniSZA
THE PROCESS TO CODE A QUESTION
As emphasized earlier, the researcher should assign numbers to each response in

the questionnaire. Example 9.1 provides the questionnaire in the research study
done by Zainudin et al. (2004). Let us see how the code is assigned for each
response.
Coding the Structured or Close-Ended Question
Steps in Coding Process

First step:
Before you begin the coding process, you should review the completed
questionnaire. Look for the type of question to code, the order of the question, the
scale used in the question and the number of responses.
Example 9.1
The example extracted from the research study done by Zainudin (2004).
Demographic Information:
1. My Gender :
Male …………………………………….. 1
Female …………………………………….. 2
2. My age:
Below 25 years ………………………………...….. 1
25 – 35 years …………………………………….. 2
36 – 45 years …………………………………….. 3
46 – 55 years …………………………………….. 4
Above 56 years ……………………………………... 5
3. Image Scale
To what extent would you agree or disagree with the following statements
concerning the services provided by this hospital? Circle 1 if you strongly
disagree, and circle 7 if you strongly agree with the statement. You can also
circle anywhere in between depending how strongly you agree or disagree with
the given statement.

UniSZA
Customer Contacts Strongly Strongly

Disagree Agree
1. The service delivered by the employees at this hospital is
1 2 3 4 5 6 7
within my expectation.
2. The employees at this hospital treat their patients with
1 2 3 4 5 6 7
respect.
4. What type of medical services do you get in this hospital?

(You may tick more than one type of service)
Out-patient services
In-patient services
Cancer treatment services
Heart diseases treatment services
Pre/post natal services
Other expertise services

UniSZA
Second step:
After you have reviewed all the factors involved, you must sort out the factors such
as the name of variable, number of question, variable label and value of responses
which are important in coding process. Below are some illustration on how to sort
out the important factors in the question 3 and 4.
Variable Name
Customer Contacts Strongly Strongly
Disagree Agree
3. The service delivered by staffs/doctors is within my 1 2 3 4 5 6 7
expectation.
4. The doctors/staffs of this hospital treat their patients 1 2 3 4 5 6 7
respectfully.
No. of Question Variable Label Value of Responses
Figure 9.2: Sorting Information for Coding Process
Third step:
Next, write down all the factors you have sorted out in a sheet of paper according
to their category. Table 9.1 below shows the examples of how to write all the
factors according to their codes. The column number, types of data and scale
measures are shown in the columns 2, 4 and 7. Refer to the spreadsheet resulted
from the “variable view” in the SPSS 16.0 software.

UniSZA
Table 9.1: The Spreadsheet of Coding Process
Question Column Variable Types Variable Values / Scale

Number Number Name of Data Label Responses Measures
Gender of
Male = 1
1 1 Gender Numeric Respondents Nominal
Female = 2
25 years & below =1
26 – 35 years =2
Age of 36 – 45 years =3
2 2 Age Numeric Interval
Respondents 46 – 55 years =4
Above 56 years =5
Service Strongly Disagree = 1

3 3 Cuscon1 Numeric within Strongly Agree =7 Interval
Expectation
Respectfully Strongly Disagree = 1
4 4 Cuscon2 Numeric Treated of Strongly Agree =7 Interval
Patients
Final step:
After you have completed furnishing the table above, it is possible to key in data
directly from the questionnaire. However, you must properly check the coding
table to ensure the accuracy of the coding especially on the variable name,
variable label, and value of responses before data entry can begin.
Coding the Unstructured or Open-Ended Question
The coding of unstructured or open-ended questions is more subjective and

complicated in which the researcher needs to assign codes on the verbatim
responses of the respondents. The purpose of coding the open-ended questions is
to reduce the large number of individual responses to a few general categories
using the numerical codes. The researcher would place similar responses into a
general category and assign the same code to represent similar category.

UniSZA
Example 9.2
The following are a few responses obtained using the open-ended questions about
frozen food.
o We do not like frozen food very often.

o I like to prepare fresh food.
o Frozen foods are not as tasty as fresh foods.
o I do not like that freezer taste.
All of these answers can be categorized under the factor of “dislike frozen
foods” and assign the code 1. The construction of category reflects the judgment
of the researcher. A major objective in the coding process is to accurately transfer
the meanings from written responses to numeric codes. Therefore, some
researchers develop and assign codes based on their experience or theoretical
consideration prior to the beginning of the fieldwork. The following are several
guidelines that the researcher might use for coding open-ended questions.
1. Differentiate Coding
Differentiating the response categories for open-ended questions is more
difficult than fixed alternative questions. The category of codes must be
mutually exclusive. Generally, the researcher performs a test tabulation in
order to identify the verbatim responses and then establish the coding
categories that reflect the judgment of the respondents. Test tabulation is a
test on a small sample of respondent that respond to a particular question to
identify the stability and distribution of responses that will determine the coding
scheme.
2. Confirm Coding
In this stage, the researcher should assign categories to use in answering the
research question although there is no respondents mention about them. For
example, the management of UiTM is concerned about the preferred menu of
food in a research study of food services at dining hall. However, there is no
respondent mentioned about the preferred menu on the open-ended question
concerning food services at dining hall. The researcher should assign this
critical issue in order to know that no one has mentioned a particular
response.

UniSZA
3. Revising Coding
After test tabulation has been conducted, the researcher might have some
useful categories in mind for answering research questions. However, the
difficulties will arise if the preliminary tabulation contained too many codes, or
the codes are too abstract, or there are only a few categories established. The
codes should be revised and reduced into smaller categories to answer the
research question.
In revising the codes, the data should be coded as detail as possible. For
example, in a study on counter services in UiTM Library, the term
“Responsible Staff” might consist of friendly staff, polite staff, helpful staff,
courteous staff, and cute staff. However, the detailed code provides too many
categories. A good researcher would generalize the heading pertinent to the
research questions.
THE CODEBOOK
In coding process, it is important for the researcher to make a codebook that

contains coding instructions and necessary information about variables in the data
pool. A codebook will help the researcher to identify and locate the variables since
it contains the following information:
1. Column number
2. Record number
3. Variable number
4. Variable name
5. Question number
6. Instructions for coding
3. Data Entry
Data entry includes tasks involved with the direct input of the coded data into
some specified software package that ultimately allows the research to
manipulate and transform the raw data into useful information. Normally, the
software packages employed are SPSS package and Microsoft Excel. For
example, the SPSS Data Editor comprises of spreadsheet in which you can
enter, edit, and view the contents of the data file. Each row of the editor
represents a case, and each column represents a variable.

UniSZA
One critical task of data entry personnel is to ensure that the data entered are
correct and mistakes free. How to detect the errors in data entry? Error
detection normally occurs after the data have been entered into computer
storage. Entry errors can be detected with error edit routines built or developed
into the data entry software. An additional approach is to visually scan the actual
data after they have been entered. A frequency data table is one of the
approaches for visually scanning entered data. All missing values should appear
with a period (dot) in the cell. It is possible to add, change, or delete values
easily after the data have been entered. Once the missing values and the
recodes are taken care of, the data are ready for analysis.
Exercises

Assigning missing values is desirable if the number of
1.
respondents with unsatisfactory responses is large.
In coding a survey instrument, category codes should be
2. assigned for critical issues even if no one has mentioned
them.
Coding of structured questions is more complex than
3.
coding of unstructured questions.
Computer packages can be programmed to identify out-of-
4.
range values for each variable.
Dummy variables are typically used to re-specify category
5.
variables.
Discarding unsatisfactory responses is desirable when the
6.
proportion of unsatisfactory responses is large.
7. A codebook generally contains the variable names.
A codebook should contain necessary information about
8.
variables in the data set, but not coding instructions.
Data entry includes the tasks of editing and coding raw
9.
data into computer.
A data table is one approach for visually scanned entered
10.
data.

UniSZA
1. The initial step in data preparation process is always to:
a. Edit the data.

b. Code the data.
c. Prepare a preliminary plan of data analysis.
d. Transcribe the data.
2. Assigning numbers to represent responses in the questionnaire is called:
a. Data editing
b. Sample selection
c. Data coding
d. Recording the response
3. Consistency checks should be used to identify data that:
a. Are out of range

b. Are logically inconsistent
c. Have extreme values
d. Show any of the above characteristics.
4. The action that should be taken in editing missing data is:
a. Back to the respondents

b. Assigning missing values
c. Excluding the respondents with missing data
d. All of the above
5. Assigning missing values is feasible when:
a. The number of respondents with unsatisfactory response is large.

b. The unsatisfactory responses for each respondent is small.
c. Responses on key variables are missing
d. The sample size is small.

UniSZA
6. A _________ is a number for each possible response to each question recorded

as a single item of data represented by a ___________.
a. Record; code
b. Code; field
c. Field; record
d. Code; recode
7. The coding of unstructured or open-ended questions should include:
a. Verbatim response
b. A summary of responses
c. Validation codes
d. Single column codes
8. A __________ contains coding instructions and the necessary information about

all variables in the data sets.
a. Project code
b. Codebook
c. Fixed field code
d. Validation code
9. A codebook generally contains:
a. The variable name

b. Instructions for coding
c. Interviewer’s name
d. Both a and b only
10. ___________ includes the tasks of direct input of the coded data into computer.
a. Data coding
b. Data editing
c. Data entry
d. Data transformation

UniSZA
Essay Question
1. Briefly explain the data management process.
2. Explain the importance of data editing in a research study.
3. What is missing data, and how does it affect the research study?
4. Discuss three ways in which missing data might be handled.
5. Explain the steps involved in coding structured or close-ended responses.
6. Discuss the process of coding unstructured or open-ended responses.
7. Discuss the importance of codebook and state the information contained in the
codebook.
8. You are hired as a project supervisor at a research consulting firm in Kuala

Lumpur. You are to supervise the data collection and preparation process for a
large survey conducted for a leading manufacturer of paper towels.
9. The data were collected via in-home personal interviews and 1200
questionnaires were returned from the field. During the process of editing and
cleaning the data, you found that 39 questionnaires have some missing
responses.
a) What activities are involved in the preliminary checking of questionnaires

that have been filled and returned from the field?
b) How should the missing responses be handled?
c) Describe the various processes of editing and coding that should be done
prior to data analysis.

UniSZA
Chapter 10: Data Analysis
CHAPTER DATA ANALYSIS

10
STEPS IN DATA ANALYSIS PROCESS
After data entry has been completed, the researcher needs to carry out data
analysis in order to achieve his research objectives, address his research
questions, and to test his research hypotheses. First of all, he has to perform
data mining for all measuring items in the study. Secondly, he has to perform
the descriptive analysis for important demographic variables to verify that
respondents are represented across their demographic characteristics. Thirdly,
he has to run the required analysis to address his research questions. Finally,
he has to perform the statistical analysis in order to test the stated hypotheses
concerning the study.
Step 1: Data Mining
This step is necessary in order to screen the data from “outliers”. This is an
important procedure to ensure that the data collected is normally distributed. The
proof of normality for the data is extremely necessary in order to meet the
requirement for employing the parametric statistical test. Otherwise the researcher
needs to employ the non-parametric statistical procedures. A common “mistake”
committed by novice researchers is running the parametric statistical analysis
without proving that the data at hand meet the required assumption for the
respective analysis to be valid. Caution should be adhered that certain statistical
analysis is not valid if the data collected do not follow normal distribution. In this
case, the proof of normality for data at hand is required before a researcher could
proceed further.
Statistically speaking, there are three assumptions that must be satisfied by

the data if the study wants to employ any parametric test. The three assumptions
are (i) normal distribution, (ii) equal sample size, and (iii) equality of variance.
However, there are alternative parametric tests available if the assumption of
equal sample size (assumption ii) and the equality of variance (assumption iii) are
not satisfied. In ascertaining that the data meet the first assumption, normally a
researcher would examine the measure of skewness for each variable in a dataset
to determine its distribution.
UniSZA
Example 10.1
We will carry out data mining process using the data obtained by Zainudin (2007)
in his study on University Education. In this example, we select one variable
namely the corporate image of a university for further analysis. In his study,
Zainudin (2007) employed 16 items to measure the corporate image of a
university.
Table 10.1 illustrates the descriptive scores for each of the 16 items based on the
responses from 872 respondents. The table also presents the standard error,
standard deviation, and the measure of skewness for every item in the study.
From Table 10.1, the measure of skewness for all items fall between -0.971 and -
0.316. The general statistical measure of skewness ranges from -3.0 and 3.0.
The measure of skewness close to -3.0 indicates the data is distributed extremely
skewed to the left, while the skewness close to 3.0 indicates the data is distributed
extremely skewed to the right. If the data is perfectly normally distributed, the
measure of skewness is 0.0. However, the measure between -1.0 and 1.0 is
considered normally distributed and acceptable to proceed with the parametric
analysis procedure.
Otherwise, if the data is not normally distributed, the researcher has to employ
non-parametric procedure for his statistical analysis. Since the measures of
skewness for all items are within the range between -1.0 and 1.0, the researcher
can conclude that the distribution of data does not depart from normality. Hence,
the required assumption for employing the parametric statistical procedures for
further analysis is satisfied.
cont...

UniSZA
Table 10.1: Descriptive Statistics for all 16 Corporate Image Items

No of Minimum Maximum Std. Std
Item Sample Score Score Mean Error Deviation Skewness
1 872 5 10 8.55 .040 1.185 -0.481
2 872 5 10 8.39 .044 1.301 -0.475
3 872 5 10 8.14 .046 1.363 -0.316
4 872 3 10 7.98 .056 1.653 -0.551
5 872 3 10 8.61 .044 1.280 -0.921
6 872 2 10 8.27 .046 1.361 -0.971
7 872 1 10 7.82 .053 1.577 -0.592
8 872 2 10 8.04 .051 1.505 -0.767
9 872 3 10 7.93 .055 1.637 -0.404
10 872 1 10 7.37 .062 1.826 -0.408
11 872 1 10 7.58 .058 1.713 -0.508
12 872 1 10 7.88 .056 1.650 -0.709
13 872 1 10 8.23 .051 1.503 -0.976
14 872 3 10 7.94 .055 1.625 -0.579
15 872 4 10 8.69 .040 1.189 -0.947
16 872 2 10 8.14 .050 1.486 -0.838

UniSZA
Step 2: Descriptive Analysis for Demographic Variables
In the descriptive analysis, the researcher wants to assess the distribution of

data across some demographic characteristics of respondents. Here, the
researcher needs to show that the respondents obtained for his study represent all
demographic characteristics of the population.
Example 10.2
Using the same dataset, the researcher analyzes the demographic variables
namely the respondents’ gender, age group, and their duration at the university.
Tables 10.2 to 10.8 show how the respondents are distributed according to the
selected demographic characteristics.
Table 10.2: Respondents’ Gender
Valid Cumulative
Gender of respondents Frequency Percent Percent Percent
Male 404 46.3 46.3 46.3
Valid Female 468 53.7 53.7 100.0
Total 872 100.0 100.0
The above table shows that the respondents are represented almost equally in
both gender categories namely male and female.
Table 10.3: Respondents’ Age Group
Valid Cumulative
Age of respondents Frequency Percent Percent Percent
20 - 22 Years 595 68.2 68.2 68.2
23 - 25 Years 277 31.8 31.8 100.0
Total 872 100.0 100.0
The above table shows that the respondents sampled only fall into two age
categories. Meaning, in this study, the respondents aged below 20 years and
above 25 years are not covered. Hence, the generalization from this study is only
valid for the students’ population aged between 20 to 25 years.
cont…
UniSZA
Table 10.4: Respondents’ Duration in the Program

Valid Cumulative
Number of semesters in campus Frequency Percent Percent Percent
2 and less than 4 267 30.6 30.6 30.6
4 and less than 5 281 32.2 32.2 62.8
5 and above 324 37.2 37.2 100.0
Total 872 100.0 100.0
The above table shows that the respondents are taken almost equally across three
durations of study at the university. However, the students less than two
semesters in campus are not included. Hence, the generalization is not applicable
for students who are less than two semesters being in the university.
Table 10.5: Monthly Family Income of Respondents
Cumulative
Parent’s income (RM) Frequency Percent Percent
Less than RM 1000 261 29.9 29.9
RM 1001 - RM 3000 222 25.5 55.4
RM 3001 - RM 5000 177 20.3 75.7
RM 5001 - RM 7000 109 12.5 88.2
More than RM 7000 103 11.8 100.0
Total 872 100.0
The above table shows that the respondents are taken across all income levels
which reflect the socioeconomic status of their parents. Meaning the generalization
from this study is valid for all income categories of students’ population.
cont…

UniSZA
Table 10.6: Parent’s Highest Education

Cumulative
Parent’s highest education Frequency Percent Percent
SPM/STPM 153 17.5 17.5
Diploma 179 20.5 38.0
Degree 213 24.4 62.4
Master/PhD 186 21.3 83.7
Other 141 16.2 100.0
Total 872 100.0
The above table shows that the respondents are taken across all education levels
which reflect the socioeconomic status of their parents. Again, the generalization
from this study is valid for all education levels of students’ population.
Table 10.7: Parent’s Occupation
Parent’s occupation Frequency Percent Valid Percent

Professional 73 8.4 8.4
Government officer 106 12.2 12.2
Teacher 119 13.6 13.7
Clerical 127 14.6 14.6
Self employed 200 22.9 23.0
Pensioner 146 16.7 16.8
Others 101 11.5 11.5
Total 872 100.0 100.0
The above table shows that the respondents are taken across all five education
levels which would also reflect the socioeconomic status of their parents. Since
respondents from all occupation levels are sampled, this study is generalizable to
respondents from all levels of occupation.
cont…

UniSZA
Table 10.8: Respondents’ “choice of university” in the Application for

Admission
Was UiTM your first choice? Frequency Percent Valid Percent

Yes 527 60.4 60.4
No 345 39.6 39.6
Total 872 100.0 100.0
The above table shows that the respondents from both choice categories namely
“first choice” and “not first choice” are taken in the study. In this study, 60.4% of
the respondents stated UiTM was their first choice university while another 39.8%
of them did not state UiTM as their first choice university.
RUNNING DATA REDUCTION PROCEDURE: THE EXPLORATORY

FACTOR ANALYSIS
When we use too many items to measure a variable, we have to reduce them
into a manageable number before further analysis can be carried out. One of
the methods normally employed for data reduction is factor analysis. In the
factor analysis procedure, items which posses similar characteristic will be
grouped together under one component. So, instead of having to deal with 16
items, the researcher only deals with a few components resulted from the
procedure.
Example 10.3
Again we will use the data from Zainudin (2007) to illustrate the factor analysis
process. The data consists of 16 items to measure one construct namely the
corporate image. Here we want to reduce the number of items (if possible) by
grouping them together into a few components using factor analysis procedure.
The Principal Component Factor Analysis (PCA) with Varimax Rotation was
performed for the 16 items of the corporate image construct. The results (Table
10.9) indicate that the Bartlett’s Test of Sphericity was significant (Chi-square =
8498.069, p-value < 0.000). The measure of sampling adequacy by Kaiser-Meyer-
Olkin (KMO) is 0.932, and it has improved above 0.6.
Cont….
UniSZA
[For information: the KMO value close to 1.0 and the significance value of
Bartlett’s Test close to 0.0 indicate the data at hand is adequate to proceed into
Factor Analysis].
Table 10.9: KMO and Bartlett's Test for Corporate Image Items
Kaiser-Meyer-Olkin Measure of
Sampling Adequacy. 0.932
Approx. Chi-Square 8498.069

Bartlett's Test of Sphericity Df 120
Significance 0.000
This KMO value of 0.932 (Table 10.9) is excellent since it exceeds the
recommended value of 0.6 (Kaiser, 1974). The two measures (KMO value close to
1.0 and the Bartlett’s test significance value close to 0.0) suggest that the data is
appropriate to proceed with its reduction procedure. The next step for the
researcher is running Factor Analysis procedure in SPSS. Table 10.10 is the
output of a Factor Analysis for the 16 corporate image items.
cont…

UniSZA
Table 10.10: Total Variance Explained for Corporate Image Items
Component Initial Eigenvalues Rotation Sums of Squared Loadings

% of Cumulative % of Cumulative
Total Variance % Total Variance %
1 7.936 49.602 49.602 4.457 27.855 27.855
2 1.313 8.207 57.809 3.480 21.750 49.605
3 1.039 6.495 64.304 2.352 14.699 64.304
4 .930 5.812 70.116
5 .720 4.498 74.613
6 .561 3.506 78.120
7 .528 3.302 81.422
8 .484 3.028 84.450
9 .421 2.634 87.084
10 .406 2.535 89.619
11 .372 2.326 91.944
12 .326 2.036 93.981
13 .305 1.905 95.886
14 .249 1.558 97.444
15 .238 1.486 98.930
16 .171 1.070 100.00
The output in Table 10.10 shows that Factor Analysis procedure has extracted
three distinct dimensions with eigenvalues exceeding 1.0. In fact the researcher
could use up to four components since the eigenvalue for the fourth component
(0.93) is also close to 1.0. Remember, the more component extracted is better as
long as their eigenvalues are reasonably close to 1.0.
The Rotated Component Matrix for Corporate Image Items
We begin our discussion on the rotated component matrix by referring to

Table 10.11. As stated earlier, the Factor Analysis procedure has extracted three
distinct components based on the eigenvalue greater than 1.0. Table 10.11
presents the factor loadings for all 16 items under each of three components. The
task now is to determine the items for the respective components.

UniSZA
According to Igbaria et al. (1995), the item which loads greater than 0.6
under one component (say component 1) and loads lower than 0.35 under the
other components should be grouped into component 1. However, for the item
which has a loading of higher than 0.6 in one component and also loading higher
than 0.35 in other components is considered to have a positive cross-loading and
should be dropped from further analysis. However, the item which has a factor
loading lower than 0.6 under all components should also be dropped from the
analysis.
Example 10.4
Table 10.11 shows the factor loading for all 16 items under three components
namely component 1, component 2, and component 3. The study will select the
item using the above criteria to be grouped under the respective component. The
factor loading in bold indicates under which component the respective item falls.
In this case, item 1, 2, 3, 15, and 16 fall under component 1; item 4, 7, 8, 9, 10, 11,
and 12 falls under component 2; and item 5, 6, and 13 fall under component 3.
Meanwhile, item number 14 is excluded from the analysis since its loading is lower
than 0.6 for all components.
Once we have identified the items and their respective component, we can
transform the measurement of these items into one mean value representing their
respective component. In this case, component 1 is the mean of item 1, 2, 3, 15,
and 16; component 2 is the mean of item 4, 7, 8, 9, 10, 11, and 12; and
component 3 is the mean of item 5, 6, and 13.
cont…

UniSZA
Table 10.11: The Rotated Component Matrix for Corporate Image Items
Corporate Image Items Corporate Image Component
Component 1 Component 2 Component 3
1 University has good quality academic staff 0.304 0.737 -0.026
2 University has a tradition of excellence 0.309 0.682 0.106
3 Faculty has a good academic reputation 0.270 0.623 0.290
4 University has a strong industrial linkages 0.627 0.321 -0.086
5 University practices an ISO 9001:2000 0.129 0.041 0.728
6 University contributes well to the society 0.227 0.162 0.714
7 University has good cooperation with its alumni 0.702 0.278 0.284
8 University has good cooperation with other universities 0.685 0.275 0.349
9 University has excellent researches 0.640 0.319 0.234
10 University has excellent publications 0.701 0.321 0.228
11 University wins awards in research and inventions 0.610 0.252 0.310
12 University organizes important seminars 0.796 0.099 0.170
13 University cooperates well with the government 0.338 0.195 0.631
14 University offers good variety of courses 0.340 0.371 0.425
15 University offers a flexible learning system 0.244 0.692 0.307
16 University has a quality management system 0.160 0.734 0.276
In fact, to make it clearly observable the item and its respective component, we
can instruct SPSS not to print the factor loading lower than 0.6 for all items. In this
case, the computer would printout only the factor loading higher than 0.6 for all
components as shown in Table 10.11a.
cont…

UniSZA
Table 10.11a: The Rotated Component Matrix for Corporate Image Items
Corporate Image Items Corporate Image Component

Component 1 Component 2 Component 3
1 University has good quality of academic staff
0.737
2 University has a tradition of academic excellence
0.682
3 Faculty has a good academic reputation
0.623
4 University has a strong industrial linkages
0.627
5 University’s management has an ISO 9001:2000
0.728
6 University contributes well to the society
0.714
7 University has good cooperation with its alumni
0.702
8 University has good cooperation with other universities
0.685
9 University has excellent researches
0.640
10 University has excellent publications
0.701
11 University wins awards in research and inventions
0.610
12 University organizes important seminars
0.796
13 University cooperates well with the government
0.631
14 University offers good variety of courses
15 University offers a flexible academic learning system

0.692
16 University has a good academic management system
0.734
The researcher could use suitable term to rename each component to reflect the
common items under that particular component. In this example, the items which
fall under component 1 are mostly related to the reputation of the university itself,
hence we could rename component 1 as “University Image”. The items which fall
under component 2 are mostly related to the academic excellence of the
university, hence we could rename component 2 as “Academic Image”. Finally, the
items which falls under component 3 are mostly related to the perception of
external organizations towards the university, hence the most appropriate name
for this component would be “External Recognition”.
Cont…
UniSZA
However, if we do not want to rename the components for any reason, we can
maintain their variable names as component 1, component 2, and component 3 in
further analysis. Remember, instead of dealing with 16 items, the data has been
reduced to only three components for further analysis.
THE RELIABILITY ANALYSIS FOR THE MEASURING ITEMS
The next step is to determine the reliability measure for the measuring items under
each component. The popular measure for reliability is the Cronbach’s alpha. The
Cronbach’s alpha of 0.6 or higher for a component reflects the measuring items
under that particular component provide a reliable measure of internal
consistency.
Example 10.5
We compute the Cronbach’s alpha for each component in the above example. As
shown in Table 10.12, the values of Cronbach’s alpha are 0.909, 0.857, and 0.688
respectively for component 1, component 2, and component 3. It is important to
note here that all reliability measures have exceeded the minimum value of 0.6 as
recommended by Nunally (1978).
Table 10.12: Reliability Statistics for Corporate Image Items
Corporate image Number of items Cronbach’s Cronbach’s Alpha

components in a component Alpha based on
standardized items
Component 1 7 0.909 0.909
Component 2 5 0.855 0.857
Component 3 3 0.685 0.688
As shown in Table 10.12, there are seven items under component 1, five items
under component 2, and three items under component 3. The next step is to
transform the individual score of these seven items into its individual mean
representing component 1 for further analysis. The same procedure goes for other
components namely component 2 and 3 respectively. In the end, we only have
three values measuring the corporate image construct instead of 16 items.

UniSZA
THE ANALYSIS TO DETERMINE THE NORMALITY OF THE DATA
After obtaining the mean scores for each component, the researcher needs to
show that the distribution of these data also do not depart from normality. As
stated earlier, one of the measures which reflect the distribution of data is
skewness. The measure of skewness between -1.0 to 1.0 indicates the data does
not depart from normality. Hence, the parametric statistical analysis can be
employed.
[Please note that another method to confirm the normality of data is the Kolmogorov-
Smirnov test statistics whereby Ho: data is normally distributed vs HA: data is not normally
distributed. We are looking for p-value higher than 0.05 so that the procedure would fail to
reject the null hypothesis to confirm that the data is normally distributed.]
Table 10.13: The Measure of Skewness of the Data Produced by AMOS

Procedure
Variables the Model min max skewness c.r. kurtosis c.r.

Component 1
4.71 10.00 -0.24 -2.91 -0.60 -3.63
(University Image)
Component 2
5.80 10.00 -0.19 -2.33 -0.64 -3.83
(Academic Image)
Component 3
5.00 10.00 -0.57 -6.86 0.57 3.41
(External Recognition)
Since all measures for the skewness are closer to 0.0 and within the range
between -1.0 to 1.0 as shown in Table 10.13, the study concludes that the
distribution of data is almost symmetry or bell-shaped. The bell-shaped distribution
indicates the data is normally distributed. Hence, the data obtained in the study
meets the required assumption for employing the parametric statistical analysis
that data comes from a normal distribution. Among the parametric statistical
analyses that can be employed are t-test, Z-test, and F-test depending on the
hypothesis test proposed for the study.

UniSZA
PARAMETRIC STATISTICAL ANALYSIS
Before proceeding further with the analysis, the researcher needs to go back to his
research objectives in order to achieve the stated objectives, to address the stated
research questions, and also to test the stated hypotheses.
Example 10.6
Again we illustrate the analysis by using the data by Zainudin (2007). The
researcher has the following objective of the study, its related research
question, and the corresponding hypotheses.
The objective of the study

To examine whether the difference in respondents’ demographic characteristics
relates to their perception towards the services provided by the university. The
variables of interest are respondent’s gender, age, parent’s income, education,
occupation, respondent’s duration in the university, and their choice of the
university when they applied for admission through the centralized university
system.
The related research question

Does the difference in respondent’s demographic (gender, age, parent’s income,
education, occupation, respondent’s duration in the university, and their choice of
the university when they applied for admission through the centralized university
system) also differ in their perceived service quality provided by a university?
Cont…

UniSZA
The corresponding hypotheses

From the stated objective and research question, the researcher should be able to
propose a few hypotheses as a direction for the study. Remember, hypothesis is
the researcher’s educated guess concerning the research question at hand.
Among the hypotheses that can be proposed for this study are:
HA1: There is a significant difference in respondents’ perceived service quality

when they are classified according to their gender.

when they are classified according to their age group.

when they are classified according to their parents’ income.

when they are classified according to their parents’ occupation.

when they are classified according to their parents’ education.

when they are classified according to their duration in the program.
Testing the Hypothesis for Mean Differences between two Groups: The
independent two-sample t-test
HA2: There is a significant difference in perceived service quality of the university

when respondents are classified according to their age group.

UniSZA
The Required Measures For Testing Hypothesis 2
Age of respondents N Mean Std. Deviation Std. Error Mean

20 - 22 years 595 8.15 1.01 0.04
23 - 24 years 277 8.12 1.16 0.07
Total 872
The T-Test For Equality Of Means Between Two Age Groups
95% Confidence
Significance Mean Std. Error
t df Interval of the
Value (p) Difference Difference
Difference
Lower Upper
0.35 870 0.73 0.03 0.08 -0.12 0.18
Since the P-value is higher than 0.05, the study fails to reject the null
hypothesis. Hence, the above research hypothesis is not supported.
Testing Hypothesis for Mean Differences for More than two Groups: The
one-way Analysis of Variance (ANOVA)

when respondents are classified according to their parents’ income.
The ANOVA table (F-test) for equality of means among 5 income levels
Mean Significance
Parents’ Income Sum of Squares df Square F Value (p)
Between Groups 9.02 4 2.26 2.04 0.08
Within Groups 960.46 867 1.11
Total 969.48 871

UniSZA

when respondents are classified according to their parents’ occupation.
The ANOVA table (F-test) for equality of means among 7 occupation

levels
Parents’ Sum of Mean Significance

Occupation Squares df Square F Value (p)
Between 8.24 6 1.65 1.95 0.11
Groups
Total 739.34 871
HA5: There is a significant difference in perceived service quality of the

university when respondents are classified according to their parents’
education.
The ANOVA Table (F-Test) For Equality Of Means Among Five Education
Levels
Parents’ Sum of Mean Significance

Education Squares df Square F Value (p)
Between 9.75 4 2.44 2.11 0.08
Groups
Total 969.48 871

UniSZA

when respondents are classified according to their duration in the program.
The ANOVA table (F-test) for equality of means among three duration
levels
Duration in the Sum of Mean Significance
program Squares df Square F Value (p)
Between 4.94 2 2.47 3.38 0.03
Groups
Total 639.40 871
Since the P-value is lower than 0.05, the study manages to reject the
null hypothesis. Hence the above research hypothesis is supported.
Once the research hypothesis is supported, the study proceeds to do the
post-hoc analysis using the Least Significance Difference method (LSD).
The multiple comparison procedure is employed to determine which pair
has a significant difference in the mean score.
The Least Significant Difference Method (LSD) among three durations
Mean
(I) Duration in the (J) Duration in the Difference Std. Significance
program program (I-J) Error Value (p)
between 4 to 5
0.10 0.07 0.16
between 2 to 3 semesters
semesters more than 5
-0.18 0.07 0.01
semesters
between 2 to 3
-0.10 0.07 0.16
between 4 to 5 semesters
semesters more than 5
0.08 0.07 0.25
semesters
between 2 to 3
0.18 0.07 0.01
more than 5 semesters
semesters between 4 to 5
-0.08 0.07 0.25
semesters
UniSZA
The study found a significant difference in the perceived service quality

between the juniors (less than 3 semesters) and seniors (more than 5
semesters). The mean for the seniors is higher than the mean for the
juniors.
TESTING THE INTER-RELATIONSHIP AMONG VARIABLES IN THE STUDY:

THE STRUCTURAL EQUATION MODELING (SEM)
Sometime the researcher develops certain theoretical framework for his study.
The theoretical framework, represented by a schematic diagram, indicates how he
theorizes the existence of relationship among the constructs involved in the study.
In the end he needs to determine whether the proposed relationship is significant
or not based on the data gathered.
Example 10.7
Zainudin (2007) also has the following theoretical framework to prove in his
study. The framework is developed based on what the researcher theorized
regarding the relationship among the constructs in his study.
Service
Quality
Students’
Satisfaction Students’
Loyalty
Corporate
Image
Figure 10.1: The Schematic Diagram Showing the Theoretical Framework

of the Study
UniSZA
The study theorized that the two constructs namely service quality provided in a
university and the corporate image of a university have significant influence on
students’ satisfaction with the university. The students’ satisfaction, in turn, has a
significant influence on their loyalty towards the university in the future. In other
words, these two constructs (service quality and corporate image) have indirect
influence on customer loyalty through customer satisfaction.
In this study, there are two independent constructs namely service quality and
corporate image. The study also has an intervening (mediating) construct namely
students’ satisfaction, and the intended dependent construct namely students’
loyalty. The specific objectives for this study are as follows:
The objectives of the study

1 To examine the relationship between service quality provided by a university
and students’ satisfaction with their learning experience in a university
2 To examine the relationship between corporate image of a university and

students’ satisfaction with their learning experience in a university system
3 To examine the relationship between service quality provided by a university

and students’ loyalty towards their university
4 To examine the relationship between corporate image of a university and

5 To examine the relationship between students’ satisfaction with their

learning experience in a university and their loyalty towards the university
The related research questions for the study (an interrogative statement)
1 Does the service quality provided by the university influence students’
perceived satisfaction with their learning experience?
2 Does the corporate image of a university influence students’ perceived

satisfaction with their learning experience?
3 Does the service quality provided by the university influence students’ loyalty
intention towards their university?
4 Does the corporate image of a university influence students’ loyalty

intention towards their university?

UniSZA
5 Does the students’ perceived satisfaction with their learning experience

influence their loyalty intention towards the university?
The hypotheses to be proven by the study (a declarative statement)

HA1: Service quality has a significant influence on students’ satisfaction
HA2: Corporate image has a significant influence on students’ satisfaction
HA3: Service quality has a significant influence on students’ loyalty
HA4: Corporate image has a significant influence on students’ loyalty
HA5: Students’ satisfaction has a significant influence on students’ loyalty
Service
Quality
HA3
HA1
Students’
Loyalty
HA2
Corporate
Image HA4 HA5
Figure 10.2: Schematic diagram showing the theoretical framework and the
hypothesized relationship among the constructs in the study
Cont…

UniSZA
The researcher’s hypothesis concerning the direct influence of one construct

on the other construct is shown in the above diagram. In the analysis part, the
researcher needs to run certain method of statistical analysis in order to
determine whether the proposed hypothesis is supported or not based on the
data at hand.
The statistical analysis for this type of relationship in social sciences is quite
complicated since the constructs involved are only hypothetical. These
constructs are latent (unobserved variables) and they are measured indirectly
using certain number of items. For example, the corporate image construct
alone is measured using 16 interval-scaled items as shown in Table 10.11.
Using the Factor Analysis procedure, those 16 items are grouped together into
three components. The same goes for other constructs (Table 10.14).
Table 10.14: The number of measuring items and its component for each construct
Number of Number of components
Name of construct measuring items after Factor Analysis
Corporate Image 16 3
Service Quality 34 3
Students’ Satisfaction 14 2
Students’ Loyalty 9 2
For information, one does not have to carry out the data reduction procedure
(Factor Analysis) if the number of items measuring the construct is less than
five. This is because the number of items is small and manageable.

UniSZA
CONVERTING A THEORETICAL FRAMEWORK INTO AMOS SYNTAX
The most popular method of analysis for this type of relationship is Structural
Equation Modelling (SEM), and the most convenient software for SEM is
AMOS Graphic. (AMOS stands for Analysis of Moments Structure). In AMOS
Graphic, the researcher draws the latent (unobserved) constructs followed by
the manifesting (observed) components.
Refer to Figure 10.2: The main constructs in this study as shown in the
theoretical framework are:
1. Service Quality (measured using 34 items),

2. Corporate Image (measured using 16 items),
3. Students’ Satisfaction (measured using 14 items),
4. Students’ Loyalty (measured using 9 items).
However, after running the data reduction procedure using Factor

Analysis, the following components emerged for each construct.
1. Service Quality (3 components renamed as human elements, academic

facilities, and infrastructure facilities of the university),
2. Corporate Image (3 components renamed as university image, academic
reputation, and external recognition obtained by the university),
3. Students’ Satisfaction (2 components renamed as satisfaction with
service and satisfaction with the process at the university),
4. Students’ Loyalty (2 components renamed as willingness to remain with
the university and willingness to recommend the university).
The main advantage of AMOS graphic compared to other methods of

analysis is the researcher can convert his original theoretical framework
directly into AMOS Graphic. Figure 10.3 illustrates the theoretical framework on
Figure 10.2 with the respective components. The figure shows how the
theoretical framework is converted into the appropriate AMOS syntax.

UniSZA
err1 err2 err3
1 1 1
Humans AcadSvc Infras
err9 err10
SERVICE
QUALITY 1 1
RetenN RecomM
err7
1
1
Process 1
STUDENTS' STUDENTS'
SATISFACTION LOYALTY
Outcome
1 1
1
Error2 Error1
err8
CORPORATE
IMAGE
UnivMge AcadMge RecogN
1 1 1
err4 err5 err6
Figure 10.3: The Theoretical Model of the Study in AMOS Graphic

UniSZA
Table 10.15: The summary of variables in the model
Component No of min max skewness

Construct Component
renamed as data
Service Quality 1 INFRASTRUCTURE 872 3.20 10.00 -0.39
Service Quality 2 ACADEMIC FACILITIES 872 4.21 10.00 -0.36
Service Quality 3 HUMAN ELEMENTS 872 4.36 10.00 -0.46
Corporate Image 1 UNIVESITY IMAGE 872 4.71 10.00 -0.24
Corporate Image 2 ACADEMIC IMAGE 872 5.80 10.00 -0.19
Corporate Image 3 RECOGNITIONS 872 5.00 10.00 -0.57
Student Satisfaction 1 PROCESS 872 5.40 10.00 -0.32
Student Satisfaction 2 OUTCOME 872 3.33 10.00 -0.54
Student Loyalty 1 RECOMMENDATION 872 4.20 10.00 -0.61
Student Loyalty 2 RETENTION 872 3.25 10.00 -0.22
Definition of Variables in AMOS Syntax
Listed below is the definition of terms for all constructs in the proposed model
(Figure 10.3). Among the latent constructs in the model are:
1. STUDENTS' LOYALTY – This is the construct of main interest in the study.

This construct is measured using two manifesting variables namely retention
(RETENN) and recommendation (RECOMM).
2. Error1 - Estimated error to emerge when service quality, corporate image, and
students’ satisfaction are included in the equation to predict students’ loyalty
towards the university. This is called an error in the equation.
3. err9 - Estimated error to emerge when the intention to remain loyal and
increase transaction is employed to measure students’ loyalty towards the
university. This is called an error in the measurement.

UniSZA
4. err10 - Estimated error to emerge when intention to provide positive words of

mouth and recommendation is employed to measure students’ loyalty. This is
also an error in the measurement.
5. STUDENTS' SATISFACTION – This is another construct of main interest in the

study. This construct is measured using two manifesting variables namely
OUTCOME and PROCESS.
6. Error2 - Estimated error to emerge when service quality and corporate image
are included in the equation to predict students’ satisfaction with the university.
This is called an error in the equation.
7. err7 - Estimated error to emerge when students’ satisfaction with service

delivery process at the university is employed to measure their overall
satisfaction with the university. This is called an error in the measurement.
8. err8 - Estimated error to emerge when students’ satisfaction with the service
outcome is employed to measure their overall satisfaction with the university.
This is called an error in the measurement.
9. SERVICE QUALITY – This construct is sometimes called predictor construct in

the model. The study measures service quality using three manifesting
variables namely human element (HUMANS), academic service facilities
(ACADSVC), and infrastructure and physical facilities (INFRAS).
10. err1 - Estimated error to emerge when human elements in service delivery is
employed to measure service quality performance. This is called an error in the
measurement.
11. err2 - Estimated error to emerge when academic facilities at the university are
employed to measure service quality performance. This is called an error in the
measurement.
12. err3 - Estimated error to emerge when the infrastructure facilities is employed
to measure service quality performance. This is called an error in the
measurement.
13. CORPORATE IMAGE - This is another predictor construct in the model. The
study measured this construct using three manifesting variables namely image
of university (UNIVIMGE), image of academic activities (ACADMGE), and
external recognitions (RECOGN).

UniSZA
14. err4 - Estimated error to emerge when reputation of the university is employed
to measure corporate image. This is called an error in the measurement.
15. err5 - Estimated error to emerge when academic excellence is employed to

measure corporate image. This is called an error in the measurement.
16. err6 - Estimated error to emerge when external recognition is employed to

measure corporate image. This is called an error in the measurement.
Once everything is in place, the researcher would run the AMOS model in
order to obtain the required statistical measures required in the study.
Automatically AMOS would give two distinct outputs namely the Standardized
Model and the Unstandardized Model. The Standardized Model is shown in
Figure 10.4 and the Unstandardized Model is shown in Figure 10.5.
AMOS Output: The Standardized Model
In the standardized model, AMOS output provides the correlation

between any two independent variables, the standard factor loading and R2 for
each component, the standardized regression coefficient linking each
independent variable to its corresponding dependent variable, and also the R2
for the regression equation. The output in Figure 10.4 shows the correlation
between Service Quality and Corporate Image is 0.78, the factor loading for
variable “Humans” is 0.81 with its R2 = 0.65, the standardized regression
coefficient of Service Quality to Students’ Satisfaction is 0.46, and the
coefficient of determination linking Service Quality, Students’ Satisfaction, and
Corporate Image to Students’ Loyalty is 0.65.

UniSZA
err1 err2 err3
.65 .91 .67
.81 .95 .82
err9 err10
SERVICE
QUALITY
.50 .70
RetenN RecomM
.05
err7 .46
.71 .84
.62
.40 .65
Process .79
STUDENTS' .66 STUDENTS'
.78 .83 LOYALTY
.91 SATISFACTION
Outcome
.21
Error2 .17 Error1
err8
CORPORATE
IMAGE
.82 .42
.90
.68 .81 .18
err4 err5 err6
Figure 10.4: The Path Diagram presents the strength and magnitude of
association among the constructs in the Study.

UniSZA
Table 10.16: The measures of strength and magnitude of association among the
constructs
Estimate of
Variables PATH Variables
Correlation
STUDENTS'_SATISFACTION ← CORPORATE_IMAGE 0.21
STUDENTS'_SATISFACTION ← SERVICE_QUALITY 0.46
STUDENTS'_LOYALTY ← CORPORATE_IMAGE 0.17
STUDENTS'_LOYALTY ← SERVICE_QUALITY 0.05
STUDENTS'_LOYALTY ← STUDENTS'_SATISFACTION 0.66
HUMANS ← SERVICE_QUALITY 0.81
ACADEMIC SERVICE ← SERVICE_QUALITY 0.95
INFRASTRUCTURE ← SERVICE_QUALITY 0.82
RECOGNITION ← CORPORATE_IMAGE 0.42
ACADEMIC IMGAGE ← CORPORATE_IMAGE 0.90
UNIVERSITY IMAGE ← CORPORATE_IMAGE 0.82
OUTCOME ← STUDENTS'_SATISFACTION 0.91
PROCESS ← STUDENTS'_SATISFACTION 0.79
RETENTION ← STUDENTS'_LOYALTY 0.71
RECOMMENDATION ← STUDENTS'_LOYALTY 0.84
The path diagram also presents the measure of association between the two
independent constructs in the study namely service quality and corporate image.
The positive association between these two exogenous constructs indicates that
the increase in either construct (service quality provided by the university) would
likely increase the other construct (corporate image of a university) and vice-versa.
In other words, it’s just like killing two birds with one stone.
Thus, any program designed and implemented by the university aimed at

improving the service quality provided by the university would indirectly help in
enhancing the corporate image of the university itself, and vice versa. Table 10.17
UniSZA
illustrates the strength of relationship between service quality provided by the

university and corporate image of a university.
Table 10.17: The measure of correlation between Service Quality and

Corporate Image construct
Variable Path Variable Estimate of Correlation
SERVICE_QUALITY ↔ CORPORATE_IMAGE 0.78
The Unstandardized Model in Figure 10.5 also presents the path diagram.
However, unlike the path diagram of the Standardized Model, the path diagram in
the Unstandardized Model presents the regression coefficients linking the
independent constructs to the dependent construct in the study. Refer to Figure
10.5 for the fitted model and table 10.18 for the corresponding text output.

UniSZA
AMOS Output: The Unstandardized Model
In Figure 10.5, the covariance between Service Quality and Corporate image
is 0.25, the actual regression coefficient between service Quality and Customer
Satisfaction is 0.44, the error variance in measurement of “Humans” is 0.43. The
significance level (P-value) for each relation is given in the text output, as shown in
Table 10.18. These values can be used to decide the relevant hypothesis.
.43 .12 .52
err1 err2 err3
1 1 1
1.00 1.22 1.16

.79 .78 .40
err9 err10
SERVICE
QUALITY 1 1
.45 RetenN RecomM

.05
err7 .44
1.00 1.10
1
Process 1.00
STUDENTS' .67 ST UDENT S'
.25 LOYALT Y
1.25 SATISFACTION
Outcome
1 1
1 .44 .27
.23
.51
Error2 .42 Error1
err8
.13
CORPORATE
IMAGE
2.83 1.00
2.53
1 1 1
.49 .20 .60
err4 err5 err6
Figure 10.5: The Path Diagram Presents the Regression Coefficients among
the Constructs

UniSZA
The path analysis among all constructs and variables in the model is
illustrated in Table 10.18. The symbol *** is given by AMOS to indicate the effect is
highly significant (P-value <0.001).
Table 10.18: The Path Analysis for All Constructs and Variables in the Model
Beta Standard Critical P-

Variables PATH Variables
Estimate Error Region value
STUDENTS'_SATISFACTION ← CORPORATE_IMAGE 0.51 0.15 3.51 ***
STUDENTS'_SATISFACTION ← SERVICE_QUALITY 0.44 0.06 7.72 ***
STUDENTS'_LOYALTY ← CORPORATE_IMAGE 0.42 0.14 2.96 ***
STUDENTS'_LOYALTY ← SERVICE_QUALITY 0.05 0.06 0.88 0.38
STUDENTS'_LOYALTY ← STUDENTS'_SATISFACTION 0.67 0.05 12.60 ***
HUMANS ← SERVICE_QUALITY 1.00
ACADEMIC SERVICE ← SERVICE_QUALITY 1.22 0.04 32.95 ***
INFRASTRUCTURE ← SERVICE_QUALITY 1.16 0.04 27.87 ***
RECOGNITION ← CORPORATE_IMAGE 1.00
ACADEMIC IMGAGE ← CORPORATE_IMAGE 2.53 0.20 12.37 ***
UNIVERSITY IMAGE ← CORPORATE_IMAGE 2.83 0.23 12.21 ***
OUTCOME ← STUDENTS'_SATISFACTION 1.00
PROCESS ← STUDENTS'_SATISFACTION 0.80 0.03 24.68 ***
RETENTION ← STUDENTS'_LOYALTY 1.00
RECOMMENDATION ← STUDENTS'_LOYALTY 1.10 0.06 19.01 ***

UniSZA
TESTING FOR THE MAIN HYPOTHESES IN THE STUDY
Service
Quality
HA3
HA1
Students’
Loyalty
HA2
Corporate HA4 HA5

Image
Figure 10.6: The Hypothesized Link under the Respective Hypothesis.
Finally, the study would like to test the main hypotheses in this research
(Figure 10.6). The following procedure would test the hypothesized links among
the main constructs in the study namely Service Quality, Corporate Image,
Students’ Satisfaction, and Students’ Loyalty as already stated above. The arrows
indicate that the study is interested to examine the significance of direct influence
of one construct to another construct.
HA1: Service quality has a significant and direct influence on customer satisfaction
The Path Analysis of Structural Equation Modeling for Hypothesis 1

Beta Standard Critical Significance
Construct PATH Construct
Estimate Error Region Value (p)
STUDENTS'_SATISFACTION ← SERVICE_QUALITY 0.44 0.06 7.72 0.001

UniSZA
Since the P-value is lower than 0.05, the above research hypothesis is supported.
The study concludes that service quality at the university has a significant and direct
influence on students’ satisfaction towards the university.
HA2: Corporate image has a significant and direct influence on customer

satisfaction

STUDENTS'_SATISFACTION ← CORPORATE_IMAGE 0.51 0.15 3.51 0.001
The study concludes that corporate image of the university has a significant and direct
influence on students’ satisfaction with the university.
HA3: Service quality has a significant and direct influence on customer loyalty

STUDENTS'_LOYALTY ← SERVICE_QUALITY
0.05 0.06 0.88 0.38
Since the P-value is higher than 0.05, the above research hypothesis is not
supported. The study concludes that service quality at the university does not have a
significant and direct influence on students’ loyalty towards the university.

UniSZA
HA4: Corporate image has a significant and direct influence on customer loyalty
The Path Analysis of Structural Equation Modelling for Main Hypothesis 4

STUDENTS'_LOYALTY ← CORPORATE_IMAGE
0.42 0.14 2.96 0.001
The study concludes that corporate image of the university has a significant and direct
influence on students’ loyalty towards the university.
HA5: Customer satisfaction has a significant and direct influence on customer loyalty
The Path Analysis of Structural Equation Modeling for Main Hypothesis 5

STUDENTS'_LOYALTY ← STUDENTS'_SATISFACTION
0.67 0.05 12.60 0.001
The study concludes that students’ satisfaction with the university has a significant and
direct influence on students’ loyalty towards the university.

UniSZA
Table 10.19: Summary of the Results of the Main Hypotheses in the Study
Result
No The Main Hypothesis Statement in the Study
Service quality of the university has a significant and

1 HA1 Supported
direct influence on students’ satisfaction
Corporate image of the university has a significant

2 HA2 Supported
and direct influence on students’ satisfaction
Service quality of the university has a significant and

3 HA3 Not Supported
direct influence on students’ loyalty
Corporate image of the university has a significant

4 HA4 Supported
and direct influence on students’ loyalty
Students’ satisfaction with the university has a

5 HA5 Supported
significant and direct influence on their loyalty
Now the researcher needs to proceed into another chapter namely the
analysis of results, conclusion, and recommendation. In this chapter the
researcher needs to address every item of his original objectives, the research
questions, and the corresponding hypotheses.
Remember, at the end of any research the researcher needs to revise his
earlier objectives to ensure that they are achieved by the study. The research is
not meaningful if it fails to achieve the stated objectives.
The same thing also applies to the research questions; the researcher needs
to revise his research questions to ensure that their answers are properly
addressed based on the result of the study. The research is not meaningful if it
fails to address or to clarify the ambiguities surrounding the problem.
As for the research hypotheses, the researcher needs to determine whether

or not the data obtained support his hypotheses. In his writing, the researcher
needs to debate whether his finding is consistent with the findings from the
previous study in the same industry as well as the findings of the study in different
industry. The researcher needs to debate his findings and provide the rationale for
being consistent or inconsistent with the literature.

UniSZA
Finally, the researcher needs to examine his theoretical framework which

indicates how he theorized the constructs in the study should relate. He needs to
debate how his theorized relationships among the constructs are supported or not
supported by the data. Again, he needs to debate the rationale behind the
stipulated result by referring to the underlying theory in the area.

UniSZA
Exercises
1) a) A researcher is in the process of preparing a statistical report for a client. He

began the analysis by conducting several hypotheses testing, such as
independent sample t-test and Chi-Square test of independence. Is this the
right thing to do? Describe your agreement or disagreement in the way the
researcher presented his analysis. Give your reason?
b) What types of scalar data (nominal, ordinal, interval, or ratio) are typically
utilized in cross-tabulation analysis?
c) In a survey, the respondents were asked to respond to a statement asking if

their work was interesting. Interpret the frequency distribution in the SPSS
output below.
“My work is interesting”
Category Label Code Frequency Percent Valid Percent

Very true 1 650 23.9 62.4
Somewhat true 2 303 11.2 29.1
Not very true 3 61 2.2 5.9
Not at all true 4 28 1.0 2.7
1,653 61.6 Missing
Total 2,715 100.00 100.00
Valid cases 1,042 Missing cases 1,673
2) The general manager of a large chain of convenience stores wants to try

different advertising methods with respect to three advertising policies. The
three policies are:
Policy 1: No advertising
Policy 2: Advertise in neighbourhood with circulars
Policy 3: Use circulars and advertise in local newspapers

UniSZA
Eighteen stores are randomly selected and divided at random into three groups
of six stores. Each group uses one of the three policies. Following the
implementation of the policies, sales figures are obtained for each of the stores
during a one-month period. The results are analyzed and obtained as follows.
Based on the SPSS output, answer the following questions.
a) Perform the normality test on sales figures based on the three advertising
policies. What can you conclude from the test? Is it appropriate to perform a
one-way analysis of variance? (Use α = 0.05).
b) At the 5 % level of significance, is there a difference in mean sales figures

between the three advertising policies? (Show all the necessary steps in
testing the hypothesis)
c) From the post hoc multiple comparisons, indicate which pairs of advertising
policies are significantly different in terms of mean sales figures? State the
p-values.
Tests of Normality
Kolmogorov-Smirnov(a) Shapiro-Wilk
Statistic df Sig. Statistic Df Sig.
Sales 0.183 18 0.112 0.914 18 0.101
ANOVA
Sales
Sum of Mean
Squares df Square F Sig.
Between Groups 139.000 2
69.500
Within Groups 87.000 15 11.983 .001
5.800
Total 226.000 17

UniSZA
Post Hoc Multiple Comparisons (Scheffe’s method)
Dependent Variable: Sales

95% Confidence Interval
Mean
(J) Difference
(I) Policy Policy (I-J) Std. Error Sig. Lower Bound Upper Bound
2 -1.500 1.390 .571 -5.27 2.27
1
3 -6.500(*) 1.390 .001 -10.27 -2.73
1 1.500 1.390 .571 -2.27 5.27
2
3 -5.000(*) 1.390 .009 -8.77 -1.23
1 6.500(*) 1.390 .001 2.73 10.27
3
2 5.000(*) 1.390 .009 1.23 8.77
* The mean difference is significant at the .05 level.
Homogenous Subsets – Sales
Scheffe’
Subset for alpha = .05

Policy N 1 2
1 21.67
6
2 23.17
6
3 28.17
6
Sig. .571 1.000
Means for groups in homogeneous subsets are displayed.
3) A pediatrician began measuring the blood cholesterol levels of her young

patients. She was surprised to find that many of them had levels over 200,
indicating increased risk of artery disease. Ten such patients were randomly
selected to take part in the nutritional program designed to lower blood
cholesterol. Two months following the commencement of the program, the
pediatrician measured the blood cholesterol levels of the 10 patients again.
The results were obtained below. At the 5% level of significance, is there
enough evidence to indicate that the nutritional program is effective in
reducing blood cholesterol level? (Show all the necessary steps in testing the
hypothesis)
UniSZA
Wilcoxon Signed Ranks Test

Ranks n Mean rank Sum of Ranks
After – Before -Ve 8 6.00 48
After – Before +Ve 2 3.50 7
After – Before = 0 0.00 0
Total 10
Test Statistics (b)
After program – Before program
Z value -2.105(a)
Asymptotic Sig. (2-tailed) P-value .035
a Based on positive ranks

b Wilcoxon Signed Ranks Test
4) The retailing manager of a supermarket chain wants to determine whether

product location has an effect on the sale of pet toys. Three different aisle
locations are to be considered: front, middle, and rear. A random sample of 18
stores is selected with 6 stores randomly assigned to each aisle location. The
size of the display area and the price of the product are constant for all stores.
At the end of a 1-month trial period, the sales volume (in thousands of dollars)
of the product in each store is recorded and the data was analyzed as follows.
i) State the median sales volume for all three locations.

ii) Which location produced the highest sales volume? Compare the
minimum and maximum sales volume of pet toys for all locations.

UniSZA
location
FRONT MIDDLE REAR
Mean 6.067 2.067 3.733
Median 5.800 1.900 3.400
Sales Variance 2.715 .427 2.011
(RM’000) Std. Deviation 1.6476 .6532 1.4180

Minimum 4.0 1.4 2.2
Maximum 8.6 3.2 6.0
5) A researcher is in the process of preparing a statistical report for a client. He

began the analysis by conducting several hypotheses testing, such as
independent t-test and Chi-Square test of independence. Is this the right thing
to do? Describe your agreement or disagreement in the way the researcher
presented his analysis and why?
6) What types of scalar data (nominal, ordinal, interval, or ratio) are typically
utilized in cross-tabulation analysis?
7) Based on the frequency tables and summary statistics below; describe the
profile of the respondents.
Gender
Valid Cumulative
Gender Frequency Percent Percent Percent
Male 54 54.0 54.0
54.0
Female 46 46.0 46.0
100.0
Total 100 100.0 100.0

UniSZA
Stay
Valid Cumulative
Frequency Percent Percent Percent
On Campus 26 26.0 26.0
26.0
Off campus with friends 51 51.0 51.0
77.0
Off campus with family 23 23.0 23.0
100.0
Total 100 100.0 100.0
Descriptive Statistics
N Minimum Maximum Mean Std. Deviation
Age 100 13 27 20.72 2.656
Valid N (listwise) 100
Smoke
Valid Percent Cumulative
Frequency Percent Percent
Yes 19 19.0 19.0 19.0
No 60 60.0 60.0 79.0
Occasionally 21 21.0 21.0 100.0
Total 100 100.0 100.0
8) Based on the frequency table and the χ2 test of association results below,
write an analysis and conclusion.

UniSZA
Gender * Smoke Cross-tabulation

Smoke
Yes No Occasionally Total
Gender Male Count % 16 26 12 54
within Gender 29.6% 48.1% 22.2% 100.0%
Female Count % 3 34 9 46
within Gender 6.5% 73.9% 19.6% 100.0%
Total Count % 19 60 21 100
within Gender 19.0% 60.0% 21.0% 100.0%
9) HyPetrol Sdn Bhd is a manufacturing company producing oil gaskets for

automobile and small tractor engine. The factory is located in Bayan Lepas
industrial area. Presently the company employs 500 employees. Three
hundred of them are locals and have been working with the company since it
started its operation in 1995. The rest of them are foreign workers from
Bangladesh and Indonesia. Recently the management has been getting
informal information that the employees are dissatisfied with the new policy
that has been implemented for the past two (2) months and this situation
seems to be affecting their productivity. The general manager requested a
survey research be conducted. Among the issues of concern is how much job
satisfaction has been affected by the new policy. Other related factors include
age, length of service, gender, opportunity for advancement, and origin (local
or immigrant). The sample size for this survey is 100.
Based on the given information,
a) Identify the target population and sample. Briefly explain how the
sampling frame is acquired.
b) State the problem statement of the study.
c) Describe a suitable research design for the study and its purpose.
d) Identify the dependent and independent variables in this study. Draw a

schematic diagram to illustrate the relationship between the variables.
e) State TWO (2) suitable hypotheses for the study.

UniSZA
Chi-Square Tests
Asymp. Sig.
Value df (2-sided)
Pearson Chi-Square 9.813ª
2 .007
Likelihood Ratio 10.625
2 .005
Linear-by-Linear Association 2.574
1 .109
N of Valid Cases 100
a. 0 cells (.0%) have expected count less than 5. The minimum expected count is 8.74.

UniSZA
Chapter 11: The Structural Equation Modeling
CHAPTER STRUCTURAL EQUATION MODELING

11
WHAT IS STRUCTURAL EQUATION MODELING?
The Structural Equation Modeling or popularly known as SEM is a second

generation statistical analysis techniques developed for analyzing the inter-
relationships among multiple variables in a model. The inter-relationships
among variables could be expressed in a series of single and multiple
regression equations. The Structural Equation Modeling technique employs the
combination of quantitative data and the correlational or causal assumptions
into the model.
SEM is a more powerful statistical technique available to solve the following

requirements:
1) Analyzing the unobserved variables or latent constructs

2) Running the Confirmatory Factor Analysis (CFA)
3) Analyzing multiple regression models simultaneously
4) Analyzing regressions with multicollinearity problem
5) Analyzing the path analysis with multiple dependents
6) Estimating the correlation and covariance in a model
7) Modeling the inter-relationships among variables in a model
SEM begins with a theory whereby the researcher intends to test the
theorized relationships among research constructs in the study. This theoretical
relationship is modeled into a theoretical framework represented by a schematic
diagram. The schematic diagram presents the hypotheses of interest in the
research. The constructs involved are measured using a set of items in a
questionnaire. The measurement scale for each item should be either interval or
ratio. Ideally, the researcher should consider a minimum of four items to measure
each construct.

UniSZA
WHAT IS THE RATIONALE FOR EMPLOYING SEM
SEM is capable of estimating a series of interdependence among constructs

simultaneously in a model. In fact SEM is a more efficient method to analyze
the correlational and causal relationships among latent constructs as well as
observed variables, estimating their variance and covariance, test hypotheses,
modeling conventional regressions, and running the Confirmatory Factor
Analysis (CFA).
In research, latent constructs are the variables which the researcher could not
measure them directly since it is only a hypothetical concept of something. Thus,
the researcher cannot model these latent constructs in the Ordinary Least Square
regression models (OLS). Some examples of latent constructs are:
Service Quality, Customer Satisfaction, Job Satisfaction, Corporate Image,
Product Image, Customer Loyalty, Purchase Intention, Consumer Behavior,
Employee Soft Skills, Perceived usefulness, Relational bond, Financial bond,
Structural bond, Relationship quality, Attitudinal loyalty, and Behavioral loyalty.
The abovementioned variables cannot be measured directly like counting the

number of kids in a family, total income of a household, monthly phone bills, daily
production, weekly price of chicken, etc. The variable which could be measured
directly is called the observed variable, while the variable which could not be
measured directly is called latent construct. These latent constructs could be
measured indirectly using a set of items in a questionnaire as shown in Table 11.1.
Table 11.1: Example of a Questionnaire to Measure Student Satisfaction

As a student of this university, I….. Strongly Strongly
Disagree Agree
1 am satisfied with the lecture schedules 1 2 3 4 5 6 7 8 9 10
2 am satisfied with the learning system 1 2 3 4 5 6 7 8 9 10
3 am satisfied with the evaluation method 1 2 3 4 5 6 7 8 9 10
4 am satisfied with academic regulations 1 2 3 4 5 6 7 8 9 10
5 am satisfied with academic services 1 2 3 4 5 6 7 8 9 10
6 am satisfied with the library references 1 2 3 4 5 6 7 8 9 10
7 am satisfied with learning facilities 1 2 3 4 5 6 7 8 9 10
8 am satisfied with the campus security 1 2 3 4 5 6 7 8 9 10

UniSZA
Other advantages of Structural Equation Modeling are:
1) Could use Confirmatory Factor Analysis to reduce measurement errors

2) Could handle the multicollinearity problem among independent variables
3) Could assess the fitness of measurement model for latent construct
4) Could analyze the model with more than one dependent variables
5) Could analyze the mediating effects of a mediator variable
6) Could analyze the moderating effects of a moderator variable
7) Could handle the correlated errors among response items
MODELING THE RELATIONSHIP BETWEEN CONSTRUCTS
Modeling the Causal Relationship
In this case, at least two measurement models involved – one for

independent construct and the other one is for dependent construct. The link
between measurement model for independent construct and measurement model
for dependent construct is called the structural model. Thus, instead of modeling
the Ordinary Least Squares regression (OLS) and analyzed using ANOVA, the
researcher is working with the Structure Equation Modeling (SEM) and analyzed
using AMOS as shown in Figure 11.1.
Note: AMOS is an acronym for Analysis of Moments Structures, one of the

software developed for analyzing SEM. Other SEM software are LISREL,
SEPATH, PRELIS, SIMPLIS, PLS, MPLUS, EQS, and SAS. However, AMOS is
widely employed by researchers for its graphic capabilities.

UniSZA
e1 e2 e3 e4 e5 e6 e7 e8 e9 e10
1 1 1 1 1 1 1 1 1 1
X11 X12 X13 X14 X15 Y1 Y2 Y3 Y4 Y5
1 1
1
X1 Y e11
Figure 11.1: The Simple Regression Equation Model for Y On X1
Note: X1 and Y are latent constructs. In Amos graphic, latent constructs are
represented by ellipses. The latent construct X1 is measured using five items
namely X11 to X15, while latent construct Y is also measured using five items
namely Y1 to Y5. These measured or observed variables are represented by
rectangles in AMOS model.
Key: X1 = Independent latent construct, while X11 to X15 is a set of 5 items to

measure construct X1. In SEM, independent construct is termed as exogenous
construct. The symbol e1 to e5 are errors in measurement for variable X11 to X15.
Y = Dependent latent construct, while Y1 to Y5 is a set of 5 items to measure

Y. In SEM, dependent construct is termed as endogenous construct. The symbol
e6 to e10 are errors in measurement for variable Y1 to Y5.
The symbol e11 is an error in the equation or the residual.
Modeling the Correlational Relationship
When both variables involved are unobserved or latent constructs, we can

model and estimate the correlational relationship between the two construct as
presented in Figure 11.2.

UniSZA
e1 e2 e3 e4 e5 e6 e7 e8 e9 e10
1 1 1 1 1 1 1 1 1 1
X11 X12 X13 X14 X15 Y1 Y2 Y3 Y4 Y5

1 1
X1 Y
Figure 11.2: The modeling the correlation between Y and X1
MODELING COMPLEX RELATIONSHIP AMONG THE CONSTRUCTS
When the relationships among variables are complex and the variables are
unobserved or latent constructs, modeling the inter-relationship among constructs
are almost impossible using OLS. However, one could employ SEM to model and
estimate the strength of relationship between constructs.
Modeling the Multiple Regressions for Latent Constructs
AMOS can model and analyze the multiple regression equations for latent
constructs simultaneously. Figure 11.3 shows how three simple regression
equations and one multiple regression equation is modeled simultaneously. At the
same time the covariance between the exogenous constructs are also estimated.

UniSZA
e1 e2 e3 e4 e5
1 1 1 1 1
X11 X12 X13 X14 X15

1
X1
1 1
e6 X21 e21 Y1 e16
1
1 1 1
e7 X22 Y2 e17
1 1
e8 X23 X2 Y Y3 e18
1 1
e9 X24 1 Y4 e19
1 1
e10 X25 Y5 e20
X3
1
X31 X32 X33 X34 X35

1 1 1 1 1
e11 e12 e13 e14 e15
Figure 11.31: The Structural Equation Model for the Latent Constructs (Y On
X1, X2, and X3)
Key: X1 = Exogenous latent construct, X11 to X15 = a set of 5 items to measure X1

e1 to e5 = error in measurement for X11 to X15
X2 = Exogenous latent construct, X21 to X25 = a set of 5 items to measure X2
X3 = Exogenous latent construct, X31 to X35 = a set of 5 items to measure X3
Y = Endogenous latent construct, Y1 to Y5 = a set of 5 items to measure Y
e16 to e20 = error in measurement for Y1 to Y5
e21= error in the equation or residual

UniSZA
Modeling the Mediator Variable in a Model
AMOS could analyze the mediating effects of latent construct in a model. The
theoretical model is illustrated in Figure 11.4. In the model, exogenous construct
X1 has five response items; mediating construct X2 also has five response items,
while endogenous construct Y has three response items respectively. Here, X2 is
modeled as a mediator construct. The researcher is interested to determine
whether X2 mediates the relationship between X1 and Y.
e6 e7 e8 e9 e10
1 1 1 1 1
X21 X22 X23 X24 X25
1
X2 e14
1
1 Y1 e11
1
X1 Y Y2 e12
1 1
1 Y3 e13
X11 X12 X13 X14 X15

e15
1 1 1 1 1
e1 e2 e3 e4 e5
Figure 11.42: Modeling the Mediating Effect of Latent Construct X2. The
Illustration of the Exogenous Variable (X1), Mediating Variable (X2),
And Endogenous Variable (Y)

UniSZA
Modeling the Moderator Variable in a Model
AMOS could model and analyze the effects of moderator variable in the
model. The theoretical model is illustrated in Figure 11.5. In the model, exogenous
construct X1 has five response items and endogenous construct Y also has five
response items. In the model, M is the moderator variable. The researcher is
interested to determine whether M moderates the relationship between X1 and
Y.
e1 e2 e3 e4 e5 e6 e7 e8 e9 e10
1 1 1 1 1 1 1 1 1 1
X11 X12 X13 X14 X15 Y1 Y2 Y3 Y4 Y5

1 1
1
X1 Y e11
Figure 11.53: The Illustration of the Exogenous Variable (X1), Endogenous

Variable (Y), and Moderator Variable M in a Model.
The moderator variable M could be respondents’ demographic

characteristics including their level of socio-economic status such as education,
occupation, and income.
ANALYZING STRUCTURAL EQUATION MODELING USING AMOS
In SEM, there are two models involved namely:
i) The measurement model

ii) The structural model.
The researcher needs to analyze the measurement model first before

modeling the structural model.
UniSZA
Analyzing the Measurement Model
In research, the researcher needs to prove the unidimensionality, validity,

and reliability of all measurement models. The unidimensionality assessment
should be made first prior to assessing validity and reliability. With Confirmatory
Factor Analysis (CFA), any item that does not fit into its measurement model
should be removed from the model.
The assessment for each element is done as follows:
1. Unidimensionality
Unidimensionality is achieved when the measuring items have acceptable

factor loadings for the respective latent construct. In order to ensure
unidimensionality of a measurement model, any item having low factor
loading should be deleted.
How low is considered to be “too low” and has to be deleted from

measurement model?
i) For a newly developed scales, the factor loading for an item should be
0.5 or higher
ii) For already established scales, the factor loading for an item should
be 0.6 or higher
The item-deletion process should be made one item at a time with the
lowest item to be deleted first. After an item is deleted, the researcher
needs to re-specify and estimate the new measurement model. The
process continues until the unidimensionality requirement is achieved.
2. Validity
Validity is the ability of instrument to measure what it supposed to be

measured for a construct. Three types of validity are required for each
measurement model is:
a. Convergent validity. This validity is achieved when all items in a

measurement model are statistically significant. The convergent
validity could also be verified through AVE (Average Variance
Extracted), and the AVE should be greater or equal to 0.5.
UniSZA
b. Construct validity. This validity is achieved when the Fitness indexes

achieved the following requirements: GFI = 0.90 or higher, CFI = 0.90
or higher, RMSEA = 0.08 or lower, and the ratio of Chisq/df is less
than 5.0.
c. Discriminant validity. This validity condition is achieved when the
measurement model is free from redundant items. AMOS will identify
the pair of redundant items in the model and reported in the MI
(Modification Indices). In the normal procedure, the researchers would
delete one of the items and re-specify the model. However, in certain
cases the researcher could set the correlated pair as “free parameter
estimate”. Another requirement for discriminant validity is the pattern
structure correlation between a pair of latent exogenous construct
should be less than 0.85.
3. Reliability
Reliability is the extent of how reliable is the particular measurement model

in measuring the respective latent construct. The assessment for reliability
of an instrument could be made using the following criteria.
i) Internal reliability – Achieved when the Cronbach’s Alpha value of

0.7 or higher (calculated in SPSS).
ii) Construct Reliability – The measure of reliability and internal
consistency of the measured variables representing a latent construct.
A value of CR > 0.6 is required in order to achieve construct reliability
(refer to the formula in Table 2).
iii) Average Variance Extracted – The average percentage of variation
explained by the items in a construct. An AVE > 0.5 is required (refer
to the formula in Table 2)
Table 2: The Formula for Computing AVE and CR
AVE= ∑ Қ2 / n Қ = factor loading of every item and

2 2 2
CR = (∑Қ) / [(∑Қ) + (∑1- Қ )] n = number of items in the model

UniSZA
Evaluating the Fitness of a Measurement and Structural Model
In SEM, there are a series of goodness-of-fit indexes that reflect the fitness of the
model to the data at hand. At the moment, there is no agreement among the
researchers which fitness indexes should be reported. Hair et al. (1995) and
Holmes-Smith (2006) recommend the use of at least three fit indexes by including
at least one index from each category of model fit. The three fitness categories are
absolute fit, incremental fit, and parsimonious fit.
The researcher could choose at least one fitness index from each category to
report depending on which literature is being referred. The information concerning
the fitness index category, their level of acceptance, and comments is presented in
Table 11.3.
Table 11.3: Index Category and the Level of Acceptance for Every Index
Name of category Name of Level of Comments

index acceptance
1. Absolute fit Chisq P > 0.05 Sensitive to sample size >200
Absolute fit RMSEA RMSEA < 0.08 Range 0.05 to 1.00 acceptable
Absolute fit GFI GFI > 0.90 GFI = 0.95 is a good fit
2. Incremental fit AGFI AGFI > 0.90 AGFI = 0.95 is a good fit
Incremental fit CFI CFI > 0.90 CFI = 0.95 is a good fit
3. Parsimonious fit Chisq/df ChiSq/df < 5.0 The value should be less than 5.0.
***The indexes in bold is recommended since they are highly reported in literatures.
The types of fitness indexes employed and the acceptable cut-off values reported
by researchers may vary depending on which literatures support they are referring.
However, the following table presents the literature support for the selected fitness
indexes.

UniSZA
Table 11.4: The Literature Support for the Respective Fitness Index
Name of category Name of Index full name Literature support

index
1. Absolute fit Chisq Minimum discrepancy Chi Square Wheaton et al. (1977)
Absolute fit RMSEA Root Mean Square of Error Browne and Cudeck (1993)
Approximation
Absolute fit GFI Goodness of Fit Index Joreskog and Sorbom (1984)
2. Incremental fit AGFI Adjusted Goodness of Fit Index Tanaka and Huba (1985)
Incremental fit CFI Comparative Fit Index Bentler (1990)
3. Parsimonious fit Chisq/df Chi Square / Degrees of Freedom Marsh and Hocevar (1985)
*** One could ignore the absolute fit index of minimum discrepancy chi-square if the
sample size obtained for the study is greater than 200 (Hair et al., 1996; Joreskog and
Sorbom, 1996).
An example a measurement model of a latent construct is presented in

Figure 11.6.
e1 e2 e3 e4 e5 e6 e7 e8 e9
1 1 1 1 1 1 1 1 1
Wl 1 Wl 2 Wl 3 Wl 4 Wl 5 Wl 6 Wl 7 Wl 8 Wl 9
Work
Load
Reference Point
Figure 11.6: The Measurement Model for Measuring Employee’s Workload
The CFA procedure, the researcher should instruct AMOS to calculate the
standardized estimate of factor loading and squared multiple correlation for every
item by clicking the respective box in the analysis properties menu. The

UniSZA
standardized estimates indicate the factor loading for each item in a measurement
model.
The AMOS output showing the factor loading and squared multiple
correlation for every item in a measurement model are presented in Figure 11.7.
e1 e2 e3 e4 e5 e6 e7 e8 e9
.50 .53 .64 .67 .50 .68 .60 .42 .56
Wl 1 Wl 2 Wl 3 Wl 4 Wl 5 Wl 6 Wl 7 Wl 8 Wl 9
.71 .77
.73 .80 .82 .83 .65
.70 .75
Response
Item R2
Work
Factor Loading Load Latent construct
Figure 11.7: The Standardized Estimate of Factor Loading for Every Item in
The Measurement Model for Employee’s Workload
Figure 11.7 presents the factor loading for each item in a measurement
model to measure the workload construct. The factor loading for a particular item
is shown near the arrow pointing to the respective item, while the value shown
above each response item is the squared multiple correlation or R2 for that
particular item. The literature suggests that any measuring item having a factor
loading less than 0.6 and an R2 less than 0.4 should be deleted from the
measurement model. However, it depends on the decision by the researcher if he
has strong reason not to drop any of the items in the model.

UniSZA
THE STEPS INVOLVED IN VALIDATING THE MEASUREMENT MODEL
1) Obtain the factor loading for all items in a measurement model

2) Delete items with factor loading less than 0.6 (suggested by the literatures)
3) Delete one item at a time (choose the lowest loading to delete first
4) Re-specify and run a new model after item is deleted (repeat step 1 to 4)
5) Obtain the fitness indexes - to assess how well the data at hand fits the model
6) If the fitness index is not satisfied, look at the modification indices (MI)
7) High value of MI (above 15) indicate the correlated errors between items
(The correlated errors indicate a pair of items is redundant of each other)
To solve the correlated errors or redundant items, the researcher could choose
one of the following two choices:
Choice 1:
i) Delete one item at a time (choose the lower factor loading to delete first)
ii) Re-specify the model and run (repeat step 1 to 7)
Choice 2:
i) Set the identified pair of correlated errors as “free parameter estimate”

ii) Re-specify the model and run (repeat step 1 to 7)
8) Obtain the Cronbach’s Alpha, CR, and AVE for the measurement model
Note: The acceptable value of factor loading, the level of fitness index, and also the
method of modification to the measurement model varies among the literatures.
RUNNING THE CONFIRMATORY FACTOR ANALYSIS
The procedure will assess the meaningfulness of every measurement model

involved in the theoretical framework of study. From here, the researcher could
execute the procedure to assess the unidimensionality, valididy, and reliability of
the measurement model.

UniSZA
The Schematic Diagram of a Model
Let say the schematic diagram of the model under study is given in Figure
11.8 and the measurement model is given in Figure 11.9.
X1
X2
Figure 11.8: The Schematic Diagram of a Model under Study
The Measurement Model Combining all Constructs
Figure 11.9 presented the measurement model for all constructs involved in
the study based on the schematic diagram shown in Figure 11.8. The CFA
procedure is executed based on the measurement model in Figure 11.9. The
researcher instructs AMOS to calculate the factor loading and R2 for every item;
and also the strength of correlation between the constructs.
The item having factor loading less than 0.6 should be deleted from the
measurement model. If the correlation between two exogenous is higher than
0.85, then discriminant validity is not achieved. The output for CFA is presented in
Figure 11.10.

UniSZA
1
X11 e1
1
1
X12 e2
1
X1 X13 e3
1
X14 e4
1
X15 e5
1
X21 e6
1
1
X22 e7
X2
1
X23 e8
1
X24 e9
1
Y1 e10
1
1
Y2 e11
Y
1
Y3 e12
1
Y4 e13
Figure 11.9: The Measurement Model for All Latent Constructs in The Study
The CFA results showing the factor loading and the squared multiple correlations
(R2) for every item. At the same time, the correlation between constructs is
presented in Figure 10.

UniSZA
.38
Factor Loading
X11 e14
2
.62 .32 R
X12 e15
Correlation
.57 .82
.91
X1 X13 e16
.88 .78
X14 e17
.92 .85
X15 e18
.87
.61
X21 e19
.93
.91
.96
X22 e20
.56
X2 .66
.44
X23 e21
.53
.28
X24 e22
.47
.91
Y1 e10
.96
.95
.98
Y2 e11 Fitness Indexes
1. ChiSq = 277.471
Y .94
.89
2. df = 62
Y3 e12 3. ChiSq/df = 4.475
4. GFI = .879
.63 5. AGFI = .823
.80
6. CFI = .943
Y4 e13 7. RMSEA = .105
Figure 11.10: The Factor Loading For All Items of the Respective Construct
As shown in Figure 11.10, the factor loadings for item X12 (from construct X1),
and item X24 (from construct X2) are below 0.6. Thus, the researcher has to delete
these two items and the new measurement model has to be re-specified. These
two items were deleted and the re-specified model is shown in Figure 11.11.

UniSZA
.36
X11 e1
.60
.83
.91
X1 X13 e3
.88 .77
X14 e4
.93
.86
X15 e5
.86
.60
X21 e6
.93
.93
.97 X22 e7
.56
X2 .65
.42
X23 e8
.46
.91
Y1 e10
.96
.95
.98 Fitness Indexes

Y2 e11 1.ChiSq = 122.335
2.df = 41
Y .94
.89
3.ChiSq/df = 2.984
Y3 e12 4.GFI = .935
5.AGFI = .895
.63 6.CFI = .977
.80
7.RMSEA = .079
Y4 e13
Figure 11.11: The Factor Loading For Other Items After X12 and X24 Were
Deleted

UniSZA
Assessing the Fitness of a Measurement Model
The factor loading for all items in Figure 11.11 are above 0.6. Thus no more
item deletion is required for this measurement model. Now, the unidimensionality
requirement has been achieved. The fitness indexes for the model are assessed in
Table 11.5.
Table 11.5: The Summary of Fitness Indexes for the Measurement Model
Name of category Name on index Index value Comments

1. Absolute fit RMSEA 0.079 Achieved the requirement
Absolute fit GFI 0.935 Achieved the requirement
2. Incremental fit CFI 0.977 Achieved the requirement
3. Parsimonious fit Chisq/df 2.984 Achieved the requirement
If the researcher is already satisfied with the fitness indexes obtained, then
the CFA for measurement model is done at this stage. The next step is modeling
the structural model, which will be discussed in next chapter. Otherwise, if the
researcher is still not satisfied with AGFI (less than 0.9), he could examine for
redundant items in the measurement model through the MI table produced by
AMOS. The MI table for the above measurement model is presented in Table
11.6.
Note: The MI value above 15 is considered high and that particular item is deemed to be
redundant.
Table 11.6: The Modification Indices Presents the Covariance between Each
Pair of Items
Parameter
M.I. Comment
Change
e12 <--> e13 26.789 0.127 MI indicates item e12 and e13 are correlated
e10 <--> e13 4.643 -0.044
e5 <--> e7 4.568 -0.032
e1 <--> e7 10.890 0.085
The MI = 26.789 (in bold) is considered high which indicate that item Y3 and
item Y4 are redundant (e12 = item Y3 and e13 = item Y4). As has been discussed
UniSZA
earlier, the researcher has two choices to remedy this problem. The two choices
are:
Choice 1: to delete one of these two redundant items and re-specify the
measurement model.
Choice 2: to set these two correlated errors to be “free parameter estimate” and
re-specify the measurement model.
Let say, the researcher elects to employ choice 2 for this particular case. The
result is presented in the following diagram (Figure 11.12).
.36
X11 e1
.60
.83
.91
X1 X13 e3
.88 .77
.93 X14 e4
.86
X15 e5
.86
.60
X21 e6
.93
.93
.97
X22 e7
.56
X2 .65
.42
X23 e8
.46
.91
Y1 e10
.96
.96
.98 Fitness Indexes

Y2 e11 1. ChiSq = 93.747
2. df = 40
Y .94
.88
3. ChiSq/df = 2.344
Y3 e12 4. GFI = .949
5. AGFI = .917
.61 .33 6. CFI = .985
.78
7. RMSEA = .065
Y4 e13
Figure 11.12: The Re-Specified Model after a Pair of E12 and E13 is Set as
“Free Estimate”

UniSZA
The fitness indexes for the model are assessed in the following table (Table 11.7).
Table 11.7: The Fitness Indexes for Measurement Model in Figure 12
Name on
Name of index category Index value Comments
index
1. Absolute fit RMSEA 0.065 Achieved the requirement
Absolute fit GFI 0.949 Achieved the requirement
2. Incremental fit CFI 0.985 Achieved the requirement
3. Parsimonious fit Chisq/df 2.344 Achieved the requirement
Note: Look how the fitness indexes have improved after the model re-specification was made.
At this stage, the CFA for the measurement model is completed. And as
usual, the business now is to address the issues of unidimensionality, validity, and
reliability before moving to next stage (modeling the structural model).
Unidimensionality: This requirement has been achieved through the item-

deletion process (also through item-constraining) and model re-specification. All
processes have been demonstrated. The correlation between exogenous
construct is less than 0.85.
Validity: This requirement could be achieved through the following processes:

1. Convergent
AVE > 0.50: Refer to the following table (Table 8)
Validity:
2. Construct Validity: All fitness indexes for the models meet the required
level
3. Discriminant All redundant items are either deleted or constrained,
Validity: and the correlation between exogenous constructs is <
0.85
Reliability: This requirement could be achieved through the following processes:

1.Internal Reliability Cronbach Alpha > 0.70: Refer to Table 8 (use SPSS)
2. CR CR > 0.60: Refer to Table 8 (calculate using a
formula)
3. AVE AVE > 0.50: Refer to Table 8 (calculate using a
formula)

UniSZA
Table 11.8: The Suggested CFA Results Reporting For The Measurement
Model
CFA Results Reporting
Cronbach CR AVE
Factor
Construct Item Alpha (Above (Above
Loading
(Above 0.7) 0.5) 0.6)
X1 X11 0.60 0.893 0.904 0.707
X13 0.91
X14 0.88
X15 0.93
X2 X21 0.93 0.869 0.894 0.743
X22 0.97
X23 0.65
Y Y1 0.96 0.946 0.955 0.844
Y2 0.98
Y3 0.94
Y4 0.78
Table 11.9: The CFA Results Summary for Discriminant Validity
Discriminant Validity
Construct X1 X2 Y
X1 0.83
X2 0.60 0.85
Y 0.56 0.46 0.92
The diagonal values in bold is the square root of AVE while other values are
the correlation between the respective constructs. The discriminant validity is
achieved when a diagonal value in bold is higher than the values in its row and
column.

UniSZA
THE ASSESSMENT OF NORMALITY FOR DATASET
Once the issues of unidimensionality, validity, and reliability have been addressed,
the researcher needs to examine the normality assessment for the data at hand
before proceed to modeling the structural model. Using the final measurement
model, select the output box named “test for normality and outliers” in order to
assess the distribution for every variable in a dataset. The following table presents
the resulted output from the abovementioned procedure. Table 11.10 presents the
normality assessment for every item involved in the measurement model.
Table 11.10: The Assessment of Normality for the Data
Variable min max skewness c.r. kurtosis c.r.

Y4 1.000 7.000 -0.561 -4.077 -0.230 -0.837
Y3 1.000 7.000 -0.786 -5.713 0.297 1.078
Y2 2.000 7.000 -0.566 -4.116 -0.274 -0.995
Y1 2.000 7.000 -0.505 -3.672 -0.429 -1.559
X23 1.000 7.000 -0.690 -5.015 0.124 0.450
X22 2.000 7.000 -0.362 -2.632 -0.165 -0.600
X21 2.000 7.000 -0.400 -2.909 -0.220 -0.799
X15 2.000 7.000 -0.507 -3.683 -0.386 -1.403
X14 2.000 7.000 -0.535 -3.891 -0.022 -0.078
X13 2.000 7.000 -0.475 -3.453 -0.437 -1.590
X11 1.000 7.000 -0.109 -.792 -0.158 -0.574
Multivariate 22.887 12.048
The researcher needs to observe two criteria in the normality assessment:
1. The measure of skewness which reflects the normality assessment for

every item. The value of skewness should fall within the range of -1.0 to 1.0
to indicate normal distribution; otherwise the data distribution for the
respective items departs from normality. The researcher should examine the
outliers, and delete certain number of extreme outliers in dataset. AMOS

UniSZA
output tabulates the list of outliers once the box “test for normality and
outliers” is selected.
2. The value of multivariate kurtosis reflects the multivariate normality
distribution for data set. The value of multivariate kurtosis should be lower
than 50.0; otherwise the assumption of multivariate normality is not satisfied.
In the case where the assumption is not met, the researcher should examine
the data outliers toward the model. The deletion of few extreme outliers might
help improve the multivariate normality. The outlines in AMOS are shown by
the Mahalanobis Distance. AMOS output will tabulate the list of observations
and their respective distance from the centroid. The centroid is the centre of
data distribution.
Otherwise, if the deletion process does not work the researcher needs to
employ bootstrapping technique in order to confirm multivariate normality. If
the bootstrapping method still fails to confirm multivariate normality of data
set, the researcher should abandon using SEM.
ANALYZING THE STRUCTURAL MODEL
First of all, refer to the theorized inter-relationships between the constructs. Most
of the times, the inter-relationships between constructs is presented in the
schematic diagram as shown below. Based on the schematic diagram, draw the
AMOS graphic for the structural model as shown in Figure 11.12.
X1
X2

UniSZA
1
e1 X11
1
1
e3 X13 X1 e14
1
1
Y1 e10
e4 X14 1 1
1
1
Y2 e11
e5 X15
Y
1
1 Y3 e12
e6 X21
1
1
1 Y4 e13
e7 X22
X2
1
e8 X23
Figure 11.12: The AMOS Graphic of the Structural Model
Based on the structural model in Figure 11.12, the researcher could instruct AMOS
to compute the values for the Standardized Estimate and also the Unstandardized
Estimate for the model. The values obtained under Standardized Estimate are
presented in Figure 11.13, while the values under the Understandardized Estimate
are presented in Figure 11.14.

UniSZA
.36
e1 X11
.60
.83
.91
e3 X13 X1 e14
.88 .91
.77 Y1 e10
e4 X14 .45 .96
.93
.98 .96
.86 Y2 e11
e5 X15
.60 .34 Y
.88
.86
.19 .94 Y3 e12
e6 X21
.93 .33
.78 .61
.93 Fitness Indexes Y4 e13
e7 X22 .97
1 ChiSq = 93.747
2 df = 40
.42 X2 3 ChiSq/df = 2.344
e8 X23 4 GFI = .949
.65
5 AGFI = .917
6 CFI = .985
7 RMSEA = .065
Figure 11.13: The Standardized Regression Weights (The Standardized Beta

Estimate)
Note:
The standardized beta estimate of X1 is 0.45, and the standardized beta estimate
for X2 is 0.19
The measure of correlation between construct X1 and X2 is 0.60.
The value of R2 is 0.34 – which indicate the contribution of construct X1 and

construct X2 in estimating Y is 34%.

UniSZA
.87
1
e1 X11
1.00
.23 .76
1 1.51
e3 X13 X1 .49
e14 .11
.28 1.40 1
1
Y1 e10
e4 X14 .68 1 1.00
.05
1.59
.20 1.03 1
1
Y2 e11
e5 X15
.45 Y .19
.20 1
1 .19 1.09 Y3 e12
e6 X21
1.00 .93 .14
1.12
.09 1
1 Fitness Indexes Y4 e13
e7 X22 1.00
1 ChiSq = 93.747
2 df = 40
1.07 X2 1.19 3 ChiSq/df = 2.344
1
e8 X23 4 GFI = .949
.81 5 AGFI = .917
6 CFI = .985
7 RMSEA = .065
Figure 11.14: The Unstandardized Regression Weights (Beta Estimate in Its

Actual Unit)
Table 11.11: The Fitness Index Assessment for the Structural Model in
Figure 11.14
Name of category Name of index Index value Comments

1. Absolute fit RMSEA 0.065 Achieved the required fit level
Absolute fit GFI 0.949 Achieved the required fit level
2. Incremental fit CFI 0.985 Achieved the required fit level
3. Parsimonious fit Chisq/df 2.944 Achieved the required fit level

UniSZA
INTERPRETING THE AMOS OUTPUT FOR STRUCTURAL MODEL
AMOS would give two sets of text output namely the standardized estimate, and
unstandardized estimate. The text output explaining results in Figure 11.13 are
presented in Table 11.11, Table 11.12, and Table 11.13. While the text output
explaining the results in Figure 11.14 are presented in Table 11.14, Table 11.15,
and Table 11.16.
Interpreting the text output: The Standardized Regression Weights
Table 11.12: The Standardized Regression Weight as shown in Figure 11.13
Path Path Standardized Beta Estimate

Y <--- X2 0.190
Y <--- X1 0.450
Y1 <--- Y 0.956
Y2 <--- Y 0.980
Y3 <--- Y 0.937
Y4 <--- Y 0.780
X15 <--- X1 0.928
X14 <--- X1 0.880
X13 <--- X1 0.909
X11 <--- X1 0.597
X23 <--- X2 0.651
X22 <--- X2 0.965
X21 <--- X2 0.927
The definition for the standardized regression weight is given below.
Extracted from Table 11.12:
Y <--- X2 0.19
When X2 goes up by 1 standard deviation, Y goes up by 0.19 standard deviations.

UniSZA
Extracted from Table 11.12:
Y <--- X1 0.45
When X1 goes up by 1 standard deviation, Y goes up by 0.45 standard deviations.
Table 11.13: The Correlation Estimate for Each Pair of Exogenous Construct
Construct Construct Estimate

X1 <--> X2 0.60
The correlation between latent construct X1 and X2 is estimated to be 0.60
*The measure of correlation indicates that the strength of relationship

between the two latent exogenous constructs (X1 and X2) is not strong. In this
case, the researcher could continue with further analysis.
***Caution: If the measure of correlation is high (above 0.85) which indicates

that the two latent exogenous constructs (X1 and X2) are strongly correlated, and
the discriminant validity failed. Thus the researcher could not treat the two
exogenous constructs separately since one is like the mirror of the other. In this
case, the researcher should drop one of these two exogenous constructs (X1 and
X2) for further analysis.
Table 11.14: The Squared Multiple Correlations
Variable Estimate (R2)

Y 0.341
Y4 0.633
Y3 0.890
Y2 0.954
Y1 0.913
X15 0.852
X14 0.780
X13 0.819
X12 0.323
UniSZA
Variable 2
Estimate (R )
X11 0.381
X24 0.280
X23 0.440
X22 0.914
X21 0.868
The definition for the squared multiple correlations are given below.
Extracted from Table 11.14: The Squared Multiple Correlations
Y 0.341
It is estimated that the predictors of Y explain 34.1 percent of its variance. In other
words, the error variance of Y is approximately 65.9 percent of the variance of Y
itself.
Y4 0.633
It is estimated that the predictors of Y4 explain 63.3 percent of its variance. In other
words, the error variance of Y4 is approximately 36.7 percent of the variance of Y4
itself.
X15 0.852
It is estimated that the predictors of X15 explain 85.2 percent of its variance. In
other words, the error variance of X15 is approximately 14.8 percent of the variance
of X15 itself.

UniSZA
Interpreting the Text Output: The Unstandardized Regression Weights
Table 11.15: The Regression Weights for the respective construct
The actual
S.E. C.R. P-value
beta values
Y <--- X1 0.662 0.106 6.233 ***
Y <--- X2 0.188 0.062 3.052 0.002
***indicate a highly significant at < 0.001
Extracted from Table 11.15: Estimate of regression weight for X2
The actual
S.E. C.R. P-value
beta values
Y <--- X2 0.188 0.062 3.052 0.002
Definition for beta 0.188: When X2 goes up by 1unit, Y goes up by 0.188 unit.
Definition for Standard Error of regression weight 0.062: The regression

weight estimate of 0.188 has a standard error of about 0.062.
Definition for Critical Ratio for regression weight: Dividing the regression
weight estimate by the estimate of its standard error gives
z = 0.188/0.062 = 3.052. In other words, the regression weight estimate is 3.052
standard errors above zero.
Definition for Level of significance for regression weight: The probability

of getting a critical ratio as large as 3.052 in absolute value is 0.002. In other
words, the regression weight for X2 in the prediction of Y is significantly different
from zero at the 0.05 level.

UniSZA
Extracted from Table 11.15: Estimate of regression weight for X1
The actual
S.E. C.R. P
beta values
Y <--- X1 0.662 0.106 6.233 ***
Definition for beta 0.662: When X1 goes up by 1 unit, Y goes up by 0.662 units.
Definition for Standard Error of regression weight 0.106: The regression

weight estimate, 0.662 has a standard error of about 0.106.
Definition for Critical Ratio for regression weight: Dividing the regression
weight estimate by the estimate of its standard error gives z = 0.662/0.106 =
6.233. In other words, the regression weight estimate is 6.233 standard
errors above zero.
Definition for Level of significance for regression weight: The probability

of getting a critical ratio as large as 6.233 in absolute value is less than 0.001. In
other words, the regression weight for X1 in the prediction of Y is significantly
different from zero at the 0.001 level.
Table 11.16: The Regression Weight for each item. Note the reference point.
Beta Estimate S.E. C.R. P

X21 <--- X2 1.000 Reference point
X22 <--- X2 0.987 0.034 28.691 ***
X23 <--- X2 0.826 0.057 14.382 ***
X24 <--- X2 0.663 0.063 10.455 ***
X11 <--- X1 1.000 Reference point
X12 <--- X1 1.057 0.119 8.910 ***
X13 <--- X1 1.457 0.116 12.601 ***
X14 <--- X1 1.359 0.110 12.409 ***
X15 <--- X1 1.528 0.120 12.744 ***
Y1 <--- Y 1.000 Reference point
Y2 <--- Y 1.031 0.023 44.765 ***
Y3 <--- Y 1.102 0.030 37.145 ***
Y4 <--- Y 1.148 0.054 21.313 ***
***indicate a highly significant at < 0.001.
Note: Since all items are significant, one could conclude that the convergent validity is achieved.

UniSZA
Table 11.17: Covariance Estimate between Variables
Estimate S.E. C.R. P

X2 <--> X1 0.480 0.065 7.357 ***
The covariance between X1 and X2 is estimated to be 0.480.
Standard error of covariance
The covariance estimate, 0.480, has a standard error of about 0.065.
Critical ratio for covariance
Dividing the covariance estimate by the estimate of its standard error gives
z = 0.480/0.065 = 7.357
In other words, the covariance estimate is 7.357 standard errors above zero.
Level of significance for covariance
The probability of getting a critical ratio as large as 7.357 in absolute value is less
than 0.001. In other words, the covariance between X1 and X2 is significantly
different from zero at the 0.001 level.
Table 11.18: The Variance Estimate for Each Variable in the Model
Variable Variance
S.E. C.R. P-value
Estimate
X2 1.200 0.112 10.695 ***
X1 0.516 0.087 5.942 ***
e14 0.749 0.067 11.133 ***
e6 0.183 0.030 6.081 ***
e7 0.110 0.027 4.072 ***
e8 1.039 0.086 12.028 ***
e9 1.352 0.110 12.307 ***
e1 0.839 0.069 12.075 ***
e2 1.212 0.099 12.188 ***

UniSZA
Variable Variance
S.E. C.R. P-value
Estimate
e3 0.241 0.028 8.678 ***
e4 0.268 0.028 9.572 ***
e5 0.210 0.027 7.691 ***
e10 0.108 0.012 8.830 ***
e11 0.059 0.010 5.785 ***
e12 0.171 0.017 9.823 ***
e13 0.870 0.072 12.034 ***
***indicate a highly significant at < 0.001.
Table 11.18 presents the variance estimate for every variable in the model.
The probability or p-values for all variables are less than 0.001 which indicate that
the variances for all variables are significantly different from zero.

UniSZA
CHAPTER
ANALYZING THE MEDIATING
12
VARIABLES
WHAT IS MEDIATING VARIABLE?
Mediating variable is the variable that mediates the effect from an independent
variable to its dependent variable. If the direct effect from variable X to variable Y
does not exist, instead the effect exist indirectly through another variable M; then in
this case M is a mediating variable. The position of mediating variable in the model is
illustrated below.
Mediating
Variable
Independent Dependent
Variable Variable
First of all, the direct effect of independent variable on dependent variable is significant.
However, when the mediating variable enters the model, the direct effect is reduced but
the indirect effect is significant. If direct effect is reduced but still significant, then the
“partial mediation” occurs. However, if the direct effect is no longer significant, then the
“complete mediation” is said to occur.
As illustrated above: X1 is an independent variable; Y is a dependent variable, and

X2 is a mediating variable in the model. The single-headed arrow indicates a causal
effect of X1 on Y, X1 on X2, and X2 on Y. In this case, X2 is a mediator variable since it
has two arrows – one arrow pointing in and the other one is pointing out. In the diagram,
the researcher can examine the direct effect of X1 on Y, the direct effect of X1 on X2,
and the direct effect of X2 on Y.
303
In this model, the researcher can assess the direct effect as well as the indirect
effect of X1 on Y. If the direct effect of X1 on Y is not significant, but instead, the indirect
effect through X2 is significant, then X2 is a mediating variable in the X1 to Y relationship.
ANALYZING THE MEDIATING EFFECTS FOR THE OBSERVED VARIABLES
Firstly, we begin by modeling the single effect of X1 on Y as shown in Figure 12.1. We

assess the direct effect of X1 on Y as shown in Figure 12.2. The output in Table 12.1
shows B1 is 0.361 and it has significant effect on Y (p-value < 0.001).
0,
e1
X1 Y
Figure 12.1: Modeling the Direct Effect Of X1 on Y
0, 1.07
e1
1
4.90, 1.36 3.87
.36
X1 Y
Figure 12.2: The Result of Direct Effect Of X1 (The Beta Coefficient Is 0.36)
304
Table 12.1: The direct effect of X1 on Y is significant (Beta coefficient 0.361)
Estimate S.E. C.R. P-value Result

Y <--- X1 0.361 0.050 7.235 0.001 Significant
Secondly, we include the mediator X2 into the model as shown in Figure 12.3.
Now we assess the direct effect of X1 on Y as shown in Figure 12.4. The output in Table
1a shows B1 is reduced from 0.361 to 0.103, and the hypothesis is no longer significant.
Here, the requirement for complete mediation is met. However, we need to test the
hypothesis for B2 and B3 to confirm.
1
X2 e2
e1
1
X1 Y
Figure 12.3: The Modeling of Mediating Variable X2 In AMOS For Observed

Variables
305
3.10
0, 1.12
X2 1
e2
0, .98
.53 .31
e1
1
4.90, 1.36 3.01
.10
X1 Y
Figure 12.1: The AMOS Output Showing Regression Coefficient between

Variables
Table 12.2: The Direct Effect as well as the indirect effect of X1 on Y

Y <--- X1 0.103 0.066 1.547 0.062 Not significant
X2 <--- X1 0.527 0.041 12.991 0.001 significant
Y <--- X2 0.312 0.050 6.282 0.001 significant
1. Hypothesis 1: X1 has significant and direct effects on Y
Refer to Table 12.2. The p-value for the test is 0.062. Thus, the regression weight
for X1 in the prediction of Y is not significant at the 0.05 level. In other words, the
direct effect of X1 on Y is not significant.
2. Hypothesis 2: X1 has significant and direct effects on X2
Refer to Table 12.2. The p-value is less than 0.001. Thus, the regression weight for
X1 in the prediction of X2 is significant at the 0.001 level. In other words, the direct
effect of X1 on X2 is significant.
306
Refer to Table 12.2. The p-value is less than 0.001. Thus, the regression weight
for X2 in the prediction of Y is significant at the 0.001 level. In other words, the
direct effect of X2 on Y is significant.
Conclusion: X2 is a mediating variable in the relationship between X1 and Y.

The type of mediation here is a called a “complete mediation” since the direct
effect of X1 on Y is not significant after X2 entered the model. Instead, the indirect
effect is significant. Thus, X1 has an indirect effect on Y through mediator variable
X2
Analyzing the Mediating Effects in for Latent Constructs
The concept of mediation in latent constructs is equivalent to that of the observed

variables as discussed above. First of all, the direct effect of X1 on Y is significant. Refer
to Figure 12.5, Figure 12.6, and the test of hypothesis concerning B1.
e14
1
Y1 e10
1 1
1
Y2 e11
Y
X1 1
Y3 e12
1 1
Y4 e13
X15 X14 X13 X11
1 1 1 1
e5 e4 e3 e1
Figure 12.2: Modeling the direct effect of X1 on Y
307
.78
e14 .11
1
Y1 e10
1 1.00
.06
.47 1.03
1
Y2 e11
.87 Y .17
1.10
X1 1
Y3 e12
1.15
1.61 1.00 .87
1.42 1.53
1
Y4 e13
Fitness Indexes
X15 X14 X13 X11 1.ChiSq = 55.676
2.df = 19
1 1 1 1 3.ChiSq/df = 2.930
.19 .28 .23 .88 4.GFI = .956
5.AGFI = .917
e5 e4 e3 e1 6.CFI = .986
7.RMSEA = .078
Figure 12.3: The direct effect of X1 on Y (Beta coefficient 0.87)
Table 12.3: The Direct Effect Of X1 on Y Is Significant (Beta Coefficient 0.868)
Beta Estimate S.E. C.R. P-value Result

Y <--- X1 0.868 0.104 8.365 0.001 Significant
The analysis for mediation begins by showing that the direct effect of X 1 on Y is
significant. The direct effect is measured through beta coefficient. In this case, B 1 is
significant (Table 12.3). When the mediating variable X2 enters the model, the value of
beta coefficient for X1 is expected to decrease, or in other words the direct effect of X 1
on Y would be reduced. The model when the mediator X2 is enters is presented in
Figure 12.7.
308
e6 e7 e8 e9
1 1 1 1
X21 X22 X23 X24
1
e15 X2
e14
1
Y1 e10
1 1
1
Y2 e11
Y
X1 1
Y3 e12
1
1
Y4 e13
X15 X14 X13 X11
1 1 1 1
e5 e4 e3 e1
Figure 12.4: The representations of mediating for latent construct X2 in AMOS

graphic
In the above diagram, X1, X2, and Y are latent constructs. The study is interested
to prove that the construct X2 is mediating the relationship between construct X1 and Y.
The output for unstandardized estimates for the model is given in Figure 12.8,
while the text output is presented in Table 12.4.
309
.45
.19 .09 1.07 1.39
e6 e7 e8 e9
1 1 1 1
X21 X22 X23 X24
1.00 .81
1.00
.64
1 .75
.77 e15 X2
e14 .11
1
.94 .19 Y1 e10
1 1.00
.06
.48 1.03
1
Y2 e11
.68 Y .17
1.10
X1 1
Y3 e12
1.59 1.15
1.00 .87
1.40 1.51
1
Y4 e13
Fitness Indexes
X15 X14 X13 X11 1.ChiSq = 151.309
2.df = 50
1 1 1 1 3.ChiSq/df = 3.026
.19 .28 .23 .87 4.GFI = .929
5.AGFI = .889
e5 e4 e3 e1 6.CFI = .972
7.RMSEA = .080
Figure 12.5: The AMOS Output Showing Regression Coefficient between The
Constructs
Note: All fitness indexes achieved the required level

***The value of beta coefficient linking X1 to Y is reduced from 0.87 to 0.68.
310
Table 12.4: The Multiple Regression Weights
Beta Estimate S.E. C.R. P-value Result

Y <--- X1 0.682 0.110 6.202 0.001 Significant
X2 <--- X1 0.941 0.108 8.712 0.001 Significant
Y <--- X2 0.187 0.061 3.073 0.002 Significant
Three hypotheses are required in order to assess the mediating effect
Refer to Table 12.4: The p-value for the test is 0.001. Thus, the regression weight
for X1 in the prediction of Y is significant at the 0.001. In other words, the direct
effect of X1 on Y is significant.
2. Hypothesis 2: X1 has significant and direct effects on X2
for X1 in the prediction of X2 is significant at the 0.001. In other words, the direct
effect of X1 on X2 is significant.
for X2 in the prediction of Y is significant at the 0.01. In other words, the direct
effect of X2 on Y is significant.
Conclusion: The construct X2 does mediate the relationship between X1 and Y.

The type of mediation here is a called a “partial mediation” since the direct effect
of X1 on Y is still significant after X2 entered the model even though the beta
coefficient for X1 is reduced from 0.87 (Figure 12.6) to 0.68 (Figure 12.8).
311
CHAPTER
ANALYZING THE MODERATING VARIABLES
13
WHAT IS MODERATING VARIABLE?
Moderating variable is the variable that moderates the strength of causal effects from
an independent variable X to its dependent variable Y. Let M be the moderator
variable in the X-Y relationship. The moderation effects of M would “alter” the causal
effects of independent variable X on dependent variable Y.
We begin by assuming that the causal effect of independent variable X on its dependent
variable Y is significant. When the moderating variable M is introduced into the model,
the causal effects would change due to some interaction effect between independent
variable X and moderator variable M. The effects of X on Y could be no longer
significant or it could be more significant after interaction. In other words, the moderator
variable M is “moderating” the strength of X-Y relationship.
Example: Let X be the amount of environmental news concerning the safe and
clean environment. The news intends to make public aware of environmental
degradation and that they save the environment by switching to green products. Let Y
be the respondents’ intention to consume green products, and let M be their level of
education. If the effect of environmental news (X) in educating the public to purchase
green products (Y) is more visible among higher educated group compared to low
educated group, then we can say that level of education (M) is the moderating variable
in the relationship between independent variable and dependent variable.
Figure 13.1 illustrates the position of moderating variable (Education) in the in the
relationship between independent variable (Environmental Awareness Campaign) and
the resulting dependent variable (Intention to Purchase Green Products).
311
Environmental Intention to
Awareness Purchase
Campaign Green Products
Level of
Education
Figure 13.1: The Schematic Diagram Showing The Position Of Independent

Variable, Dependent Variable And Moderating Variable In A Model
THE INTERACTION EFFECTS OF MODERATING VARIABLE
As has been said earlier, although moderation implies a weakening of a causal effect, a
moderator can also reverse or enhance that particular effect. A key part of moderation
is the measurement of causal effect X on Y for different values of moderating variable
M. In statistics, the effect of X on Y for a fixed value of M is referred as the simple effect
of independent variable on its dependent variable.
Let X = independent variable, Y = dependent variable, M = moderating variable.

Let assume that the linear relationship between X and Y is significant; and the model is
additive. Then, the moderation effect is modeled in the multiple regression equation as
follows:
Y = B0 + B1X + B2M + B3XM + e
The coefficient B3 measures the interaction effect between independent variable

X and moderating variable M. Note that coefficient B1 measures the simple effect of X
when M = 0 (no interaction effects involved). Then, the test of moderation is
operationalized by the product term XM.
In order to prove that M is a moderating variable in the X-Y relationship, the

researcher needs to test the hypothesis that the coefficient B3 is significantly different
from zero.
312
Types of moderation
1) Complete Moderation
A complete moderation is said to occur if the causal effect of independent variable X

on dependent variable Y would change from “significant” to “non significant” when
the moderator variable M comes into the model. In other words, B1 is not
significant and B3 is significant.
2) Partial Moderation
If the causal effect of independent variable X on dependent variable Y is still

significant, then the partial moderation is said to exist in the X-Y relationship. In
other words, B1 is still significant and B3 is also significant.
3) No Moderation
If the test of hypothesis fails to prove that B3 is significant then there is no

moderation effect of M on the X-Y relationship.
MODELING THE MODERATING EFFECTS FOR OBSERVED VARIABLES

Figure 13.2 illustrates how the observed variables are represented in AMOS graphic. In
the diagram, X is the independent variable, M is the moderating variable, while XM is
the product of X and M. The above information can be modeled in the following
regression equation: Y = B0 + B1X + B2M + B3XM + e1 and can be converted into
AMOS model as shown in Figure 13.2.
X
e1
1
M Y
XM
Figure 13.2: The Representation of Moderating Variable M in AMOS Graphic for

Observed Variable
313
As shown in Figure 13.2, three hypothesis testing is required in the model namely:
1. The X-Y relationship (testing for B1) – we indicate as Hypothesis 1

2. The M-Y relationship (testing for B2) – we indicate as Hypothesis 2
3. The XM-Y relationship (testing for B3) – we indicate as Hypothesis 3
The moderating effects of variable M is said to exist in the X-Y relationship if the XM - Y
relationship (B3) or Hypothesis 3 is significant while the M-Y relationship (B2) or
Hypothesis 2 is not significant.
As for B1 or Hypothesis 1, there are two possibilities could occur:
1) If Hypothesis 1 is not significant – then the complete moderation occurs

2) If Hypothesis 1 is significant – then the partial moderation occurs.
314
ANALYZING THE MODERATING EFFECTS FOR OBSERVED VARIABLES
Suppose the researcher is interested to assess the moderating effects of age of workers
in the relationship between the monetary incentives given to them and their monthly
productivity. The AMOS model illustrating the researcher’s theoretical argument is given
in Figure 3. All variables in the model are directly observed.
Incentives
e1
1
Age Productivity
I.A
Figure 13.3: The Model Representation in AMOS Graphic for Testing Age as
Moderator Variable
Key: Independent variable = Monetary Incentives (can be interval or ratio scale)

Moderating variable = Age of workers (can be nominal or ordinal scale)
Dependent variable = Monthly Productivity (can be interval or ratio scale)
The corresponding AMOS output for the above model is given in Figure 4
315
.48
Incentives .60
.59 e1
.24 1
.49
Age .89 Productivity

.66
.68 -.88
.20
I.A
Figure 13.4: The AMOS Output Showing the Beta Coefficients, Variance, and
Covariance
Hypothesis 1: The Monetary Incentives has significant effects on their productivity
Table 13.1: Testing the causal effects of Incentives on Productivity

Productivity <--- Incentives 0.59 0.068 8.636 *** Significant
Table 13.1 show the p-value is less than 0.001 (***). Thus, the regression weight for
Incentives in the prediction of Productivity is significant.
In other words, the hypothesis that the causal effects of incentives on
productivity are significant has been supported.
316
Hypothesis 2: The Workers’ age level has significant effects on their productivity
Table 13.2: Testing the causal effects of Workers’ Age on their Productivity
Estimate S.E. C.R. P Result

Productivity <--- Age 0.89 0.61 1.451 0.072 Not Significant
Table13.2 show the p-value is 0.072. Thus, the regression weight for Age in the
prediction of Productivity is not significant.
In other words, the hypothesis that the causal effects of age on productivity are
significant is not supported.
Hypothesis 3: The workers’ age moderates the effects of incentives on productivity
Table 3: Testing the moderating effects of Incentives*Age on Productivity

Estimate S.E. C.R. P Result
Productivity <--- I.A -0.88 0.186 -4.742 *** Significant
Table 13.3 show the p-value is 0.001 (***). Thus, the regression weight for I.A in the
prediction of Productivity is significant.
In other words, the hypothesis that the moderating effects of workers’ age (M) on
the relationship between incentives (X) and their productivity (Y) are significant
has been supported.
Note: The type of moderation that occurs in this case is partial moderation since
the hypothesis for the main effect is also significant.
317
MODELING THE MODERATING EFFECTS FOR LATENT CONSTRUCTS
Analyzing the moderating effects for the model with latent constructs is very
complicated. The normal modeling procedure in Regression Analysis (using interaction
terms between independent and moderator variable) is not practical with latent
constructs since it would cause problems with model convergence as well as distortion
of standard errors. In the end, it resulted in model misfit and the procedure stops.
The position of moderator in the relationship between X1 and Y is shown in Figure 13.5.
e1 e2 e3 e4 e5 e6 e7 e8 e9 e10
1 1 1 1 1 1 1 1 1 1
X11 X12 X13 X14 X15 Y1 Y2 Y3 Y4 Y5
1 1
1
X1 Y e11
Moderator
Figure 13.5: Modeling the Moderator Variable in the Path Between X1 and Y
Alternatively, the Multi-Group Confirmatory Factor Analysis (Multi-Group CFA) has

been suggested as an alternative method for assessing the effect of moderator variable
in the model. The researcher only needs to identify the path of interest where the
moderator variable is to be assessed. This particular path would be constrained with
parameter = 1 and the model is termed as the constrained model. The procedure will
estimate two models separately. One is the constrained model (with parameter is
constrained to 1) while the other one is the unconstrained model (with no parameter
constrained). The step by step process of multi-group CFA is discussed below.
318
ANALYZING THE MODERATOR FOR LATENT CONSTRUCTS: THE MULTI-GROUP
CFA
There are few steps involved in performing multi-group CFA:
1) Split data equally into two groups based on the median of the moderator variable.
2) Save data into two separate files. Rename the two groups as group 1 and group 2.
3) Obtain the final model for the study. Identify the path of interest where the effect of
moderator variable is to be assessed.
4) Develop two separate AMOS models. Rename as model 1 and model 2 respectively.
5) In Model 1, put a constraint to fix the relationship parameter between constructs of

interest to be equal to 1. Rename model 1 as the constrained model.
6) In Model 2, allow the relationship parameter to be freely estimated. Rename model 2

as the unconstrained model.
7) Execute model 1 (constrained model) using both data 1 and data 2 separately.
8) Execute model 2 (unconstrained model) using both data 1 and data 2 separately.
9) Obtain the difference of Chi-Square value between the constrained model and the
unconstrained model separately for data 1 and data 2.
10) If the Chi-Square value between constrained and unconstrained model differs by
more than 3.84, then the moderator variable has a significant moderating effect on
the relationship between the constructs in the selected path.
Example: Suppose that we are modeling the effect of two independent variables
namely X1 and X2 on a dependent variable namely Y (Figure 13.6). One of the
objectives of this research is to assess the moderation effect of moderator variable
namely education in the relationship path between X1 and Y.
319
e1 e3 e4 e5
1 1 1 1
X11 X13 X14 X15 (Moderator)

Path of interest
1
e14
X1 1
Y1 e10
1 1
1
Y2 e11
Y
1
Y3 e12
1
Y4 e13
X2
X24 X23 X22 X21
1 1 1 1
e9 e8 e7 e6
Figure 13.6: The Path Of Interest To The Researcher Here Is From X1 To Y Where
The Effect Of Moderator Variable Namely Education Is To Be
Examined.
The path of interest where the moderation tests to be carried out is shown in
Figure 13.6. First of all, sort data in ascending order based on respondents’ level of
education. Then the data is split equally based on the median of education and saved
into two separate data files. Data 1 is renamed as low education group, while data 2 is
renamed as high education group.
320
Secondly, put a constraint on the path of interest in the model to be equal to “1” as
shown in Figure 13.7. This model is renamed as the constrained model.
e1 e3 e4 e5
1 1 1 1
X11 X13 X14 X15

Constrained
1
Parameter = 1
constrained e14
X1 1
Y1 e10
1 1
1
1
Y2 e11
Y
1
Y3 e12
1
Y4 e13
X2
1
X24 X23 X22 X21
1 1 1 1
e9 e8 e7 e6
Figure 13.7: The Constrained Model: The Coefficient In The Path Of Interest (X1 To
Y) Where The Effect Of Moderator Variable To Be Examined Is
Constrained To 1.
Thirdly, using the same model, remove the constraint in the path as shown in Figure
13.8. This model is renamed as the unconstrained model.
321
e1 e3 e4 e5
1 1 1 1
Parameter
X11 X13 X14 X15
Unconstrained
1
e14
X1 1
Y1 e10
1 1
1
Y2 e11
Y
1
Y3 e12
1
Y4 e13
X2
X24 X23 X22 X21
1 1 1 1
e9 e8 e7 e6
Figure 13.8: The Unconstrained Model: The Coefficient In The Path Of Interest (X1
To Y) Where The Effect Of Moderator Variable To Be Examined Is
Not Constrained.
Finally, obtain the estimate for both the constrained model and also the
unconstrained model using the same dataset (data 1: low education group). The output
is presented in Figure 13.9 for the constrained model and in Figure 13.10 for the
unconstrained model. The procedure for testing moderation is carried out as shown in
Table 13.4.
322
.87 .19 .24 .25
e1 e3 e4 e5
1 1 1 1
Constrained Slope
X11 X13 X14 X15
1.78
1.00 1.65
.73
1.76
e14 .09
X1 1
Y1 e10
1 1.00
1.00
.39 .04
1
1.02 Y2 e11
.43 Y 1.09
.20
1
Y3 e12
.07 1.05
1.31 .72 .12
1
Y4 e13
X2 Fitness Indexes
1. ChiSq = 107.979 Chi-Square
1.00 2. df = 50
.62 .83 1.01 and its df
3. ChiSq/df = 2.160
4. GFI = .905
5. AGFI = .851
X24 X23 X22 X21 6. CFI = .970
7. RMSEA = .086
1 1 1 1
1.41 .96 .04 .18
e9 e8 e7 e6
.51
Figure 13.9: Data 1: The Constrained Model. The Value of Chi-Square Is 107.979,
Df=50
323
.86 .19 .24 .25
e1 e3 e4 e5
1 1 1 1
Unconstrained Slope
X11 X13 X14 X15
1.55
1.00 1.44
.72
1.54
e14 .09
X1 1
Y1 e10
1 1.00
.69
.51 .04
1
1.03 Y2 e11
.49 Y 1.09
.20
1
Y3 e12
.14 1.06
1.31 .72 .12
1
Y4 e13
X2 Fitness Indexes
1. ChiSq = 104.520
.62
1.00 2. df = 49 Chi-Square
.83 1.01 3. ChiSq/df = 2.133 and its df
4. GFI = .906
5. AGFI = .851
X24 X23 X22 X21 6. CFI = .971
7. RMSEA = .085
1 1 1 1
1.41 .96 .05 .18
e9 e8 e7 e6
.51
Figure 13.10: Data 1: The Unconstrained Model. The Value of Chi-Square Is 104.520, Df=49
324
Table 13.4: The Measures Required For Performing Moderation Test Using Data 1
Constrained Unconstrained Chi-Square Result on Result on

Model Model Difference Moderation Hypothesis
Chi-Square 107.979 104.520 3.459 Not Significant Not Supported
DF 50 49 1
GFI 0.905 0.906
AGFI 0.851 0.851
CFI 0.970 0.971
RMSEA 0.086 0.085
CMIN/DF 2.160 2.133
Not Supported
Ha: Respondent’s education moderate the causal effects of X1 on Y
***The moderation effect is not significant since the difference in Chi-Square

value between the constrained and unconstrained model is less than 3.84.
The difference in Chi-Square value is (107.979 - 104.520) = 3.459, while the

difference in Degrees of Freedom is 50 – 49 = 1. For the test to be significant, the
difference in Chi-Square value must be higher than the value of Chi-Square with 1
degree of Freedom, which is 3.84
The procedure for performing the test of moderation for the same variable
(education) using another dataset (data 2) is carried out in Table 13.5. The test of
hypothesis is expected to produce the same result.
325
.92 .27 .29 .14
e1 e3 e4 e5
1 1 1 1
X11 X13 X14 X15
1.73 1.61
1.93
.79
1.00
e14
X1 1
.12
Y1 e10
1 1.00
1.00
.32
1
1.04 Y2 e11 .06
.34 Y 1.09
.17
1
Y3 e12
.17
1.08 1.21 .15
1
Y4 e13
1.07
X2 Fitness Indexes
1. ChiSq = 75.987
2. df = 50
.66 .79 1.00
.97 3. ChiSq/df = 1.520
4. GFI = .928
5. AGFI = .888
X24 X23 X22 X21 6. CFI = .985
7. RMSEA = .058
1 1 1 1
1.38 1.17 .12 .20
e9 e8 e7 e6
.40
Df=50
326
.91 .27 .29 .14
e1 e3 e4 e5
1 1 1 1
X11 X13 X14 X15
1.53 1.41
1.70
.78
1.00
e14
X1 1
Y1 e10 .12
1 1.00
.70
.42
1
1.04 Y2 e11 .06
.39 Y 1.10
.17
1
Y3 e12
.23
1.08 1.21 .15
1
Y4 e13
1.07
X2 Fitness Indexes
1. ChiSq = 73.575
2. df = 49
.66 .79 1.00
.98 3. ChiSq/df = 1.502
4. GFI = .930
5. AGFI = .888
X24 X23 X22 X21 6. CFI = .985
7. RMSEA = .057
1 1 1 1
1.38 1.17 .12 .20
e9 e8 e7 e6
.40
Figure 13.12: Data 2: The Unconstrained Model. The Value Of Chi-Square Is

73.575, Df=49
327
Table 13.5: The Measures Required For Performing Moderation Test Using Data 2

Chi-Square 75.987 73.575 2.412 Not Significant Not Supported
DF 50 49 1
GFI 0.928 0.930
AGFI 0.888 0.888
CFI 0.985 0.985
RMSEA 0.058 0.057
Chisq/df 1.520 1.502
Ha: Respondent’s education moderate the causal effects of X1 on Y Not Supported
***The moderation effect is not significant since the difference in Chi-Square

value between the constrained and unconstrained model is less than 3.84. The
difference in Chi-Square value is (75.987 - 73.575) = 2.412, while the difference in
Degrees of Freedom is 50 – 49 = 1. For the test to be significant, the difference in Chi-
Square value must be higher than the value of Chi-Square with 1 degree of Freedom,
which is 3.84
The test of hypothesis for moderation found that the moderator variable
“respondents’ education” does not moderate the causal effects of X1 on Y. The results
of hypothesis testing are consistent for both datasets.
Example: Suppose that the researcher is also interested to determine whether the
same moderator variable (respondents’ education) moderate the causal effect of X2 on
Y – the same moderator variable but testing on different path. The same procedure that
has been executed earlier is applied.
328
e1 e3 e4 e5
1 1 1 1
X11 X13 X14 X15
e14
X1 1
Y1 e10
1 1
1
Y2 e11
Y
1
Y3 e12
1
1
Y4 e13
X2
1
X24 X23 X22 X21 Constrained

Parameter = 1
1 1 1 1
e9 e8 e7 e6
constrained
Figure 13.13: The Constrained Model: The Coefficient in the Path of Interest (X2 to
Y) Where the Effect of Moderator Variable to Be Examined Is
Constrained To 1.
Recall: The parameter constraint is fixed in the path where the moderation effect
will be examined, and the data is split equally based on median of the moderator
variable. In the above example, the path of interest is from X2 to Y and the moderator
variable to be tested is respondents’ education.
The output for the constrained and unconstrained models is presented in Figure
13.14 and Figure 13.15 respectively. The test of moderation is carried out in Table 13.6.
329
.87 .18 .26 .24
e1 e3 e4 e5
1 1 1 1
X11 X13 X14 X15
1.57
1.00 1.44
1.21
1.56
e14 .09
X1 1
Y1 e10
1 1.00
-.10
.51 .05
1
.99 Y2 e11
.46 Y 1.06
.20
1
Y3 e12
1.00 1.02
1.12 .73 .12
1
Y4 e13
X2 Fitness Indexes
1. ChiSq = 185.363
1.00 2. df = 50
.69 .91 1.08 3. ChiSq/df = 3.707
4. GFI = .848
5. AGFI = .763
X24 X23 X22 X21 6. CFI = .930
7. RMSEA = .131
1 1 1 1
1.39 .94 .06 .20
e9 e8 e7 e6
.49
Df=50
330
.86 .19 .24 .25
e1 e3 e4 e5
1 1 1 1
X11 X13 X14 X15
1.55
1.00 1.44
.72
1.54
e14 .09
X1 1
Y1 e10
1 1.00
.69
.51 .04
1
1.03 Y2 e11
.49 Y 1.09
.20
1
Y3 e12
.14 1.06
1.31 .72 .12
1
Y4 e13
X2 Fitness Indexes
1. ChiSq = 104.520
1.00 2. df = 49
.62 .83 1.01 3. ChiSq/df = 2.133
4. GFI = .906
5. AGFI = .851
X24 X23 X22 X21 6. CFI = .971
7. RMSEA = .085
1 1 1 1
1.41 .96 .05 .18
e9 e8 e7 e6
.51
Figure 13.15: Data 1: The Unconstrained Model. The Value of Chi-Square Is

104.520, Df=49
331
Table 13.6: The measures required for performing moderation test using
Data 1

Chi-Square 185.363 104.520 80.843 Significant Supported
DF 50 49 1
GFI 0.848 0.906
AGFI 0.763 0.857
CFI 0.930 0.971
RMSEA 0.131 0.085
Chisq/df 3.707 2.133
Ha: Respondent’s education moderate the causal effects of X2 on Y Supported
***The moderation is significant since the difference in Chi-Square value between

the constrained and unconstrained model is more than 3.84. The difference in Chi-
Square value is (185.363 – 104.520) = 80.843, while the difference in Degrees of
Freedom is 50 – 49 = 1. For the test to be significant, the difference in Chi-Square value
must be higher than the value of Chi-Square with 1 degree of Freedom, which is 3.84
The test of hypothesis for moderation found that the moderator variable
“education” moderates the causal effects of X2 on Y.
The procedure for performing the test of moderation for the same variable
(education) using another dataset (data 2) is carried out in Table 13.7. The test of
hypothesis is expected to produce the same result that the respondents’ level of
education does moderate the causal effects of X2 on Y.
332
.91 .26 .30 .14
e1 e3 e4 e5
1 1 1 1
X11 X13 X14 X15
1.53 1.40
1.69
1.04
1.00
e14
X1 1
.12
Y1 e10
1 1.00
.06
.42
1
1.00 Y2 e11 .06
.36 Y 1.06
.18
1
Y3 e12
1.00
.85 1.18 .15
1
Y4 e13
1.06
X2 Fitness Indexes
1. ChiSq = 123.410
2. df = 50
.75 .90 1.00
1.10 3. ChiSq/df = 2.468
4. GFI = .884
5. AGFI = .819
X24 X23 X22 X21 6. CFI = .957
7. RMSEA = .097
1 1 1 1
1.36 1.15 .11 .26
e9 e8 e7 e6
.39
Df=50
333
.91 .27 .29 .14
e1 e3 e4 e5
1 1 1 1
X11 X13 X14 X15
1.53 1.41
1.70
.78
1.00
e14
X1 1
.12
Y1 e10
1 1.00
.70
.42
1
1.04 Y2 e11 .06
.39 Y 1.10
.17
1
Y3 e12
.23
1.08 1.21 .15
1
Y4 e13
1.07
X2 Fitness Indexes
1. ChiSq = 73.575
2. df = 49
.66 .79 1.00
.98 3. ChiSq/df = 1.502
4. GFI = .930
5. AGFI = .888
X24 X23 X22 X21 6. CFI = .985
7. RMSEA = .057
1 1 1 1
1.38 1.17 .12 .20
e9 e8 e7 e6
.40
Figure 13.17: Data 2: The Unconstrained Model. The Value of Chi-Square Is

73.575, Df=49
334
Table 13.7: The measures required for performing moderation test using Data 2

Chi-Square 123.410 73.575 49.835 Significant Supported
DF 50 49 1
GFI 0.884 0.930
AGFI 0.819 0.888
CFI 0.957 0.985
RMSEA 0.097 0.057
Chisq/df 2.468 1.502
Ha: Respondent’s education moderate the causal effects of X2 on Y Supported
***The test is significant since the Chi-Square difference between the constrained
and unconstrained model is greater than 3.84. Recall: The Chi-Square value with 1
degree of freedom is 3.84.
Referring to Table 13.7: All fitness indexes for the unconstrained model is
significantly better (smaller Chi-Square) than the constrained model, indicating that the
two group’s coefficient differ. The results show strong support for the hypothesis that
education moderates the relationship between latent exogenous construct X 2 and its
corresponding latent endogenous construct Y.
Once the moderation effect is established, the study might be interested to

determine in which group (low education group or high education group), the
relationship between X2 on Y is more pronounced?
To address this particular research question, the researcher needs to obtain the
standardized parameter estimate for the coefficient linking X2 and Y for both datasets.
The result is presented in Figure 13.18 for low education group, and Figure 13.19 for
high education group.
335
COMPARING THE GROUP EFFECT FOR MODERATOR VARIABLE
Suppose one has the following research question to address: In which group (low
education or higher education), the effect of moderator variable (education) is more
pronounced? To address this RQ, the researcher needs to obtain the standardized
estimate of the path of interest for both datasets. The procedure is demonstrated in
Figure 13.18 and Figure 13.19.
e1 e3 e4 e5
.37 .87 .81 .83
X11 X13 X14 X15
.93
.61 .90
.91
e14 .92
X1
Y1 e10
.48 .96
.97
.98 Y2 e11
.60 Y .93
.86
Y3 e12
.15 .79
.34 .63 .31
Y4 e13
X2 Fitness Indexes
1. ChiSq = 104.520
.94 2. df = 49 Standardized
.51 .70 .98 3. ChiSq/df = 2.133
.26 .49 .97 .88 4. GFI = .906
Slope
5. AGFI = .851
X24 X23 X22 X21 6. CFI = .971
7. RMSEA = .085
e9 e8 e7 e6
.44
Figure 13.18: The Standardized Beta for Low Education in Path X2 to Y (0.15)
336
The effect of X2 on Y is not significant for “low Education” group
Standardized beta p-value Result

Y <--- X2 0.15 0.077 Not Significant at 0.05
e1 e3 e4 e5
.32 .78 .74 .90
X11 X13 X14 X15
.88 .86
.56 .95
e14 .90
X1
Y1 e10
.42 .95
.96
.98 Y2 e11
.58 Y .94
.89
Y3 e12
.22 .34 .62
.79 .35
Y4 e13
X2 Fitness Indexes
1.ChiSq = 73.575
.50 2.df = 49 Standardized
.61 .95 .92 3.ChiSq/df = 1.502
4.GFI = .930
Slope
.25 .37 .89 .84
5.AGFI = .888
X24 X23 X22 X21 6.CFI = .985
7.RMSEA = .057
e9 e8 e7 e6
.32
Figure 13.19: The Standardized Beta for High Education in Path X2 to Y (0.22)
337
The effect of X2 on Y is significant for “High Education” group
Standardized beta p-value Result

Y <--- X2 0.22 0.011 Significant at 0.05
***Since the estimated beta coefficient or the slope measuring the causal effect of
X2 on Y differs significantly between the two groups namely low education and high
education, the researcher could conclude that “education” moderates the relationship
between X2 and Y.
In this case, the effect of “high education group” (slope 0.22) is more pronounced
compared to “low education group” (slope 0.15)
***Remember- when the two slopes differ, the two linear equations would interact
with each other at some point (interaction means moderation). The types of moderation
have is full moderation since one of the slope (slope 0.15) is not significant in testing
of hypotheses while the other slope (slope 0.22) is significant.
As a summary: The effect of X2 on Y effects significantly depending on the

respondents’ level of education. The effects (X2 on Y) is not significant for the law
education group, while the effects is significant for the high education group.
Note: If both groups (low education and high education) are significant then the
partial moderation occurs.
338
Suggested Solution for the Selected Exercises in the Chapter
Solution for Chapter1
true false
1 x
2 x
3 x
4 x
5 x
1 C
2 A
3 D
4 C
5 B
6 D
7 A
8 B
9 D
Essay Questions
To study how the amount of advertisement spent would cause the

1a Causal
volume of sales
b Descriptive To describe the reaction of society
c Descriptive To describe the customer profiles
To determine how the rate of interest would influence the price of
d Causal
stocks
Basic research – because this study is about general theory

2 Situation a concerning absenteeism and the probable factors. Furthermore, this
study is carried out by an academician
Situation b Applied research – because this study is about solving
specific problem at hand in a company
4
i Causal
ii Descriptive
iii Descriptive
iv Descriptive
8 Descriptive study. Scope is the woman’s dress boutique within 40 km radius.
339
Population of the study is customers and potential customers of woman dress.

Sampling method is convenience sampling. Data collection is either personal
interview or self-administered questionnaire
Basic research: It is concerned with advancing scientific knowledge and tends to
be theoretically based and obtains data to refute or support theories or related
hypotheses about certain phenomena. It is also often the case that there is no
practical usefulness but mainly served to produce fundamental theories.
10
Applied research: It is focused on specific and current practical problems. Applied
research may be conducted to learn how to improve for example, management
systems to ensure a more effective and productive workforce. The findings are
frequently used by someone other than the person who conducted the research.
340
Solution for Chapter 2
true false
1 x
2 x
3 x
4 x
5 x
6 x
7 x
8 x
9 x
10 x
1 D
2 C
3 A
4 A
5 C
6 D
7 D
8 A
9 C
10 D
11 B
12 B
13 D
14 D
15 A
16 A
17 D
18 A
19 D
20 B
341
Essay Questions
10
a. Employees are not motivated to work. The possible causes of no motivation
include no salary increment given and also they are not being rewarded
accordingly by the management.
Dependent variable: Employees motivation level
Independent variables: Gender, Job Position, Amount of Salary, Length of
Service, Recognition, and Awards.
b. Research Questions:
i) What is the employees’ level of motivation?
ii) Does motivation increase with the increase in salary?
iii) What is the relationship between length of service and employees’
motivation level?
iv) Is there a difference in mean motivational level between male and female
employees?
v) Is motivation level greater among those who have served longer than
those who have served in a short period of time?
c. Objectives of the study:
i) To determine the level of motivation among employees.
Analysis: Frequencies and percentages of high, medium and low
motivation level.
ii) To determine the differences in mean motivational level between
male and female employees.
Analysis: t-test for two independent groups.
iii) To determine the relationship between length of service and
employees’ motivation level.
Analysis: Cross-tabulation and Chi-square test of association
d. Demographic profile:
Gender: Male ( ) Female ( )
Salary (per month) : RM________
Highest level of education: __________
Length of service: ___________
1. Strongly disagree 2. Disagree 3. Neutral 4. Agree 5. Strongly disagree
I am happy working in this company. ( )

This company gives me the opportunity to present my ideas. ( )
I am happy with my current salary level. ( )
I am enthusiastic in tackling new tasks in my job. ( )
342
1 a. Chapter 1
 Introduction which include the background of study and organization in which
the research took place (if any)
 State research objectives, research questions and hypotheses
 Scope of study
 Significance of study
 Limitation of study
Chapter 2
 Literature Review
Chapter 3
 Research Methodology
- Research Design
- Sampling Design
- Instrument
- Process of Data Collection
- Types of Analyses
1 b. Through scientific observations: Observe the buying behavior of customers and

record any unusual purchase trends in the past. Past sales record and distribution
of the magazines to vendors need to be observed and recorded. Indication of
decline in sales need to be highlighted.
Through informal interviews with magazine vendors and buyers at the vendor
newsstands will provide some insights into the possible factors that may influence
the low purchasing power among the corporate officers.
Yes, it is appropriate to proceed using descriptive research because once the
problem is defined; solutions to the problem need to be obtained. And this is done
using survey method through the use of questionnaire as the instrument to collect
information from the respondents.
343
1 c. What are the factors that may influence corporate officers to purchase the
magazine? What is perception of corporate officers towards the magazine?
1 d. 1. Which section of the magazine appeals to you most?

2. Which section of the magazine is of least interest to you?
3. Please indicate your opinion on the overall presentation of the magazine based
on the following:
Poor Fair Good Excellent
1) Quality of paper 1 2 3 4
2) Cover Design 1 2 3 4
3) Special Features 1 2 3 4
4) True Stories 1 2 3 4
5) Advertisement 1 2 3 4
6) Current issues 1 2 3 4
2.  Identify the variables of interest in the study. State how the study believes
these variables should relate to one another. This should be done based on
the theorized relationships among the variables.
 Elaborate the nature and direction of the relationships that can be theorized
from the findings of previous research.
 Provide a clear explanation of why the study would expect these
relationships to exist. The arguments could be drawn from the previous
research findings.
 Draw a schematic diagram of the theoretical framework so that the reader
can see and easily comprehend the theorized relationships.
3. Dependent variable: A variable that is to be predicted or explained.

(exam scores, income, satisfaction score)
Independent variable: A variable that is expected to influence the dependent
variable, also called predictor variable. (gender, education level, race)
Moderating variable: It is the one that has a strong contingent effect on the
independent-dependent variable relationship.
344
true false
1 x
2 x
3 x
4 x
5 x
6 x
7 x
8 x
9 x
10 x
1 B
2 A
3 B
4 D
5 A
6 A
7 C
8 D
9 C
10 C
11 D
12 B
13 C
14 B
15 A
16 B
17 D
345
Essay Questions
1
a Total students (population) = 12000
Total first year students (subpopulation) = 4000
Total sample (required) = 3000
Thus, the proportion to be sampled from first year = (4000/12000) x 300 = 100
b Both simple random sampling and systematic sampling method can be used. This
is due to the availability of sampling frame (list of students) and also the
characteristic within subpopulation is homogeneous.
c First of all, select six random numbers either through row or column arrangements.
Supposed the selected numbers are as shown in row 1 below. The amendment for
each number can be made as follows:
Take the first one digit from the first number, take the first two digits from the
second number, take the first three digits from the third number, and take all four
digits from the fourth number. For the fifth number, we take the last three digits,
and the sixth number we take the last two digits.
The
9156 7425 6535 2478 2810 1795
9156 7425 6535 2478 2810 1795

Now we have the following six numbers
9 74 653 2478 810 95
selected respondents (sample) would be according to the following student list
namely 9, 74, 95, 653, 810, and 2478.
Note: The amendment for each number depends on the creativity of the researcher since all
the numbers selected are already random.
d Another method can be made by generating random numbers through computer.
We can ask the computer to select randomly 70 numbers ranging from 1 to 2800.
Random 70 1-2800.
e The method of data collection for this study can be made through either face to
face interview with the selected students or by giving them the self-administered
questionnaire personally to complete.
8 a) Method 1: Using personal interview where researcher meet and interview the
selected students at their respective faculties in the campus using questionnaire
that contain mostly structured questions with several unstructured questions.
Method 2: Self-administered questionnaire – researcher distributed the

questionnaire to the selected students at their respective faculties and waited for
the students to fill up the questionnaire between 10-15 minutes time.
346
b) Method 2 is expected to capture a large amount of responses because the

questionnaire can be distributed to a large number of respondents at the same
time. The strategy is to gather as many respondents as possible in one time. It is
also less time consuming because there is no presence of interviewer which
could slow down the interview process.
c) Non-sampling error that could arise from the study are:
Respondent error
- non-response error such as respondent not at home or not at place of
interview
- response bias
Administrative error
- sample selection error
interviewer error
9. Cluster sampling: A probability sampling in which each element in the

population has an equal chance of being selected as sample element. It is an
economically efficient sampling technique in which the primary sampling unit is
not the individual element in the population but a large cluster of elements.
 Prepare a sampling frame
 Divide the population into clusters-characteristics between clusters are
homogenous while within clusters are heterogeneous.
 Choose clusters at random from a group of clusters based on calculated
sample size.
 For PSU, use all elements from the selected clusters.
 For SSU, select elements at random from the selected clusters using
SRS or SS.
10 a) Sampling frame of this study is the name list of the customer who fills the
application forms for the credit card.
- The application form provides the full list of the population of interest. The
information from the application form was then transferred to the bank
database system.
- The sampling frame can be obtained from the name list from the bank
system or from the application form.
b) Sampling technique → simple random sampling
- It is the most appropriate sampling technique since there is no requirement
for evaluating the credit amount charged according to certain factors.
- From the name list, the computer will generate the possible respondent’s
name according to the number of sample needed or chosen.
c) Data collection method – through mail
- We could just send a mail with a questionnaire in it for the respondent’s to
answer.
- It is not convenient to ask them to come to the bank or to go and meet them
at their house.
347
Advantages
- Cost less than other data collection method.
- Do not need many workers or people involve in the data collection and will
not waste the researcher time rather than go and meet them at their house.
Disadvantages
- The possible respondent may not get the mail or he or she did not want to
answer the questionnaire.
- Take a lot of time to be collected back to the researcher.
d) Questionnaire
- Let the respondent to answer the questions and it obtains the simplest
answer or data.
348
true false
1 x
2 x
3 x
4 x
5 x
6 x
7 x
8 x
9 x
10 x
1 B
2 A
3 A
4 D
5 A
6 D
7 B
8 A
9 D
10 C
Essay Questions
1
a It would determine the method of data analysis
b i) Qualitative – ordinal scale
ii) Qualitative – ordinal scale
iii) Quantitative – Interval scale
2 a) Primary data
b) All repair workshops in Kota Bharu, Kelantan
c) The list of all repair workshops in Kota Bharu. The list could be obtained from
the office of Kota Bharu City Council
e) Personal meeting with the respondents to explain the purpose of the study and
obtain their consent before giving them the self-administered questionnaire to
answer at their convenient time. This method would increase the rate of response
and also the accuracy of information.
349
18. Nominal: A scale in which the numbers or letters assigned to certain objects as
labels for identification or classification in a dataset.
Example 1 = Male, 2 = Female.
Ordinal: A scale that arranges objects or alternatives according to their
magnitudes – ordered arrangement.
Ratings of car service: Excellent, Good, Fair or Poor.
Interval: A scale that arranges objects or alternatives according to their
magnitudes; distinguishes ordered arrangement in units of equal intervals.
Measurement of psychological attributes such as feelings, motivation, etc.
Not happy Moderately happy Very happy
-1 0 1
The researcher can comment on the magnitude of differences but cannot
determine the actual strength or attitudes toward an object.
Temperature: -10ºC 0ºC 10ºC 20ºC
It cannot be said that 10ºC is twice as cold as 20ºC. 0ºC represents the lack of
temperature, but a relative point on the Celsius scale. It is a meaningful zero
(not absolute).
Ratio: A scale having absolute rather than relative quantities and possessing an
absolute zero, where there is an absence of a given attribute.
Money and weight possess an absolute zero and interval properties.
19 a) Nominal, ratio, nominal, nominal, ratio, ratio
b) Pie/simple bar, histogram/stem and leaf/ box plot, Pie/simple bar, Pie/simple
bar, histogram/stem and leaf/ box plot, histogram/stem and leaf/ box plot.
350
true false
1 x
2 x
3 x
4 x
5 x
6 x
7 x
8 x
9 x
10 x
1 D
2 B
3 C
4 B
5 B
6 D
7 D
8 A
9 D
10 D
Essay Questions
2 a) Non-probability sampling, i.e. Quota Sampling.

b) The sampling technique used is stratified random sampling. There is indication of a
sampling frame being used, in this case would be the sales record, however, it is not
clear as to whether the dealers can be considered as the stratification variable. Also
the distribution of questionnaire may not be sufficient to represent the entire
population of buyers since only 100 questionnaires are given to the dealers. There is
a tendency for certain group of customers not represented because the distribution
of questionnaire may not represent certain characteristics of buyers.
Data collection method: It is expected that the questionnaires are distributed by mail
and there is a tendency for very low response from the buyers.
351
3. Cross-sectional: A study in which data are collected at one point in time such as a
public opinion survey, experimental study & case study. It is also known as a snap-
shot research.
Longitudinal: A study in which data are collected at more than one point in time. It
may consist of a series of cross-sectional surveys conducted with different people,
but asking the same or similar questions or it may be conducted with the same
people who are repeat-interviewed or surveyed.
4.
Factors to consider when choosing between observation and survey methods.
Observation Survey
- Purpose of Study (sensing the - Purpose of study when the
phenomena) study involved measuring social
- When the study does not involve phenomena descriptively using
interviewing respondents but a certain developed instrument;
instead requires direct ability to reach the target
observation of the event sample simultaneously.
occurring such as observing - Control of data collection
students’ interactions in the - Speed of data collection
classroom. - Time factor
- Control of data collection - Cost
- Time factor
- Cost
5. To overcome respondents’ inability and unwillingness to answer a questionnaire,
a) ensure that respondents are informed about the study

b) ask questions that would allow them to remember certain event
within a short period of time
c) ask questions that would not require much effort to answer
d) ask questions in a simple language
e) avoid questions with technical jargons
f) avoid lengthy questions
g) avoid asking questions of sensitive issues
h) give clear and direct instructions on how to answer the
questionnaire
6. An internet survey is a self-administered questionnaire posted on a Web site.

Respondents provide answers to questions displayed on screen by highlighting a
phrase, clicking an icon, and providing their responses. Internet survey is preferred
over mail survey in a situation where a researcher wishes to conduct a study
involving respondents who have access to an internet. The study may be interested
to assess respondents’ opinion concerning current issues.
352
true false
1 x
2 x
3 x
4 x
5 x
6 x
7 x
8 x
9 x
10 x
1 B
2 D
3 A
4 C
5 D
6 C
7 A
8 D
353
true false
1 x
2 x
3 x
4 x
5 x
6 x
7 x
8 x
9 x
10 x
1 D
2 D
3 B
4 A
5 D
6 D
7 A
8 B
9 A
10 D
11 D
12 A
13 A
14 C
15 C
16 D
17 A
18 C
19 A
20 D
354
Essay Questions
1a Dear alumnni of class 2007-2008:

I am the President of our Alumni Association. Currently I am doing a survey to
gather information concerning our members of class 2007-2008.
You have been selected as a respondent in this survey. Your sincere cooperation is
needed in providing the required information for our association. In this survey, I
would like to obtain information regarding your achievements, current activities and
future aspirations. Please read carefully all instructions pertaining to every section
and answer every question sincerely. All responses provided will be used to furnish
the database of our association. Thank you very much for your cooperation.
Sincerely,
………………………
Dr Zainudin Hj Awang (President)
b Should avoid being biased towards their political affiliation, types of their social
activities and their future aspirations which might be sensitive in certain extent.
5 Five common mistakes in developing a questionnaire (Any)
 Avoid the use of complex language
 Avoid Leading questions
 Avoid loaded questions
 Avoid double-barrel questions
 Avoid burdensome questions
 Avoid asking sensitive questions
6 Factors that lead to poor questionnaire design:

- Questions did not translate the information needed
- Unable to overcome respondents’ inability to answer
- Unable to overcome respondents’ unwillingness to answer
- Questions are not suitable for the chosen interview methods
Good question
- How many times did you shop at Giant supermarket in the past two weeks?
________
Poor question
- How many times did you shop at Giant supermarket?
(no mention of time frame is a bad question)
** Any sensible answers are accepted.
355
7. (i) How much do you spend per week on groceries?

˂ RM 50.00 □1
RM 50.00 – RM 99.99 □2
RM100.00 – RM149.99 □3
RM150.00 – Rm199.99 □4
˃ RM200.00 □5
(ii) How satisfied are you with the car rental service?
Very dissatisfied Dissatisfied Neutral Satisfied Very satisfied
1 2 3 4 5
(iii)
Which brand of hand phone would you prefer to buy?(Tick √ one only)
Nokia □1
Samsung □2
Motorola □3
Sony Ericsson □4
8
Structured questions Unstructured questions
Advantage:
Advantage:
- It allows quick response from the
- It enables the respondents to
respondents
express general attitudes and
- Less time to administer the
opinions
questions
- Less biasing influence on
- Coding of responses is cheaper
response
and less time consuming
Disadvantage:
Disadvantage:
- Potential for interviewer bias is
- Respondents have less freedom
high
of expressing general attitudes
- Coding of responses is costly and
- More biasing influence on
time consuming.
response
9. Six steps involved in a questionnaire construction:
a) Specify the information needed

b) Decide on the interviewing method (personal, telephone, mail, internet
interviews)
c) Decide on the number of questions that should be asked to obtain the desired
information
d) Question should be able to overcome respondent’s unwillingness to answer
e) Question should be able to overcome respondent’s unwillingness to answer
f) Choose appropriate question structure 9srtuctured or unstructured)
356
true false
1 x
2 x
3 x
4 x
5 x
6 x
7 x
8 x
9 x
10 x
1 A
2 C
3 D
4 D
5 B
6 B
7 B
8 B
9 D
10 C
8 a. Activities involved in preliminary checking of questionnaire:

- check that all questionnaires are returned at the collection point by the
enumerators
- check for identification of serial number
- check for consistency in answering the questions sequentially
- check for missing responses of blanks
- check for errors in the response
b. Missing responses can be handled as follows:
- assign missing values to the missing responses
- assign a mean or average value of all the responses onto the missing values
357
- if the proportion of missing responses is large (˃ 50%), discard the

questionnaire
if the proportion of missing responses is small (˂ 10%) in a large sample size,
discard the questionnaire
c. Editing process
- check for errors in response
- check for blank or missing responses
- check for consistency in responses
Coding
- assign relevant codes (numeric or string) to the categories of responses
- label the codes according to the questions
assign different codes for missing values
358
1)
a. Disagree with the process of preparing the report because the researcher should
begin the analysis with the basic descriptive analysis of each question or variable
of interest. It should then be followed by some bivariate descriptive analysis and
inferences. This includes conducting hypotheses testing, such as t-test and chi-
square test of independence.
b. Nominal and Ordinal.
c. From a total of 2,715 respondents, it was found that 1,653 did not respond to the
question and hence were excluded from the analysis. Thus valid percentage will
be reported.
The table shows that majority of the respondents, i.e. 62.4% (650 out of 1,653)
indicated that their work is interesting. This is followed by 29.1% (303) of
respondents who were slightly less favorable on the statement.
2 a) H0 : Sales data is normally distributed

H1 : Sales data is not normally distributed
Based on S-W test, p-value = 0.112 ˃ 0.05, do not reject Ho.

This concludes that sales data is normally distributed.
Therefore, it is appropriate to perform a one-way analysis of variance.
b) H0: No difference in mean sales figures between the three advertising policies (μ1
= μ2 = μ3)
H1: At least one mean sales is different
From the ANOVA table, since F-statistics = 11.983 and p-value = 0.001 ˂ 0.05,
therefore Ho is rejected. This concludes that at least one mean sale figures is
different.
c) Based on the post hoc multiple comparisons, the following pairs of policies are
found to be significantly different in their mean sale figures:
Policy 1 and Policy 3 (p = .001)
Policy 2 and Policy 3 (p = .009)
3 H0: The nutritional program is not effective in reducing blood cholesterol level
(μA ≥ μB)
H1: The nutritional program is effective in reducing blood cholesterol level (μA ≤ μB)
From the Wilcoxon signed ranks test, Z-test = -2.105 and p-value = 0.035 ˂ 0.05,
359
therefore H0 is rejected. This concludes that there is enough evidence to indicate

that the nutritional program is effective in reducing blood cholesterol level where
blood cholesterol level is lower after the program.
4 (i) Median sales for front location: 5.80

Median sales for middle location: 1.90
Median sales for rear location: 3.40
(ii) Front location produce the highest sales volume
Front location has the highest maximum and minimum sales volume at 8.6 and
4.0, respectively. Rear location gives the second highest max and min sales
volume at 6.0 and 2.2, respectively, followed by middle location with the least max
and min sales volume.
5 Disagree with the process of preparing the report because the researcher should
begin the analysis with the basic descriptive analysis of each question or variable
of interest. It should then be followed by some bivariate descriptive analysis and
inferences. This includes conducting hypotheses testing, such as t-test and chi-
square test of independence.
6 Nominal and Ordinal.
7 Majority is males (54%) and stays off-campus (51%) with friends. The age of the
respondents ranges from 13 (it’s a mistake here but too late to change) to 27
years old with an average of 20.72 years and a standard deviation of 2.656 years.
Slightly less than 20 per cent (19%) smokes regularly while a higher percentage
(21%) smokes occasionally
8 The table indicates that there is a higher proportion among male students who
smoke regularly (29.6%) or occasionally (22.2%) compared with female students
(5.5% and 19.6%, respectively). This observation is confirmed by the results of
the Chi-square test of association which shows that there is a significant
difference in the smoking habit between males and females at the 5% level of
significance (p-value=0.007˂0.05)
9 a) Target population: All employees working at HyPetrol Sdn Bhd.
Sample: Representatives of local and foreign workers
Sampling frame can be acquired from the Manager of the Human Resource Dept
of the company.
Problem Statement: There is a concern for low productivity in the manufacturing
b) of oil gaskets for automobile and small tractor engine. Based on informal
information, the management found that the outcome of low productivity relates to
employees’ dissatisfaction with the new company policy.
c) A suitable research design is descriptive research via survey method.
Purpose of descriptive research is to describe characteristics of a population or a
360
phenomenon and to determine the answers to who, what, when, where, and how
questions. This is done through survey method where it is used to obtain
information from respondents either through personal interview, telephone, mail,
and other suitable methods.
d)
Dependent – Job Satisfaction
Independent – Age, length of service, gender, opportunity for advancement and
origin.
Independent variables
Age Dependent variable
Length of service Job Satisfaction
Gender
Opportunity for advancement
Origin (local for foreign)
There is a difference in mean job satisfaction between the local and foreign
e) workers
- There is a relationship between age group and job satisfaction level of workers
- Foreign workers are more satisfied with their job than the local workers
- There is a difference in mean job satisfaction between different age groups.
Or any other appropriate hypotheses
361
Index
absolute fit, 278

actual problem, 18, 25
actual responses, 110
adequate alternatives, 178
administrative information, 179
alternative hypothesis, 33
ambiguous words, 176
an absolute zero point,123
an exploratory study, 19
Analysis of Moments Structure (AMOS), 245, 293
Analysis of Variance (ANOVA), 115
approximately normally distributed, 85
assessing the questionnaires, 210
associational relationship, 40
Average Variance Extracted (AVE), 277
Bartlett’s Test of Sphericity, 229, 230

basic research, 8
bias wordings, 177
bibliographic citations, 28
bibliography search, 156
binomial test, 115
bipolar labels, 121
branching questions, 181
business periodicals, 160
captive audience, 93
categorical scale, 117, 118
causal factors, 41
causal research design, 40, 41
cause and effect relationship, 40
CD-ROM database, 158
census, 45, 84
Central Limit Theorem, 85
characteristics of interest, 5, 94, 110
Chi-Square test for correlation, 49, 115
classification information, 179
close-ended questions, 171
cluster sampling, 92
coding, 76 , 217
coefficient of determination, 249
coefficient of variation, 115
cohort analysis, 43
comparative scale, 117
comparison scale (ranking scale), 122
complete mediation, 306
complete moderation, 313
computerized database, 157
computerized search, 158
conclusive evidence, 37, 39, 43
conference proceedings, 28
confidentiality, 13
confirm coding, 216
confirmatory factor analysis, 268, 276
constant sum scaling, 124
constrained model, 320, 327
construct of primary interest, 69
construct validity, 277
construct reliability,
content in a questionnaire, 169, 172
continuous variable, 107
convenience sampling, 93, 94
corporate image, 9, 24, 70
corrective action, 2
correlational study, 40, 271
confirmatory factor analysis (CFA), 269
constrained model, 319, 320, 323
construct validity, 276
covariance, 252, 301
convergent validity, 276
credible scientific finding, 11
critical ratio, 299
critically and objectively, 3
Cronbach’s alpha, 234
cross sectional design, 43, 77
customer loyalty, 9, 61, 110
customer satisfaction, 3, 24, 61
data analysis, 7, 32, 206

data coding, 210
data collection process, 73, 207
data editing, 207, 218
data management process, 206
data mining procedure, 76, 154
data reduction procedure, 229, 244
data transcribing, 1, 18, 37, 76
declarative statement, 33, 62
defining the managerial problem, 22
degree of sampling error, 85
demographic characteristics of the population, 225
demographic variables, 30
dependent variable, 21, 68, 69
depth interviews, 46
descriptive analysis, 225
descriptive research, 38
designing a sensitive question, 177
developing a questionnaire, 32, 169, 174
diagnosing a situation, 37
dichotomous choice, 178
differentiate coding, 216
direct observation, 110
direction for the study, 10
discrete variables, 107
discriminant validity, 277, 280, 290
distinct dimensions, 230, 231
distribution of sample, 84
dramatic and general statements, 59
editing, 76
eigenvalue, 230
endogenous construct, 272
error in the equation, 247
error in the measurement, 248
establish rapport, 179
existing phenomenon, 38
exogenous constructs, 252, 273, , 277, 297
experimental research, 40
explicit directions,181
exploratory research, 37
external sources, 155
factor analysis, 49, 115, 229

factor loadings, 231, 280
factorial ANOVA, 49
field editing, 207
first-hand information, 103
fitness indexes,288
focus groups interviews, 26, 103
forced choice scale, 124
framework, 12, 32
F-test for multiple regression, 49
F-test for regression, 49
F-test for variance, 49
F-test, 235
full-text databases, 157
funnel approach, 180
general bibliography, 159

general theory, 29
generalizable, 228
good questionnaire, 168
graphical rating scale, 122, 124
group effects, 335
heterogeneous population, 90
historical information, 9, 104, 153
homogeneity, 72
homogeneous characteristics, 72, 88, 90
hypothesis,10, 235
hypothesized link, 255
hypothetical constructs, 244
I
identifying Process, 169

inconsistent response, 209
incremental fit, 278
independent variable, 21, 68
in-depth interview, 26
indirect influence, 242
inferential statistics, 49
informal interview, 25
informed consent, 13
in-house editing, 207
instruments, 26,46
interaction effect, 311, 312, 313, 338
internet access, 159
internal validity, 277
interpretation of output, 48
interrogative statement, 33
interval properties, 114
interval scaled data, 48, 122
interval scaled measurement, 113, 244
intervening variable, 21
interviewer identification, 168
introductory statements, 59
isomorphic, 111
job satisfaction, 9
judgmental sampling, 94
Kaiser-Meyer-Olkin (KMO), 229
laboratory experiments, 104

lack of fit, 85
latent (unobserved) constructs, 245, 272, 280, 318
leading question, 177
Least Significance Difference, 240
level of awareness, 8
level of confidence, 86
level of measurement, 111
library search, 159
Likert scale, 119
list of population, 83
literature review, 27, 67
literature search, 9, 23, 26, 28
longitudinal design, 44, 77
M
magnitude of difference, 85, 112

mailed questionnaires, 109
main factors, 8, 29
main hypotheses, 255
managerial problems, 2
manifesting (observed) components, 245
MANOVA, 49
market basket-analysis, 155
market potential, 5
measure of association, 252
measure of skewness, 224
measurement model, 282, 286
measurement process, 46
measurement scale employed, 48
measuring Instruments, 74
mediating construct, 70, 307
mediating variable, 21, 68, 274, 303
method of data analysis, 12, 22
methods of data collection, 23
methods of sampling, 23
missing data, 209
mode, 47
median, 47
moderating variable, 22, 275, 311, 319, 336
moderating effect, 313, 314, 317, 318
moderation test, 324
multicollinearity, 268
Multi-Group CFA, 318, 319
multiple cross-sectional designs, 43
multiple-choice questions, 171
multivariate associational question, 49
multivariate kurtosis, 291
nature of the research problem, 39

nominal scale data, 48, 111, 118
non-experimental research, 37
non-numerical data, 105
non-parametric methods, 119, 223
non-probability sampling, 82, 83, 84, 86, 93
non-sampling error, 84
normally distributed, 75, 291
null hypothesis, 33
numerical data, 107, 121
objective of the study, 18, 61, 206

observation methods, 45
off-line databases, 158
one-way ANOVA, 49
online database, 28, 38, 157
open-ended question, 171, 178, 215
open-ended response, 178
opinion from the experts, 25
ordinal scaled data, 48, 112
original records, 103
outline of research process, 12
paired comparison scaling, 123

parametric methods, 119
parametric statistical test, 223
parsimonious fit, 278
partial mediation, 310
path analysis, 252, 254
periodicals, 160
personal interviews, 103
personal judgment, 86
personalised words, 177
phenomena, 33
pilot survey, 26, 93
pilot testing, 46
population, 12, 29, 45, 90
positive attitudes, 181
positive word of mouth, 69
post-hoc analysis, 240
post-hoc research, 41
Post-Hoc test of difference, 49
predictor construct, 248
prejudice, 75
presentation of findings, 48
pre-test questionnaires, 93, 160
primary data sources, 9, 103, 105
Principal Component Factor Analysis, 229
probability sampling method, 82, 83, 87
probability sampling procedure, 88
probability-sampling plan, 85
problem definition process, 19, 25
problem identification process, 19
problem statement, 60
product-moment correlations, 115
professional journals, 160
proposed hypothesis, 76
proposed model, 247
public opinion surveys, 160
published journals, 9
pure research, 9
P-value, 256
qualitative research, 26, 105, 106

quantitative variable, 107
quasi-experimental research, 41
questionnaire, 74, 167
quota sampling, 94
R
random numbers, 73, 88

randomly distributed, 94
randomly selected, 72, 75
ranked order scale, 124
rating scale, 117
ratio scale, 48, 114
raw data, 44, 45
records of eyewitnesses, 103
reference database, 157
regression, 115
re-inventing the wheel, 11
relationship among variables, 29
relative magnitude, 112
relative quantities, 114
relevant hypothesis, 252
reliability measures, 49, 234
Reliability Statistics, 234, 276
replicable, 11
representativeness of a sample, 72, 87
required assumption, 223
research assistants, 74
research design, 12, 29
research ethics, 13
research findings, 43
research hypotheses, 23
research instrument, 103
research objectives, 2, 22, 23, 71, 82
research process, 18, 57
research proposal, 48, 57
research questions, 4, 23, 48, 61, 76, 82,
research schedule, 78
researcher’s hypothesis, 244
respondent identification, 168
respondents for the study, 74
respondents, 13
respondents’ perception, 171
response bias, 74
response options, 178
response rate, 109
revising coding, 217
rotated component matrix, 231, 232
sample size, 85
sample, 29
sampling adequacy, 229
sampling and data collection procedure, 73
sampling design, 82
sampling distribution of the mean, 85
sampling distribution, 75
sampling error, 45, 72, 84
sampling frame, 73, 81, 88, 92
sampling method, 72
sampling methodology, 37
sampling procedure, 88
sampling technique, 80, 81
sampling units, 85, 95,
schematic diagram, 241
scientific method, 2, 38, 85
scope of a research project, 12, 77, 83
screening alternatives, 37
secondary data, 103, 153
secondary research, 104
second generation technique, 268
second-hand data, 104
selection bias, 94
self-addressed envelope, 74
self-administered questionnaires, 44, 109
SEM, 268
semantic differential scale, 120
semantic meaning, 120
sense of authority, 182
sensitive question, 180
service quality, 24, 61, 62, 70
significance of the research, 57
significance of the study, 66
significance value, 230
significant and direct influence, 242, 257
simple random sampling, 88, 89
sincere response, 113
skewness, 291
snowball sampling, 95
source databases, 157
specific question, 180
squared multiple correlation, 297
standard deviation, 47, 224, 249
standard error, 47, 224
standardized estimate, 294, 337
standardized model, 249
stapel scale, 121
statement of problem, 59
statistical analysis procedure, 49
statistical analysis techniques, 84, 244
statistical information, 3
statistical measures,47
statistical reliability, 72
statistical significance, 85
statistical test, 12, 75
statistical validity, 72
statistically invalid, 84
strata, 90
stratified sampling method, 72, 83
Structural Equation Modelling (SEM), 49, 245, 292
structure of questionnaire, 170
structure of response, 169
structure to the interview, 167
structured questionnaire, 44, 93, 103
structured questions, 167
subgroups in a population, 43, 94
symptom to a problem, 18, 19
systematic procedure, 210
systematic random sampling, 83
systematic sampling technique, 89
systematic search, 1
T
target information, 39, 179

target population, 71, 72, 81, 84
technical jargon, 176
technical review, 29
tentative statement, 62
test statistics, 48
the research problem, 71
theoretical foundation, 12
theoretical framework of the study, 23, 47, 69, 241
theorized relationship among variables, 68
time intervals, 44
tolerance for error, 82
true value of the parameter, 86
t-test for each regression coefficient, 49
t-test for mean, 49
t-test for Pearson’s correlation, 49
t-test, 115, 235
types of measurement, 22, 111
types of variables, 22
unbiased, 72
unconstrained model, 321, 324, 327
underlying problem, 3, 25, 48
underlying theory, 159
unidimensionality, 276
unit of analysis, 20
univariate associational question, 49
univariate descriptive question, 49
univariate difference question, 49
unstandardized regression weights, 298
unstructured questions, 171
validity, 26, 153, 276

variability of the population characteristics, 86
variable of interest in the study, 21, 34, 47, 91
Varimax Rotation, 229
wordings of a question, 169, 174
Z-test, 235
Bibliography
Zainudin, A. and Zaihan, O. (2004). The Impact of Corporate Image on Perceived Quality in
Healthcare Industry. Proceeding of Conference of Social Science Research (CSSR
2005). Institute of Research Development and Commercialization UiTM,
Zainudin, A., Rustam, S. and Farouk, M.Z. (2005). Service quality and customer satisfaction
in the environment of higher education: A study on the public and private university
students in Malaysia, Institute Research Development and Commercialization, IRDC,
UiTM.
Zainudin, A., W. Fauzi, W.M. and Azuhari, C.M. (2005b). Achieving Students’ Satisfaction
in Higher Education: Closing the gaps. International Conference on Quantitative
Sciences and its Application (ICOQSIA 2005). Faculty of Quantitative Sciences,
UUM.
Zainudin, A. (2007). The Effects of Service Quality and Corporate image on Students’
Loyalty in Postgraduate Education: The case study of a local university, Unpublished
Thesis, Universiti Malaysia Terengganu (UMT), Malaysia.
Zainudin, A, Azahar, Y, and Masayu, F (2008). The influence of corporate reputation on
competitive advantage to the firms in household furniture industry. Paper presented at
The International Furniture Conference IFCx, PWTC Kuala Lumpur 28-30 July
2008.
Anderson, J.C., & Gerbing, D.W. (1988).Structural Equation Modeling in Practice: A review
of recommended two steps approach. Psychological Bulletin, 103(3), 411-423.
Arbuckle, J.L. (2005). AMOS 6.0 User’s Guide. Spring House PA: Amos Development
Corporation.
Bagozzi, R. P., & Yi, Y. (1988). On the evaluation of structural equation models. The
Academy of Marketing Science. 16(1), 74-94.
Baron, R.M., & Kenny, D.A. (1986). The moderator-mediator variable distinction in social
psychological research: Conceptual, strategic and statistical considerations. Journal of
Personality and Social Psychology, 51: 1173-1182.
Bentler, P. & Chou, C. (1987). Practical Issues in Structural Equation Modeling. Sociological
Methods and Research. 69(16): 78-117.
Bentler, P.M. & Bonett, D.G. (1980). Significance tests and goodness of fit in the analysis of
covariance structures. Psychological Bulletin. 88: 588–606.
Bentler, P.M. (1990). Comparative fit indexes in structural models. Psychological Bulletin
107: 238–246
Bollen, K.A. (1986). Sample size and Bentler and Bonett’s non-normed fit index. Psychometrika.
51:375–377.
Bollen, K.A. (1989b). A new incremental fit index for general structural equation models.
Sociological Methods and Research 17: 303–316.
Byrne, B.M. (2001). Structural Equation Modeling with Amos: Basic Concepts, Applications,
and Programming. Mahwah, NJ: Erlbaum.
Byrne, B.M. (2010). Structural Equation Modeling With AMOS: Basic Concepts,
Applications, and Programming 2nd Edition. Routledge. New York.
Cronbach. L.J. (1951). Coefficient Alpha and the Internal Structural of Tests. Pychometrica.
16(3): 297-334
Edwards, J.R., & Lambert L. S. (2007). Methods for integrating moderation and mediation: A
general analytical framework using moderated path analysis. Psychological Methods,
12: 1-22.
Frazier, P.A., Tix, A. P. & Barron, K. E. (2004). Testing moderator and mediator effects in
counseling psychology research. Journal of Counseling Psychology, 51: 115-134.
Hair, J.F., Black, W.C., Babin, B.J., Anderson, R.E., & Tatham, R.L. (2006). Multivariate
data analysis. New Jersey: Pearson Prentice Hall.
Hair, J.F., Black, W.C., Rabin, B.J. & Anderson, R.E. (2010). Multivariate Data Analysis (7th
Ed). Englewood Cliffs, NJ: Prentice Hall.
Hair, J.F.,Anderson, R.E., Tatham, R.L., & Black, W.C. (1995). Multivariate Data Analysis
with Readings (4th Ed). Englewood Cliffs, NJ: Prentice Hall.
Hayes, A.F., & Matthes, J. (2009). Computational procedures for probing interactions in OLS
and logistic regression: SPSS and SAS implementations. Behavior Research Methods,
41: 924-936.
Holmes-Smith, P., Coote, L., & Cunningham, E. (2006). Structural Equation Modeling:
From the Fundamental to Advanced Topics. Melbourne: Streams.
Jöreskog, K.G. & Sörbom, D. (1984). LISREL-VI user’s guide (3rd ed.). Mooresville, IN:
Scientific Software.
Kenny, D.A., & Judd, C.M. (1984). Estimating the nonlinear and interactive effects of latent
variables. Psychological Bulletin, 96: 201-210.
Klein, A.G., & Moosbrugger H. (2000). Maximum likelihood estimation of latent interaction
effects with the LMS method. Psychometrika, 65: 457-474.
Kline, R.B. (2005). Principles and Practice of Structural Equation Modeling (2nd ed.).New
York: The Guilford Press.
Kraemer, H.C., Wilson G.T., Fairburn, C.G., & Agras, W.S. (2002). Mediators and
moderators of treatment effects in randomized clinical trials. Archives of General
Psychiatry, 59: 877-883.
Kraemer, H.C., Stice, E., Kazdin, A., Offord, D., & Kupfer, D. (2001). How do risk factors
work together? Moderators, mediators, independent, overlapping, and proxy risk
factors. American Journal of Psychiatry, 158: 848-856.
Mahadzirah, M. & Zainudin, A. (2009). Building corporate image and securing students’
loyalty in the Malaysian higher learning industry. Journal of International Management
Studies. Vol. 4 (1): 30-40.
Muller, D., Judd, C. M., & Yzerbyt, V. Y. (2005). When moderation is mediated and
mediation is moderated. Journal of Personality and Social Psychology, 89: 852-863.
Norsalihah, A.R., Zainudin, A., and Nor Hazreeni, H. (2010). The effects of information and
communication technology on security compliance. Journal of Statistical Modeling and
Analytics. Vol 1 (1): 28-44.
Ping, R.A. (1996). Latent variable interaction and quadratic effect estimation: A two-step
technique using structural equation analysis. Psychological Bulletin, 119, 166-175.
Salamiah A. Jamal (2011). Perceived Value, Satisfaction and Visitor Psychographics in

Community-based Homestay Tourism: Behavioural Intentions Perspective.
Unpublished Thesis: Faculty of Hotel and Tourism Management, UiTM Malaysia
Shammout, A.B. (2007). Evaluating Extended Relationship MarketingModel for Arab Guests
of Five Star Hotels. Unpublished Thesis: Faculty of Business and Law, Victoria
University Melbourne.
Tanaka, J.S. & Huba, G.J. (1985). A fit index for covariance structure models under arbitrary
GLS estimation. British Journal of Mathematical and Statistical Psychology 38: 197–
201.
Zainudin, A. (2010). The importance of corporate image of a university in the marketing of

postgraduate programmes. Asian Journal of University Education. Vol. 6 (1): 13-28.
Zainudin, A. & Kamaruzaman, J. (2009). The effects of corporate reputation on the

competitiveness of Telecommunication service providers. International Journal of
Business and Management. Vol. 4 (5): 173-178.
Zainudin, A., Junaidah, H.A., & Nazmi, M.Z. (2010). Modeling job satisfaction and work
commitment among lecturers. Journal of Statistical Modeling and Analytics. Vol 1 (22):
45-59
Zainudin, A. (2010). Research Methodology for Business and Social Science. Shah Alam:
Universiti Teknologi MARA Publication Centre (UPENA).
Zainudin, A. (2007). Modeling the Loyalty of Outgoing Undergraduates towards
Postgraduate Education at their University. Thesis: Faculty of Management and
Economics, Universiti Malaysia Terengganu.

3 Ebook Research Methodology & Data Analysis

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

3 Ebook Research Methodology & Data Analysis

Uploaded by

Copyright:

Available Formats

RESEARCH

Universiti Sultan Zainal Abidin (UniSZA)

All rights reserved.

RESEARCH METHODOLOGY AND DATA ANALYSIS 2ND EDITION

Author : Zainudin Awang

Research Methodology and Data Analysis offers a comprehensive guide for

Zainudin Bin Awang, PhD

2 THE RESEARCH PROCESS 19

3 WRITING THE RESEARCH PROPOSAL 58

Chapter 2: The Review of Literature Related to the Study

Chapter 3: The Research Methodology

4 THE SAMPLING DESIGN 82

5 DATA, MEASUREMENT, AND SCALES 105

6 THE PRIMARY DATA SOURCES 136

7 THE SECONDARY DATA SOURCES 154

8 DESIGNING A QUESTIONNAIRE 167

10 THE DATA ANALYSIS 223

11 STRUCTURAL EQUATION MODELING (SEM) 268

12 ANALYZING THE MEDIATING VARIABLE 303

13 ANALYZING THE MODERATING VARIABLE 311

SUGGESTED SOLUTION FOR EXERCISES

CHAPTER INTRODUCTION TO RESEARCH

Research is a process of searching information concerning the underlying

Why is research required? Is it because there is a “problem” to be solved?

a. The existing problem in the organization (declining sales, high complaints)

b. The anticipated problem in the future due to changes in lifestyles and

c. The theoretical gaps or knowledge gaps. Findings from previous research

WHAT IS THE IMPORTANCE OF RESEARCH?

Prof Dr Zainudin Awang 1

Research provides valuable information that guides manager, employees and

As an example, a research is required to identify the factors that cause the

THE NEED FOR RESEARCH

In managing the organization, the manager as a decision maker would come

The essence of research is important to managers in which research helps them

Prof Dr Zainudin Awang 2

Knowledge on research would also help managers to utilize the available

Having adequate knowledge and experience in research would help

A good manager would ask for a research report before making an

THE ROLE OF RESEARCH IN BUSINESS

Decision making process requires systematic and organized efforts to investigate

Prof Dr Zainudin Awang 3

THE DETERMINANT FOR CONDUCTING RESEARCH PROJECT

1. The availability of time frame.

1. The Availability of Time Frame

A systematic research takes time to accomplish. However, in many instances,

Prof Dr Zainudin Awang 4

decision. Making a decision blindly without adequate information might land an

2. The Availability of Data

Adequate information is very important for a manager in making decision. The

3. Nature of Decisions to Be Made

The value of research depends on the nature of the managerial decision to be

Prof Dr Zainudin Awang 5

to do a small modification in their product because they are confident of long-

4. The Value of Decision versus the Cost of Research

We have discussed the various benefits of research. However, a substantial

Time constraints Availability of data

The Value of research Yes

Figure 1.1: The Decision to Conduct or Not to Conduct Research

Prof Dr Zainudin Awang 6

THE TYPES OF RESEARCH

1. Applied Research or Action Research

Every business organization requires continuous research and development

Prof Dr Zainudin Awang 7