Professional Documents
Culture Documents
BBRC4103
Research Methodology
BBRC4103
RESEARCH
METHODOLOGY
Assoc Prof Dr Ahmad Shuib
Dr Thinagaran Perumal
Assoc Prof Dr Nagarajah Lee
Project Directors:
Module Writers:
Developed by:
Table of Contents
Course Guide
ix-xiv
Topic 1
1
2
3
4
6
7
12
13
Topic 2
Research Process
2.1 Research Process
2.2 Process of Identifying the Problem
2.3 Data for Research
2.4 Analysing and Interpreting Data
2.5 How to Choose a Topic
Summary
Key Terms
14
14
18
19
20
20
24
25
Topic 3
Review of Literature
3.1 What is Literature Review?
3.2 Importance of Literature Review
3.3 Review of Literature Procedures
3.4 Common Weaknesses
3.5 Evaluating Journal Articles
Summary
Key Terms
26
27
28
30
33
34
36
36
Topic 4
Sampling Design
4.1 Sampling Concept
4.2 Justification for Sampling
4.3 Criteria of a Good Sample
4.4 Types of Sampling Designs
4.4.1 Probability Sampling Design
4.4.2 Types of Probability Sampling
4.4.3 Non-probability Sampling Design
4.4.4 Types of Non-probability Sampling
37
37
39
40
41
42
44
46
46
iv
TABLE OF CONTENTS
4.5
Sample Size
4.5.1 Qualitative Approach
4.5.2 Statistical Approach
Summary
Key Terms
48
49
49
54
55
Topic 5
56
57
58
59
60
61
65
65
68
70
71
72
75
76
Topic 6
77
78
79
79
81
83
85
85
86
89
90
92
92
93
94
95
95
96
97
100
103
TABLE OF CONTENTS
Topic 7
104
105
105
105
106
107
108
109
111
112
113
114
116
118
118
Topic 8
119
120
121
121
122
123
124
124
125
131
Data Analysis
9.1 Data Screening and Editing
9.1.1 Data Editing
9.1.2 Field Editing
9.1.3 In-house Editing
9.1.4 Missing Data
9.2 Coding
9.3 Data Entry
9.4 Data Transformation
9.5 Data Analysis
9.5.1 Descriptive Statistics
134
135
135
136
136
137
138
139
141
142
144
Topic 9
132
133
vi
TABLE OF CONTENTS
9.6
Topic 10
What is a Hypothesis?
9.6.1 Null and Alternate Hypotheses
9.6.2 Directional and Non-directional Hypotheses
9.6.3 Sample Statistics versus Population
9.6.4 Type I and Type II Errors
9.6.5 Steps in Hypothesis Testing
9.7 Inferential Statistics
9.7.1 Testing for Significant Differences between
Two Means Using the t-test (Independent Groups)
9.7.2 Testing for Significant Differences between
Two Means Using the t-test (Dependent Groups)
9.7.3 Testing for Differences between Means Using
One-way Analysis of Variance (ANOVA)
9.7.4 Correlation Coefficient
Summary
Key Terms
148
148
149
150
150
152
152
155
162
163
163
167
168
170
170
171
156
157
158
160
161
173
173
COURSE GUIDE
COURSE GUIDE
ix
INTRODUCTION
BBRC4103 Research Methodology is one of the courses offered by the Faculty of
Business and Management at Open University Malaysia (OUM). This course is
worth 3 credit hours and should be covered over 8 to 15 weeks.
COURSE AUDIENCE
This course is offered to undergraduates students who need to acquire
fundamental knowledge in research methodology.
As an open and distance learner, you should be able to learn independently and
optimise the learning modes and environment available to you. Before you begin
this course, please confirm the course material, the course requirements and how
the course is conducted.
STUDY SCHEDULE
It is a standard OUM practice that learners accumulate 40 study hours for every
credit hour. As such, for a three-credit hour course, you are expected to spend 120
study hours. Table 1 gives an estimation of how the 120 study hours could be
accumulated.
COURSE GUIDE
Study
Hours
60
10
Online participation
12
Revision
15
20
120
COURSE OUTCOMES
By the end of this course, you should be able to:
1.
2.
3.
4.
5.
COURSE SYNOPSIS
This course is divided into 10 topics. The synopsis for each topic is as follows:
Topic 1 describes the concept of science, the scientific research process in problem
solving and the need to formulate a good hypothesis.
Topic 2 describes the processes in doing research starting from the identification
of a specific problem, the identification and definition of the concepts, and the
identification of the methodology of the research.
COURSE GUIDE
xi
xii
COURSE GUIDE
Topic 10 focuses on writing research report and principles of research writing, the
content of a research proposal, the guidelines for writing research proposal,
common weakness in research proposal and ethics in research.
COURSE GUIDE
xiii
Key Terms: This component can be found at the end of each topic. You should go
through this component to remind yourself of important terms or jargon used
throughout the module. Should you find terms here that you are not able to
explain, you should look for the terms in the module.
References: The References section is where a list of relevant and useful
textbooks, journals, articles, electronic contents or sources can be found. The list
can appear in a few locations such as in the Course Guide (at the References
section), at the end of every topic or at the back of the module. You are
encouraged to read or refer to the suggested sources to obtain the additional
information needed and to enhance your overall understanding of the course.
PRIOR KNOWLEDGE
There is no prerequisite requirement for learners prior to taking this subject.
ASSESSMENT METHOD
Please refer to myINSPIRE.
REFERENCES
Black T. R. (1999). Doing quantitative research in the social sciences: Sage
Publication, London.
Boaden, R.C. & Biklen, S.K. (2003). Qualitative research for education: An intro to
theories and methods (4th ed.). Pearson, New York.
Cooper, D. R. & Schindler, P. S. (2007). Business research methods. (10th ed). New
York: McGraw Hill.
Uma Sekaran. (2003). Research methods for business. A skill building approach.
(4th ed). New York: Wiley.
xiv
COURSE GUIDE
Topic
Scientific
Thinking in
Research
LEARNING OUTCOMES
By the end of this topic, you should be able to:
1.
2.
3.
INTRODUCTION
The purpose of science is to expand knowledge and discover the truth. By
building theory, researchers undertake research to achieve this purpose.
Prediction and understanding are the two purposes of theory and they usually
go hand in hand. To make a prediction, one must know and understand why
variables behave as they do and theories provide this explanation. A theory is a
coherent set of general propositions used as principles to explain the apparent
relationships of certain observed phenomena. The scientific method is a series of
stages used to develop and refine theories.
Scientific methods and scientific thinking are based on concepts. Concepts are
invented so as to enable us to think and communicate abstractions. Higher-level
concepts are used for specialised scientific explanatory purposes that are not
directly observable. Concepts and constructs are used at the theoretical levels
while variables are used at the empirical level. The scientific research process is
used to develop and test various propositions using inductive-deductive
reflective thinking. Scientific research uses an orderly process that combines
TOPIC 1
1.1
SCIENTIFIC METHOD
(b)
(c)
(d)
(e)
It is logical, meaning that conclusions are drawn from the results based on
logic; and
(f)
TOPIC 1
It is noted here that the difference between hard science and soft science is
control over confounding variables. For example, in business, there are factors
which may be beyond the control of managers, so there has to be some trade-off
between the rigours of science and the pragmatics of business. There has to be
some give and take between the desires of the businesspeople and the desires of
the researchers.
Although this will lead to error, as long as the researcher informs the decisionmaker of the limitations, and the results are qualified based on the limitations,
the research should go on to produce the information. Good scientific research
also follows the principle of parsimony, that is, a simple solution is better than a
complex solution. Parsimonious research means applying the simplest approach
that will address the research questions satisfactorily.
SELF-CHECK 1.1
What are the characteristics of good science?
1.2
Common Sense
Information and knowledge can be gained by relying on what everyone
knows and what just makes sense. Common sense is valuable in daily
living but it can allow logical fallacies to slip into thinking. Common sense
may contain contradictory ideas that many people may not notice because
the ideas are used at different times. Common sense can originate from
tradition; it is useful and may be correct but it may contain errors,
misinformation and contradiction. It may be prejudiced because of beliefs
and socio-cultural differences. One can avoid making wrong decisions by
accepting the truth that a deficiency of knowledge in common sense exists.
To reduce this deficiency, one has to generate the right kind of knowledge
and common sense knowledge needs to be examined systematically to find
the actual cause. The actual cause can be found by setting up experiments
for systematic testing or continually collecting data to examine the repeat
Copyright Open University Malaysia (OUM)
TOPIC 1
Personal Experience
When something happens, you feel it, you experience it and you accept it as
true. Personal experience or seeing is believing is a forceful source of
knowledge. But personal experience can lead one astray. What may appear
true may actually be due to a slight error or distortion in judgment. People
make mistakes or fall for illusions. They may believe what they see or
experience but these may be full of errors. Personal experience is reinforced
by four basic errors:
(i)
(ii)
(iii) Premature closure: This often operates with and reinforces the first
two errors. Premature closure occurs when people feel they have all
the answers and do not need to listen, seek information or raise
questions any longer.
(iv) Halo effect: This happens in many forms whereby people
overgeneralise from what they interpret to be highly positive or
prestigious.
1.3
TOPIC 1
(a)
(b)
(c)
TOPIC 1
From this tentative theory, prediction and hypotheses are derived for further
investigation or testing. The process of further investigation or testing will
continue until the theories and laws derived are refined. The refined laws or
theories are tentative. If an anomaly is found when a new observation does not fit
into a current body of knowledge or the theories or laws are proven wrong, a
modification has to be carried out. The process will continue again and again
when new knowledge is generated from new observations.
1.4
Research involves the use of theory. In the process of designing a research, theory
may or may not be expressed explicitly, although in the presentation of the
findings and conclusion, relationships with theories will be explicitly made.
(a)
Inductive Model Moves from the particular to the general, from a set of
specific observations to discovery of a pattern that represents some degree
of order among all given events; the logical model in which general
principles are developed from specific observations.
(b)
Deductive Model Moves from the general to the specific, from a pattern that
might be logically or theoretically expected and observations that test whether
the expected pattern actually occurs; the logical model in which specific
expectations of hypotheses are developed on the basis of general principles.
Figure 1.2 illustrates the differences between the inductive and deductive models.
TOPIC 1
ACTIVITY 1.1
By providing appropriate examples, discuss the deductive and
inductive methods.
1.5
It narrows the range of facts needed to study; any problem can be studied
in many different ways. A theory can suggest the ways that are likely to
yield the greatest meaning;
(b)
(c)
TOPIC 1
(d)
(e)
Concepts
A concept is a bundle of meanings or characteristics associated with certain
events, objects, conditions and situations. Concepts may be developed
because of frequent, general and shared usage over time. It may be acquired
through experience. Some concepts are unique to a particular culture and
not easily translated into another language.
In research, concepts used must be precise and comprehensible; hypotheses
are designed using concepts, measurement concepts are used to collect
data, new concepts may be invented to express ideas. The success of
research depends on the ability of researchers to conceptualise ideas and
how well others understand the concepts used. Concepts represent
progressive levels of abstractions; the degree to which the concepts do not
have objective referents. A shirt is an objective concept while personality is
a concept with a high degree of abstraction; such concepts are called
constructs.
(b)
Constructs
A construct is an image or idea specifically invented for a given research
and/or theory-building purpose. Constructs are developed by combining
simpler concepts, especially if the idea or image we want to convey is not
directly subject to observation. Intelligent quotient (IQ) is constructed
mathematically from observations of the answers given to a large number
of questions in an IQ test. No one can directly or indirectly observe IQ but it
is a real characteristic of people.
(c)
Definitions
If the meaning of the concept is confused, the value of the research may be
destroyed. If the concepts used give different meanings to different people,
it indicates that the parties are not communicating on the same wavelength.
A concept may be defined with a synonym. For research purposes, the
definition must measure concepts, thus, needing a more rigorous definition.
Operational definition is a definition stated in terms of specific testing
criteria or operations; the terms must have empirical referents (must be able
to count, measure or gather information in an objective manner). The
definition must specify the characteristics to study and how to observe the
characteristics. An effective operational definition ensures that two or more
Copyright Open University Malaysia (OUM)
TOPIC 1
Variables
At the theoretical level, constructs and concepts are used to relate to
propositions and theory; at this level, constructs cannot be observed. At the
empirical level, propositions are converted into hypotheses and tested; at
this level, the concepts are termed as variables. The term variable is used
as a synonym for construct or the property being studied. Quantitative
variables usually take numerals or values as the indicator of the degree of
level. The followings are some commonly used quantitative variables:
(i)
(ii)
Propositions
Propositions are statements about concepts which may be judged as true or
false if they refer to observable phenomena.
(f)
Hypothesis
Hypothesis is a proposition that is formulated for empirical testing:
(i)
(ii)
10
TOPIC 1
Model
A model is a representation of a developed system used to study some
aspects of the system or the system as a whole. It is different from theory
because theory explains relationships in the system whereas a model is a
representation of the relationships in the system.
(h)
Framework
A framework is an abstract representation of a phenomenon. It describes
the variables studied and the relationships among the variables. It can be
represented graphically in a diagram. Thus, in the early stage of a research,
a theoretical framework is usually constructed based on initial studies or
literature search. The theoretical framework is used to explain the
relationships that need to be investigated and tested in research. A
framework that has been successfully tested will be considered as the final
framework. A research will report the research findings by presenting the
final framework.
(i)
Process
A process is developed for a specific purpose in a business organisation. It
aims to make some change in the organisation. For example, lets say a
company implements a process to improve its quality performance. This
process may involve changes in the structure (for instance, someone is
transferred to a different department) or operations (for example, the
quality inspection procedure is modified) of the organisation. In research, a
process is developed to help solve an organisations problem or improve its
performance. The output of this research will be in the form of a new
process rather than a framework or model. A process is also called a tool,
procedure, method or system.
Copyright Open University Malaysia (OUM)
TOPIC 1
11
SELF-CHECK 1.2
1.
2.
3.
SELF-CHECK 1.3
1.
2.
No.
Question
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
True
False
12
TOPIC 1
TOPIC 1
Concept
Model
Construct
Generalisation
Deductive models
Process
Definition
Proposition
Empirical
Replicable
Framework
Variables
Hypothesis
13
Topic
Research
Process
LEARNING OUTCOMES
By the end of this topic, you should be able to:
1.
2.
3.
INTRODUCTION
Research usually involves a multi-stage process. Although the actual number of
stages may vary, research must include formulating and identifying a topic,
reviewing literature, planning a strategy, collecting data, analysing data and
writing a report. In discussing the research process, the presentation depicts a
stage by stage and straightforward rational discussion, although in real working
conditions of doing a research project, this is unlikely to be the case. The
researcher may have to revisit each stage more than once because each stage is
interrelated and may influence or be influenced by other stages. Each time a
researcher revisits a stage, he may have to reflect on the associated issues and
refine his ideas; in addition, he has to consider ethical and access issues during
the process.
2.1
RESEARCH PROCESS
The research process usually starts with interest in a certain event, situation,
object or just wanting to know about something. Research is the process of
gathering the information needed to answer certain questions and thereby
helping in solving problems faced by an individual, firm, organisation or society.
For information to be useful, it must be good. To get good information, the
Copyright Open University Malaysia (OUM)
TOPIC 2
RESEARCH PROCESS
15
Problem Identification
The first stage of research is to identify problems or issues and to justify the
need for research. There are many sources of research problems such as
personal interest, personal experiences, social problems, world trends, new
development in technology or society, etc.
16
TOPIC 2
RESEARCH PROCESS
(b)
(c)
Literature Review
Literature review includes the purposes of the research, the search
strategies and plan of how to undertake the research and write the review.
(d)
(e)
Research Design
In this stage, a range of research methods are available for conducting your
research. The choices are between quantitative and qualitative methods.
Sometimes the use of combined research methods is encouraged.
(f)
Data Collection
Before collecting data, you need to think about the sampling method.
Qualitative research will usually adopt a theoretical sampling method
while quantitative research will adopt probability or non-probability
sampling. You have to decide what data need to be collected such as
primary or secondary data. You will also need to think about how to access
these data and what method you will use to capture these data. There are
many ways you can collect data such as observation as well as semistructured or structured interviews. Before collecting data, research
instruments such as questionnaires need to be developed.
(g)
TOPIC 2
RESEARCH PROCESS
17
(b)
Know what those who have more information about the problem feel about
the situation and the technological prospects;
(c)
(d)
(b)
The notion that the results of the study could solve all problems;
(c)
(d)
(e)
SELF-CHECK 2.1
Identify the purpose of the research process and the main factors to
make it successful.
18
TOPIC 2
2.2
RESEARCH PROCESS
(b)
(c)
(d)
Identifying the evidence to check the questions and the hypotheses; and
(e)
Investigative Questions
Once the research problem has been identified, the researcher has to think of the
problem in a more specific or focused way; this is the investigative question.
These are questions that the researcher must ask in order to get the most
satisfying conclusion regarding the research question. The specific questions will
help in determining the types of data to be collected.
Measurement Questions
These are questions that are actually asked of respondents in order to obtain
necessary data for analysis; these are questions that appear in the questionnaire.
If the research uses an observational approach, the measurement questions take
the form of records of the observations of the subject made by the researcher.
ACTIVITY 2.1
What are the systematic/scientific steps needed to carry out a research?
TOPIC 2
RESEARCH PROCESS
19
ACTIVITY 2.2
The general manager of the company you work for calls you to his office.
He is very worried about the companys engineering department as the
turnover rate is quite high for technicians. He asks you to do a survey
among other major companies in the region to learn how they take care
of the problem of high turnover of technicians.
(a)
(b)
2.3
Data are facts which the researcher gets from the environment. Data may take
numerical or non-numeric forms of information and evidence that have been
carefully gathered according to sets of rules and established procedures. Data
may be obtained using simple observations at a specific crossroad to modern
technologically enhanced survey from big giant corporations all over the world.
The technique used to collect data will determine the methods by which data is
collected. Among the techniques used to record raw data include questionnaires,
observational forms, laboratory notes, instrument calibration logs, financial
statements and standardised instruments. Data is used in order to reject or to
accept hypotheses; and as evidence or empirical information that represent the
concept.
The characteristics of data can be examined in terms of:
(a)
Level of abstractness They are more metaphorical than real, for example,
profits cannot be observed directly but the effects can be recorded.
(b)
(c)
(d)
20
TOPIC 2
RESEARCH PROCESS
Secondary data
Data that have been collected and processed by one
researcher and reanalysed for a different purpose by another researcher.
(b)
Primary data Data that has close proximity to the truth and control over
error, so careful designing for the collection of the data becomes pertinent.
2.4
2.5
One of the problems faced by students when it comes to research is choosing the
right topic. Getting the right topic will help in designing the suitable steps in
carrying out the research effectively. Below are some of the important
terminologies one should understand before deciding on the research topic:
(a)
TOPIC 2
RESEARCH PROCESS
21
Example 2.1
A subject/general problem must lead to a good topic one that raises
some questions which have not been answered to the satisfaction of all
authorities on the topic.
Remember the purpose of research: To explain/describe, or to
illustrate/explore, or to argue for/determine causal relationship, or to
forecast/control. Research is more than mere reporting or just finding
information. The researcher should be able to evaluate the information
and ideas discovered and to arrive at a clear, well-thought conclusion
that gives the reader something to think about or to use in solving
problems.
(b)
(c)
22
TOPIC 2
RESEARCH PROCESS
Hypothesis
The predictions (of the eventual thesis), made sometime
before reading the sources, as to what the research will reveal about the
topic i.e. what answers are expected to be found for the major questions
raised by the topic. As can be seen, the hypothesis (educated guess) can
help the researcher to find exactly what information (data, methods) is
needed as quickly and efficiently as possible, by keeping attention focused
on a limited number of specific aspects of the topic. A carefully worded
hypothesis can greatly reduce problems of searching for sources and
extracting from them the most useful information. In other words, the
hypothesis points to the right direction by indicating the specific questions
that need answers. The information/answer that either agrees or disagrees
with the hypothesis will bring the researcher closer to the truth, which is
the thesis of the researcher.
Forming the hypothesis should be done while choosing the topic. This is
because the topic involves unanswered questions and the hypothesis
predicts the possible answers. The hypothesis can thus test the
thoroughness of the research. The hypothesis should not be defended by
using only those that support it; for validity of the conclusions. Different
sources representing different viewpoints should be considered. The
mission of the research is to present readers with the full picture so that
they will have enough information to evaluate the conclusions.
TOPIC 2
RESEARCH PROCESS
SELF-CHECK 2.2
Tick True or False for each statement below:
No.
Question
1.
2.
3.
4.
5.
6.
7.
8.
True
False
(b)
(c)
(d)
23
24
TOPIC 2
2.
3.
RESEARCH PROCESS
(b)
(c)
(d)
(b)
(c)
(d)
TOPIC 2
RESEARCH PROCESS
Concept
Empirical framework
Conceptualism
Exploratory study
Construct
Halo effect
Deductive models
Hypothesis
Definitions
Inductive model
25
Topic
Review of
Literature
LEARNING OUTCOMES
By the end of this topic, you should be able to:
1.
2.
3.
4.
INTRODUCTION
The review of literature is not properly understood by some learners. Some have
the opinion that literature review means collecting and compiling facts for the
research being undertaken. In fact, the literature review process needs analytical
thinking, critiquing ability and empirical approach. Review of literature is an
integral part of the entire research process. When you undertake a research
process, review of literature will help you to establish the theoretical roots of your
field of interest, clarify your ideas and develop your methodology. The review of
literature also helps you to integrate your findings with the existing body of
knowledge. You must remember that one of your important responsibilities in
research is to compare your findings with those of others, and that is why review
of literature plays a very important role in the research process.
TOPIC 3
3.1
REVIEW OF LITERATURE
27
The aim of literature review is to highlight what has been done so far in the
field of interest and how your findings relate to earlier research. The review of
literature also indicates the following:
(a)
Approaches;
(b)
Methods;
(c)
(d)
Statistical procedure.
ACTIVITY 3.1
List some obstacles that learners may face in doing a literature review
for their theses or research reports. Discuss your answer during your
tutorial.
28
3.2
TOPIC 3
REVIEW OF LITERATURE
TOPIC 3
REVIEW OF LITERATURE
29
Figure 3.2 lists the main reasons why literature review is important.
(b)
(c)
30
TOPIC 3
REVIEW OF LITERATURE
It is fundamental that you know what others are doing in your field of
interest or similar topics as well as understand theories that have been put
forward and gaps that exist in the field.
(d)
(e)
SELF-CHECK 3.1
3.3
1.
2.
It is important for you to have a specific idea of what you want to research before
embarking on literature review. There is danger in reviewing literature without
having a reasonably specific idea of what you want to study. It can condition
your thinking about your research and the methodology you might prefer,
resulting in a less innovative choice of research problem and methodology.
Therefore, try to draft your main idea before reviewing literature. Generally,
there are four steps in literature review, as depicted in Figure 3.3.
TOPIC 3
REVIEW OF LITERATURE
31
(ii)
32
TOPIC 3
REVIEW OF LITERATURE
Note the theories put forward, critiques, methods used (sample size,
data used, measurement procedure);
(ii)
(iii) Find differences of opinions among researchers and jot down your
opinions about their validity; and
(iv) Examine the gaps that exist in the body of knowledge.
(c)
(d)
Assumptions of research
Copyright Open University Malaysia (OUM)
TOPIC 3
(ii)
REVIEW OF LITERATURE
33
(iii) Hypotheses
(iv) Research designs applied
(v)
Variables selected
3.4
(b)
(c)
There is some evidence to suggest that students sometimes do not read the
original works and instead take someone elses work and cite it as though they
had read the primary source.
SELF-CHECK 3.2
1.
2.
34
TOPIC 3
3.5
REVIEW OF LITERATURE
Writing your literature review is essential as it enables you to interpret the works
of other researchers. How do you go about evaluating journal articles or
proceedings? The procedure for evaluating journal or research articles is shown
in Figure 3.4.
What was the research about? Are the objectives or aims of the study
specified clearly?
(ii)
TOPIC 3
(b)
REVIEW OF LITERATURE
35
You should keep in mind that the author is assuming that the reader
is an expert in the field and has some background knowledge about it.
(ii)
This section describes the methods used to collect data and the
background of the subjects.
(ii)
This section describes the connection between the results and the
research questions or hypotheses.
(ii)
(iii) Were the results clearly reported and presented? (e.g. use of tables
and graphs)
(iv) Did the results reflect predictions made in the Introduction section?
(e)
(ii)
36
TOPIC 3
REVIEW OF LITERATURE
ACTIVITY 3.2
Select three journals in the research area you are interested in and
identify the main contributions of those papers.
Literature review shows what has been done in the research topic and how
the intended study relates to earlier research.
Literature review consists of research findings as well as propositions and
opinions of researchers in the field.
Literature review delimits the study, relates the methods used by other
researchers as well as recommendations of earlier works and provides the
basis for the intended research task.
All journal and research articles reviewed should be critically evaluated.
Literature review can reveal methods of dealing with the research problem
that may be similar to the difficulties you are facing.
Literature review will increase your confidence in your research topic if you
find other researchers have an interest in this topic and have invested time,
effort and resources in studying it.
Abstracts of articles
Research journal
Body of knowledge
Review of literature
Citation indices
Theoretical framework
Indices of journals
Topic
Sampling
Design
LEARNING OUTCOMES
By the end of this topic, you should be able to:
1.
2.
3.
4.
INTRODUCTION
This topic introduces strategies to collect primary data. The process of collecting
primary data must be identified properly based on the purpose and objectives
of the research. Data used to answer research questions must come from the
appropriate population in order to be useful. If data is not collected from
the people, events or objects that can provide the correct answers to solve the
problem, then the process of collecting the data is a waste. The process of
selecting the right individuals, objects or events for study is known as sampling.
4.1
SAMPLING CONCEPT
Whatever the research questions and objectives of the study may be, a researcher
must collect data to answer them. If the researcher collects and analyses data
from every possible member, this is known as a census. However, most
researchers are faced with the limitation of resources, time, and often access,
which made it impossible to collect or analyse all the data. Sampling techniques
provide a range of methods that enable the researcher to reduce the amount of
data needed, by considering only data from a subgroup rather than from all
possible cases or elements (refer to Figure 4.1).
Copyright Open University Malaysia (OUM)
38
TOPIC 4
SAMPLING DESIGN
In order to ensure that the data collected is representative, a few terms related to
the concept of sampling must be understood.
(a)
Population
Total collection of elements or cases in which to make
inferences; it refers to the entire group of people, events or things of interest
that the researcher wants to study.
(b)
Element
taken.
(c)
Census
(d)
(e)
Sample
A subset of the population; it is made up of some members
selected from the population. These are some, not all, elements of the
population that form the sample.
(f)
(g)
Sampling Frame
TOPIC 4
SAMPLING DESIGN
39
ACTIVITY 4.1
How can sampling techniques help to obtain good research results?
4.2
The reasons for using a sample are many; in research investigations involving
several hundreds and even thousands of elements, it would be impractical to
collect data, test or examine every element. Consider the cost of using a census,
the time and the human resources needed; they are prohibitive. The quality of the
information obtained from a sampling study is likely to be more reliable than
from a census; this is mostly because fatigue is reduced and fewer errors will
result in collecting the data, especially if a large number is involved. In some
situations, sampling is required. For example, in testing the life of an electric
bulb, it would be impossible to test the entire population or if burn them, then
there would be none to sell.
40
TOPIC 4
SAMPLING DESIGN
SELF-CHECK 4.1
1.
2.
4.3
TOPIC 4
SAMPLING DESIGN
41
ACTIVITY 4.2
Although a researcher cannot get 100% accuracy in the research
findings, why is it still important to have a good sample design?
4.4
Probability sampling
With probability sampling, the chance or probability of each case being
selected from a population is known and is usually the same. It is based on
the concept of random selection, which is a controlled procedure that
assures each population element, or case is given a known non-zero chance
of selection. By using a probability sample, it is possible to answer research
questions and achieve objectives of estimating characteristics of the
population from the sample. Thus, probability sampling is often used in
surveys and experimental research. Researchers use a random selection of
the elements to reduce or eliminate sampling bias.
(b)
Non-probability sampling
In non-probability sampling, the probability of each case being selected
from the total population is not known; and it is impossible to answer
research questions or address objectives that require statistical inferences
about the characteristics of the population. Although generalisations could
still be made from non- probability samples about the population, it cannot
be done on statistical grounds. For this reason, non-probability sampling is
often used in a case study research.
42
TOPIC 4
SAMPLING DESIGN
SELF-CHECK 4.2
How many types of sampling designs are there in a research study
and which one is the most often used?
4.4.1
(b)
TOPIC 4
SAMPLING DESIGN
43
and checking the data. Thus, the determination of the sample size within
this compromise is influenced by:
(i)
The confidence level in the data that is the level of certainty that the
characteristics of the data collected will represent the characteristics of
the total population.
(ii)
The margin of error tolerated the level of accuracy required for any
estimate made from the sample.
(d)
SELF-CHECK 4.3
If Harun calculated that the adjusted minimum sample size was 439
and his estimated response rate was 30%, what would his actual
sample size be?
44
TOPIC 4
4.4.2
SAMPLING DESIGN
(b)
Systematic Sampling
In systematic sampling, an element of the population is selected at the
beginning. With a random start of a range of 1 to k and following the
sampling fraction is selected for every kth element. The sampling involves
the selection of the sample at regular intervals from the sampling frame.
This sampling technique is simple to design and is flexible; it is easier to
use and more efficient than simple random sampling. It has an added
advantage of being easy in determining the sampling distribution of mean
or proportion. As it is not necessary to construct a sampling frame, it is less
expensive than simple random sampling.
Periodicity within the population may skew the sample and results. For
instance, assume the sampling fraction is k = 4, and the list contains the
names of every male followed by a female. If the first selection is a male,
then the sample will contain only male respondents. Consequently, the
sample will be biased. If the population list has a monotonic trend, listing
from the smallest to the largest element, a biased estimate will result based
on the starting point.
(c)
Stratified Sampling
This is a modification of the random sampling in which the population is
divided into two or more mutually exclusive subpopulations or strata;
based on one or a number of attributes. Then, the random selection (simple
or systematic) is used on each strata. Results may be weighted and
Copyright Open University Malaysia (OUM)
TOPIC 4
SAMPLING DESIGN
45
Cluster Sampling
The population is divided into internally heterogeneous subgroups, each
with a few elements in it. The subgroups are selected according to some
criterion of ease or availability in data collection; within subgroups there is
heterogeneity but between subgroups there is homogeneity. Samples are
taken from some randomly selected subgroups for further study.
It often involves large samples since there must be sufficient data to stratify
or cluster the population. However, if the method is indiscriminately used,
it will increase costs.
(e)
Multistage Sampling
This is a sampling method that employs more than one sampling strategy.
It usually starts with cluster sampling since it is a method of selecting a
group rather than individual elements. Having selected the groups
(clusters), the individual elements representing the groups are determined
using other probability techniques mentioned above.
SELF-CHECK 4.4
What are the factors influencing the choice of the following sampling
designs?
(a)
(b)
46
TOPIC 4
4.4.3
SAMPLING DESIGN
(b)
Another reason for choosing non-probability sampling is the lower cost and
time factor. Probability sampling is time consuming and expensive. If the
non-probability sampling is carefully controlled, it can produce acceptable
results.
(c)
(d)
The non-probability technique may be the only feasible method if the total
population is not available for the study or not known. In such cases, the
sampling frame will not be available to choose the elements. It may
not be possible to determine completely that the respondent of the mail
questionnaire is actually the person selected or the true cross section of the
population.
4.4.4
TOPIC 4
SAMPLING DESIGN
47
(c)
Quota Sampling
The design is based on the premise sample that will represent the
population; as the variability in the sample for various quota variables is
the same as that in the population. The logic is that certain relevant
characteristics describe the dimension of the population, thus making the
design a type of stratified sample in which the selection of cases within the
strata is entirely non-random. If the sample has the same distribution of
these characteristics, then it is quite likely to be a representative of the
population.
The quota sampling has several advantages over the probability sampling.
It is in particular less costly and can be set up rather quickly. It does not
require a sampling frame and may be the only technique that can be used if
other techniques are not available. It is most useful if the population is
large; since the sample size is governed by the need to have sufficient
responses in each quota to enable subsequent analyses to be undertaken,
hence the total sample size may be more than 2000.
A major weakness of the quota sampling design is that the assumption of
the quota being representative is arguable as there is no assurance that each
variable under study represents the population characteristics. Available
data used as a basis for the determination of quota may be outdated or
inaccurate, thus without relevant or sensible quotas, data collected may be
biased. The number of control variables that are used may be limited and is
often left to the choice of the field workers.
48
(d)
TOPIC 4
SAMPLING DESIGN
Snowball Sampling
This design is usually used when the respondents are difficult to identify
and located through some referrals from people who know them. The
respondents may or may not be chosen initially through probability. The
initial individuals are used to locate other individuals who have similar
characteristics, and who, in turn identify others. The referral approach can
help to reach particularly hard to find respondents; however, it may get
only individuals similar in characteristics to the introducers. The design
may result in a highly homogeneous group.
ACTIVITY 4.3
In a situation where the respondents live in rural areas, what is the
most effective type of sampling that can be used?
4.5
SAMPLE SIZE
The sample size is more often than not determined by judgement as well as
calculation. In many cases, the types of statistical analyses used would determine
the minimum sample size for each individual category. As a rule of thumb, a
sample size of 30 is the smallest number in each category within the overall
sample that is acceptable.
The sample size is the number of elements to be studied in the research project.
Determining the size is one of the great challenges of many junior researchers.
Some of the major considerations in sample size determination are namely:
(a)
(b)
(c)
(d)
(e)
TOPIC 4
4.5.1
SAMPLING DESIGN
49
Qualitative Approach
4.5.2
10
10
30
28
50
44
100
80
200
132
300
169
500
217
1,000
278
2,000
322
3,000
341
4,000
351
5,000
357
8,000
367
10,000
370
20,000
377
50,000
381
Statistical Approach
50
TOPIC 4
SAMPLING DESIGN
effect size, alpha, beta, and the population standard deviation in the calculation.
The sample size determination for confidence intervals presented in this manual
was adapted from Malhotra (1999) while the effect size approach was adapted
from Brewer (1996).
Sample Size Considerations using the Confidence Interval Approach
This approach is based on the construction of confidence intervals around the
sample means or proportions using standard error formula.
(a)
(ii)
(iii) Determine the z value associated with the level of confidence set in (b)
using the z-distribution table.
(iv) Determine the standard deviation of the population ( ). This is
determined based either on some secondary sources, empirically
derived from pilot tests, or defined judgmentally by the researcher.
(v)
Calculate the sample size using the formula for the standard error of
the mean
2
z2
D
n
(b)
(ii)
(iii) Determine the z value associated with the level of confidence set in (b)
using the z-distribution table.
(iv) Estimate the population proportion. This can be done based on
information from previous studies, derived empirically from pilot
tests, or judgmentally defined by the researcher.
(v)
Calculate the sample size using the formula for the standard error of
the proportion.
) z2
(1
D2
TOPIC 4
(c)
SAMPLING DESIGN
51
Alpha is the probability of rejecting the Null when the Null is indeed true.
Because the focus of hypothesis testing is to minimise the errors in making
a decision, an adequately small value of alpha is essential for the results to
be meaningful.
Power is the probability of correctly rejecting the Null. Since power refers to
correct rejection for the rejection to be meaningful, the power should be set
substantially high.
Effect size (ES) is the degree of association between the variables under
investigation. If the study is concerned with differences between two
populations, then the effect size refers to the magnitude of difference that
make it meaningful. A small effect size will allow the researcher to detect
even a small effect of the phenomenon. For example if it is hypothesised
that there is a true difference between male and female employees in terms
of their job satisfaction levels, a small difference in the mean scores of these
two populations (if the null is rejected) is good enough to provide evidence
of practical importance if the effect size is set to be small. A small effect size
is able to detect even small true differences if there is a difference between
the null and the alternate hypothesis.
Using this method, mainly in hypothesis testing, the minimum sample size
is defined as a function of alpha, power and the effect size (Brewer, 1996;
Cohen, 1977).
For one sample hypothesis testing, the minimum sample size is defined as
N =
Where, N
=
=
=
ES =
[(Z + Z ) /ES]2
minimum sample size
alpha (probability of type I error)
beta (probability of type II error)
effect size
52
TOPIC 4
SAMPLING DESIGN
=
=
=
ES =
2[(Z + Z ) /ES]2
minimum sample size
alpha (probability of type I error)
beta (probability of type II error)
effect size
SELF-CHECK 4.5
Tick True or False for each statement below:
No.
Question
1.
2.
3.
4.
5.
6.
True False
TOPIC 4
SAMPLING DESIGN
2.
3.
Population
(b)
Universe
(c)
Parameter list
(d)
Sample frame
Consistency
(b)
Accuracy
(c)
Precision
(d)
Reliability
Cluster sampling
(b)
(c)
Stratified sampling
(d)
53
54
TOPIC 4
SAMPLING DESIGN
SELF-CHECK 4.6
For each of the following research questions, it has not been possible to
obtain a sampling frame. Suggest the most appropriate non-probability
sampling technique to obtain the necessary data, giving reasons for
your choice.
(a)
(b)
(c)
(d)
The logic of sampling is that there are similarities among the elements in a
population that can adequately represent the characteristics of the total
population.
Some of the elements may underestimate the true value of the population,
but others may overestimate the value. The combination of these estimates
gives the statistics; which give a true value estimated population.
A good sample should be accurate; there is little or no bias or systematic
variance.
A good sample must be precise; the sampling error is within acceptable limits
for the purpose of the study.
The choice of the sampling design depends on the objectives and the research
questions of the study.
The size of the sample depends on the accuracy of the results required, the
confidence level of the study and the resources available to collect and
analyse the data.
TOPIC 4
SAMPLING DESIGN
55
The probability sample design is the ideal design, since it allows the
determination of the level of error likely to be produced. It is often time
consuming and expensive.
Stratified and systematic sampling are modifications to simple random
sampling.
A sampling frame is needed to apply probability sampling.
If the sampling frame is not possible, a non-probability design can be applied.
The non-probability techniques are a compromise between accuracy and cost
of collecting data. Non-probability sampling has many advantages especially
ease of use and reducing cost of data collection. In some instances, probability
sampling is the only feasible method of data collection.
Census
Population
Population case
Topic
Measurement
and Scales
LEARNING OUTCOMES
By the end of this topic, you should be able to:
1.
2.
3.
4.
INTRODUCTION
This topic begins with an explanation of conceptualisation and
operationalisation. The definition of concepts and the methods of measuring the
concepts will help the researcher to determine the methods of collecting and
analysing data. The process of defining concepts is important in a research to
ensure that readers have the same understanding as the researcher; this will
prevent any confusion or misunderstanding by readers in interpreting the
meaning of the concept.
Once the concept is defined, it is necessary to identify the methods to measure
the concept. Measurement of the variables is an integral part of the research
process and is an important aspect of a research design. Unless the variables are
measured in some way, the researcher will not be able to test the hypothesis and
find answers to complex research issues.
TOPIC 5
5.1
57
CONCEPTUALISATION
(b)
58
TOPIC 5
Conceptualisation may differ among researchers but definitions are specific and
unambiguous. Therefore, even if one disagrees with the definitions, he has a
good idea of how to interpret the results because the definitions are clear and
specific.
ACTIVITY 5.1
How do you define the concept of socio-economic status in terms of
nominal definition and operational definition?
5.2
OPERATIONALISATION
Once the concepts have been identified, the next step is the process of developing
the specific research procedures/operations that will result in empirical
observations representing those concepts in the real world.
The process of linking a conceptual definition to a specific set of measurement
techniques or procedures is called operationalisation. These are procedures to
measure a concept either through a collection of data from a survey research or
by conducting observation research. The following example explains this.
Example 5.1
Operationalising the concept of an individual/person: Variable Individual
Attributes Gender characteristics (male/female)
Nominal Definition An individual is either a male or female
Operational Definition If B defines/represents an individual
Mapping out attributes:
1 represents an individual who is a male
0 represents an individual who is a female
Thus, for B1, B2, B3, B4, B5, B6:
B1 is measured as 1 if B1 is a male
B2 = 0 if B2 is a female
B3 is measured as 1 if B3 is a male
B4 = 0 if B4 is a female
B5 = 1, B6 = 0
TOPIC 5
59
5.3
VARIABLES
At the theoretical level, concepts and constructs are used; whereas at the
empirical level, the constructs are transformed into variables. Thus, variables are
the construct or property to be studied. A variable consists of logical groupings
or sets of attributes/values.
An attribute is the intensity or strength of attachment to attitudes, beliefs
and behaviours associated with a concept. It is a characteristic or quality of a
concept/symbol to which numerals or values are assigned.
Two important characteristics of a variable are:
(a)
(b)
(b)
(c)
(d)
(e)
Intervening IVV shows the link between IV and DV; it acts as a DV with
respect to an IV and as an IV with respect to a DV.
ACTIVITY 5.2
What are the relationships between IV, DV and IVV? How does the
inclusion of MV change or affect the relationship?
60
TOPIC 5
5.4
MEASUREMENT
The concepts used in a research are divided into objects or properties. Objects are
things such as shirts, hands, computers, shoes, books and papers. Things that are
not so concrete such as genes, nitrogen, attitudes, stocks and peer-group pressure
are also included as objects. Properties or attributes, on the other hand, are the
characteristics of the objects.
An individuals physical characteristics are indicated in terms of weight, height
and posture. An individuals psychological attributes are shown in terms of
attitudes and intelligence. The social characteristics of the person include
leadership ability, social status or class affiliation. The object and the
characteristics can be measured in a research study.
Measuring the properties indicators of the objects makes the measurement of the
objects or characteristics more sensible. It is easy to see that A is older than B, and C
participates more than D in a group discussion. Indicators such as age, working
experience and number of reports done can be easily measured. Hence, they are so
commonly accepted that one considers the properties to be observed directly.
However, properties such as an individuals ability to solve problems,
motivation for success, political affiliation and sympathetic feelings are more
difficult to measure. Since they cannot be measured directly, they have to be
gauged by making inferences to the presence or absence of certain behaviour or
attitude by observing some indicators or pointer measurement.
Essentially, the measuring process consists of giving numbers or symbols to
empirical events based on a set of rules. The process of making the measurement
involves three steps: selecting observable objects or properties; using numbers or
symbols to represent aspects of the events or objects; and applying a mapping
rule to connect the observation to the symbol. Thus, some mapping rules are
devised to transfer the observation of the property indicators using these rules.
The accepted rules in using numbers to map the observation of the indicators
include:
(a)
Order of numbers
another number;
(b)
(c)
The number series has a unique origin indicated by the number zero.
Copyright Open University Malaysia (OUM)
TOPIC 5
61
SELF-CHECK 5.1
Why is it necessary to define the concepts of research clearly?
5.4.1
Level of Measurement
Once the operationalisation of the concepts has been established, the concepts
need to be measured in some manner. A scale is a tool or mechanism by which
individuals are distinguished based on the variables of interest in the study. The
scale or tool could be gross or fine-tuned. A gross scale broadly categorises
individuals on certain variables. A fine-tuned scale differentiates individuals on
the variables with varying degrees of sophistication.
Using these rules of order, distance and origin of the data are classified into the
following types of scales:
(a)
SELF-CHECK 5.2
Give three examples of nominal scale.
62
(b)
TOPIC 5
Least Preferred
Types of Questions
(c)
2.
Preferred
3.
Most Preferred
Ranking of Preference
(a)
Objective questions
(b)
Subjective questions
(c)
Combination of both
TOPIC 5
63
Example 5.3
Using the scale below, please indicate your choice for each of the items
that follow, by circling the number that best describes your feeling.
1.
Strongly disagree
2.
Disagree
3.
Neutral
4.
Agree
5.
Strongly Agree
(a)
(b)
(c)
(d)
The interval scale has equal magnitude of differences in the scale point. The
magnitude of difference represented by the space between 1 and 2 on the
scale is the same as the magnitude of difference represented by the space
between 4 and 5, or between any other two points. Any number can be
added to or subtracted from the numbers on the scale. Assuming the
magnitude of the difference is still retained, if a 6 is added to all five points
on the scale, the interval scale will become 6 to 11; the magnitude of the
difference between 7 and 8 is still the same as the magnitude of the
difference between 10 and 11. Thus, the origin or the starting point could be
any arbitrary number.
The interval scale taps the differences, the order and the equality of the
magnitude of the differences in the variable. It is a more powerful scale
than the ordinal and nominal scales. It allows the measuring of the central
tendency, mean, dispersion, range, standard deviation and variance.
(d)
64
TOPIC 5
Measure of
Central
Tendency
Difference
Order
Distance
Unique
Origin
Nominal
Yes
No
No
No
Mode
Ordinal
Yes
Yes
No
No
Median
Interval
Yes
Ratio
Yes
Yes
Yes
Yes
Yes
Measure of
Dispersion
No
Standard
deviation,
Arithmetic
variance,
mean
coefficient of
variation.
Yes
Standard
Arithmetic/ deviation,
geometric variance,
mean
coefficient of
variation.
Example 5.4
(c)
How many books have you read in the last two weeks?
(d)
How many times have you visited a shopping complex in the last
month?
The measures of central tendency of the ratio scale could be either the arithmetic
or the geometric mean; and the measure of dispersion could be the standard
deviation, variance or the coefficient of variation.
ACTIVITY 5.3
1.
2.
TOPIC 5
65
SELF-CHECK 5.3
What are the essential differences among the nominal, ordinal,
interval and ratio scales?
SCALING TECHNIQUES
5.5
Four different types of scales are used to measure the operationally defined
dimensions and elements of a variable. It is necessary to know the methods of
scaling; the process of assigning numbers or symbols to elicit the attitudinal
responses of subjects towards objects, events or persons. There are two main
categories of attitudinal scales rating scale and ranking scale.
5.5.1
Rating Scales
Rating scales have several categories and are used to elicit responses with regard
to the object, event or person studied. The following are some examples of rating
scales often used in social science research.
(a)
Yes
No
66
(b)
TOPIC 5
Train station
(ii)
Grocery outlet
Others
Time
(ii)
Readers Digest
Vogue
(vi) Family
(vii) Others (specify)
(d)
TOPIC 5
67
(e)
(i)
Strongly Agree
(ii)
Agree
(iv) Disagree
(v)
Strongly disagree
(f)
Good
Unfair
Dirty
Traditional
Numerical Scale
The numerical scale is similar to the semantic scale, with the difference that
numbers on a five-point or seven-point scale are provided, with the bipolar
adjectives at both ends. The scale used is an interval scale.
How do you feel about the idea of war?
Bad
Fair
Clean
Modern
1
1
1
1
2
2
2
2
3
3
3
3
4
4
4
4
5
5
5
5
Good
Unfair
Dirty
Traditional
68
(g)
TOPIC 5
(h)
100
Staple Scale
The staple scale provides simultaneous measures of the direction and
intensity of the attitude towards the items under study. The characteristics
of interest to the study are placed at the centre and a numerical scale
ranging from +3 to 3 are put on either side of the item. The scale gives an
idea on the gap of the individual response to the stimulus. It does not have
an absolute zero, thus it is an interval scale.
Please indicate how you would rate the restaurant with respect to each
of the characteristics mentioned below, by circling the appropriate
number.
Services
Cleanliness
Prices
5.5.2
3
3
3
2
2
2
1
1
1
+1
+1
+1
+2
+2
+2
+3
+3
+3
Ranking Scale
TOPIC 5
69
Paired Comparison
In using this scale, the respondents are asked to choose among a small
number of objects; two objects at a time. This helps to assess preferences
because the respondents can express attitudes unambiguously by choosing
between two objects. The number of paired comparisons that will be judged
by the respondents for n objects is {(n)(n 1)/2}. If n = 4, then the number
of paired comparisons will be {(4)(4 1)/2 = 6}. The greater the number of
objects, the greater the number of paired comparisons that will be
presented to the respondents. This will tire the respondents mentally. This
technique is good if the number of objects is small.
Example 5.5
For each pair of national parks, place a check beside the one you most
prefer if you had to choose between the two.
(b)
(i)
(ii)
(iii)
(iv)
(v)
Forced Choice
This choice enables the respondents to rank objects relative to one another,
among alternatives provided. This is easier for the respondents, especially if
the number of choices to be ranked is limited in number.
70
TOPIC 5
Example 5.6
Please rank the following daily newspaper you would like to subscribe
in order of preference, assigning 1 for the most preferred choice and 5
for the least preferred.
(i)
(ii)
Utusan Malaysia
Harian Metro
Comparative Scale
This scale gives a point of reference to assess attitudes towards the current
object, event or situation under study. The technique is ideal if the
respondents are familiar with the standard.
Example 5.7
Compared to your previous visit to this holiday destination, your
present visit is:
More enjoyable
1
5.6
Less enjoyable
4
MEASUREMENT QUALITY
TOPIC 5
5.6.1
71
Three major criteria are often used to determine the quality of a measurement
tool: reliability, validity and practicality.
Reliability and validity are associated with how concretely connected the
measures are to the constructs because perfect reliability and validity are
impossible to achieve. It is important to establish the truthfulness, the credibility
or the believability of findings, with no random or systematic errors. Thus,
reliability and validity are considered as the scientific criteria of the
measurement.
Reliability is related to the consistency of the measurement, which means the
recurrences are measured with an identical method or under very similar
conditions. If a particular technique is applied repeatedly to the same object and
yields the same result each time, then this indicates consistency. The criteria take
into account the degree to which the measurement is free of random error.
Reliability can be assessed by posing the following questions (Easterby-Smith et
al., 2002):
(a)
(b)
(c)
Is there transparency in how sense was made from the raw data?
(b)
Maturation Effects The time passage of the relationship can influence the
cause and effect among variables and cannot be controlled. The maturation
effects are a function of processes operating within the respondents as a
result of the time passage. Examples of maturation processes include
growing older, getting tired, getting bored and feeling hungry.
Copyright Open University Malaysia (OUM)
72
TOPIC 5
(c)
(d)
Instrumentation Effects
The effects on validity may occur because of
changes in the measuring instrument between pre-test and post-test.
Example 5.8
Relationship between reliability and validity are shown using this example:
You use a bathroom scale to measure your weight. If the scale measures your
weight correctly, then the scale as a measuring tool is both reliable and valid.
If the scale is tampered and consistently gives an overweight of 6 kg every
time it is used, it is reliable but not valid. If the scale gives an erratic weight
reading from time to time, it is neither reliable nor valid.
Practicality is correlated with the operational requirement of the measurement
process. The criterion of practicality involves the aspects of economy,
convenience and interpretability. To achieve a high degree of reliability and
validity, one may require high expenditure that may be beyond the budget for
research; thus there has to be some form of trade off between the ideal measures
and the budget. Data collection techniques are always dictated by budget
constraints and other economic factors.
The measuring device should also be easy to administer; the design of the
instruments used should allow easy comprehension and have complete and clear
instructions. If the instrument is to be administered by people other than the
designer, then it must also be easy to interpret.
5.7
In an ideal situation, a study design should be able to control the precision and
ambiguity of the measurement. However, an ideal situation is impossible.
Therefore, the next best thing to do is to go for the reduction of errors. The
researcher should be aware of the sources of potential errors such as systematic
and random errors.
(a)
TOPIC 5
73
Situational Factors
Any condition that may place strains on the interview can have serious
effects on the interview-respondent rapport. If the interview is carried out
in the company of other people, friends, relatives and children, the
responses can be distorted by others joining in, distractions or by others just
merely being there. Some may feel they are being intruded upon and thus
may not willingly give their responses.
(c)
(d)
74
TOPIC 5
SELF-CHECK 5.4
What are the four major sources of measurement errors? Give an
example of how each source can affect the measurement results in a
face-to-face interview.
SELF-CHECK 5.5
Tick True or False for each statement below.
No.
Question
1.
2.
3.
4.
5.
6.
Classifying
someone
as
employed
or
unemployed treats employment as a nominal
variable.
True
False
(b)
(c)
(d)
TOPIC 5
2.
3.
75
(b)
(c)
(d)
What to measure
(b)
(c)
How to measure
(d)
76
TOPIC 5
The ratio scale indicates the magnitude and proportion of the differences.
The data becomes more precise when we move from the nominal to the ratio
scale and allow the use of more powerful statistical tests.
Sound measurement must meet the criteria of validity, reliability and
practicality.
Validity reveals the degree to which an instrument measures what it is
supposed to measure.
A measure is reliable if it provides consistent results each time it is used.
Reliability is a partial contributor to validity but a measurement tool may be
reliable without being valid.
A measure meets the criteria of practicality if it is economical, convenient and
interpretable.
Conceptualisation
Operational definition
Dichotomous scale
Operationalisation
Internal measure
Ranking scale
Measurement
Ratio measure
Nominal definition
Staple scale
Nominal measure
Variables
Ordinal measure
and Secondary
Data
LEARNING OUTCOMES
By the end of this topic, you should be able to:
1.
2.
3.
4.
5.
6.
INTRODUCTION
The type and amount of data collected depends on the nature of the study
together with its research objectives. If the study is exploratory, the researcher is
likely to collect narrative data through the use of focus groups, personal
interviews or observation of behaviour or events. These types of data are known
as qualitative.
Qualitative approaches to data collection are typically used at the exploratory
stage of the research process. Their role is to identify and/or refine research
problems that may help to formulate and test conceptual frameworks. Such
studies normally involve the use of smaller samples or case studies.
78
TOPIC 6
6.1
SURVEY RESEARCH
Survey research is a common tool for applied research. Surveys can provide
a quick, inexpensive and accurate means to obtain information for a variety
of objectives. The typical survey is a descriptive research study that has the
objective of measuring awareness, knowledge, behaviour, opinions and the like.
Surveys can also be used to collect data for explanatory or analytical research to
enable researchers to examine and explain relationships between variables; in
particular cause and effect relationships. The term sample survey is often used
because a survey is expected to obtain a representative sample of the target
population.
Surveys are popular because they allow the collection of a large amount of data
from a sizeable population in a highly economical way. This data is standardised
and often obtained by using a questionnaire to allow for easy comparison. In
addition, the survey strategy is perceived as authoritative by people in general.
Every day, a news bulletin or a newspaper article reports the results of a new
survey indicating a certain percentage of the population that thinks or behaves in
a particular way. The reliability and validity of the findings in survey depends on
the quality of the instrument used.
Among the popular instruments in survey research are questionnaire and
observation inventory.
Methods of collecting survey data fall into two broad categories: self-completion
and interviewer-administered.
Self-completion methods include mail and electronic surveys. Intervieweradministered methods involve direct contact with the respondents through
personal interviews, including face-to-face, telephone and computer dialogue.
TOPIC 6
79
SELF-CHECK 6.1
Explain the difference between questionaire and observation inventory.
Explain the use of these instruments by providing appropriate
examples.
6.2
PERSONAL INTERVIEW
6.2.1
Structured Interview
In a structured interview, the interviewer uses an interview sequence with
predetermined questions. For each interview, the interviewer is required to
use the same interview sequence and to conduct the interview in the same
Copyright Open University Malaysia (OUM)
80
TOPIC 6
Semi-structured Interview
In this approach, the researcher is free to exercise his or her own initiative
to follow up with the interviewee for his or her responses. The interviewer
may want to ask related, unanticipated questions that were not originally
included in the interview. This approach may result in discovers of
unexpected and insightful information, thus it may enhance the findings.
The semi-structured interview has an overall structure and direction but
allows more flexibility to include unstructured questioning. Perhaps the
best-known semi-structured interview approach is the focus group. Focus
groups are semi-structured interviews that use an exploratory research
approach and are considered as part of qualitative research. Focus groups
are structured within a list of topics and/or questions prepared by
moderator. However, they can be unstructured if the moderator allows
participants to answer questions in their own words and encourages them
to elaborate on their responses.
(c)
Unstructured Interview
An unstructured interview is conducted without an interview sequence.
This allows the researcher to elicit information by engaging the interviewee
in an open discussion on the topic of interest. A particular advantage of this
approach is that the researcher has the opportunity to explore in-depth
issues raised during the interview.
Unstructured interviews are used when a research is directed towards an
area that is relatively unexplored. By obtaining a deeper understanding of
the critical issues involved, the researcher is in a better position to not only
better define the research problem but also to develop a conceptual
framework for the research. This will then form the basis for subsequent
empirical research to test the ideas, concepts and hypotheses that emerge.
TOPIC 6
(d)
81
In-depth Interview
An in-depth interview is an unstructured one-to-one discussion session
between a trained interviewer and a respondent. Respondents are usually
chosen carefully because they have some specialised insight. For example, a
researcher exploring employee turnover might conduct an in-depth
interview with someone who has worked for five different restaurants in
two years. Like a focus group, the interviewer first prepares an outline that
guides the interview (this is the structured part of an in-depth interview).
The responses are usually unstructured. Indeed, an in-depth interview
allows deeper probing than a focus group. The researcher probes into a
response to identify possibly hidden reasons for a particular behaviour. Indepth interviews can be very useful in clarifying concepts. Administering
in-depth interviews is similar to coordinating a focus group.
SELF-CHECK 6.2
You have been asked by the management to carry out a study on
sexual harassment at the workplace after the female employees
expressed their concerns on the matter. Which method would you
choose to collect data?
6.2.2
82
(b)
TOPIC 6
choose
compulsory
method
to
make
Length of Interview
If the research objective requires a lengthy questionnaire, personal
interviews may be the only alternative. Generally, telephone interviews last
fewer than 10 minutes, whereas a personal interview can be much longer,
Copyright Open University Malaysia (OUM)
TOPIC 6
83
perhaps an hour and a half. A rule of thumb for mail surveys is that they do
not exceed more than six pages.
(d)
Complete Questionnaires
Social interaction between a well-trained interviewer and a respondent in a
personal interview increases the likelihood that a response will be given to
all items on the questionnaire. The respondent who is bored with a
telephone interview may terminate the interview at his or her discretion by
hanging up the phone. A respondents self-administration of a mail
questionnaire requires more effort. Rather than writing a long explanation,
the respondent may fail to complete some of the questions on the selfadministered questionnaire. Failure to provide the answer to a question is
less likely to occur with an experienced interviewer and face-to-face
interaction.
(e)
(f)
High Participation
While some people are reluctant to participate in a survey, the presence of
an interviewer generally increases the percentage of people willing to
participate in the interview. Respondents are generally not required to do
any reading or writing. All they have to do is talk. Most people enjoy
sharing information and insights with friendly and sympathetic
interviewers. Personal interviews can be conducted at the respondents
home or office or other places. The locale for the interview generally
influences the participation rate. Interestingly, personal interviews are
being conducted in shopping malls even though research has shown that
the refusal rate is highest when respondents are shopping in a mall.
6.2.3
There are numerous advantages to personal interviews but there are some
disadvantages as well. Respondents are not anonymous and therefore are
reluctant to provide confidential information to another person. There is some
evidence that the demographic characteristics of the interviewer influence
respondents answers. For example, one research study revealed that male
interviewers produced a larger variance than females in a survey where 85
percent of the respondents were female. Older interviewers and interviewing
older respondents produced more variance than other age combinations,
Copyright Open University Malaysia (OUM)
84
TOPIC 6
Cost
Personal interviews are generally more expensive than mail and telephone
interviews. The geographical proximity of respondents, the length and
complexity of the questionnaire, and the number of non-respondents can
affect the cost of the personal interview.
(b)
Anonymity of Respondent
A respondent is not anonymous and may be reluctant to provide
confidential information to another person. Researchers often spend
considerable time and effort to phrase sensitive questions so that social
desirability bias will not occur. For example, the interviewer might show a
respondent a card that lists possible answers and ask him or her to read a
category number rather than verbalise sensitive answers.
(c)
Callbacks
When a person selected to be in the sample cannot be contacted on the first
visit, a systematic procedure is normally initiated to call him or her back at
another time. Callbacks, are the major means of reducing non-response
error. The cost of an interviewer calling back on a sampling unit is more
expensive (per interview) because subjects who were initially not at home
are generally more dispersed geographically than the original sampling
units.
TOPIC 6
85
Callbacks are important because individuals who are away from home at
point of call (working women) may vary from those who are at home (nonworking women, retired people, etc).
ACTIVITY 6.1
1.
2.
6.3
TELEPHONE INTERVIEW
6.3.1
86
(b)
TOPIC 6
6.3.2
Speed
In telephone interviewing, the speed of data collection is a major
advantage. For example, union officials who wish to conduct a survey on
members attitudes towards a strike may conduct a telephone survey
during the last few days of the bargaining process. Rather than taking
several weeks for data collection by mail or personal interviews, hundreds
of telephone interviews can be conducted overnight. When the interviewer
enters the residents answers directly into a computerised system, the rate
of data processing escalates.
TOPIC 6
87
(b)
Cost
As the cost of personal interviews continues to increase, telephone
interviews are becoming relatively inexpensive. Telephone interviews cost
approximately 40 percent less than the cost of personal interviews. Costs
are further reduced, when travelling costs are eliminated and the
interviews are centralised and computerised.
(c)
(d)
Cooperation
In some neighbourhoods, people are reluctant to allow a stranger to come
even to the doorstep. The same individual, however, may be willing to
cooperate in a telephone survey. Likewise, interviewers can be reluctant to
conduct face-to-face interviews in certain neighbourhoods, especially
during evening hours. Some individuals will refuse to participate and the
researcher should be aware of potential non-response bias. The likelihood
of an unanswered call and not-at-home respondent varies by the time of
day, the day of the week and the month of the year.
(e)
Callbacks
Situations like an unanswered call, a busy signal or a respondent who is not
at home require a callback. Telephone callbacks are less expensive than
personal interview callbacks. Houses with telephone answering machines
are more common nowadays. Although their effect has not been studied
extensively, it is clear that many individuals will not return a call to help
someone conducting a survey. Some researchers argue that leaving a
proper message on an answering machine will produce return calls. The
message left on the machine should explicitly state that the purpose of the
call is not sales-related. Others believe no message should be left on the
machine because respondents can be reached eventually if the researcher
calls back. Many people do not allow their answering machines to record
Copyright Open University Malaysia (OUM)
88
TOPIC 6
100 percent of their calls. If enough callbacks are made at different times
many respondents can be reachable through the telephone.
(f)
Representative Samples
When the study group consists of the general population, researchers may
face difficulties in obtaining a representative sample based on listings in the
telephone directory. In most developed countries, the majority of the
households have telephone connections. The poor and those in rural areas
may be a minor segment of the market but unlisted phone numbers and
new numbers not printed in the directory are a greater problem. Unlisted
numbers fall into two groups: those unlisted because of mobility and those
unlisted by choice.
(g)
(h)
Limited Duration
One major disadvantage of the telephone interview is the length of the
interview is limited. Respondents who feel they have spent too much time
in the interview will simply hang up. Refusal to cooperate with interviews
is directly related to interview length. A major study on survey research
found that for interviews of 5 minutes or less, the refusal rate was 21
percent. For interviews of 6 12 minutes, the refusal rate was 41 percent. For
interviews of 13 minutes or more, the refusal rate was more than 47 percent.
A thirty minutes frame is the maximum time most respondents will spend,
unless they are interested in the survey subject. (In unusual cases, a few
interested respondents may put up with longer interviews.) A good rule of
thumb is to plan telephone interviews for a ten-minute period long
(Struebbe, 1986).
ACTIVITY 6.2
Do you think that the interviewers can get accurate information from
telephone interviews? What if the respondents give biased answers?
How can the interviewers be certain?
TOPIC 6
89
SELF-CHECK 6.3
What are the major advantages and disadvantages of the telephone
interview method?
6.4
SELF-ADMINISTERED SURVEY
ACTIVITY 6.3
Suggest a good approach to attract a respondent to do a selfadministered survey. Why do you think it is good?
90
TOPIC 6
6.4.1
Mail Survey
A mail survey is a self-administered questionnaire sent to respondents
through the mail. This method presents several advantages and
disadvantages.
(i)
Geographical Flexibility
Mail questionnaires can reach a geographically dispersed sample at
the same time and incur a relatively low cost because interviewers are
not required. Respondents in isolated areas (like farmers) or those
who are otherwise difficult to reach (like executives) can be easily
contacted by mail.
Self-administered survey questionnaires can be widely distributed to
a large number of employees, allowing the diagnose of organisation
problems to be accomplished quickly at a low cost. Questionnaires
can be administered during group meetings. An hour long period
may be scheduled during the working day so that employees can
complete a self-administered questionnaire. These meetings generally
allow the researcher to provide basic instructions to a large group
(generally fewer than 50 people) and to minimise data collection time.
They also give the researcher the opportunity to debrief subjects
without spending a great deal of time and effort.
(ii)
TOPIC 6
91
E-mail Surveys
Questionnaires are now being distributed electronically via electronic mail
(e-mail). E-mail is a relatively new method of communication, however,
there are still so many individuals who have no access to it yet. Yet, certain
circumstances allow for e-mail surveys, such as internal employee surveys
Copyright Open University Malaysia (OUM)
92
TOPIC 6
or surveys of retail buyers who regularly deal with the organisation via email. The benefits of this method include cheaper distribution and
processing fees, faster turnaround time, more flexibility, and less paper
chasing.
(c)
Internet Surveys
A typical Internet survey appears when a computer user intentionally
navigates a particular website. Questions are displayed on the website. The
respondent typically provides an answer by highlighting an answer or by
clicking an icon. In some instances, the visitor cannot venture beyond the
survey page without providing information for the organisations
registration questionnaire. When cooperation is voluntary, response rates
are low and participants tend to be more deeply involved with the subject
of the research than the average person.
SELF-CHECK 6.4
1.
2.
6.5
Secondary data include both raw and published summaries and can include both
quantitative and qualitative data. They are used in descriptive and explanatory
research. The secondary data may be raw data, where there has been little or no
processing at all, or it can be compiled data that has been processed or selected
and summarised. Secondary data is mostly used in business and management
case studies.
6.5.1
This type of secondary data is used with primary data collection methods or with
other secondary data and is often used in historical research. If the researcher
uses secondary data exclusively, then it is called archival research. However
historical research may also use recent data as well as historical data.
TOPIC 6
93
6.5.2
Types of Documents
Personal Documents
Personal documents such as diaries, letters, notebooks and personal files in
computers can be used as a primary source of data. Personal documents
can be used to trace history or events that happened in the past as well as
the opinions and feelings of individuals.
(b)
Public Documents
Public documents are good sources of data. A great deal of information can
be obtained from public documents such as government reports, economic
growth of a sector, official statistics on manufacturing growth, investor
records, etc. Public documents not only provide a large amount of
quantitative data but also a potential source of a lot of textual material.
(c)
Internal Documents
Internal documents are available from most organisations. Some of the
internal documents are widely available in public such as a companys
annual reports, press releases, catalogues and product brochures,
advertisements and other information on the company website. Other
internal documents that are not publicly available are minutes of meetings,
newsletters, companys notice board, memos, letters, working procedures,
technical drawings, production and maintenance schedules, quality reports,
and inventory records. Internal documents can be used to describe the
company performance and to analyse its performance, strength and
weaknesses. However, the most difficult part of data collection is to get
access to the company.
Copyright Open University Malaysia (OUM)
94
TOPIC 6
(d)
Mass Media
Mass media source of data can be categorised as newspapers, magazines,
radio broadcasting, television programmes, films and banners. Using mass
media as a source of data, credibility and authenticity is frequently an issue
of debate. This is mainly because the evidence is usually an issue. Articles
written in the newspaper or magazine are usually unclear, biased and
without proper justification. Researchers have to make sure that they
always follow the proper scientific process in doing research.
(e)
Internet
The Internet is a rapidly growing source of information. More people are
getting access to the Internet today and are using it as a quick reference.
However, like mass media,
information from the Internet can be
questionable. The authenticity and credibility of the Internet source is an
issue. This is mainly because anyone could put up anything on the Internet.
6.5.3
These are data collected by questionnaires that have been analysed for their
original purpose. The data may be compiled in the form of data tables or as a
computer-readable matrix of raw data. Survey-based secondary data may be
obtained by census, regular surveys or ad-hoc surveys.
The government usually carries out censuses where participation is obligatory.
The purpose is to collect data on the population to meet the needs of government
departments and local departments. The data collection is usually well-defined,
well-documented and of high quality. Individual researchers and organisations
can access these data for their own researches.
Regular surveys are those undertaken repeatedly over time or at regular intervals
by various organisations. They may be used for comparative purposes,
monitoring purposes or general purposes by public organisations, nongovernmental organisations or private firms. The data may have gone through
detailed analyses and the results of the surveys may be kept in many different
forms. Data collected by certain private firms or organisations may not be
accessible to individual researchers if the information produced from the surveys
is sensitive in nature.
If secondary survey data is available in sufficient detail, it can be used to answer
research questions and meet the objectives of the studies. In many cases, the data
may need to be rechecked because results from some of the survey-based
secondary data take at least a couple of years to be published.
Copyright Open University Malaysia (OUM)
TOPIC 6
95
Ad-hoc surveys are usually one-off surveys and undertaken for specific
purposes. Organisations, government and independent researchers may carry
out the surveys on an ad-hoc basis. To get the relevant data requires substantive
search because of the nature of the ad-hoc surveys. The data from ad-hoc surveys
may be kept in aggregate form, thus the data may have to be reanalysed.
6.5.4
6.5.5
Triangulation
ACTIVITY 6.4
1.
2.
3.
96
TOPIC 6
6.6
Disadvantages
(b) Unobtrusive
Quickly obtained and of higher
quality.
(c) Unsuitable
aggregations
and
definitions
Aggregation
and
inappropriate
definition of data cause difficulties in
combining different data sets.
new
A researcher interested in small farm tractors finds that the secondary data
on the subject is broader, less pertinent in category and encompasses all
agricultural tractors. Moreover, the data was collected five years ago.
Copyright Open University Malaysia (OUM)
TOPIC 6
97
(b)
(c)
(d)
The Daily Gold Index reports the stock market indicator series. This
secondary data source reflects the prices of 50 non-randomly selected blue
chip stocks. This data is readily available and inexpensive, thus the source
of information may not suit the needs of individuals concerned with the
typical companies listed on the KLSE.
ACTIVITY 6.5
6.7
1.
2.
Internal Sources
There is an endless list of potential sources for secondary data. Internal
sources may be a good start for the researcher to begin searching secondary
data. Internal sources refer to data previously collected by or for the
organisation itself. The data are compiled in the form of previous primary
data collection as well as routine record inventories. Other useful internal
sources can be found in employee annual evaluation reports, salesperson
itineraries, sales invoices, company financial reports and records, customer
complaints, billing records, bank ledgers and previous strategic planning
documents.
98
(b)
TOPIC 6
External Sources
After the potential sources of internal secondary data are looked through,
the researcher must consider the external data sources. Countless volumes
of secondary data are available from both non-profit and profit
organisations. With advanced technologies for data searching, these sources
can be easily accessed and searched with an electronic search engine. The
key to a successful computer search is using useful key words in a search
engine. Most libraries have access to several search engines that can
identify potentially relevant research studies and/or data. Individuals and
private companies may also subscribe to an online database vendor for a
fee. Some provide access to print articles from trade periodicals, academic
journals and general business magazines. Others provide access primarily
to statistical data.
Secondary data is abundant online. All one needs is a good search engine
and a little imagination. Many libraries have access to many search engines
that charge a fee to use them. Table 6.2 shows a few examples.
Table 6.2: Examples of Online Sources of Secondary Data
Sources
Addresses
Ministry of Agriculture
http://agrolink.moa.my
www.statistics.gov.my
www.census.govmain/www/stat_int.ht
ml
www.bnm.gov.my
www.miti.gov.my
agrolink.moa.my/dof/statdof.html
Tourism Malaysia
www.tourism.gov.mytourism.gov.my/sta
tistics/statistics.asp
www.dca.gov.my/homeng.htm
jpbpo@stats.gov.my, hadi@stats.gov.my
www.planet-venture.de/seiten/stat.htm
Yale
www.library.yale.edu/socsci/egcmalay.
html
University:
Economic
Growth
TOPIC 6
members.tripod.com/pugahome/diario.
htm
www.dotmy.com/
irb11.tripod.com/irbhome
ideas.uqam.ca/EDIRC/malaysia.html
SELF-CHECK 6.5
What are the differences between internal sources and external
sources?
ACTIVITY 6.6
Suppose you are interested in a statistical overview of aquaculture (fish
farming) as part of an environmental analysis for a prospective
entrepreneurial business venture. How would you search for
information?
99
100
TOPIC 6
SELF-CHECK 6.6
Tick True or False for each statement below:
No.
Question
True
1.
2.
3.
4.
5.
6.
7.
False
from
TOPIC 6
101
102
TOPIC 6
Secondary data is gathered and recorded prior to (and for purposes other
than) the current needs of the researcher. It is usually historical and already
assembled, and does not require access to respondents or subjects.
Primary data is data gathered for the specific purpose of the current research.
The main advantage of secondary data is that it is almost less expensive than
primary data.
Secondary data can generally be obtained rapidly and may include
information available to the researcher.
The main disadvantage of secondary data is that it is not designed
specifically to meet the researchers needs. Therefore, the researcher must
examine secondary data for accuracy, bias and soundness.
One method for doing this is to crosscheck different sources of secondary
data.
One of the main sources of secondary data for business research is internal
proprietary sources such as accounting records.
External data is created, recorded, or generated by an entity other than the
researcher s organisation. The government, newspapers and journals, trade
associations and other organisations produce information.
Traditionally, this information has been distributed in a published form
either directly from producer to user or indirectly through intermediaries
such as through the public library.
Modern computerised data archives, the Internet and electronic data
interchange systems have changed the distribution channels for external data.
Due to the rapid changes in computer technology, they are now almost as
easily accessible as internal data. Hence, the distribution of multiple types of
related data by single-source suppliers has radically changed the nature of
research using secondary data.
TOPIC 6
Survey
Telephone Interview
Personal Interview
Mail Survey
Self-Administered Survey
Secondary Data
Structured Interview
Triangulation
Unstructured Interview
Questionnaire Survey
103
Topic
Experimental
Research
Designs
LEARNING OUTCOMES
By the end of this topic, you should be able to:
1.
2.
3.
4.
INTRODUCTION
According to Christensen, research design refers to the outline, plan or strategy
specifying the procedure to be used in seeking an answer to the research
question. It specifies such things as how to collect and analyse the data. The
design of an experiment will show how extraneous variables are controlled. The
design will determine the types of analysis that can be done to answer your
research questions and the conclusions that can be drawn from your research.
The extent to which your design is good or bad will depend on whether you are
able to get the answers to your research questions. If your design is faulty, the
results of the experiment will also be faulty. How do you go about getting a good
research design that will provide answers to the questions asked? It is not easy
and there is no fixed way of telling others how to do it. The best that can be done
is to examine different research designs and to point out their strengths and
weaknesses, and leave it to you to make the decision.
TOPIC 7
105
7.1
Research design can be thought of as the structure of research i.e. it is the glue
that holds all of the elements in a research project together. In experimental
research, a few selected symbols are used to show the design of a study.
O = Observation or Measurement (e.g. mathematics score, score on an
attitude scale, weight of subjects, etc.).
O1, O2, O3 On = More than one observation or measurement.
R = Random assignment: subjects are randomly assigned to the various
groups.
X = Treatment which may be a teaching method, counselling techniques,
reading strategy, frequency of questioning and so forth.
7.2
WEAK DESIGNS
7.2.1
One-shot Design
This is a simple design where the researcher makes a single observation without
any follow-up measure or comparison with another homogenous group. For
example, you want to determine whether praising primary school children
makes them perform better in arithmetic as in Figure 7.1. You measure arithmetic
achievement with a test. To test this idea, choose a class of Year 4 pupils and start
Copyright Open University Malaysia (OUM)
106
TOPIC 7
praising the pupils. You will find that their performance in mathematics is
significantly improved.
You conclude that praise increases the pupils' mathematics score. This design is
weak for the following reasons:
(a)
Selection Bias: It is possible that the pupils you selected as subjects were
already good in mathematics.
(b)
7.2.2
Maturation: If the time frame between the pre-test and post-test is long, it is
possible that the subjects may have matured because of developmental
changes.
(b)
Testing: Sometimes the period between the pre-test and the post-test is too
short and there is a possibility that subjects can remember their pre-test
session and give inaccurate responses.
TOPIC 7
107
ACTIVITY 7.1
Twenty pupils who had poor scores in arithmetic were taught
arithmetic using the Zandox method. Three weeks later, when they
were tested, their arithmetic scores improved. Thus, the Zandox
method improves their arithmetic performance.
1.
2.
7.2.3
The main weakness of the previous two designs is the lack of a comparison
group and the vague correlation between the practice of (praising) and
increased mathematics score. In the Non-Equivalent Post test Only Design, an
attempt is made to include a comparison group (i.e. control group) that did not
receive any praise as in Figure 7.3. The dashed lines separating the
experimental group and the control group indicate that the children were not
randomly assigned to the two groups. Hence, the two groups are non-equivalent.
Matching can be used but there is no assurance that the two groups can be
equated. The only way one can have assurance that the two groups are equated is
to assign the children randomly.
108
TOPIC 7
The three designs described are weak research designs because they do not
allow for extraneous factors that might influence the outcome of the
experiment to be controlled within the research construct. For example, if the
attitude towards mathematics and additional tuition classes in mathematics
are not controlled, it may not be possible to conclude that praise
(treatment) affects mathematics performance (dependent variable). Also,
poor research designs do not attempt to randomly assign subjects to the
groups. This introduces extraneous factor affecting the dependent measure.
Random assignment controls for both known and unknown extraneous
variables that might affect the results of the experiment.
SELF-CHECK 7.1
1.
2.
ACTIVITY 7.2
A teacher assigns one class of pupils to be the experimental group and
another class as the control group. Both groups are given a science
post-test. The pupils in the experimental group are taught by their
peers, while pupils in the control group are taught by their teacher.
1.
2.
7.3
TOPIC 7
(a)
(b)
109
What is the difference between the two designs? The after-only design relies only
on a post-test while the before-after design (as the name suggests) relies on both
a pre-test and a post-test.
7.3.1
The After-Only Research Design gets its name from the fact that the dependent
variable is measured only once after the experimental treatment. In other words,
the post-test is administered once to the experimental group and the control
group as provided in Figure 7.5.
110
TOPIC 7
students in the control group were not taught using the inductive approach.
Instead, students in this group were taught the same science content using the
traditional didactic approach (chalk-and-talk method).
In the above example, the experimental and control groups consist of two
different sets of students. This procedure is called a between-subjects design (also
sometimes known as an independent or unrelated design). One advantage of this
design is that the students are less likely to get bored, with the study because
each set of students is exposed to only one condition. In a similar vein, the
research is less susceptible to practice and order effects. However, you will need
more students to participate in your research. There is also a need to ensure that
both groups of students are homogeneous in any confounding variables that
might affect the outcome of the study. This is because different students bring
different characteristics to the experimental setting. Even though we randomly
assign students to experimental and control conditions, we might allocate
students with one characteristic to one condition by chance, and this might
produce confusing results.
Another research procedure in the after-only design is the within-subjects
design (sometimes known a repeated measures or related design). In this design,
the same students are exposed to two or more different conditions under
comparison. For example, you wish to study the effects of content familiarity on
reading comprehension performance. You can assign the same students to read
two types of passages, one familiar and the other unfamiliar, and then analyse
their comprehension performance.
One obvious advantage is that you need fewer students to participate in your
research. Besides, you will have much greater control over confounding variables
between conditions because the same students are used in both conditions. By
large, the same individual will bring the same characteristics to the conditions.
However, it is not all rosy in the within-subjects design. First, since the same
students are exposed to different conditions, they might get bored by the time
they are given the experimental treatment in the later condition. Besides, there is
an increased likelihood of practice and order effects.
One way to eliminate these effects is to introduce counterbalancing into your
design. In counterbalancing, you get half of your students to complete the first
condition followed by the second condition. You then get the remaining half of
your students to do the two conditions in the appositive order; the second
condition is given first followed by the first condition.
TOPIC 7
7.3.2
111
112
TOPIC 7
SELF-CHECK 7.2
1.
2.
SELF-CHECK 7.3
1.
2.
3.
7.4
QUASI-EXPERIMENTAL DESIGN
So far, we have examined both weak and strong experimental research designs.
However, in social science research (e.g. education) there are times when
investigators face situation in which all the requirements of a true experiment
cannot be met.
For example, sometimes it is not possible to assign students to groups which are
a requirement of strong experimental research. Due to logistic reasons, it is
challenging to randomly assign subjects to groups and so a whole class may have
to be used in the research. Is it still possible to do an experiment despite these
limitations? The answer is yes, you can use a quasi-experimental design.
According to Christensen and Johnson, a quasi-experimental design is an
experimental research design that does not provide for full control of potential
confounding variables. In most instances, the primary reason that full control is
not achieved is because participants cannot be randomly assigned.
TOPIC 7
7.4.1
113
The fact there is no random assignment means that subjects in the experimental
group and control group may not be equivalent for all variables. For example,
you could have more poor performing students in the control group compared to
the experimental group. Hence, it may be difficult to establish whether the better
performance of the experimental group is due to the treatment or because there
are more high performing students in the group.
In the non-equivalent control-group design, both groups are given first a pre-test
and then a post-test (after the treatment is given to the experimental group). The
pre-test score and the post-test score are compared to determine if there are
significant differences.
When you cannot assign subjects randomly, you can be sure that extraneous
variables or factors will influence the experiment and threaten its internal
validity. Do you leave it alone or do you take action regarding the external
threats?
Knowing that extraneous factors will creep into a quasi-experiment, a good
researcher will take steps to ensure that the subjects in the experimental group
and control group are as similar as possible, especially pertaining to important
variables such as academic ability, attitude, interest, socioeconomic status and so
forth. How do you address this issue?
114
TOPIC 7
Cook and Campbell proposed the following steps to enhance the internal validity
of the non-equivalent control-group design or quasi-experiments in general:
(a)
Selection: Ensure that subjects in the experimental and control groups are
matched in terms of important variables that may affect the results of the
experiment. For example, match subjects in terms of academic ability, IQ,
attitudes, interests, gender, socioeconomic background and so forth.
(b)
Testing: Ensure that the time period between the pre-test and post-test is
not too short that subjects are able to remember the questions given to them
earlier.
(c)
History: Ensure that events outside the experiment do not affect the
experiment. The problem is most serious when only subjects from one of
the groups are exposed to such events (e.g. motivation talks, private
tuition).
(d)
Instrumentation: Ensure that the pre-test and the post-test are similar. If a
different test is used, you should make sure that the two tests are
equivalent in terms of what it is measuring (i.e. high reliability and
validity).
7.4.2
TOPIC 7
115
116
TOPIC 7
SELF-CHECK 7.4
1.
2.
3.
7.5
During World War II, Nazi scientists conducted some gross experiments such as
immersing people in ice water to determine how long it would take them to
freeze to death. They also injected prisoners with newly developed drugs to
determine their effectiveness and many died in the process. These experiments
were conducted by individuals living in a demented society and they were
universally condemned as being unethical and inhumane.
Research in education involves human as subjects: students, teachers, school
administrators, parents and so on. These individuals have certain rights, such as
the right to privacy that may be violated if you attempt to obtain answers for
many significant questions. Obviously, this is a dilemma for the researcher as to
whether to conduct the experiment and violate the rights of subjects, or abandon
the study. Surely, you have heard people say: I guess we are the guinea pigs in
this study! or We are your white rats!.
Any researcher conducting an experiment must ensure that the dignity and
welfare of the subjects are maintained. The American Psychological Association
published the Ethical Principles in the Conduct of Research with Human
Participants in 1982. The document listed the following principles:
(a)
(b)
(c)
The principal researcher is responsible for the ethical conduct of the study
and be responsible for assistants or other researchers involved.
(d)
The researcher should make it clear to the subjects before they participate in
the study regarding their obligations and responsibilities. The researcher
TOPIC 7
117
should inform subjects of all aspects of the research that might influence
their decision to participate.
(e)
(f)
(g)
(h)
ACTIVITY 7.3
1.
1.
2.
3.
4.
5.
Dim
45
11
Bright
12
44
118
TOPIC 7
post-test
True experimental designs enable the researcher to maintain control over the
situation in terms of assignment of subjects to groups.
Examples of true designs are after-only research design, factorial design and
before-after research design.
A quasi-experimental design does not provide for full control of potential
confounding variables.
Examples of quasi-experimental designs are non-equivalent control-group
design and interrupted time series.
Researchers conducting experiments involving human subjects should
respect the confidentiality of subjects.
After-only design
Before-after design
One-shot design
Experimental design
Quasi-experimental design
Factorial design
Non-equivalent design
Topic
Qualitative
Research
Methods
LEARNING OUTCOMES
By the end of this topic, you should be able to:
1.
2.
3.
4.
INTRODUCTION
The term qualitative research is a general definition that includes many
different methods used in understanding and explaining social phenomena with
minimum interference in the natural environment. Qualitative research begins by
accepting that there are many different ways of understanding and making sense
of the world. You are not attempting to predict what may happen in the future.
You want to understand the people in that setting (e.g. What are their lives like?
What is going on for them? What beliefs do they hold about the world?) In short,
qualitative research relates to the social aspects of our world and seeks to find
out answers for the following questions:
Why do people behave the way they do?
How are opinions and attitudes formed?
How are people affected by the events occurring in their surroundings?
How and why cultures have developed in the way they have?
What are the differences between social groups?
Copyright Open University Malaysia (OUM)
120
8.1
TOPIC 8
The qualitative research method involves the use of qualitative data such as
interviews, documents and observation, in order to understand and explain a
social phenomenon. Qualitative research methods originates from social sciences
to enable researchers to study social and cultural-oriented phenomena. Today,
the use of qualitative method and analysis is extended to almost every research
field. The method generally includes respondent observation, interviews and
questionnaires and the researchers impression and perception.
A good definition is given by Denzin and Lincoln (1994):
A qualitative research focuses on interpretation of phenomena in their
natural settings to make sense in terms of the meanings people bring to
these settings.
The qualitative research method involves data collection of personal experiences,
introspection, stories about life, interviews, observations, interactions and visual
texts which are significant in peoples life. Qualitative research typically serves
one or more of the following purposes (Peshkin, 1993) (Figure 8.1):
TOPIC 8
121
8.2
8.2.1
Action Research
122
TOPIC 8
(b)
8.2.2
Case Study
TOPIC 8
123
Case studies can be in single or multiple designs. Single case design is ideal for
studying extreme cases in order to confirm or challenge a theory. Additionally it
is also used in that cases a researcher did not have access previously. However, it
is important for a researcher to be careful in interpreting what is being observed.
A multiple case design is appropriate when a researcher is keen to use more than
one case to gather data and draw upon a conclusion based on the facts. The
multiple case design confirms the evidence which enhance the reliability and
validity of a research work.
8.2.3
Ethnography
124
TOPIC 8
8.2.4
Grounded Theory
Grounded theory uses a prescribed set of procedures for analysing data and
constructing theoretical model from them. A good definition given by Glaser and
Strauss, (1967) states:
The discovery of theory from data systematically obtained from social
research.
Although it originates from social research, the method is now widely used in
other fields as well.
They also defined that a category emerges from the data and may stand by
itself as a conceptual element. The term grounded refers to the idea whereby a
theory emerged from the study is derived from and grounded in data collected
in the field rather than taken from research literature.
Grounded theory is very useful when current theories about a phenomenon
are either in-adequate or non-existent (Creswell, 1998). Data collection for
this method is field-based and is likely to change over the course of the study.
Interviews play a major role in this method but some other techniques like
observation, multimedia resources and documents may also be used.
8.2.5
Content Analysis
TOPIC 8
125
content analysis requires thorough planning from the very beginning. Research
problem or research questions need to be specified from the beginning.
Some steps in content analysis are:
(a)
(b)
(c)
(d)
A researcher should scrutinise and sort the materials based on the defined
characteristics.
SELF-CHECK 8.1
Identify types of qualitative research methods.
8.3
Organise data into several forms (i.e. database, sentences or individual words);
Copyright Open University Malaysia (OUM)
126
TOPIC 8
(b)
Peruse the data sets several times to gain a complete picture or overview of
what it contains as a whole. During the process, a researcher should jot
down short notes or summarise of the key points that suggest possible
categories or interpretations;
(c)
(d)
Finally, integrate and summarise the data for the audience. This step also
may include hypotheses that state the relationships among those categories
defined by the researcher. The data summary could be represented by
table, figure or matrix diagram.
The stages in the analysis of qualitative data are shown in Figure 8.5. It usually
begins with familiarisation of the data, transcription, organisation, coding,
analysis (grounded theory or framework analysis) and reporting (though the
order may vary).
Familiarisation
The first step of data analysis is familiarisation in which you listen to tapes
and watch video material, read and re-read field notes, and make memos
and summaries before formal analysis begins. This is especially important
when besides you, others are also involved in data collection. You have to
get familiar with the field notes they made (perhaps try to decipher their
handwriting!).
Copyright Open University Malaysia (OUM)
TOPIC 8
127
(b)
Transcription
Almost all qualitative research studies involve some degree of transcription.
What is transcription? Transcription is the process of converting audio or
video-recorded data obtained from interviews and focus groups as well as
handwritten field notes into verbatim form (i.e. written or printed) for easy
reading. Why do you have to do this? If you were to analyse directly from an
audio or video recording, there is the likelihood that you may include those
sections that seem relevant or interesting to you and ignore others. With a
transcript of everything that you observed and recorded (audio, video or field
notes), you get the whole picture of what happened and the chances of your
analysis being biased is minimised.
(c)
Organisation
After transcription, it is necessary to organise your data into sections that
is easy to retrieve. What does this mean? Say for example, in your study
you interviewed 10 teachers (30 minutes each) on their opinion about the
leadership style of their principal. It is advisable that you give each teacher
a pseudonym (e.g. Elvis, Jagger, Dina not their real name) or referred to
by a code number (e.g. T1, T2..T10). You need to keep a file that links the
pseudonym or code number to the original informants which are to be kept
confidential and destroyed after completion of the research. Names and
other identifiable material should be removed from the transcripts.
The narrative data you obtained from the 10 teachers need to be numbered
depending on your unit of analysis. In other words, you have to determine
whether you intend to analyse at the word level, sentence level or
paragraph level and they have to be numbered accordingly. Make sure that
the unit of text you use can be traced back to its original context.
You have at your disposal TWO approaches to analyse the data:
(d)
(i)
(ii)
Coding
Coding is the process of examining the raw qualitative data in the
transcripts and extracting sections of text units (words, phrases, sentences
or paragraphs) and assigning different codes. This is done by marking
sections of the transcript and giving a numerical reference, symbol,
descriptive words or category words. Most of the text (or transcript) will be
Copyright Open University Malaysia (OUM)
128
TOPIC 8
marked and given different codes which will be later refined or combined
to form themes or categories.
(e)
(f)
TOPIC 8
129
data analysis allows the researcher to set the categories and themes from
the beginning of the study. However, this approach also allows for
categories and themes that may emerge during the data analysis process
which the researcher had not stated at the beginning of the study.
Once the categories to themes have been pre-determined, specific pieces of
data are identified which correspond to the different themes or categories.
For a change, let us take an example from the medicine field. You may want
to know, for instance, about how people who had a heart attack
conceptualise the causes of the attack. From existing literature, you may
know that these can be divided into physical causes, psychological causes,
ideas of luck, genetic inheritance and so forth. You interview people who
have had a heart attack and from the interview transcript you search the
data for materials that are coded under these headings.
Using the headings, you can create charts of your data so that you can
easily read across the whole data set. Charts can be either thematic for each
theme or category across all respondents (cases) or by case for each
respondent across all themes:
(i)
Thematic Chart
THEME
Psychological
cause
(ii)
Case 2
Case 3
Case Chart
CASE 1
Theme 1
Genetic inheritance
Theme 2
Physical cause
In the chart boxes, you could put line and page references to relevant
passages in the interview transcript. You might also want to include some
text e.g. key words or quotations as a reminder of what is being referred to
(see (i) and (ii)). For example, under the theme Psychological Causes, Case 2
talks about stress in the workplace while Case 3 talks about business
failure.
130
TOPIC 8
Next, let us look at the data analysis spiral, as illustrated by Creswell, 1998, in
Figure 8.6.
ACTIVITY 8.1
1.
2.
TOPIC 8
8.4
131
There are some differences between the quantitative and qualitative approaches
in research methodology. In information and communication technology, both
methods play a significant role in facilitating the entire research process and
leading to desirable results or outcomes. Qualitative research tends to focus on
the subject or respondents instead of the perspective of the researcher. This is
also termed as the emic or insider perspective against etic or outsider
perspective. A researcher is always the main person in data collection and
analysis in qualitative approach as compared to questionnaires or tests in case of
quantitative approach.
Qualitative method also involves field work where a researcher must participate in the
setting especially for observation and interviews with respondents of the research
topic. Table 8.1 lists the differences between qualitative and quantitative research.
Table 8.1: Differences between Qualitative and Quantitative Research
Qualitative
Quantitative
Focus
Quality (features)
Philosophy
Phenomenology
Positivism
Method
Ethnography/Observation
Experiments/Correlation
Goal
Understand, meaning
Design
Flexible, emerging
Structured, predetermined
Sample
Small, purposeful
Data collection
Interviews, observation,
documents and artefacts
Analysis
Findings
Comprehensive, description
detailed, holistic
Precise, numerical
Researcher
Immersed
Detached
132
TOPIC 8
takes the form of communication of the respondents itself, extracts from research
documents, multimedia resources like audio and video recordings. These also
support the finding of a study.
SELF-CHECK 8.2
1.
2.
inductive
whereas
quantitative
follows
deductive
TOPIC 8
133
Action research
Inductive approach
Case study
Interviews
Content analysis
Ethnography
Qualitative methods
Grounded theory
Topic
Data Analysis
LEARNING OUTCOMES
By the end of this topic, you should be able to:
1.
2.
3.
4.
INTRODUCTION
The goal of most research is to provide information. There is a difference
between raw data and information. Information refers to a body of facts that is in
a format suitable for decision-making, whereas data is simply recorded measures
of certain phenomena. Raw data collected in the field must be transformed into
information that will provide answers to the managers questions. The
conversion of raw data into information requires that the data be edited and
coded so that it can be transferred to a computer or other storage medium. This
topic introduces the processes of data analysis. These comprise several
interrelated procedures that are performed to summarise and rearrange the data.
Researchers edit and code data to provide input that results in tabulated
information that will answer the research questions. With this input, researchers
could logically and statistically describe research findings.
TOPIC 9
9.1
DATA ANALYSIS
135
After data has been collected and before it is analysed, the researcher must
examine it to ensure its validity. Blank responses, referred to as missing data,
must be dealt with in some way. If the questions are pre-coded, then they can
simply be transferred into a database. If they are not pre-coded, then a system
must be developed so that they can be keyed in the database. The typical tasks
involved are data editing, which deals with missing data, coding, transformation
and data entry.
ACTIVITY 9.1
What is raw data? How is it different from primary and secondary data?
9.1.1
Data Editing
Before the collected data can be used, it must be edited. It must be inspected
for completeness and consistency. Some inconsistencies may be corrected at
this point. Editing also involves checking to see if respondents understood the
question or followed a particular sequence they were supposed to follow in a
branching question. For example, assume the researcher is using an experimental
design with two treatments. One treatment is designed to be a supportive work
environment and the other treatment is a much less supportive environment.
To verify that a respondent interpreted the treatment properly, the researcher
may conduct a manipulation check. After a respondent has answered the
questions, he or she is asked to comment on both work environments. If the
respondent indicates both work environments are equally supportive, it means
he or she did not respond appropriately to the treatment. In such situations, the
researcher may choose to remove that particular respondent from the data
analysis because he or she did not see the difference in the two work
environments.
Finally, editing may result in the elimination of questionnaires. For example, if
there is a large proportion of missing data, then the entire questionnaire may
have to be removed from the database. Similarly, a screening question may
indicate that you want to interview only persons who own their own home.
However the response in a questionnaire may indicate a particular respondent is
a renter. In such cases, the questionnaire must not be included in the data
analysis.
Copyright Open University Malaysia (OUM)
136
TOPIC 9
9.1.2
Field Editing
DATA ANALYSIS
The process of editing can be done in the field. The purpose of field editing on
the same day as the interview is to detect technical omissions (such as a blank
page on the interview questionnaire), check legibility of handwriting and clarify
responses that are logically or conceptually inconsistent. If a daily field edit is
conducted, a supervisor who edits completed questionnaires will be able to
question the interviewers who can recall the interview well enough to correct the
problem.
The number of no answers or incomplete responses to some questions can be
reduced with the rapid follow-up stimulated by a field edit. The daily field edit
also allows possible recontacting of the respondent to fill in omissions before the
situation has changed. Moreover, the field edit may indicate the need for further
interviewer training. For example, the field editor should check open-ended
responses for thoroughness of probing and correct following of skip patterns.
When poor interviewing is reflected by lack of probing, supervisors may further
train the interviewer.
9.1.3
In-house Editing
TOPIC 9
DATA ANALYSIS
137
9.1.4
Missing Data
Missing data can impact the validity of the researchers findings and therefore
must be identified and the problems resolved. Missing data typically arise
because of data collection or data entry problems. The researcher must assess
how widespread the missing data problem is and whether or not it is systematic
or random. If the problem is of limited scope, the typical solution is to simply
eliminate respondents and/or questions with missing data. When missing data is
more widespread, the researcher must deal with it differently by removing
respondents with missing data. The sample size may become too small to
provide meaningful results.
Several possible approaches can be taken to deal with missing data:
(a)
Identify the respondents and variables that have a large number of missing
data points. These respondents and/or variables are then eliminated from
the analysis.
(b)
138
TOPIC 9
DATA ANALYSIS
(c)
Assign to the item the mean value of the responses of all those who have
responded to that particular item.
(d)
Give the item the mean of the responses of this particular respondent to all
other questions measuring this variable.
(e)
Give the missing response a random number within the range for that scale.
(f)
9.2
CODING
If scanner sheets for collecting questionnaire data are used, such sheets facilitate
the entry of the responses directly into the computer without manual keying in of
the data. However, if this cannot be done, then it is perhaps better to use a coding
sheet first to transcribe the data from the questionnaire and then key in the data.
This method, in contrast to flipping through each questionnaire for each item,
avoids confusion especially when there are many questions and a large number
of questionnaires involved.
Responses could be coded either before or after the data is collected. If at all
possible, it is best to code them ahead of time. Coding means assigning a number
to a particular response so the answer can be entered into a database. For
example, if a five-point Agree-Disagree scale is used, then it must be decided if
Strongly Agree will be coded with a 5 or a 1. Most researchers will assign the
largest number to Strongly Agree and the smallest to Strongly Disagree; for
example; 5 = Strongly Agree and 1 = Strongly Disagree, with the points in
between being assigned 2, 3 or 4. A special situation arises when the researcher
has a two-category variable like gender. Some researchers use a coding approach
that assigns 1 = male and a 2 = female. It is recommended that in such instances a
coding approach be used that assigns 1 to one of the categories and 0 to the other
category. This enables greater flexibility in data analysis and is referred to as
using dummy variable coding.
When interviews are completed using a computer-assisted approach, the
responses are entered directly into the database. When self-completed
questionnaires are used, it is good to use a scanner sheet because the responses
can be directly scanned into the database. In other instances, however, the raw
data must be manually keyed into the database using a personal computer. A
most popular software known as SPSS includes a data editor that looks like a
spreadsheet that can be used to enter, edit and view the contents of the database.
Missing values typically are represented by a dot in a cell (.), so they must be
coded in a special way as was indicated earlier.
Copyright Open University Malaysia (OUM)
TOPIC 9
DATA ANALYSIS
139
Human errors can occur while completing the questionnaire, coding it or during
keying in data. Therefore, at least 10 percent of the coded questionnaires, as well
as the actual database, are checked for possible coding or data entry errors.
Questionnaires to be checked are selected by a systematic, random sampling
process.
Coding in qualitative research will be different from of quantitative research.
Research findings in raw form need to be classified and transformed into
categories or variables. Raw data in qualitative research could hardly be
associated with numbers. Thus, a researcher cannot assign numbers to the data.
The concept of open, axial and selective coding systems can be used for
qualitative data handling, interpretation and theory development.
In qualitative research, researchers usually end up with too much data. As a
result, data is coded to prevent data overload and to enable further analysis for
theory development. Open coding system is used to identify categories that are
derived from the concepts generated in a research. In this coding system, the
dimensions and properties of the concept are identified from the raw data in
order to group certain concepts into certain categories. These categories reflect
the concepts in a more abstract or higher order concept. Axial coding is used to
relate categories to their sub-categories. In other words, axial coding is used for
linking concepts at their dimensions and properties level in order to provide a
more precise and complete explanation of the categories.
Selective coding system is used as the basis for theory development. In this
coding system, categories are rearranged and reorganised in order to relate them
to a core concept. This core concept will form a framework or model to explain
the phenomenon being studied. The framework or model built upon categories
and subcategories is an important milestone for theory development because it
facilitates the process of further data collection to test the framework or model.
ACTIVITY 9.2
Explain the benefits of coding and when to use it in brief.
9.3
DATA ENTRY
140
TOPIC 9
DATA ANALYSIS
remains a mainstay for researchers who need to create a data file immediately
and store it in a minimal space in a variety of media medium.
Optical scanning instruments, the ever-present choice of testing services, have
improved efficiency. Examinees darken small circles, ellipses or sets of parallel
lines to choose a test answer. Optical scanners process the marked-sensed
questionnaires and store the answers in a file. This technology has been adopted
by questionnaire designers for the most routine data collection. It reduces the
number of times data is handled, thereby reducing the number of errors that are
introduced.
The cost of technology has allowed most researchers access to desktop or
portable computers or a terminal linked to a larger computer. This technology
enables computer-assisted telephone or personal interviews complete with
answers to be keyed in directly for processing, eliminating intermediate steps
and errors. The increase in computerised random-digit dialling encourages other
data collection innovations.
Voice recognition and response systems, while still far from mature, are
providing some interesting alternatives for the telephone interviewer. Such
systems can be used with software programmed to call specific three-digit
prefixes and generate four-digit numbers randomly, reaching a sample within a
set geographical area. Upon getting a voice response, the computer branches into
a questionnaire routine. Currently, the systems are programmed to record the
verbal answers but voice recognition systems are improving rapidly and soon
this system will be able to translate voice responses into data files.
Field interviewers can use portable computers or electronic notebooks instead of
the traditional clipboards and pencils. With a built-in communications modem or
cellular link, their files can be sent directly to another computer in the field or to a
remote site. This lets supervisors to inspect data immediately or simplify the data
processing at a central facility. Bar code readers are used in several applications
at point-of-sale terminals, inventory control, product and brand tracking, and at
busy rental car locations to facilitate the return of cars and generate invoices. This
technology can be used to simplify the interviewers role as a data recorder.
Instead of writing (or typing) information about the respondents and their
answers by hand, the interviewer can pass a bar code wand over the appropriate
codes. The data is recorded in a small, lightweight unit for translation later.
Even with these time reductions between data collection and analysis, continuing
innovations in multimedia technology are being developed by the personal
computer business. The capability to integrate visual images, audio and data may
soon replace video equipment as the preferred method for recording an
Copyright Open University Malaysia (OUM)
TOPIC 9
DATA ANALYSIS
141
ACTIVITY 9.3
People nowadays are attracted to SMS service provider advertisements,
be it for mobile ring tone services or contests. Even local TV stations use
this service to get information on certain survey questions. Why has this
phenomenon become so widely acceptable by the public even when the
charges are expensive?
9.4
DATA TRANSFORMATION
The process of changing data from its original format to a new format is called
data transformation. This process is usually done so that data can be easily
understood or achieved to meet some other research objective. For example,
when the data is measured using scales, quite often the statements are given in
negative as well as positive formats. In such cases, the researcher will reverse
code the questions that are negatively worded so a summated scale can be
calculated to interpret the results. If a scale of 5 is used, a 5 will be transformed to
a 1 and a 4 to a 2; a 3 does not have to be changed.
Data transformation is usually done to reduce bias when ages of respondents are
being studied. To reduce the biased response, respondents are asked the year
they were born. In such cases, the researcher would have to simply transform the
birth year to obtain the age of the respondents. Data transformation is required
when the researcher wants to create a new variable by respecifying the data
according to logical transformation. In many cases, the Likert scales are
combined into a summated rating. Usually, the transformed variable involves
combining the scores (raw data) for several attitudinal statements into a single
summated score.
The researcher could also calculate an average summated score that involves
dividing the total summated score by the number of variables. For example, if
three 5-point statements are used, the summated score might be 4 + 4 + 5 = 13.
Using the average summated score, the result becomes 4 + 4 + 5 = 13/3 = 4.3.
142
TOPIC 9
DATA ANALYSIS
SELF-CHECK 9.1
9.5
1.
2.
DATA ANALYSIS
The objectives of data analysis can be viewed from three aspects: to have a feel of
the data, to test the goodness of the data and to test the hypotheses developed for
the research. Getting a feel of the data can be achieved by checking the mean, the
range, the standard deviation and the variance in the data. These statistics will
give the researcher a good idea of how the respondents have reacted to each item
in the questionnaire and how effective the items and measures are.
Suppose the researcher notices that the item in the data set does not have a good
spread (range) and shows little variability. The researcher can deduce that the
question may not be understood by the respondents due to improper wording or
the respondents may not fully understand the intent of the question. If the
respondents have given similar answers to all items, the researcher may want to
check for biases (e.g. if the respondents have stuck at only certain points on the
scale). The objective of descriptive analysis is to portray an accurate profile of
persons, events or situations. The analysis could be an extension or a beginner for a
piece of exploratory research. Table 9.1 summarises data presentation by data type.
TOPIC 9
143
DATA ANALYSIS
USAGE OF DATA
Descriptive
Ranked
QUANTIFIABLE
Continuous
Discrete
Bar Chart
(data may
need
grouping)
Histogram/
Frequency
Polygon (data
must be
grouped)
Bar Chart/
Pictogram (data
may need
grouping)
Line Graph/
Bar Chart
Line Graph/
Histogram
Line = Graph/
Bar Chart
Pie Chart/Bar
Chart(data may
need grouping)
Histogram/Pie
Chart (data
must be
grouped)
Pie Chart/Bar
Chart (data may
need grouping)
Frequency
Polygon,
Histogram
(data must
be grouped)
or Box Plot
Frequency
Polygon, Bar
Chart (data may
need grouping) or
Box Plot
the
trend
for
144
TOPIC 9
DATA ANALYSIS
Source: Adapted from Saunders. M., Lewis. P., and Thornhill. A. (2003)
ACTIVITY 9.4
What are the factors that determine the choice of data analysis?
Compare your answers with those of your classmate.
9.5.1
Descriptive Statistics
In the first part of the topic, we have discussed how responses could be coded and
entered. When nominal measurements are involved, each category is represented by
its own numerical code. With ordinal data, the items rank, reflecting a position in
the range from the lowest to the highest, is entered into the system. The same is true
with interval-ratio scores. When this data is tabulated, it may be arrayed from the
lowest to the highest scores on the scales. Together with the frequency of occurrence,
the observations form a distribution of values.
Many variables of interest have distributions that approximate a standard normal
distribution. It is a standard of comparison for describing distributions of sample
data and is used with inferential statistics that assume normally distributed
variables. The characteristics of location, spread and shape describe distribution.
Their definitions, applications and formulas fall under the heading of descriptive
statistics. Although the definitions will be familiar to most readers, the review
takes the following perspective on distributional characteristics:
(a)
(b)
(c)
TOPIC 9
DATA ANALYSIS
145
(a)
Measures of Location
The common measures of location, often called central tendency or centre
include the mean, median and mode.
(b)
Mean
Mean is the arithmetic average. It is the sum of the observed values in the
distribution divided by the number of observations. It is the location
measure most frequently used for interval-ratio data but can be misleading
when the distribution contains extreme scores, large or small.
Mean of the population (or population mean) is denoted as ( ) and is
defined as:
Formula:
x/N
(c)
Median
Median is the midpoint of the distribution. Half of the observations in the
distribution fall above and the other half fall below the median. When the
distribution has an even number of observations, the median is the average
of the two middle scores. The median is the most appropriate locator of
centre for ordinal data and has resistance to extreme scores, thereby making
it a preferred measure for interval ratio data particularly those with
asymmetric distributions.
(d)
Mode
Mode is the most frequently occurring value. When there is more than one
score that has the highest yet equals frequency, the distribution is bimodal
or multi modal. When every score has an equal number of observations,
there is no mode. The mode is the location measure for normal data and a
point of reference along with the median and mean for examining spread
and shape.
(e)
Measures of Spread
The common measures of spread (alternatively referred to as measure
dispersion or variability) are the variance, standard deviation, range,
interquartile range and quartile deviation. They describe how scores cluster
or scatter in a distribution.
(f)
Variance
Variance is the average of the squared deviation scores from the
distributions mean. It is a measure of score dispersion about the mean. If
Copyright Open University Malaysia (OUM)
146
TOPIC 9
DATA ANALYSIS
all the scores are identical, the variance is 0. The greater the dispersion of
scores, the greater is the variance. Both the variance and the standard
deviation are used with interval ratio data. The symbol for the sample
variance is (s2), and the population variance is the Greek letter sigma
squared ( 2)
Formula:
Sample variance,
s2
(Xi Xj)2
n 1
Population variance,
2
(g)
(Xi Xj)2
N
Standard Deviation
Standard deviation is the positive square root of the variance. It is perhaps
the most frequently used measure of spread because it improves
interpretability by removing the variances square and expressing
deviations in their original units. Like the mean, the standard deviation is
affected by extreme scores. The symbol for the sample standard deviation is
(s), and a population standard deviation is ( ).
Formula:
s
Variance
(h)
Range
Range is the difference between the largest and smallest score in the
distribution. Unlike the standard deviation, it is computed from only the
minimum and maximum scores. Thus, it is a very rough measure of spread.
With the range as a point of comparison, it is possible to get an idea of the
homogeneity (small std. dev.) or heterogeneity (large std. dev.) of the
distribution. For homogeneous distribution, the ratio of the range to the
standard deviation should be between 2 and 6. A number above 6 would
indicate a high degree of heterogeneity. The range provides useful but
limited information for all data. It is mandatory for ordinal data.
(i)
Interquartile Range
Interquartile range (IQR) is the difference between the first and third
quartiles of the distribution. It is also called the midspread. Ordinal or
Copyright Open University Malaysia (OUM)
TOPIC 9
DATA ANALYSIS
147
ranked data use this measure in conjunction with the median. It is also used
with interval-ratio data if there are asymmetrical distributions or for
exploratory analysis. Recall the following relationships: the minimum value
of the distribution is the 0th percentile and the maximum is the l00th
percentile.
The first quartile (Q1) is the 25th percentile; it is also known as the lower
hinge when used with box plots. The median, or (Q2), is the 50th percentile.
The third quartile (Q3) is the 75th percentile; it is also known as the upper
hinge. The IQR is the distance between the hinges.
(j)
Semi-interquartile Range
Semi-interquartile range, is expressed as:
Formula:
Q = (Q1 Q3)/2
The semi-interquartile range is always used with the median for ordinal
data. It is helpful for interval-ratio data of a skewed nature. In a normal
distribution, a quartile deviation (Q) on either side encompasses 50 percent
of the observations. Eight (Qs) cover approximately the range. Qs
relationship with the standard deviation is constant (Q = 0.6745s) when
scores are normally distributed.
(k)
Measure of Shapes
The measure of shape, skewness and kurtosis describe departures from
the symmetry of a distribution and its relative flatness (peakedness),
respectively. They are related to statistics known as moments, which use
deviation scores
moment.
(i)
(ii)
148
TOPIC 9
DATA ANALYSIS
9.6
WHAT IS A HYPOTHESIS?
(b)
(c)
(d)
(e)
9.6.1
The null hypothesis states that there is no change or difference in the group
means. It is based on the notion that any change from the past is due entirely to
random error. We are saying that the population value has not changed from one
time to another or the sample statistic does not vary significantly from an
assumed population parameter.
TOPIC 9
DATA ANALYSIS
149
One Sample
The mean brand preference score of male teachers aged 35 to 40 is 85.
Two Samples
There is no difference in the mean brand preference score between male and
female teachers aged 35 to 40.
The alternate hypothesis (or sometimes called the research hypothesis) is the
hypothesis that contradicts the null. It is commonly written as Ha. The alternative
hypothesis can indicate the direction of the differences or relationship, or assume
a neutral position. If direction is emphasised (indicated in the alternative
hypothesis), we called it one tailed-test. Otherwise, the test will be a two-tailed
test. The following are examples of alternate hypothesis.
One Sample
The mean brand preference score of male teachers aged 35 to 40 is not equal
to 85.
Two Samples
There is a difference in the mean brand preference score between male and
female teachers aged 35 to 40.
9.6.2
150
TOPIC 9
9.6.3
DATA ANALYSIS
SELF-CHECK 9.2
Explain the difference between sample statistics and population
parameters.
9.6.4
The assumption that there is always a risk of inference when a researcher studies
a population may be incorrect. Thus, in research, error can never be completely
avoided and statistical tests that the researcher performs to accept or reject the
null hypothesis may be incorrect. Researchers, therefore, need to be aware of two
types of errors associated with hypothesis testing: Type I Error and Type II Error.
TOPIC 9
(a)
DATA ANALYSIS
151
Type I Error
Type I Error, referred to as alpha ( ), occurs when the sample results lead
to rejection of the null hypothesis when it is true. The probability of this
type of error, also referred to as the level of significance, is the amount of
risk regarding the accuracy of the test the researcher is willing to accept.
Thus, the level of significance is the probability of making an error by
rejecting the null hypothesis.
Depending on the research objectives and situation, researchers typically
consider either <.05 or <.01 as an acceptable level of significance. The
researchers are willing to accept some risk they will incorrectly reject the
null hypothesis but that level of risk is specified before the research project
is carried out. If the research situation involves testing relationships where
the risk of making a mistake is high, the researcher would specify a higher
level of significance, for example <.01.
In examining the relationship between two chemicals that might explode or
the failure rate of an expensive piece of equipment, the researcher would
not be willing to take much risk. On the other hand, when examining
behavioural relationships or when the risk is less costly, then only the
researcher is willing to take more risk. In some situations, the researcher
may even accept a 0.10 level of significance.
(b)
Type II Error
The second type of error is referred to as Type II Error. Type II Error occurs
when, based on the sample results, the null hypothesis is not rejected when
it is, in fact, false. Type II Error is generally referred to as beta ( ) error.
Usually, the researcher specifies the alpha error ahead of time but the beta
error is based on the population parameter (mean or proportion) and/or
the sample size.
A third important concept in testing hypotheses for statistical significance is
the statistical power of the test. The power of a test is the ability to reject the
null hypothesis when in fact the null hypothesis is false. The statistical
power of a test can be described as (1
), the probability of correctly
rejecting the null. The probability of a Type II Error is unknown but it is
related to the probability of a Type I Error.
Extremely low levels of Type I error ( ) will result in a high level of Type II error
( ), thus it is necessary to reach an acceptable compromise between the two types
of error. Sample size can help control Type I and Type II Errors. Generally, the
researcher will select the sample size in order to increase the power of the test
and to minimise Type I error and Type II error.
Copyright Open University Malaysia (OUM)
152
TOPIC 9
DATA ANALYSIS
The trade off between Type I and Type II Errors has a practical dimension as
defined by the costs incurred for each error. Often a change in the status quo is
associated with a great cost (the risks or gambling the future of the firm on a new
technology, a new investment in an equipment, etc.) Since the change must be
beneficial, the risks associated with alpha should be kept very low. However, if it
is essential to detect changes from a hypothesised mean, the risk of a beta would
be more important. Thus, a higher less critical alpha level would be chosen.
9.6.5
(b)
(c)
(d)
Collect data from a sample and compute the statistical test to see if the level
of significance is met.
(e)
Accept or reject the null hypothesis. Determine whether the deviation of the
sample value from the expected value would have occurred by chance
alone (five times out of one hundred).
9.7
INFERENTIAL STATISTICS
TOPIC 9
DATA ANALYSIS
153
(a)
Mean
Mean is also known as average. A mean is the sum of all scores divided by
the number of scores. The mean is used to measure central tendency or
centre of a score distribution generally. For example, the mean for the
following set of integers: 3, 4, 5, 7 and 6 = 5.
154
TOPIC 9
DATA ANALYSIS
(b)
Standard Deviation
A standard deviation tells us how close the scores are centred around the
mean.
As shown in Figure 9.3, when the scores are bunched together around the
mean, the standard deviation is small and the bell curve is steep. When the
scores are spread away from the mean, the standard deviation is large and
the bell curve is relatively flat.
To explore better what standard deviation means, refer to Figure 9.4. The mean is
20 and the standard deviation (SD) is 5. Figure 9.4 represents the score obtained
on grid test for two organisation terminals using cluster computing with the
same mean of 20.
Copyright Open University Malaysia (OUM)
TOPIC 9
DATA ANALYSIS
155
(a)
One standard deviation (SD = 5) from the mean in either direction on the
horizontal axis accounts for around 68% of the organisation in this group.
In other terms, 68% terminals obtained 15 and 25 optimal time.
(b)
Two standard deviations (5 + 5 = 10) away from the mean account roughly
95% of terminals. In other words, 95% terminals obtained are between 10
and 30 optimal time.
(c)
Three standard deviations (5 + 5 + 5 = 15) away from the mean account for
roughly 99% terminals. In other words, 99% terminals obtained are
between 5 and 35 optimal time.
9.7.1
Let say you are conducting a study to compare the effectiveness of the use
of service discovery protocol (independent variable) in enhancing network
appliances detection in home networks. The mean score and standard deviation
for the application test are shown in a table and you want to test the null
hypothesis.
H0: There is no significant difference between the experimental group and the
control group in terms of enhancing network appliances detection.
To solve this, you may use the statistical approach called t-test to obtain the
t-value for independent means. In this case, independent means that the two
groups consist of different subjects. The t-test gives the probability that the
difference between the two means is caused by chance. For testing the
significance, you will need to set a risk level called the alpha level. In social
science research, the alpha level is set at 0.05. This means that the obtained result
which is significant at .05 level could occur by chance only 5 times in a trial of
100.
Table 9.2 Means and Standard Deviations Obtained for
the Experimental and Control Groups
N
Mean
Standard Deviation
Experimental group
10
13.8
2.10
Control group
10
11.4
1.96
156
TOPIC 9
DATA ANALYSIS
(a)
Table 9.2 displays t-value of 2.65 obtained. If you are using statistical
software like SPSS or SAS, the probability value is given (i.e. p < 0.02). We
could also refer to the table of critical values to find out whether the t-value
is large enough to say that the difference between the groups is not likely to
have been a chance finding.
(b)
We can also determine the degrees of freedom (df) for the test which is the
sum of the terminals in both groups minus 2 (i.e. n 2). By the given alpha
level, the df and the t-value, we can refer to the t-value in the table of
critical values.
(c)
Refer to Table 9.3. The obtained t-value (2.65) is bigger than the critical
value (2.1009) for 18 degree of freedom (20 2 = 18). From this, we can
conclude that the differences between the means for two organisations is
significantly different at the 0.05 level of significance.
Table 9.3: Extract from the Table of Critical Values
(d)
df
p = 0.05
p = 0.01
17
2.1098
2.8982
18
2.1009
2.8784
19
2.1009
2.8609
9.7.2
Let us say you would like to conduct a study to compare the effectiveness of the
use of service discovery protocol (independent variable) in enhancing network
appliances detection (dependent variable) in ONE home network (IEEE 802.11).
You gave a pre-test and after testing the protocol with the IEEE 802.11 network,
you give a post-test. Here, the same group of subjects are tested twice. The mean
score and standard deviation obtained for network detection are in Table 9.4.
You want to test the null hypothesis.
H0: There is no significant difference between the pre-test mean and the post-test
mean in terms of network appliances detection enhancement.
Copyright Open University Malaysia (OUM)
TOPIC 9
DATA ANALYSIS
157
Table 9.4: Means and Standard Deviation Obtained for the Pre-test and Post-test Scores
Mean
Standard Deviation
Pre-test
9.90
1.66
Post-test
10.90
0.99
(a)
By using the t-test for dependent groups, we can obtain the value of 1.94. In
this case, dependent means that the two means are obtained from the same
groups.
(b)
From Table 9.5, we can highlight that for 9 degrees of freedom, the critical
value is 2.2622, which is larger than the t-value 1.94. We can conclude that
the means are NOT significantly different at the 0.05 level of significance.
Table 9.5 Extract from the Table of Critical Values of t
9.7.3
df
P = 0.05
p = 0.01
2.3060
3.3554
2.2622
3.2498
10
2.2281
3.1693
Mean
Standard Deviation
Buffer Method 1
10
14.6
1.83
Buffer Method 2
10
15.6
2.22
Buffer Method 3
10
18.0
2.10
158
TOPIC 9
DATA ANALYSIS
Sum of Square
df
Mean Squares
Treatment
61.066
30.533
7.1811
0.003
Within
114.800
27
TOTAL
175.866
29
Not significant
From Table 9.8, we can conclude that there is no significant difference between
the performance using Buffer Method 1 and Buffer Method 2. Buffer Method 3
performed significantly better at significance level 0.05. Buffer Method 3 also
outperformed Buffer Method 1 at the 0.01 level of significance.
9.7.4
Correlation Coefficient
To find the relationship or correlation between two variables, the approach used
is called correlation coefficient. For example, in a research work, you collected
data on bandwidth rate and also jitter. You may want to find out if there is a
correlation between bandwidth rate and jitter in network performance. It is
important to note that correlation has direction and can be positive or negative.
The Pearson product-moment correlation coefficient (represented by r) is used
to show the strength of relationship between two variables. A coefficient can
range from r = +1.00 to 1.00. Figure 9.5 shows what the coefficient means.
TOPIC 9
DATA ANALYSIS
159
SELF-CHECK 9.3
1.
2.
160
TOPIC 9
DATA ANALYSIS
SELF-CHECK 9.4
1.
When would you use a longitudinal survey rather than a crosssectional survey? Discuss.
2.
3.
Discuss some ethical issues you think could arise during survey
research.
After data is collected and before it is analysed, the researcher must examine
it to ensure its validity. Blank responses, referred to as missing data, must be
dealt accordingly.
If the questions were pre-coded, then they can be inserted into a database. If
they were not pre-coded, then a system must be developed so they can be
included in the database.
The typical tasks involved are editing, dealing with missing data, coding,
transformation and entering data.
Descriptive analysis refers to the transformation of raw data into an
understandable form so that their interpretation will not be difficult.
Summarising, categorising, rearranging and other forms of analysis obtain
descriptive information.
Tabulation refers to the orderly arrangement of data in a table or other
summary format. It is useful for indicating percentages and cumulative
percentages as well as frequency distributions. The data may be described by
measures of central tendency, such as the mean, median or mode.
Cross-tabulation shows how one variable relates to another variable to reveal
differences between groups. Such cross-tabulations should be limited to
categories related to the research problem and purpose. It is also useful to put
the results into percentage form to facilitate inter-group comparisons.
A hypothesis is a statement of assumption about the nature of the world.
A null hypothesis is a statement about the status quo.
Copyright Open University Malaysia (OUM)
TOPIC 9
DATA ANALYSIS
161
ANOVA
Mode
Descriptive Statistics
Median
Correlation
Normal Distribution
Comparing Differences
Standard Deviation
Inferential Statistics
T-test
Mean
Variance
Topic
10
Proposal
Writing and
Ethics in
Research
LEARNING OUTCOMES
By the end of this topic, you should be able to:
1.
2.
3.
4.
5.
INTRODUCTION
Before embarking on a research project, students must have an overall research
plan that indicates the research problem to be studied, research objectives,
significance of the research, strategies to obtain answers to the research problems
and the research project implementation schedule. This overall research plan is
called a research proposal. It is very important that students write a good
research proposal because without a good research proposal the student may not
get a research supervisor or even an approval to carry out the research. The
student must also know about ethics in research because that governs the
researchers behaviour in terms of what should do and should not be done in the
research project.
TOPIC 10
10.1
163
This topic will address guidelines for writing a research proposal for your
research work. All students are required to write a research proposal before
venturing into research work.
A research proposal is a document written to state your proposed research
direction and the study you intend to do. A research proposal serves to advise
your academic supervisor or potential provider of research contract on your
conceptualisation of the total research process that you propose to undertake and
examine its suitability.
A research proposal is an overall plan, scheme, structure and strategy to
obtain answers for your research problems which constitutes your research
project. A research proposal can be rejected if your supervisor or graduate
commitee finds it is poorly devised.
Therefore, you need to have a well-planned research proposal prior to
undertaking any research task. If the proposal is well-designed, it would be
much easier for you to outline the entire research processes as well as help you to
prepare a thesis or dissertation in a sequential manner.
10.2
164
(a)
TOPIC 10
Introduction
(i)
In this section, you should provide an overview of the issues that you
intend to study. The content of introduction should be brief, precise
and straight to the point. After giving an overview on the scope of the
research, you will need to narrow it down to the specific area of your
concern.
(ii)
(iii) Define the problem statement in your proposal. This will give a clear
perception to the reader on the issue you are going to solve. However,
the problem statement you write in the proposal may only be
tentative at the point of proposal preparation because the research has
not been carried out yet.
(iv) Use simple sentences and make sure you narrow down the research
issue to focus on a very specific field.
(b)
Research Objectives
(i)
List out your research objectives and it should indicate the central
theme of your research that you intend to study.
(ii)
(c)
Research Questions
Research questions are questions formulated to address the research
objectives and to break the research objectives into smaller parts. Research
questions will influence research methodology and the type of analysis to
be performed.
(d)
(ii)
TOPIC 10
(ii)
165
You can also provide justification if the methodology you are going to
use has some degree of novelty and your research would contribute
to new knowledge.
(iii) Some other criteria you can mention are the variables that you are
going to use and the expected outcomes of your research and its
influence on the model or design.
(e)
Literature Review
(i)
(ii)
Try to read a thesis in similar area that you are investigating to get the
feel of it.
(f)
Methodology
(i)
(ii)
You may explain some reasons why you are using certain theory
or models, whether your research approach will be qualitative or
quantitative, or a combination of both.
166
(g)
TOPIC 10
Project Schedule
Provide a Gantt Chart or timetable specifying how long it will take to
complete your research work. Also indicate how long you will take for data
collection, analysis and writing up the final report or document. Some
research proposals require indication of milestone dates to give clear
picture on the expected accomplishment.
TOPIC 10
(h)
167
References
(i)
(ii)
SELF-CHECK 10.1
1.
2.
10.3
There are several steps and guidelines to be followed during research proposal
preparation.
(a)
(b)
(c)
While writing the first proposal, present your ideas by narrowing it down
sequentially and focus more on presenting the information in an interesting
manner. You must remember that your proposal should be expressed
clearly and states the overview of your research intention.
168
TOPIC 10
(d)
Write and explain about your research problem at the beginning of the
proposal content (i.e. Introduction section). This is important as to give
readers attention since the entire research process is driven by the research
problem.
(e)
Write about the methodology you are going to implement briefly and
precisely. It is a good practice to outline methods and source of data in the
proposal stage itself. This will put your proposal in a better position in
order to determine its worth and potential contribution.
SELF-CHECK 10.2
Identify the guidelines involved in constructing a research proposal.
10.4
As a learner, you must ensure that your research proposal meets the requirement
and guidelines specified by your institution or university. This is partially
important to help you plan your research process and ease you towards the
preparation of the dissertation or thesis. There are some common weaknesses
encountered by researchers during research proposal preparation (Allen, 1960).
Proposals submitted by researchers and students for academic projects tend to
have weaknesses as described in Table 10.1.
Table 10.1: Weaknesses in Research Proposal
No.
Weaknesses
(a)
Research Problem
Justification
Explanation
Project description is unfocused and research direction is
unclear.
Research problem is not significant enough to support
or provide justification in initiating new knowledge
contribution.
The research problem is of interest only to a particular
group and has many limitations.
The research problem may be so complex than the author
actually thinks.
The hypotheses are doubtful or rest on insufficient
evidence.
Copyright Open University Malaysia (OUM)
TOPIC 10
(b)
Research
Methodology
169
(c)
Proposal Author
SELF-CHECK 10.3
1.
weaknesses
encountered
2.
during
170
TOPIC 10
10.5
RESEARCH ETHICS
10.6
(b)
(c)
(d)
(e)
(f)
Effects on participants of the way the researcher uses, analyses and reports
the data; and
(g)
Privacy is accepted as the key to ethical issues that the researcher has to confront
in carrying out any research project. Almost all aspects of ethics, for example,
consent, confidentiality, participant reactions, and the effects of the way the
researcher uses, analyses and reports research findings have the capacity to
affect, or are related to, the privacy of participants.
TOPIC 10
10.7
171
During the early stages of the research process, the researcher may have to seek
access to agencies, organisations or even individuals. Ethical problems may arise
when participants are not clear of the objectives of the research. Ethical issues
may be related to a researcher attempting to apply pressure to grant access.
Issues relating to privacy may arise when a researcher tries to get access by
telephone calls at inappropriate times or approach the participants during odd
hours. Access to secondary data may also have ethical consequences, for
example, when the researcher obtains personal data of individuals who have not
consented to be involved in the project.
Consent to participate in the research project may not be a straight forward
matter because when someone agrees to participate, it does not necessarily imply
consent.
The nature of consent can be differentiated as shown in Figure 10.3.
172
TOPIC 10
ACTIVITY 10.1
In your opinion, why do ethical issues become a major concern in the
research process? Explain how to handle this issue.
SELF-CHECK 10.4
Tick True or False for each statement below:
No.
Question
1.
2.
3.
4.
5.
6.
7.
8.
9.
True
False
TOPIC 10
173
Ethics
Project Schedule
Proposal
Significance of Study
MODULE FEEDBACK
MAKLUM BALAS MODUL
OR
2.
Thank you.
Centre for Instructional Design and Technology
(Pusat Reka Bentuk Pengajaran dan Teknologi )
Tel No.:
03-27732578
Fax No.:
03-26978702