OVERVIEW OF RESEARCH

Meaning of Research
✓ Scientific research entails collection and analysis of data in order to solve a research problem.
It is a systematic, controlled, empirical and critical investigation of a hypothetical
proposition about the presumed relationships among natural phenomena, intended to produce
useful, verifiable and generalizable knowledge.
✓ It is a systematic way of asking questions, a careful enquiry or examination to discover
new information or relationships, and expand or verify existing knowledge.
Process Process

- Systematic - Valid
Research is a - Guided Search for - Reliable
- Intensive - Generalizable - knowledge.

Figure 1.1. Conceptualization of research.

Research process is:

i. Systematic because it follows defined and ordered procedure.
ii. Guided since it is intended to answerer specific questions, and focuses only on those
questions. Any other issues outside the posed questions are irrelevant.
iii. Intensive because it consumes resources in terms of time, capital, uses equipment and
personnel.
On the other hand, a research product is:
i. Valid if knowledge obtained is based on verifiable data.
ii. Reliable if that product id reproducible at another time or by another researcher, or by
another method.
iii. Generalizable if the knowledge produced through research at one point can be useable in
other areas or in other populations.

Research is usually:
i. Directed towards finding a solution to a problem.
ii. Leads to the development of generalizations, principles or theories that will be used to
predict future occurrences. Hence, it is more than gathering information.
iii. Based upon observable experiences or empirical evidence. It rejects spiritual revelations,
dogmas and accepts only that which can be verified.

✓ Research is a systematic way or attempt to provide answers to questions. Such questions may
be abstract and general as in the case of basic research, or highly concrete and specific
as in applied research. The goal of research is to explain, predict andcontrol phenomenon.
This goal is based on the assumption that all behavior and events are orderly and that
they are effects which have discernible causes. Scientific research generally has six major
steps:
1. Recognition and definition of the problem in clear and measurable terms.
2. Development of a hypothesis (alternative solutions).
3. Development of techniques and instruments of measurement.
4. Collection of data.
Research Seminar
Oso Willis Yuko
1
5. Analysis and interpretation of the data collected.
6. Conclusion based on the data.

Research based on the principle of:

o Precession of measurements: All measurements used in research must be accurate and exact.
The data collected and the instruments used to collect them must be accurate.
o Reliability: if the same procedure or methods is repeated, it should produce similar
results. It must be consistent in terms of producing same results under similar circumstances.
o Validity: a study should accurately measure what it is intended to measure in terms of content,
theory, criterion, or on the face of it.
o Ethics: a study should be based on socially accepted standards and practices.

Research generally:
o Accepts certain critical assumptions.
o Requires collection and interpretation of data.
o Is by nature cyclical, or helical.
However, research is NOT:
o Mere information gathering.
o Transportation of information from one location to another.
o Mere rummaging for information.
Research is the process of attempting to achieve systematically and with support of data, an
answer to a question, a solution to a problem, or a greater understanding of a phenomenon.

Characteristics of Research
At any one time in any educational institution, there are always different teachers teaching
different subjects to different classes all at the same time. Yet, each is still called a teacher,
and what each does is still called teaching. Why is this so? It is so because there are common
elements in what they are doing. This is also true of research. There are elements common to
all activities called research. We shall now examine each characteristic in more detail.
1. Research is systematic: This means that research is structured (i.e. there are specific
rules, steps and procedures for carrying it out). You cannot just start from any point and hope
to produce valid, reliable and dependable results. Research follows ordered steps, and it
can only be accomplished in that order. For example, it is not possible to define a problem
that has not been identified; and it is not possible to collect data unless the tools for collecting
that data have been developed. It is this orderliness that makes research systematic.
2. Research is logical: Ideal research follows a system that employs logic at every point. By
logical examination of the procedures employed in an experiment, and in the context of
internal validity, a researcher is able to check the validity of the conclusions drawn. Every
step in a research process and every activity must be selected based on justifiable grounds and
not on faith or dogma.
3. Research is empirical: Research is reality referent. Much deduction may precede its
application, but data is the end result of research procedures. Research must be based on

Research Seminar
Oso Willis Yuko
2
 data. Any conclusions drawn must be supported by day. It is this requirement (of data or
evidence) that makes research as an empirical process.
4. Research is reductive: When a researcher applies analytic procedures to collected data,
the confused individual events and objects are reduced to more understandable categories and
concepts.
5. Research is replicable and transmittable: Individuals, other than the researcher himself
should be able to use the results of the study. The processes and procedures should themselves
be transmittable, enabling others to replicate them and assess their validity. This is
transmitabilitywhich means that the results of research can be used in other areas and by
other people who were not part of the study. Replicability is the fact that the same research
can be done again by another researcher, or at a different time, or be a different method and
will still produce fairly similar results.

Purpose of Research
The main role of research is generation of knew of knowledge to the already existing, or
verification of already existing body of knowledge. There are other ways of gaining knowledge,
but all these other methods have inherent weaknesses. It is the fact that all other methods have
inherent weaknesses that make research so important.

Types of Research
There are different types of research depending on whether they are classified by purpose or
on method. The most popular classification is by purpose. This produces five main types of
research: basic, applied, participatory action research (participatory action research), research and
development (R & D), and operations research (OP). However, some authors have argued
that there are only two types of research: basic and applied, and that the other types of research
are but modification of these two types. You can take your decision as we discuss each type of
research in this section.

Basic Research
o Basic research (also known as fundamental or pure research) is found mostly in the field
of pure sciences. It is the work of scientists and others who pursue their investigations without
conscious goals apart from the desire to unravel the secrets of nature.
o It is concerned with finding solutions to questions that are intellectually challenging, but
which may not have immediate practical application. It is conducted purely for the
purpose of theory development and refinement: for the sake of science.
o It tests theories, studies relations and attempts to understand phenomena, but with little or no
thought of the applications of the results of such studies to immediate practical problems.
Basic research seeks only to discover basic truth and principles, although it yields findings
that eventually can be applied to practical problems outside the domain of science. Its
emphasis is on problems as abstractions not as utilitarian matters.

Applied Research
This is research designed to improve process or progress. It carries the findings of basic research
to appoint at which they can be applied to meet specific needs through testing of
Research Seminar
Oso Willis Yuko
3
theoretical concepts in actual problem situations. It is conducted for the purposes of applying
or testing theory and evaluating its usefulness in solving educational problem. Applied
research includes most of the features of basic research, but differs from it in terms of its
goals. Applied research is directed towards the practical application of scientific knowledge and
towards the solution to specific problems.

Action Research
✓ Actionresearch developed as a reaction to the weaknesses of the basic and applied
research discussed above.Action research, or more specifically, participatory action research
(PAR) is concerned with the immediate solutions a concrete problems. It is mainly
concerned with a local problem and is always conducted in local settings. Its aim is social
responsibility and improvement of practice with focus on social applicability. The primary
goal of action research is generation of a solution to a given societal problem, but not
necessarily contribution to science.
✓ Participatory Action Research is “a family” approach to research, having at its core the joining
together of participants within a project or program, and whose is confined mainly to those
conducting it. As opposed to pure and basic research, participatory action research is not
principally concerned with whether the results are generalizable to other settings, although
they often are. Because of this, participatory action research lacks the rigid controls that
characterize other types of research.
✓ Participatory action research is a research process where people collaboratively examine their
situations, identify and analyze their strengths and weaknesses, reflect on the results and take
action to improve their situations according to their needs. The major elements of
participatory action research are collaborative assessment, collaborative action, and
collaborative analysis and reflection. This process is represented as shown in Figure 1.1.

Figure. 1.1. The participatory action research process.

o Participatory action research is an iterative process in which the end of each cycle marks the
beginning of another cycle. The path of each participatory action research cycle depends on
the results of analysis and evaluation results indicate that the solution is effective, then
appropriate strategies must be designed to maintain and improve it: this
Research Seminar
Oso Willis Yuko
4
 requires another participatory action process. If, however, evaluation suggests that the
solution is ineffective, then the process must start all over again, either to redefine the
problem, or another solution or implementation. This is what makes participatory action
research iterative.

Research and Development (R&D)

Research and development is a branch of research is concerned with the development of effective
programs for use in the market rather with the formulation or testing of theories. In research and
development, both the problem and the model solution exist: the role of the research is toproduce
thatsolution to the specification of the intended consumer. So research and development is
conducted when a user knows what he or she wants but cannot produce or manufacture that
product by himself or herself. A researcher is then contracted to producethat exactmodel
according to detailed instructions.

Operations Research
Operations research is a type of research concerned with identifying service delivery problems
and test solutions to them. It is a design used to determine the courses of action available that can
best accomplish the objectives most effectively and efficiently.

Research Topic
o A topic is the subject to be researched, analyzed and interpreted. It is a thematic statement of
what the research is all about. Typically, a researcher selects a research topic by identifying
a wide area of interest or concern (research area) and then narrows that area to a manageable
set of research questions. The researcher focuses the direction of the research and positions
the central concept for the research at an early stage through selection of a research topic. A
research topic is a brief statement, of about 12 words, with a subject and an object but without
detailed wording and unnecessary explanations or addendum.
o It is a concise and specific statement that encompasses succinctly all the essential
elements of the study. It may arise as a problem to be solved or as an issue to be better
understood, or it may arise as conditions which the researcher wants to improve, or as
difficulties to be eliminated or questions for which immediate answers are needed.

Research Seminar
Oso Willis Yuko
5
 THE RESEARCH PROCESS
Identifying a Research Topic
The first step in identifying a research topic is to determine a broad problem area that is
closely related to one’s interest or professional goals or to one’s area of in-depth
specialisation such as say, staff motivationetc. These are broad statements and person reading
them could still ask, for example: what is it about private secondary education? The
researcher then seeks out specific research problems within those broad areas that could form the
basis of investigation. It is not possible, for example, to investigate the whole of staff motivation.
One can only study particular aspects of them. Those aspects selected for study become the
research topic.

Sources of Research Topics

One major problem for beginning researchers is where to obtain research topics. Students spend
a lot of time struggling to locate research topics because they imagine that a topic is a very special
issue that can only come through divine intervention. Yet research topics exist around them: in
their families, classrooms, places of work, and everywhere in their general environment. The
common sources of research topics include:
1. Personal Experiences: A research topic can be suggested by one’s experiences and personal
observation of certain relationships for which no satisfactory explanation exists. These may
include routine ways of doing things which may be based simply on tradition or authority
without scientific justification. For example, why should school shoes be black? Or why
should teachers correct students’ work in red ink? A researcher can seek to explain why these
things are the way they are and suggest a topic such as effect of shoe colour on students’
achievement in mathematics.
2. Existing Theories: Theories are general principles whose applicability to specific problems
or situations are not known unless empirically tested. It is therefore only through research
that one can determine whether the generalisations embodied in the theory can be
translated into specific recommendations for practice. For example it may be an established
theory that students learn better through induction than through lecture. But for a teacher to
know for certain that his or her students will learn better by induction, the teacher has to try
the method with the students. The teacher may therefore want to out the Effect of Teaching
Method on Students Achievement.
3. Formal Needs Assessment:Careful assessment of needs may reveal problems that may
require research to solve. Research topics may result from such problems. For example.
A needsassessment may have identified lack of mathematics texts as a need. A study may
therefore be conducted on Effect of Mathematics Texts on Students Achievement in
Mathematics.
4. Literature Sources: Nearly every research that has been reported contains recommendations
for future research, and research topics may result from such recommendations. Students can
obtain research topics by just consulting the recommendations sections of reported studies
in general areas of interest. A researcher may also wish to replicate a study that has been
reported, and there are various valid reasons for replicating of studies. Such reasons may
form the focus for further research.

Research Seminar
Oso Willis Yuko
6
Considerations in Selecting a Research Topic
Once a researcher has identified a topic, it does not automatically follow that the topic will be
successfully studied and reported. Several other subsidiary factors needs to be considered before
finally selecting a topic of study. It is important that you consider the following issues when
selecting a research topic:
o Personal interest in the topic: You should not choose a topic in which you have no
interest, or have only little interest.
o Importance of the research: A topic that is not significant may not attract funding or
readership as it may not constitute a worthwhile contribution to knowledge.
o Time available: A good topic should be able to be accomplished within the set time.
o Researcher skills: Youneed to consider whether you havethe necessary skills required to
study the topic you have suggested, especially data analysis and interpretation skills.
o Cost: It is necessary to consider the cost of conducting and reporting a study and be sure
that it can be accommodated within the constraints of the available resources.
o Ethical issues: Some topics raise ethical issues which you may not easily overcome. Such
topics should be avoided.

Qualities of a Good Research Topic

A research topic is different from the other topics for class discussions, because a research
topic must meet the following specific criteria. A good research topic should:
1. Summarize the main idea of the research simply and with style. It is a concise statement
of the theme of the study, a research topic and should identify the actual variables or
theoretical issues under investigation and the relationship between them.
2. Be fully explanatory on its own, and must compress easily into a running herad.
3. Avoid words that serve no purpose, which increase length and which can be misleading.
Words such as - method, results, a study of, an investigation, a survey, among others - are
unnecessary because they compound the topic without adding much value to it.
4. Avoid using abbreviations because they reduce accuracy.
5. Not exceed 12 words, though different bodies have their rules on this.

Research Background
o A research background is not just a story about the problem or the topic; it is, as the name
suggests, the background or more precisely, the foundation of the study. It must introduce the
status and the theoreticalorientations of the study and provide a rationale of the study through
adducing evidence and conditions that exist to create urgencyon the problem and the need to
study it in order to solve the problem or contribute to its solution.
o A research background should portray the history and the character of the problem, and
the issues at hand, and expose the facts that surround the problem that must be tackled through
undertaking the study.
A good research background should:
1. Introduce the problem of the study and anchor the study in the context of what is already
known.
2. Provide definitions of the terms as used in the study and ensure that the researcher and the
readers attach the same meanings to the same terms throughout the study.

Research Seminar
Oso Willis Yuko
7
3. Lay the foundation and inform the readers of what to expect in the rest of the study.
4. Form the basis of the whole study. Whatever has not been introduced at the background is
irrelevant to the study and cannot be introduced in the subsequent sections of the study.

o A well-written research background has four main dimensions: the historical, the theoretical,
the conceptual and the contextual dimensions as summarized in the scheme below.

Research Background

Historical Theoretical Conceptual Contextual

Describes Presents a Literal Working Operational Status of the
historical rationale for the Primary or usual Researcher’s External and operational
perspective of the need to or general selected observable variables in the
concepts under investigate the definitions. definitions. features of the target population.
investigation. variables in the terms as
research topic determined by the
researcher.

Research Problem
o A research problem is a terminology for a topic that has been narrowed down and refined
appropriately for research. It is the why of the study. It is a hypothesis or a question of interest
to a researcher which can only be tested or answered through the collection and analysis of
fresh data.
o A research problem goes beyond surface problems which do not demand collection and
interpretation of data. Research is usually conducted because a deficiency or a need has been
identified.
o A research problem refers to what the researcher has identified as existing in literature, theory
or practice - a condition which needs to be probed and a solution found. It may arise
from issues, difficulties, current practices or problems that need to be solved or better
understood, or it may arise from deductions from existing theories related to the current social
or political issues, practical situations or personal experiences. It may be an issue that a
researcher wants to describe, an event that has been observed and needs to be explained, or a
problem for which a solution is needed. Or it may be a state of affairs that needs to be
changed, or anything that is not working as well as it ought to, or a difficulty to be
eliminated.

Identifying a Research Problem

✓ One of the most difficult tasks for beginning researchers is how to identify a research problem.
Students usually have so many ‘problems’ in mind, but such problems are not always research
problems. A research problem is different from ordinary or general problems: it is a need or
a discrepancy that can only be solved through collection and analysis of data.
✓ Some of the common pointers to research problems are issues which attract attention and may
need investigation, unanswered questions, missing links, imbalances, unsatisfactory states of
affairs, and problems of adjustment to technological advances. The list is

Research Seminar
Oso Willis Yuko
8
 endless. It is up to the researcher to look into these areas and identify problems that
require research to solve. But how exactly does one identify a research problem?
✓ To identify a research problem, a researcher should answer YES to the following
questions:
1. Is there a perceived inconsistency or discrepancy between “what is” (the actual) and “what
should be” (the ideal)? In other words, are you seeing what you want to see or are you seeing
something different? If your answer is NO, ten there is no problem because things are just as
they should be. You now need to look elsewhere for a problem. But if the answer is YES,
then there is a problem, but it may not necessarily be a research problem. The YES scenario
should lead to the second question.
2. Is there a question about why there is a discrepancy? In other words, is the discrepancy
understood? If there is no question about the discrepancy, then there is no research
problem because the discrepancy has been, or can be explained from the available facts
and data, and there is no need for research. But if the answer is YES, then the discrepancy is
not understood, and it cannot be explained using the already available data. Hence a problem
exists, but it may not necessarily be a research problem. The YES scenario should lead
to the third question.
3. Are there possible and reasonable solutions or answers to the discrepancy? If the answer
is NO, then the problem cannot be investigated through research since it is not possible to
generate or formulate hypothesis. Hence it is not a research problem. But if the answer YES,
then there is a research problem, and these possible answers now form the basis of
investigation.
✓ The questions can be visualized a long a problem tree flow diagram in figure 2.1.
Examine -Theory/Practice/Experience/Environment

Qn. 1. Is there a discrepancy between the actual status (what is) and what should be (the ideal)?

No Yes [A problem exists]

No Problem Qn. 2. Is there a question about why there is a discrepancy?

No Yes [A problem exists]

Solution already exists: No research problem Qn. 3. Are there possible & reasonable answers to the discrepancy?

No Yes [A problem exists]

No research problem. The problem A research problem exits. The possible

cannot be investigated through research. answers form the basis of investigation.

Qualities of a Good Research Problem

A well stated research problem statement should:
1. Clarify what exactly is to be solved or to be determined. It should be written in a clear,
non-technical language avoiding jargon, clearly identifying an area of concern. It must
carefully fit into the broader context of current theories and relevant research without making
assumptions or unsupported claims or statements.
2. Contain the variables and the target population that is to be studied.
3. Restrict the scope of the study to a specific question or questions, hypothesis or
hypotheses; and be sufficiently limited in scope to be manageable.
Research Seminar
Oso Willis Yuko
9
4. Be feasible and amenable to research; not be too wide and data for solving it should be
available. Whether or not data is available is crucial because usually a researcher may not
have time to collect fresh data for the background information. Hence, a researcher needs
to be sure of the sources of data and the means of accessing it.
5. Not raise ethical issues. Ethical issues are paramount in research and a good research problem
should not raise or overlook ethical concerns.
6. Be original. Originality is important in research because knowledge cannot be advanced
by duplicating solutions to problems that have already been solved.
7. Be significant: it should result in advancement of knowledge or significant improvement
of life. There should be some pay off if the uncertainties about the problem are resolved
or at least reduced. A good problem should explore an important question, address a
recognizable and significant need, and make a useful contribution to knowledge. It must
be current and timely.

Research Purpose
✓ The general objective, or as it is also commonly known, the purpose or aim of a study is a
description of the general intention of a study. The general objective, the purpose or goal
of as study elaborates on the information implied in the topic of the study, and presents a
quick overview of the study. It incorporates the problem, target population, the design,
and broad expectations of the studyin ambitious and largely immeasurable terms.
✓ In stating a purpose, a researcher does not promise the expected results or contributions
will occur, but only establishes the direction of the study and captures, in a single line or
paragraph, the essence of the study.
✓ A purpose statement provides a specific and accurate synopsis of the overall intent of the
study and forms the central and controlling idea of the study.

Objectives
o Objectives state the immediate intentions of the study, indicate the variables which will
be examined and measured.
o They constitute promises by the researcher that certain activities will take place, that
certain specific variables will be examined, and certain specific targets will be realized.

Why State Objectives?

o Objectives are critical in any research process. Just a ship without an intended can go
anywhere in the waters, so is a study without objectives. It may attempt to achieve everything
and in the end achieve nothing constructive. Objectives are useful in research because:
1. They provide focus for the study by specifying its intended outcomes. Because the researcher
has an end in mind and can work consciously towards that end.
2. They narrow the scope of the study to specific issues. This helps the researcher to
generate specific hypotheses and to develop research questions.
3. They enable the researcher to select a research design or method. Objectives that deal
with views, for example, require a different method from those that deal with effects; and

Research Seminar
Oso Willis Yuko
10
 those that deal with associations require different designs from those that deal with lived
experiences.
4. They enable the researcher to select suitable tools for data collection and appropriate methods
for data analysis.
5. They enable the researcher to evaluate the outcomes of the study. It is by comparing the
intended outcomes against the actual that a researcher can tell whether or not the study
was successful.

Qualities of a Good Research Objective

It is very important that a researcher states good objectives as poorly stated objectives can
misguide the study. A good research objective has three basic characteristics: (i) it must be
SMART or even SMART(ER). SMART is an acronym forSpecific, Measurable, Attainable,
Realistic and Time-bound or timeliness.
1. An objective is specificifit communicates clearly and explicitly, the intended outcome of the
study. It should be short, concise and precise, and should use words which do not lend
themselves to multiple interpretations. An objective should have a clear focus and should
relate only one operation of the independent variable to the dependent variable, and vice-
versa.
2. An objective is measurableif it is stated in terms of observable outcomes; and if it is expressed
in terms of external characteristics or determinants on which data can be collected. Objectives
should be written using activeverbs(words that describe action such as find out, assess,
determine, describe, establish, etc). But use of words without terminal characteristics such as
know, understand, feel, enjoy, master should be avoided as they do not signify readily
observable or easily measurable attributes.
3. An attainableis feasible and manageable in the sense that it can be answered through
collection of data. An objective should not be stated on which data cannot be collected.
Anobjective is manageable if it can be accomplished within the constraints of the
available finances, equipment, personnel and even time.
4. An objective is realisticif it is of reasonable level of difficulty. An objective should
neither be too hard nor be too simple. At any level and for every study, a researcher
should generate objectives that are within the scope of the study.
5. An objective is time-boundif the period within which it will be achieved is stated, and if it can
be accomplished within the stated time. Usually researchers have set times within which to
submit their reports. Do not state objectives which cannot be achieved within the stipulated
period. An objective must also be timely; it should neither come too late nor too early.
The challenge of the researcher therefore is to (i) formulateobjectives that are achievable
within the available time; and (ii) state the time frame within which each objective is to be
achieved.

ResearchQuestions
✓ Research questions and hypotheses are usually used alternately. If questions are posed,
hypotheses are usually not stated at the same time. A researcher poses questions when he
or she is not sure of the possible solutions to the research problem. So questions are posed
such that if they are answered, the responses will help achieve the objectives.
Research Seminar
Oso Willis Yuko
11
✓ But if the researcher has clues on the possible answers and is just in doubt, then
hypotheses are stated. Any ordinary question can be converted into a research question by
posing it in such a way that it cannot be answered unless data is collected on it.
✓ A question such -should children attend preschool? Or does basic salary affect teacher
motivation? - can be answered by simply expressing a yes or no opinion, and possibly
providing a justification for the opinion so expressed. However if the same questions are
posed as - Do children who attend preschool have a better social adjustment than those who
do not- then it becomes difficult to answer it directly by yes or no unless data is collected on
the social adjustments of children who attend preschool and those whom do not. Further if
the question - does basic salary affect teacher motivation-is posed as - do teachers with
different basic salaries have the same motivation- then data has to be collected on basic salary
and on motivational levels of teachers before the question can be answered conclusively. It is
this requirement for collection of data that makes a question a research question.

Research Hypotheses
✓ A research hypothesis is a proposal about a solution to a problem: It is an “intelligent” or
an “educated” guess, or a tentative solution to a problem. Hypotheses are declarative
statements on the relationships between two or more variables on the nature of some
phenomenon.
✓ They form the elements for testing to enable the researcher make an opinion about certain
phenomenon being investigated, and contain the variables which the researcher wishes to
consider or test in order to observe their actions or reactions so that meaningful
conclusions can be drawn.
✓ A hypothesis is some prediction of some sort, regarding the possible outcomes of a study.
✓ Hypotheses, as opposed to questions, are stated if the researcher has an idea of the
possible solutions to the problem under investigation, and only wants to find out if the
‘solutions’ hold under the certain circumstances or not. While purpose indicates the
anticipated contributions of the study, and specific objectives indicate what will be done
or measured, hypotheses specify the expected relationships between the variables.

Types of Hypotheses
o Hypotheses may be stated either in a null (non-directional) form and symbolized as (Ho),
or in alternative and directional forms and symbolized as HA andH1respectively.
o A null hypothesis states that there is no relationship between the variables being studied
or that no difference will be found in the dependent variable due to manipulation of the
independentvariable. It is referred to as null because sincescientists tend to state that
differences between a sample and the population is a sampling error and significantly different
from zero.
o An alternativehypothesis generally states a relationship between the variables being
investigated, or that a difference will occur in the dependent variable due to the manipulation
of the independent variable.

Research Seminar
Oso Willis Yuko
12
Advantages and Disadvantages of stating Research Hypotheses
A well stated hypothesis enables the researcher to:
1. Think more deeply about the possible outcomes of the study. Elaborating a question by stating
a hypothesis leads to a more detailed understanding of what the question implies and exactly
what variables are involved.
2. Make specific predictions based on prior evidence or theoretical arguments. If these
predictions are borne out by subsequent research, the entire procedure gains both in
persuasiveness and efficiency.
3. Focus the research on specific areas of investigation.

Hypotheses have their weaknesses too. For example;

1. Hypotheses lead to bias on ten part of the researcher. Once hypothesis are stated, the
researcher may be tempted to arrange the procedure to bring about the desired results.
2. Hypotheses are unnecessary or inappropriate in some types of research, like in qualitative
research designs.
3. Focusing attention on hypothesis may prevent the researcher from noting other
phenomena that might be relevant or important to the study. For example once hypotheses are
stated, they become the sole focus of the study and all other issues, no matter how significant,
become irrelevant.

Research Frameworks
A framework is a structure of an idea or a concept and how it is put together. There are three
major frameworks in research: theoretical, conceptual, and contextual frameworks.

Theoretical Framework
A theoretical framework has its origin in a theory. A theory is a set of properly argued ideas
intended to explain a phenomenon. It is a system of explaining a phenomenon by specifying
variables in the phenomenon and the laws that relate the variables to each other. A theory is
an interrelated set of constructs (or variables) formed into prepositions or hypotheses that specify
the relationship among variables in terms of direction and magnitude. A theory generally consists
of broad generalizations (or laws) and variables.
In developing a theoretical framework, the researcher should, in a brief passage:
- Name the theory and its author.
- Indicate where the theory has been used and who used it.
- Describe the propositions or hypotheses of the theory.
- Explain the variables in the study.
- Discuss the implications of the theory to the study (i.e. how it relates the independent
variable to the dependent variable.)
- Identify some alternative theories and why they are inadequate for this study. This helps
to justify the choice of the theory for the study.
- Identify some weakness of the selected theory.

Research Seminar
Oso Willis Yuko
13
.Conceptual Framework
✓ A conceptual framework is a scheme of concepts (or variables) which the researcher will
operationalize to achieve set objectives. It is a schematic (or diagrammatic) presentation
of a theory. The theory is presented as a model where variables and the relationships between
them are translated into a visual picture to illustrate the interconnections between the
independent, extraneous, and dependent variables as well any other variable significant
to the study.
✓ A conceptual framework is a basic visual structure consisting abstract blocks representing the
observational experiential and analytical or synthetical aspects of a process or a system
being conceived: the interconnections of these blocks complete the framework of certain
expected outcomes.
o A theoretical framework and a conceptual for the same study will present or describe the same
variables and portrays the same information, but while a theoretical framework is descriptive,
a conceptual framework is pictorial. Hence a conceptual framework is essentially a theory. It
is important to note that even a well and carefully designed studies could be misleading if
the underlying context of assumptions are wrong. So the systems of concepts, assumptions,
expectations, beliefs and theories that support and inform a study must be correct.

Extraneous

Independent Variable Dependent Variable

++

EV1
IV1 DV1
EV2

EV3 DV2
IV2

++ DV3

IV3
DV4

IV4

Figure 2.7. Conceptual framework for remuneration and motivation.

Research Seminar
Oso Willis Yuko
14
 LITERATURE REVIEW
o One of the requirements of scientific research is a demonstration by the researcher of thorough
understanding of the issues and facts surrounding the problem under investigation. In order
to understand the problem at hand, a researcher must extensively read many information
materials that relate directly to the topic under investigation. This is referred to as literature
review.
o Literature review is the systematic identification, location, retrieval, analysis and
evaluation of documents containing information that are related to the research problem.
It also entails analyses of casual observations and opinions related to the planned
research. A researcher should develop a thorough understanding of, and insight into, previous
works and trends that have been recorded pertaining to the research problem.

Why Review Literature?

A good literature review:
1. Delimits the research problem and defines and refines it. Literature reviewisolates key
issues pertinent to a study and provides the knowledge needed toconvert a tentative
problem into a detailed and concise plan of action. It provides the dimensions and the
limits of theproblemarea and the extent to which the answer is alreadyknown, thus
delimiting the problem.
2. Providesinsight into methods, measures, subjects and techniques alreadyused, or
proven useful or otherwise by otherresearchers, and thus leads to improvement of design
selection.
3. Preventsunintended duplication of previous studies by identifying what has already been
done that relates to the research problem. Literature review tells the researcher what has
been done and what needs to be done. Studies that have been done provide the
researcher with reliable information on what should be done.
4. Provides the means of getting into the frontiers of a particular field of knowledge.
Until one has learnt what has been done and what remains to be done, it is not
possible to develop a study that will advance knowledge in that particular field. It is
the literature review that forms the basis upon which a significantresearch should be built.
Without an effective literature review, the rest of the research work is likely to be
shallow and naïve.

Sources of Information for Literature Review

There are three main sources of information for a literature review: the preliminary, the primary
and the secondary sources.

Preliminary Sources
o Preliminary or general sources refer to materials which a researcher consults first to help him
or her locate other sources related to the research problem. Thepreliminarysources are in
themselves not information that the researcher is seeking for literature review, but are just
pointers to that information; they direct the researcher on where to get that information. A
table of contents is an example of a preliminary source.

Research Seminar
Oso Willis Yuko
15
Primary Sources
o Primary sourcesare direct descriptions of an occurrence by individuals who actually
experienced the occurrence. In research, this generally means description of a study by
persons who actually carried it out. Primary sources include journals, reports, dissertations
and theses, conference papers, etc. primary sources are preferred over other sources because
they provide detailed accounts of events in specialized fields. But they have a narrow scope
because they provide a lot of details in a small area.

Secondary Sources
o Secondary sources are materials where authors discuss or describe the works of others. They
are works presented by people who were not direct observers or participants in the events
being described, but are merely reporting on the work of someone else. The most popular
secondary sources are textbooks, and Encyclopedia. In a textbook on Management for
example, an author may describe several studies that have been done on education as a way
of illustrating an ideaor a concept. But the author did not personally conduct the studies
referred to.
o Secondary sources are preferred over other sources because they combine knowledge
from many primarysources into single publications. They also eliminate and simplify
much of the technical materials that may not be of interest to the general readers, and provide
a quick and relatively easy method of obtaining a good overall understanding of a field.
However, with a secondary source, one cannot be sure of the changes that have been
from the original source by the secondary author. In the process of combining and simplifying
many primary sources, a secondary source may leave out information that the reviewer may
need to know. Therefore literature review should, whenever possible, be based on primary
source.

Research Seminar
Oso Willis Yuko
16
 RESEARCH VARIABLES
o A variable is a characteristic (or an attribute) that can take a variety of forms (or values)
at different times, or in different people, or in different places, or in different circumstances.
Some phenomena can be measured or observed in two or more categories. These different
forms or categories are called attributes. Because the phenomenon varies (or exists in two
or more categories), it is called a variable. So variable is any characteristic with a variance.
o Examples of variables include educational status, marital status, gender, religion, ethnic
group, ability, temper, among others. If the values of a variable are expressed in numbers
to indicate the amount, degree, quantity or magnitude of the attribute, it is called a quantitative
variable. If a variable is expressed in terms of qualities, then it is called a qualitative variable.
But it is possible for one variable to be expressed in qualitative and in quantitative forms at
the same time - as shown in the below.
o A variable can active or inactive. An inactive variable cannot be manipulated (e.g.
gender). Active variables can be manipulated (e.g. teaching method). A variable is said to be
manipulateable if it can be made to assume different forms by the researcher.

Types of Variables
The Independent Variable
An independent variable refers to the conditions that a researcher controls (or changes) in
order to test its effect on some outcome. It is the variable which the researcher chooses to
study and manipulates in terms of amount or level, in order to assess its effect another
variable.

Dependent (or Criterion) Variable

o An independent variable is usually presumed to affect another variable. That other
variable the independent variable is presumed to affect is called the dependent variable.
Generally, the nature of the dependant variable depends on what the independent variable
does to it. Thus, a dependent variable manifests observable changes attributable to the
influence of an independent variable.
o Since a dependent variable depends on an independent variable, a dependent variablemust
change accordingly whenever an independent variable changes. In other words, a dependent
variable has no control over its behavior as it is dictated by the behavior of the independent
variable. Because it wholly dependent on an independent variable, it is called a dependent
variable. Further comparisons between the dependent and independent variables are
summarized in Table 4.3.

Table 4.3
Relationship between the Independent and the Dependent Variables
Factor Independent Variable Dependent Variable
Cause-effect Presumed cause Presumed effect
Timing Antecedent - happens before Outcome - happens after
Prediction Variable predicted from Variable predicted to

Research Seminar
Oso Willis Yuko
17
Experiment Manipulated in experiments Measured or observed in experiments
Change What the researcher changes What is supposed to change as a result

Extraneous Variables
o In any situation, there are always some unwelcome things or issues. In a family,or in
aschool,or in any organization, there may be unwantedpersons, behaviorcircumstances.
Because they are not wanted, every effort is made to keep them away or to neutralize
their effects. In a research situation also, there are always some unwantedvariables. These
unwanted variables are what are referred to as extraneous variables, or intervening, or
nuisancevariables.
o In an ideal study, only the independent variable should influence the dependent variable. But
this is not usually the case because there are always other variables that could also influence
the dependent variable, and hence the outcome of a study. However at any one time, a
researcher can only study a few of these variables in one study. Those other variables that
can also influence the results of the study, but which the researcher does not wish to study at
the moment are called extraneous, or intervening, or nuisance variables.
o The illustration between independent, dependent and extraneous variables can be
visualized as shown in Figure 4.1.

Extraneous Factors

Age
Gender
Level of education
Alcohol - IV Memory – DV (Ideal Path)
Health of the students.
IQ
Learning environment.
The teacher
Teaching method

Extraneous Factors

Fig. 4.1: Relationship between independent, dependent, and extraneous variables.

o All other routes or effects other than the route of the effect of alcohol on memory ( ) are
extraneous variables. Extraneous variables are actually independent variables, which the
researcher does not wish to investigate at the moment. They must be controlled so that
they do not influence the outcome of the study.

Control Variables
o Extraneous variables must be controlled. If they are not controlled, then they will also
influence the dependent variables and confound the results of the study. To control an
extraneous variable is to reduce its effect on the dependent variable to a bare minimum so that
it does not significantly influence the outcome of the study. Once an effect of the independent
on the dependent variable has been reduced to a bare minimum, it is then said to have
been controlled, and it becomes a control variable.
Research Seminar
Oso Willis Yuko
18
o A control variable is an extraneous variable whose effects on the dependent variable has been
held constant, or reduced to a bare minimum, or removed statistically (or by some other
means) so that it does not have a significant influence on the dependent variable of the study.
In other words, it is an extraneous variable that has been manipulated so as not to have a
significant effect on the dependent variable.

Moderator Variable
A moderator variable affects the relationship between the independent variable and the dependent
variable, but does not have a direct role in the cause and effect relationship. An example is a
catalyst in a chemical reaction.A moderator or mediator variable modifies the length of time the
independent variable takes to influence the dependent variable, but it does not alter, in any way,
the overall effect of the independent variable on the dependent variable. For example, people get
drunk faster when they drink in a bright light than when they drink under a dim light. Thus while
a person may take three hours to get drunk with two bottle of beer in dim light, the same person
may take only 45 minutes to get drunk to the same intensity with two bottles of beer when
taken in bright light. But light alone does not make the person drunk. So light is a moderator
or a mediator variable.

Continuous Variable
A continuous variable is a variable that can take unlimited number of values between any two
points. It is a variable whose valuesvary along a continuum, and which can assume any value
within a specific range. An example of a continuous variable is weight. Between 4 kg and 6
kg, there are infinite numbers of possible weights because weight can be expressed as a fraction.

Discrete Variable
A discrete or a categorical variable canassume only definite whole numbers, not fractions,
between any two points. It is a variable whose individual values fall on a scale with distinct gaps,
and defined sets of categories. An example of a discrete variable is the number of people.
Between 4 and 6 people, there can only be one value - 5 people and not 5.5, or any other number.
Other examples of discretevariable are gender, occupation, teaching method, and eye color.

Levels of Measurement
Measurement takes place in many places, at many times and by different people. A grandmother
may not know how to read time from a watchbutshe can tell the time by looking at the shadows
and the position of the sun. A mother may know how much water to use for tea and for ugali
without having to use complicatedscientific measurement tools. A carpenter, however, must use
scientificmeasurementtools to cutand fit the right sizes of wood, or his pieces may not fit together.
All these people are involved in measurement of some sort, but at different levels. Hence there
are differentlevels of at which a measurementcantake place.

Research Seminar
Oso Willis Yuko
19
There are four main levels of measurement: the nominal, ordinal, interval and ratio levels or scales
of measurements. The levels differ in their ability to classify and order variables, their ability to
provide equal interval and to absolute zero in variables.

Nominal Scale
The nominal scale is the lowest level of measurement; and it is concerned with just labeling,
classification or identification. Numbers in this scale are used purely for identification
purposes and they do not have magnitude.Admission number of students, the number of a student
in a class register, the numbers on players’shirts in a football team, post office box numbers, and
telephone numbers are some examples of a nominal scale. These numbers are arbitrary and do
not covey quantitative information. In a nominalscale, only names or labels are meaningful as a
means of making distinction between categories, but there is no quantification in them. A student
who is number one in the class register is notnecessarilythe best student in the class. The numbers
only serves as a name and is purely for identification purposes. Nominal measurements therefore
do not allow for comparison between the variables being measured. A researcher may chose
to code males - 1 and females - 2, but this does notmean that males are better than females, or
that females have more weight than males. It just indicates that 1 refers to males and 2 to
femalesbutthereare quantities in the numbers.

Ordinal Scale
The ordinalscale is a slightly higher scale than the nominal measurement because it introduces
order in measurements through the concept of relativity. For example, a student who is ranked
number one after a test, is better than one who is rankednumber two, and one ranked fifth is better
than one rankled six. There is magnitude attached to the numbers in this scale, which makes it
possible to identify variables and also arrange them in order of magnitude. However, the
difference between the values of any two consecutive entries in a variable such between 1 and 2,
or 5 and 6 are not the same. Suppose students A to F sat a mathematics test and obtained the
scores indicated in Table 4.3.

Table 4.3
An Example of Ordinal Measurement
Student Score Position
1
A 80 80 – 78 = 2
B 78 2
C 70 3
D 52 4
5
E 48 48 – 31 = 17
F 31 6

Student A, who scored 80% is number 1 and is better than student B who scored 78% and located
position 2. Butthe differences between the scores of Student A and student B is 2, and this is not
equal to the difference between the scores of Student E and Student F which is 17, yet 2 and 17
are both differences between two consecutive numberswhich vary be an equal

Research Seminar
Oso Willis Yuko
20
interval (1 & 2, and 5& 6). Even though it is possible to compare variables in terms of greater or
less than, there is no specificity because the amounts separating the observations are not indicated,
and even if they could be determined, they are not necessarily the same.So different
variablesmay occupy the same positions but havedifferent values. For example, two students
from different schools can both be first in their respective schools, but they may have
different grades. In the Olympics, each race always has a gold medal winner, but all winners run
different times.

Interval Scale
o Measurements on interval scale are more advanced than those taken on the nominal and
ordinal scales.Interval measurements are taken on an ordered scale in which intervals
have equidistant measures. The differences in score units are precisely defined to
represent the same amount of variable at all points on the scale. The same unit of measurement
taken on an interval scale represents the same quantity of the variableregardless of where the
measurement is taken. Temperature is an example of an interval scale. The interval between
25oC and 35oC and the interval between 73oC and 83oC both represent 10oC. A temperature
of 10oCdescribe the same degree of hotness whetherthe measurement is taken in Kenya or
anywhere in Europe; or whether the measurement is taken from water or from a metal
plate. Further, water heated to 100oC will boil regardless of how the heating is done.
o A major problem of the interval scale is that it does not have absolute zero; the value zero in
this scale is just arbitrary. So even though some point is marked is marked zero on the scale,
the value zero does not mean the absence of the characteristic being of measured. For
example, 0oC on the temperature scale does not mean that there is no heat (or energy) in the
materials whose temperature has been measured. Further, the absence of a true zero means
that doubling or tripling rule does not apply. A student who score 40% is not half as
intelligent as one who score 80%. Sincethereis no true zero, it is not possible to make
statements about how many times a measurement on this scale or higher or lower than another.

Ratio Scale
Measurements on the ratio scale are the most advanced, and are mostly used to measure
quantitative variables. A ratio measurement has all the three characteristics described above, but
it also has an absolute or true zero. A zero in a ratio measurement means the total absence of the
characteristic being measured. Because of the true zero, it allows for doubling or tripling
rule, and this makes it more advanced than the interval scale and indeed than all other scales of
measurements. In this scale, the difference between a 12 year old and a 24 year old is the same
as the difference between a 36 year old and a 48 year old. But a 36 year old is thrice as old as
a 12 year old. Other examples of ratio scale are weight, height, time, etc.

Research Seminar
Oso Willis Yuko
21
The difference between the scales described above can be summarized in Table 4.5 below.
Table 4.4
Characteristics of Levels of Measurements
Characteristics of Measurement
Scale Can Classify Can Order Has Equal Interval Has Absolute Zero
Nominal Yes No No No
Ordinal Yes Yes No No
Interval Yes Yes Yes No
Ratio Yes Yes Yes Yes

Methods of Controlling Variance

The different types of variables encountered in research and the relationships between them were
discussed in the preceding sections. There are always several variables in a research setting; the
challenge of any researcher is to ensure that the observed changes on the dependent variable
are due to the manipulation of the independent variable, and to other (extraneous) variables.
Otherwise, the study will be confounded. To avoid confounding a study, a researcher must
control the effects of the extraneous variables.

Summary of Variance Control Techniques

Table 4.7
Summary of Variance Control Techniques
Technique Main Characteristics When Appropriate to Use
1. Randomization A random selection of cases is The sample is fairly large to create
carried out in order to select a equivalent representative samples that
sample from the population. If there are similar in all the relevant variables
are different conditions in the study, that could influence the dependent
then the cases are randomly variable.Randomization is the simple best
assigned tom then different way to control the many extraneous
conditions. variables (known and suspected). It
should be used as a first priority
whenever possible.
2. Eliminating Extraneous Extraneous variables are physically The extraneous variable can be easily
Variables removed from the study. eliminated.Some extraneous variables
(e.g. noise, interruptions of physical
natures, environmental changes such as
temperature can be eliminated
completely from the study.)
3. Holding Factors The values of the extraneous An extraneous variable cannot be
Constant variable are kept the same for all eliminated. It significantly reduces the
cases or groups of cases (i.e. thus effect that an extraneous variable can

reducing the value of the measure of the variable in each

extraneousvariable to a constant.)
4. Matching Participants Two scores (or more), at a time,
on an Extraneous having fairly the same value, are
Variable randomly distributed to each group
to produce approximately equal

Research Seminar
Oso Willis Yuko
22
have on the dependent
variable.

To make
the cases
initially
comparable
on an
identified
extraneous
variable.

Research Seminar
Oso Willis Yuko
23
Technique Main Characteristics When Appropriate to Use
group.
5. Building Extraneous Extraneous variables are added into When an extraneous variable cannot be
Variables as the study by making them eliminated or held constant, or matched
Independent Variables independent variables; the influence on cases.
of the extraneous variable on the
dependent variable is also
determined.
6. Balancing Extraneous The effects of the extraneous There are so many unforeseen (real
Variables variables are equated by using a &suspected) extraneous variables in a
control group, but groups all are study.
initially made equivalent before the
study begins, through
randomization. It is assumed that
any other factor other than the
independent variable will affect all
groups in the same way.
7. Counter-Balancing The same group is first exposed to There are two conditions to be
Extraneous Variables one treatment, and then after some investigated in a study, but there are a
time, the same group is again limited number of casesmaking it
exposed to another treatment. impossible to have entirely different
groups.
8. Statistical Control The dependent variable scores are o Groups are not equivalent even
adjusted to eliminate the effect of an after randomization, and all other
extraneous variable using a techniques are not suitable.
statistical procedure.

Research Seminar
Oso Willis Yuko
24
 SAMPLING
Sampling is the process of selecting a “few cases” from a target population in order to
provide information that can be used to make judgments about a much larger number of
cases. The “few cases” is called the sample, and the whole group from which the “few” has been
drawn is the population or universe.

Aims of Sampling
The aim of sampling is to enable the estimation of a population from a part of it. Sampling is
necessary to:
1. Shorten the time that would otherwise be spent studying the whole population.
2. Reduce costs since fewer cases and resources are required than if the whole population is
used.
3. Gain greater accuracy in data collection and analysis, since the few cases involved are easily
amenable to a verity of data collection and analysis tools and techniques.
4. Eliminate the need to spend excessive resources to study a whole population when the
same information can be obtained from a part of the population. This is a doctor does not need
a litter of blood from a patient to test for malaria. The information contained in a litter
of blood can be obtained from just a blood smear.

Qualities of a Good Sample

A good sample is:
1. Representativeness: It should resemble the population closely on all characteristics on which
judgment is to be based..
2. Sample Size. A sample must attain a certain minimum size relative to the population
based on a particular level of precision. Too small a sample may not capture all the major
characteristics of the population, while too large a sample is wasteful.
3. Errors: A sample should be free from errors due to bias or distortion.

Disadvantages of Sampling
Despite the usefulness of sampling in research, it suffers from several weaknesses, and posses
some challenges to the research process.
1. Firstly, a sample is merely an estimate and no matter how accurate and exhaustive, it can
never fully describe the population. A completely representative sample is impossible to
attain.
2. Secondly, in studies where the treatment leads to destruction of the sampling units, sampling
becomes very uneconomical. But if the study leads to preferential treatment of aparticular
group, then it gives unfair advantage to that group.
3. The sampling process requires adequate knowledge of sampling techniques, of
statisticalanalyses and calculation of probableerrors, which not all researchers have.
4. Scientific sampling is a complex process in terms of administration and other logistics.

TermsRelatedtoSampling
The following terms are useful in understanding sampling.

Research Seminar
Oso Willis Yuko
25
Census A complete enumeration of the population. In census, each individual
member of the population is used in the study.
Target Population The larger population (universe) from which a sample has been drawn
and to which the researcher intends to generalize his or her findings.
It is an aggregation of study elements and refers to all members of a
real or hypothetical set of people, events, or objects to which we
wish to generalize the findings. Remember that a population refers to
whatever entity being studied and not necessarily to people. The
target population can be finite like in the case of students in a school;
or infinite like in the case of people in a market.
Accessible Population An accessible population is part or a subset of the target population
that a researcher can actually reach, and from which the researcher
actually draws a sample. It is an aggregation for which a sample is
selected. In order to generalize conclusions from an accessible to a
target population, the accessible population should closely resemble
the target population in respect to those variables relevant to the study.
This is called population validity. The relationship between a target,
accessible, and the sample populations can be visualized as depicted
in Figure 5.1.

Target Population

Accessible Population

Sample

Figure. 5.1. Relationship between target, accessible and sample.

Stratum Refers to a layer, or a subgroup or a sub-population within a
population when a set of specifications divide the population into
mutually exclusive entities. Males constitute a stratum when gender
is used to categorize human population.
Sample A small group of subjects or participants selected procedurally
from the target or accessible population to represent the latter. It is
a portion of the target or accessible population which the researcher
will actually study.
Sampling Frame An exhaustive, comprehensive, correct, reliable and appropriate list
of members of the defined population from which the sample is to be
selected. It is a list of members of the target/accessible population
from which a researcher draws a sample.
Sampling Unit A single member of a population, which must be non-overlapping. A
student is a sampling unit when a sample is to be selected from a
population of university students.
Element The unit of analysis about which data is collected, and that forms

Research Seminar
Oso Willis Yuko
26
 the basis of analysis. It is an individual member of a sample.
Random Selection The process of picking cases from the target population without
any bias (or by chance) to represent the population.
Random Assignment The placement of cases to different groups without bias (by chance)
for information-gathering. This is usually done when a researcher has
more than one condition to investigate in a study.

Probability Sampling Methods

SimpleRandomSampling
In simple random sampling, each member of the population has an “equal” and “an
independent” chance of being included in the sample.
o The chances are independent if the selection of one member does not, in any way, affect
the chances of another member being selected. The fact that a member, A, has been selected
does not in any way prevent a member, B, from being selected.
o The chances are equal if the selection of a member is not influenced by, or is not
dependent on, other factors other than being a member of the population. Since all members
are part of the population, they all have equal chances on this basis. And because of the
equal chance, those individuals that are finally included in the sample are selected only by
chance.
o The chances are known if the probability of each member can be determined with
certainty prior top selection.

Aim of SimpleRandomSampling
The aim of simple randomsampling is to produce a random sample that is representative of
the population. A sample is random if:
1. Itis selected from the population in such a way that every member of the population has a
chance of being selected. If the chance is the same for each member of the population,
then it is called a simple random sample.
2. Each possible sample of the same size has an equal probability of being selected.
3. There is no systematic variation in the sample that would make it different from other
samples. If any differences occur, then they are only due to chance and not bias.
o Representative does not imply that the sample is identical to the population. In fact, a sample
can never be equal, or identical to the population from which it was drawn. Rather,
repetitiveness means that a sample is approximately the same as the population on the major
characteristics of interest.
o Approximately implies some degree of difference between the sample and the population, the
size of which is closely related to the size of the sample.
o The difference between the characteristics of the sample and those of the population is called
sampling or random error.

Stratified Sampling
o Stratified sampling is a sampling technique that identifies the major and significant sub-
groups within a population and then selects elements from each sub-group to form a
sample with a view of reproducing the population.
Research Seminar
Oso Willis Yuko
27
o The purpose of stratified random sampling is to divide a population of study into homogenous
sub-groups and represent each sub-group in the sample in proportions equivalent to its size in
the population. This ensures that each subgroup characteristics are taken into account, thus
raising the validity of the study.
o A population is heterogeneous if it has distinct and mutually exclusive subgroups within
the main population (i.e. when it is not possible for one member to belong to more than
one sub-group at the same time). When faced with such a situation, researcher instead of
selecting the sample randomly from the target population, first divides the population into
homogenous strata or layers, and then select cases from each stratum by random sampling
procedure. This is stratified sampling. T

Types of Stratified Sampling

o In proportionatestratified sampling, a sample is randomly selected from each stratum in
proportions to the size of the stratum to the population. In this case, the size of each
stratum in the sample is a simple fraction (or percentage) of its size in the total
population.
o Sometimesresearchers do not wish to get involved with complex mathematics. Soa researcher
may just decide and assign sizes to sub groups at his or her discretion, and based on
other criteria without mathematical rations. This is disproportionatestratified sampling. In
disproportionate stratified sampling, the sizes of subgroups in the sample are not direct rations
of their sizes in the population. The researcher simply assigns the largest percentage in the
sample to the stratum with the largest size in the population without using mathematical
equations.
o Disproportionate stratified sampling is most appropriate when:
i. A researcher doe not attach much significance to the results of a study, as in the case
of pilot and exploratory studies.
ii. When a researcher has a strong conviction that particular group has the
relevantinformation or is more affected by the issue under investigation, and therefore
require more representation, while at the same time, there is a strong conviction that
the views of other groups are equally important.

Strengths and Weaknesses

The major weaknesses of stratified sampling are that:
1. It does not indicate exactly how the individual cases from each stratum should be
selected. Stratification only indicates the sizes of each stratum required in the sample, but
it does not indicatehowthese cases should actually be selected. This is open to the discretion
of the researcher.
2. It requires that all population and subgroupcharacteristics be known. However, this is usually
not an easy task especially when the population is large or when there are many stratification
variables to be considered.
3. Logistics for the selection are usually complex in stratified sampling than in simple
random sampling since all population parameters must be taken into account.
But despite its shortcomings, stratified sampling is a very important strategy in sample selection
because:
Research Seminar
Oso Willis Yuko
28
1. It ensures that each sub-group is represented in the sample by at least one case. This
raises the validity of the findings because it takes into account the heterogeneity of
populations.
2. Since each subgroup is represented, stratified sampling is a sure way of accounting for the
differences between the subgroups characteristics.
3. It has proportionate strategy and this ensures equitable distribution in the sample and increase
validity of findings.

Area (Cluster) Sampling

o A cluster or an area is any group (geographical or otherwise) where an intact group with
similar characteristics can be found. Cluster or area sampling is a random selection of clusters
from the larger population of clusters.Situations may sometimes arise where it is not possible
or cost effective to obtain anexhaustive, comprehensive, correct, reliable and appropriate list
of all the members of the population, like when the population is large, or widely dispersed.
In such a case, selecting a simple random sample poses administrative problems. So under
such and similar conditions, a researcher may divide the population into subgroups (as in
the case of stratified sampling), then the subgroupsas the sampling frame. Some of the groups
are then selected at random. Once a subgroup is selected through a random procedure, all
the members of that group are studied. This means that the random sampling is at the group
rather than at the individual level. Groups or areas or clusters are selected through simple
random procedure and once a group or a cluster is selected, all the members of the selected
groups or clusters constitute the sample.
o The major advantage of cluster sampling is that it is cheap, fast and easy to administer.
But is rather inaccurate as it involves more than one sampling error, especially if
multistage sampling is used. Moreover, it produces a larger sampling error than simple
random sample of the same size because the clusters tend to be similar within the cluster, thus
reducing representativeness of the sample. Analysis of data collected from cluster samples
also requires non-conventional analysis methods which are rather subjective.
o A Multistageand two-stage random sampling are improved versions of cluster sampling
technique. They combine cluster sampling with individual participant sampling. A multistage
technique begins by selecting clusters at random, and then proceeds to select individuals
within the clusters until the required number of individual cases is reached. This process
is continued until the individual sample is obtained by the simple random sampling technique.

Systematic Sampling
o Systematic sampling is usually regarded as a simplified version of simple random sampling.
In systematic sampling, the first number (n) of the sample is randomly selected from a
randomized list and from that point on, every other nth member on the list is picked in
equal fractions of n. Suppose a researcher wants to select a sample of 10 from a population
of 30 respondents using the systematic sampling technique, the researcher should first obtain
the sampling interval or sampling ratio (fraction). A sampling interval is the distance (in
the list) between each of the individuals selected for the sample, and rounded off to the nearest
whole number. For the case here, it is:

Research Seminar
Oso Willis Yuko
29
 Sampling Interval = Population = 30 = 3
Sample Size 10
The sampling frame is then divided into groups of three (sampling interval) starting from the first
member in the list. The total number of groups produced by this procedure should be equal
to the sample size (in this case 10) as demonstrated below.

1 4 7 10 13 16 19 22 25 28
2 5 8 11 14 17 20 23 26 29
3 6 9 12 15 18 21 24 27 30

o The first member (n) is randomly selected from the first group on the list. After the first
member of the first group is selected, every other nth member of each group is picked for the
sample. Suppose the second member (n = 2) from the first group has been selected through a
simple random procedure, then every second member of each group should be picked for the
sample. The units 2, 5, 8, 11, 14, 17, 20, 23, 26 and 29 are systematically selected for the
sample.

Non-Probability SamplingMethods
ConvenienceSampling
✓ Convenience sampling is also referred to as accidental or as man-on-the-street sampling
technique. In this technique the sample is selected from those who happen to be available
at the time of data collection, a first-come-first-served basis, until the desired number is
attained. The researcher simply picks from those who happen to be available at the time
of data collection, without any conscious effort to select particular case or cases, and without
attempting to construct a sampling frame. Those who happen to be available at the time
of data collection constitute the accessible population, and whoever is available and willing
can be selected for the study.
✓ The problem with convenience sampling technique is that but cannot produce a representative
sample.

Purposive Sampling
✓ Purposive (or judgmental) sampling is a selection technique where a researcher uses personal
judgment regarding the cases from which information is to be collected, and deliberately
picks cases that are typical of the population being studied. The researcher uses personal
judgment and logic to select only those cases that best suit the purposes of the study, or
those believed to have the information sought. The researcher hand-picks the individual cases
on the basis of their typicality - and builds up a sample that is satisfactory to the specific
needs of the study. Just like convenience sampling, purposive sampling cannot produce a
random sample.

Quota Sampling
✓ Quota sampling is usually referred to as a non-probability stratified sampling, because it
attempts to obtain representatives of the various elements of the total population in the
proportions in which they occur in the population. Just like in stratified sampling,
theproportions of the various strata in the population are determined and the sample is
Research Seminar
Oso Willis Yuko
30
 drawn to have the same percentage as the population. But the sample is not drawn at random.
✓ A quota sample is not drawn at random. Once the quota for a particular category is
determined, then any non-probability technique (purposive, convenience, or snowballing) can
be used to fill the quota.

Networking Sampling (Snowball)

✓ This technique is usually employed to locate cases that are difficult or impossible to
locate through other means. Networking sampling strategy takes advantage of social networks
and the fact that friends tend to have some similar characteristics. In the first stage, a case,
or few cases, having the requisite characteristic(s)are identified and studied. Once the case(s)
with the requisite characteristics are identified, they are then asked for assistance in locating
more cases with characteristics similar to theirs. This process is continued until the desired
sample is attained.The main weakness of snowball sampling is the time it takes to builds the
requisite sample size. Further, the sample produced by this method cannot be representative.
If people are given a chance to invite others, they will always have a reason to invite one
and not the other, and such prejudices will definitely be reflected in the sample.

The major features of the common sampling techniques are summarized in Table 5.5.

Research Seminar
Oso Willis Yuko
31
Table 5.2
Summary of Common Sampling Techniques
Sampling Technique Main Features Strengths Weaknesses Purpose Justification
Selects individuals without - Ideal procedure for selecting - Easily affected by attrition. To select a random and - Gives each member of
bias from a a representative sample. - Tedious to construct an representative sample. the population known,
target/accessible - Results are generalizable. exhaustive sampling frame in equal and independent
population in such a way - High population validity. some situations. chance of being
that each unit has a - No guarantee of selected.
Simple Random known, equal and representativeness especially - Produces a random
independent chance of if the population is sample.
being selected in the heterogonous.
sample. - Low precision.
- Can be expensive in some
studies.
Identifies subgroups in a - Subgroups are - Cumbersome because sub- - To group a population - Ensures proportionate
population and their proportionately represented. group characteristics must be into homogenous representation of
proportions and - Accounts for the difference in known, and logistics of subgroups. subgroups in the sample,
selectscases randomly subgroup characteristics. selection complex. - To ensure equitable and accounts for
from each sub group to - High precision. - More expensive than simple representation of each differences in subgroup
Stratified form a sample. random. subgroup ion the characteristics.
- Difficult to identify sample.
appropriate identification
variables.
- Unfeasible or expensive on
many stratification variables.
Probability
1-Stage: Selects group - Cheaper and easier to - Less accurate because there is To select groups instead of - Cheaper and faster than
rather than individual implement than simple more than one sampling individuals when an simple random.
members. Once a group is random. error. exhaustive v sampling frame - Suitable if a sampling
selected, all the cases inn - Data analysis requires cannot be constructed. frame cannot be
then group are studied. unconventional means, hence obtained.
subjective.
2-Stage: Selects group - Slightly more accurate than 1-stage
rather than individual because of random selection.
members. Once a group is
Cluster
selected, individual
members are selected at
random from the group.
Multistage: selects - Cheaper and faster than Cannot claim to produce a Used when it is either - Cheaper and faster than
progressively smaller simple random. representative sample. impossible or impractical to simple random.
groups until the individual - Substitutes for simple compile an exhaustive - Substitutes for simple
members of the sample are random. sampling frame. random.
selected through a random
procedure.
Select every nth member of - Easy and cheaper to - Suffers from periodicity. - Substitutes from simple - Easy and cheaper to
a population from a implement than simple and - List must be randomized to and stratified sampling implement than simple
Systematic randomized list of the stratified sampling. avoid periodicity. techniques. and stratified sampling.
population. - Low generalizability.

Research Seminar
Oso Willis Yuko
32
Sampling Technique Main Features Strengths Weaknesses Purpose Justification
The researcher - Selects typical and useful - May produce biased sample. - To collect focused - Selects typical and
Purposive or deliberately determines cases only. - Low generalizability. information. useful cases only.
Judgmental which cases to include in - Saves time and reduces
the sample. costs.
Selects, on-first-come- - Collects data on the spur of - Cannot produce a - Suitable for pilot and - Collects data on the spur
first-served basis, those the moment to take advantage representative sample. exploratory studies. of the moment without
who happen to be of unexpected events. - Not useful in descriptive and - Suitable for infinite rigidity of procedure.
Convenience or
available, and appropriate - Easy to implement. causal studies. populations when it is - Takes advantage of
Accidental
at the spur of the moment. not possible to construct those who happen to be
an exhaustive sampling there, or of unexpected
frame. events.
Selects a sample on the - The only way to locate cases - Time consuming, and cannot - To locate cases that are - Locates cases that are
Non-Probability basis of social networks on that are difficult to trace. produce a representative difficult to trace. difficult to trace.
Snowball the premise that cases with sample.
similar characteristics tend
to belong together.
Identifies subgroups in the - Proportionate representation - Cumbersome because - To group a population - Ensures proportionate
population and their of each subgroup. subgroups characteristics into homogenous representation of
proportions and selects - Easier to implement than must be known. subgroups. subgroups in the sample.
from each subgroup (but stratified sampling. - Low reliability - To provide equitable - Accounts for differences
Quota not randomly) to form a - Sample can be controlled for representation of each in subgroup
sample. Uses only non- certain characteristics. subgroup ion the characteristics.
probability techniques to sample. - Useful in large scale
select individual cases - To plan a major study. surveys.
from each quota.

Research Seminar
Oso Willis Yuko
33
The Sample Size
• Drawing conclusions about a population based on a sample is never totally satisfactory since
a researcher can never be sure that the sample is representative of the population. Some
differences between the sample and the population are bound to exist. However, if the sample
is randomly selected and is of a sufficient size, then such differences are likely to be
insignificant and incidental.
• A sample size is the number of cases or elements that should be studied and from which reliable
inferences can be made about the population. It is represented by as s or by an n, and the
population or universe by an N, or µ. A researcher must determine the least size of a sample
that will provide sufficient data to answer the research questions.

Factors influencing Sample Size

Generally, a sample size will depend on:
1. Homogeneity or heterogeneity of the population: The more heterogeneous the population, the
larger the sample should be so as to capture all the diverse characteristics of the population.
2. The breakdown planned in tabulation: if a complex tabulation is planned, then a large sample
is required to ensure that there is reasonable number of subjects in each category.
3. Types of sampling techniques to be used or the research design itself (See below).
4. Resources available: scarcity of resources may necessitate the use of a smaller sample than
recommended.

Determination of Sample Size

A sample size can be determined through non-statistical, statistical, and by through statistical
tables.

Non-StatisticalEstimations
In this approach, the sample size is decided at the discretion of the researcher, taking into account
several factors in the study, but without applying any approved mathematical formula. The issues
that are considered includethe type of research; hypotheses: number of variables in the study;
methods of data collection; importance of the study; resource constraints.

Statistical Methods
• There are mathematical formulae available for determining the required sample sizes. A
sample size can be determined with statistical precision depending on the sampling error, the
degree of confidence and the desired degree of representativeness of the sample.
• Krejcie and Morgan (1970) have provided the formula for estimating the sample size needed
for a population (N) as;

s = χ2 NP (1 – P)
d2 (N – 1) + χ2 (1 – P)

Research Seminar
Oso Willis Yuko
34
Use of Statistical Tables
• There are tables available determining the required sample size. However, these tables only
deal with finite populations. As such, the number of subjects in the population intended for
study must be known. Table 5.10 as reproduced here was developed by Krejcie and Morgan
(1970) and has been widely used to determine sample sizes for finite populations.

Table 5.10
Sample Size for Finite Populations
N s N s N s N s
10 10 150 108 460 210 2200 327
15 14 160 113 480 214 2400 331
20 19 170 118 500 217 2600 335
25 24 180 123 550 226 2800 338
30 28 190 127 600 234 300 341
35 32 200 132 650 241 3500 346
40 36 210 136 700 248 4000 351
45 40 220 140 750 254 4500 354
50 44 230 14 800 260 5000 357
55 48 240 148 850 265 6000 361
60 52 250 152 900 269 7000 364
65 56 260 155 950 274 8000 367
70 59 270 159 1000 278 9000 368
75 63 280 162 1100 285 10000 370
80 66 290 165 1200 291 15000 375
85 70 300 169 1300 297 20000 377
90 73 320 175 1400 302 30000 379
95 76 340 181 1500 306 40000 380
100 80 360 186 1600 310 50000 381
110 86 380 191 1700 313 75000 382
120 92 400 196 1800 317 100000 384
130 97 420 201 1900 320
140 103 440 205 2000 322
Note. N = population size: s = sample size. Adapted from “Introduction to research ”, (pp. 55-
56), by Kathuri, N. J. & Palls, D. A., Egerton University Book Series.

Research Seminar
Oso Willis Yuko
35
 VALIDITY AND RELIABILITY
Validity
o The common definition of validity is the ability of a test to measure what it purports to measure.
Validity is concerned with one basic question: are we measuring what we are supposed to
measure or are there some systematic errors that might interfere with the measurement?
However, this general definitiondoes not take into account the fact that there is more
thanonekind of validity. Therefore the best way to ask a question on validity is not whether a
test is valid or not, but rather, whether the test is valid for the purpose to which it isintended.
Validity is the appropriateness, meaningfulness and usefulness of the information that a
researcher makes based on thedata collected.

Internal Validity
o Internal validity is the extent to which the results of a study can be accurately interpreted.
Internal validity is concerned with the extent to which the effects of the extraneous variables
have been controlled so that the observed effects on the dependent variable can be attributed
to the independent variable only and not any other factor.
o It is the claim that all observed relationships between the variables are meaningful in their own
rights, rather being due to something else. To achieve high internal validity, a researcher should
properly screen out and control all extraneous variables, and other contaminants to reduce their
effect on the dependent variable to minimum.

External Validity
o External validity is the extent to which research results can be generalized to other populations.
o Population validity is the extent to which the sample from which data is to be, or was collected,
correctly represents the population to which the findings of the study should apply. Without
population validity, findings of a study cannot be applied to contexts other than the one in
which the study was conducted.

Ecological validity
o Ecological validity is equivalent to naturalness of the research setting. The more a study setting
reflects a real life situation, the higher its ecological validity. A study has a high ecological
validity if it generalizes beyond the experimental environment to more realistic formal or field
settings. Ecological validity will be affected if the independent variable is not described
explicitly and if there is multiple treatment interfaces.

Content Validity
o Content validity is the representativeness of the items on the instrument as they relate to the
entire domain of the questions being asked. Usually, when collecting data, it is not possible

Research Seminar
Oso Willis Yuko
36
 or necessary to ask all the possible questions that there are to be asked in the research, or to
include all the items that there are, in a questionnaire or in an interview.
o Content validity is concerned with one basic question: has the conceptual space been covered?
It measures the appropriateness, the comprehensiveness and the adequacy of the content of a
study.

Criterion Validity
o Criterion (or absolute) validity refers to the relationship between the scores on the instrument
in question and the scores on another instrument or another external variable referred to as the
criterion.
o A criterion is a second test or another procedure presumed to measure the same variable. It
measures the correlation between an item (or items) and some standard or benchmark which is
known to be a good indicator of the concept being measured.

Construct Validity
o Construct (or theoretical) validity is to the degree to which the totalities of evidence obtained
from research or from an instrument are consistent with theoretical expectations.It describes
how well items are tapping into the underlying theory or model of behavior, or the extent to
which an instrument measures a hypothetical construct. Construct validity is particularly useful
in studies that propose to test hypothesis.

Face Validity
o Face validityrefers to the external appearance and appeal of a research instrument or research
design as a whole. It is the degree to which a research instrument or design looks valid to some
research subject or to somebody untrained in research or measurements. It is simply an
assertion from the researcher claiming that what was intended to be measured has been
measured. It is possible and common for instruments (or even the whole studies) to elicit
undesirable responses and produce invalid results because the respondents do not take
themseriously due to lack of face validity.

Threats to Validity
Reactivity (The Hawthorn Effect)
When research respondents or subjects are aware that they are being studied, they tend to modify
their behavior to produce the information which the researcher is looking for.

Maturation
Systematic sociological, physiological, or psychological changes in the subjects may sometimes
cause them to behave differently. These natural developments in subjects or respondents in the
course of a study are referred to as maturation.

Research Seminar
Oso Willis Yuko
37
History
Research usually extends over a period of time. This provides opportunity for other events in the
research environment which may influence the results of the study to occur. Sometimes one or
more unanticipated and unplanned, or even planned events occur in the course of the study and
affect the behavior of respondents. Such events which occur during the course of a study and may
affect its outcome are called history.

Regression
If a group is selected because of unusually low (or high) performance, it will on average, score
closer to the mean on subsequent tests or measurements, regardless of what transpires in the mean
time. If a sample is selected because of extreme scores, then it is very likely that outcomes will
show averages towards the mean due to regression rather than due to treatment.

Sample Size
The smaller the sample, the lower the validity. Hence it is important to select a sample that will
give credibility to his or her study (For more discussion, see chapter 5).

Mortality (Attrition)
It is quite common in research for subjects or respondents already included in the sample to be lost
as the study progresses. They may be lost when they leave (or migrate), are absent at the time of
data collection, die or refuse to participate. Mortality is a case where subjects or respondents
already included in the sample are lost before the study is concluded. This is more serious if
mortalityrate is differential with one group losing more subjects than the other.

Selection
The way respondents or subjects are selected for a study may result the individuals or groups
differing from one another in unintended ways that are related to the variables to be studied,
especially if a random sample is not used. Differential selection of subjects for differentlevels of
the independent variable and any other factors that may make the groups different from the
beginning of the study is referred to as selection bias or subject characteristic threat.

Reliability
o Reliability is generally defined as stability over time and stability over rates, and is concerned
with consistency and replicability of study results.
o A reliable study should produce fairly the same results if it is done repeatedly assuming, of
course, that there is no change in information between successive studies.
o A study is reliable if its results can be depended on, being consistent and predictable over time
and use: that is if, after repeated measurements, it produces more or less the same results.
o The concepts of reliability can be illustrated by analogy in Figure 6.1.

Research Seminar
Oso Willis Yuko
38
 ***
Type I: Good reliability and poor validity. The shots are consistently clustered at the same
place(good reliability) though not at the center. The
center (target) is consistently missed, hence poor

Type II: Good reliability and good validity. All shots are clustered together (good reliability) and they
**
* all hit the target (good validity).

Type III: poorreliability and poor validity. * All shots are scattered and they all miss the target.

Figure. 6.1. Relationship between validity and reliability.

o A study cannot be valid if it is not reliable, but it can be reliable without being valid. A study
can consistently produce wrong results. The fact that it produces consistent results makes it
reliable, but the fact that it produces incorrect results makes it invalid.

Techniques of Measuring Reliability

Test-Re-Test Method
In a test-re-test method, a researcher administers the same instrument twice to the same group at
different points in time. The results from both administrations are then correlated. This is called
test-re-test reliability and yields a coefficient of stability.

Equivalent Forms Methods

These are also known as alternate, multiple or parallel forms reliability. The equivalent forms
developed as a reaction to the weaknesses of the test re test method described above, any two or
more instruments are parallel or alternate if they have the same number of items, have same means,
variances, inter correlations, and if all other variables that are not significantly different from each
other such that it does not matter which one is used.

Internal Consistency Methods

Internal consistency methods, as opposed to the ones already discussed, require only single
administration of an instrument, which is then systematically divided into two equal halves, and
correlated. The researcher analyses data from a single instrument under the assumptions that all
the items on the instrument are intended to measure the same concept.

Research Seminar
Oso Willis Yuko
39
Reliability in Qualitative Research
While many qualitative researchers use many of the procedures described above, some take the
position that validity and reliability, as discussed here, are either irrelevant or not suited to their
research efforts because they are attempting to describe a specific situation or event as viewed by
a particular individual. They emphasize instead honesty, believability, expertise and integrity of
the researcher.

Research Seminar
Oso Willis Yuko
40
 RESEARCH DESIGNS
o A research design is the plan or the overall strategy for conducting the research. It refers to the
procedures used by a researcher to explore relationships between variables, to form subjects
into groups, to administer treatments and analyze data. It is a means of ensuring that a research
process is sufficiently systematic and scientific, and that the results obtained are applicable in
real life.
o There are two overall research methodologies (or designs): qualitative and quantitative.
Generally, quantitative research designs are characterized by data (or findings) which are
expressed in numerical forms, and by analyzed using suitable statistical methods. Qualitative
research designs are characterized by data (or information) that is described verbally or non-
numerically.

Table 7.1
Major Differences between Qualitative and Quantitative Research Designs
Source of Difference Quantitative Designs Qualitative Designs
1. Approach to research - Scientific and formal. - Scientific and neutral.
2. Research Data - Reduced to scores. - Presented in a narrative
description.
3. Knowledge View - Concrete, conceptual and precise. - Experiential, contextual and
Can be tested. Seeks proof. best expressed by the owners.
4. Process of Inquiry - Linear, from stage to stage. - Cyclic and sophisticated.

5. Mode of Inquiry - Experimental, manipulates - Multimodal, multi stranded.

conditions.
6. Use oh Hypotheses - Manipulates variables to prove - Understands patterns of
hypothesis. occurrences in order to
identify variables.
- Generates hypotheses.
7. Main Questions - Asks how much, how many. - Asks why, who, when, etc.

8. Types of Samples - Uses a random sample. - Uses theoretical samples and

bounded cases.
9. Mode of Analysis - Statistical. - Interpretative.
10. View of Reality - Objective, singular and independent - Subjective and multiple.
of the researcher. - As seen by the researcher.
11. Relationship of the - Withdrawn and independent of that - Interacts with that being
Researcher being researched. researched.
12. Role of Value - Value free and unbiased. - Value laden and biased.
13. Purpose of Research - Deductive: cause-effect. - Inductive: mutual,
Generalizations leading to simultaneous studying of
explanations, understanding, etc factors. Generation of
(Nomethetic). theories.
- Particularization of cases
(Ideographic).

Research Seminar
Oso Willis Yuko
41
Choosing a Research Design
o In planning to conduct a study, a researcher may develop a liking for one design. However, it
would be unwise to use one approach exclusively. A mixed method approach is recommended.
o Sequential mixed designs are procedures where a researcher seeks to elaborate or expand the
findings of one method with another. A researcher may begin with a qualitative method to
identify the key or main or major issues pertinent to the study and then plan a detailed study to
investigate the issues so isolated.
o Concurrent mixed designs combine quantitative and qualitative data in order to provide a
comprehensive analysis of a research problem. In such a case, both quantitative and qualitative
data are collected at the same time and then integrated in the interpretation of the overall results.
o In transformative mixed designs, the results of one study are intended as a leveraging pad.

Table 7.2
Guide to Choosing a Researcher Design
Criteria Quantitative Design Qualitative Design
1. Researcher’s World - The world is objective and - The world is subjective with
View measurable in specific terms. multiple realities.
Reality is single. - Reality is contextual.
2. Training and - Technical writing skills, - Literally writing skills, computer
Experience computer, statistical skills, text analysis skills.
library skills.
3. Psychological - Comfort with rules and - Comfort with lack of specific rules
Attributes regulations for conducting and procedures for conducting
research. research.
- Low tolerance for ambiguity. - High tolerance for ambiguity.
- Time short. - Time for study lengthy.
4. Nature of the - Previously studied by other - Exploratory research,
Problem researchers. - Variables unknown,
- A body of literature exists. - Context important.
Known variables. - Lacks theory base for the study.
- Existing theories.
5. Type of Audience - Individuals accustomed to, or - Individuals accustomed to, or
supportive of quantitative supportive of qualitative studies.
research studies.
6. Level of Control - Full or sufficient control - Limited control

Characteristics of a Good Research Design

A good research design must be:
1. Free from Bias: The data and the statistical methods used should not vary in any systematic
way except as would be expected from random fluctuations. Any differences must be
attributable to the independent variable. Systematic errors occur in given directions and cannot
be estimated statistically.

Research Seminar
Oso Willis Yuko
42
2. Free from confounding effects: A good research design eliminates confounding of variables,
or keeps it to minimum so that effects can be separated and results interpreted without
confusion. Controlling extraneous variables also ensures internal validity.
3. Collect precise data for statistical hypothesis tests: A good research design should produce
adequate data with sufficient precision to adequately test those hypotheses that require
statistical testing.

Research Seminar
Oso Willis Yuko
43
 Amoud University - Somaliland

RESEARCH

QUANTITATIVE QUALITATIVE

Experimental Non Experimental Case Study Ethnography Phenomenology Grounded Theory

True Experimental Quasi Experimental Pre Experimental Expost Fcato Descriptives

Exploratory Correlational Survey Historical

Pretest-Posttest Posttests Only Solomon-4-Group Factorial

Simple Partial Multiple Point bi Serial Phi

Time Series Intact Group Non-Equivalent One-Shot Case Study

Figure 7.1. Description of research designs. Literature Review Expert Survey Case Study Cross-sectional Longitudinal

Panel Cohort Trend

Research Seminar
Oso Willis Yuko
43
Quantitative Research Designs
o Quantitative research designs are broadly divided into experimental and non-experimental
designs.
o Experimental designs generally involve control and manipulation of variables in order to
establish cause-effects relationships, the knowledge of which is used to predict and control
events. Experimental designs are carefully designed to control all variables except the
independent variables.
o Non-experimental designs do not manipulate variables, and they do not establish cause-effect
relationships. They are basically used to:
1. Describe and explain events, situations, phenomena, etc as they exist or as they once existed.
2. Evaluate products, events, concepts, etc.
3. Develop new innovations.

Experimental Research Designs

o An experimental research is a study conducted to examine and establish cause-effect
relationships between variables in a controlled setting. An experimental research determines
the relationship between a dependant variable of interest and an independent variable that the
researcher has hypothesized to be causally related. The independent variable is then
manipulated by the researcher under carefully controlled conditions to determine if it is
producing a change in the dependent variable.
o The basic characteristic of an experimental research design that distinguishes it from the other
research designs is the manipulation of the independent variable. Manipulation means that the
researcher determines (directly or indirectly) the various forms (or levels, amounts, etc) that
the independent variable may take and which groups will get which kinds of treatment.

Characteristics of Experimental Research Designs

An experimental research has three basic characteristics:
1. The researcher manipulates the independent variable. This is to say that the researcher
deliberately determines (directly or indirectly) what forms the independent variable takes, and
which groups receive which forms of treatment.
2. There are at least two groups (one experimental and one control group) included in the study.
The experimental group receives treatment while the control group does not (but it may receive
a routine treatment).
3. There is randomization of subjects. Subjects in the study are first randomly selected from the
target population and then randomly assigned to experimental and control groups, or groups
(E and C) are randomly assigned to subjects.
o Experimental research designs have advantage over other designs because they can achieve
higher internal validity as well as establish causal claims. Only experiments can test precise
predictions because they are controlled and the researcher can manipulate the independent
variables as he or she wishes.

Research Seminar
Oso Willis Yuko
44
Other Terms in Experimental Research
1. Pretest: Since experiments are usually concerned with measuring changes, there should
always exist two values from which the magnitude of change can be determined. Before a
researcher can introduce an intervention, he or she should first determine the initial level of the
variable of interest. The determination of the initial value of the variable before the intervention
is called a pretest
2. Treatment: This is the intervention or the change that a researcher introduces in the
independent variable to see its effects on the dependent variable.
3. Posttest: After the intervention, a researcher would want to know if the treatment (or induced
change) has caused a change in the dependent variable. To do this, a researcher has to
determine the level of the dependent variable after treatment. This is called posttest. It is the
determination of the final value of the dependent variable after treatment.

Ex-Post Facto Research Design

Expost facto or casual comparative research design refers to an “experiment” in which a
researcher, rather than administer a treatment, examines the effects of a naturalistically occurring
treatment after that treatment has occurred. It is a pseudo (or false) experimental research design
where a researcher, rather than administer a treatment, substitutes for the treatment with a naturally
occurring condition after the condition has occurred. The researcher attempts to relate this after-
the-fact-treatment to an outcome, by looking for a natural cause without having to directly
manipulate the independent variable. Expost facto research designs are suitable if:
1. The variables being investigated cannot be manipulated (e.g. gender).
2. The variables or should not be manipulated (e.g. brain damage, HIV infection),
3. The variables are simply not manipulated though they could be manipulated (e.g. teaching
method, broken homes, intelligence).

Suppose a researcher wants to study the effect of broken homes on the performance of students at
school, the researcher would, in an ideal situation, randomly select a sample of students, break
their homes and then measure their performance. But it is not ethically right break up homes. The
researcher would select students whose homes are already broken and administer a performance
test on them. This is expost facto design, where a researcher rather than administer a treatment,
looks for an already naturalistically existing condition without having to administer the treatment.

Survey Research Design

o Survey studies, or as they are also called, normative or status studies, are present oriented
studies designed to investigate populations by selecting samples to analyze and discover
occurrences, and provide quantitative descriptions of some part of a population within a slice
of time. Surveys basically explore, describe and explain opinions, attitudes, preferences and
perceptions of groups of people of interest to a researcher.

Research Seminar
Oso Willis Yuko
45
o Surveys involve a researcher asking (usually) a large group of people questions related to the
issue at hand. Survey research uses mainly questionnaires to collect basic descriptive
information from a broad sample, and interviews to gather in depth responses usually from
small samples or on sensitive topics, and to elicit deeper responses that cannot be adequately
captured by questionnaires.
o Surveys gather facts (rather than manipulate variables) to discover, clarify and describe
incidences, distribution, or interrelationship among variables to fully explain the phenomenon
involved.

Cross-Sectional Survey
A cross-sectional survey is a kind of survey that collects data from a target population at one point
in time. The data is collected from various cases at the same time - although the time taken to
collect this data may vary between a day and a few weeks. The researcher goes to the population
and collects data from a cross-section of the population at one point in time, and then makes a
report based on the data collected at once. The researcher has no intention of going back to the
same population to collect the same data for the dame study. Cross-sectional survey designs are
preferred over other designs because they save time and cost which may be incurred in repeated
data collections. And since it collects data from several cases at the same time, it ensures that
variety of views are over the same issue are captured in a short time.

Longitudinal Survey
o Longitudinal (also called developmental) survey collects data from the same target population
at different points in time in order to study changes over time. Data is collected over time and
at specified points in time. The cases in a longitudinal study are followed over a long period
to track changes on the some issue of interest through collecting same data from the same
cases (or case) over extended period of time.
o The major problem with longitudinal surveys is that it takes a long time to collect data. Since
the same data is collected from the same case(s) over an extended time, there is a danger of
losing some cases, or the case(s) may become fatigued due to repeated treatment or
observation. Nevertheless, it is a very effective way to study changes and trends of behavior
over time.

Panel Longitudinal Survey

In a panel longitudinal survey, a sample is selected from a target population at the beginning of a
study, and the same sample is surveyed at different times in the course of the study. The same
sample is used every time data is collected.

Cohort Longitudinal Survey

Cohorts are longitudinal survey designs that study a specific target population over time through
sampling. A population of study is defined and data is collected from fresh samples of the same
size drawn from the defined population which does not change through time. A population
Research Seminar
Oso Willis Yuko
46
remains the same if members who may have joined it after the first sample was drawn are purposely
excluded from the study.

Trend Longitudinal Survey

In a trend survey design, data is collected over time and at specified points in time by drawing new
samples at each observation point from a target population which keeps changing. The target
population keeps changing as new members join and other leave.

Table 7.20
Differences between the Longitudinal Survey Designs
Design Main Features Population Sample Weaknesses Strengths
Panel Studies the same Remains the same. Remains the same. Easily affected by Easy to note
sample over time. attrition. Puts changes and offer
strain on the explanations.
researcher to trace
particular cases.
Cohort Studies different Remains the same. Changes. May not generate Easy to recognize
sample of the same new information. stability and
size from the same changes in
target population. behavior.
Trend Studies different Changes. Changes. It is difficult to Keeps tracks of
samples of the attain saturation. changes in the
same size from a target population.
dynamic
population.

Correlational Research Designs

Correlational studies are typical associational studies which determine whether, and to what extent
an association exists between two or more paired and quantifiable variables. Correlational studies
establish relationships (or lack of them) and use such relationships to make predictions between
two or more characteristics from the same group. In its simplest form, a correlation research design
involves collecting data on two or more quantifiable variables from the same group and then
computing a correlation coefficient (Ґ). Correlational studies describe the (i) direction, (ii) nature
and (iii) strength of an association between two or more variables.

Exploratory Research Designs

An exploratory research is more of a pilot study, and is commonly conducted when a researcher
has little knowledge of the variables of interest. It is more of a preliminary research effort in
planning a further rigorous research. In exploratory research, a researcher explores the field to
establish which variables are involved. The researcher freely follows through any new leads and
moves the study to a new area. Exploratory studies employ the following techniques either singly
or in a combination.

Research Seminar
Oso Willis Yuko
47
o Literature review. This is the secondary analysis of information already available, to focus
the structure, process and relationship of the research already done.
o Expert survey interviews with experts in the field of research area.
o Case study technique as a method of exploration.

Qualitative Research Designs

Qualitative research designs are ideal when:
i. The concept being investigated is ‘immature’ due to conspicuous lack of relevant theories
and previous research.
ii. The available theories are inaccurate, inappropriate, incorrect or biased.
iii. There is need to explore and describe the phenomenon and develop a theory.
iv. The nature of the phenomenon is suited to qualitative measures.

Characteristics of Qualitative Research

Research, whether qualitative or quantitative, is intended to generate valid, reliable and
generalizable knowledge. But qualitative research, unlike quantitative research, focuses more on
the process, meanings, instruments and fieldwork.
i. Qualitative research is concerned more with the process rather than the product. It is
particularly interested in how things occur rather than what exactly occurs, though it does not
ignore what occurs.
ii. Qualitative research derives meaning out of experiences and in context. It is concerned with
how people make sense out of their lives, experiences and the structures of the world.
Qualitative researchers want to know what people think and why they think the way they do.
They attempt to capture the thinking of a people from the people’s perspective as opposed to
merely reporting what the researcher thinks.
iii. The primary instrument of data collection in qualitative research is the researcher rather than
some animate inventory. Qualitative research data is mediated through human experiences
rather than through inventories, questionnaires or machines.
iv. Fieldwork is usually available if not inevitable in qualitative research. A qualitative researcher
physically goes to the people, settings and sites to observe and or record behavior as it occurs
in its natural setting.
v. Qualitative research is inductive and builds abstractions and theories rather than test theories.
Qualitative researchers do not form hypotheses before hand and then try to test them. Rather
they tend to play-it-as it-goes. Unlike quantitative research that puts together puzzles whose
pictures are already known, qualitative research constructs pictures that take shape as they
collect and examine parts.

Research Seminar
Oso Willis Yuko
48
Table 7.22
Differences between Qualitative and Quantitative Research Processes
Quantitative Qualitative
- Problem identification and definition. - Problem identification and definition.
- Development of hypotheses. - Identification of study participants.
- Design of data collection tolls. - Data collection: the researcher is the primary data
collection tool.
- Data collection. - Generation of hypotheses.
- Data analysis and interpretation. - Data analysis and interpretation.
- Conclusion - Conclusion

Triangulation
o Triangulation (or convergence, or integration, or synthesis, or qualitative and quantitative
method, or multi-method, or combined method or mixed methods), refers to combination of
two or more data sources or methods of investigation into one study to converge on a single
construct.
o Triangulation generally entails collecting same data from a variety of sources using a verity of
techniques then comparing the similarities between the data obtained from various sources and
techniques. There are various approaches to triangulation, but all about examine different
perspectives of the same thing to notice the differences and similarities between the
perspectives.

Qualitative Research Designs

Case Study Research Designs
o Case studies are similar to survey studies except they use relatively smaller samples and have
greater depths. Case studies are descriptive studies that provide detailed information about the
characteristics or attributes of interest through intensive and holistic descriptions and analyses
of single entities. They explore single entities or phenomena (the case) bounded by time and
activity through collecting detailed information from a verity of data collection procedures for
a specified period of time. The case is a particular phenomenon (or a bounded system), an
instance of some concern or an intrinsically interesting scenarios that would lead to achieve
full understanding of the phenomenon as much as possible.
o A case is a specified entity (e.g. person, region, situation, etc) taken as an example through
which an issue is to be better understood. A case is a specific entity singled out several others
for detailed investigation or analysis. The entity is selected because:
i. It is unique and seams to offer insight than other would be cases because it is succeeding where
others have failed, or it is failing where others have succeeded, or it is just out of ordinary.
ii. It is an instance of concern.
iii. It is intrinsically interesting.
iv. Selected through a random procedure.

Research Seminar
Oso Willis Yuko
49
Ethnography Research Designs
o Ethnography is the design used to learn about what is actually happening as it is seen and lived
by a people rather than what would happen given some specified conditions. Ethnographers
study intact cultural groups in natural settings during for long period of time by collecting,
primarily, observational data in order to learn about culture of the people who actually live in
that culture.
o Ethnographers learn about a people by letting them describe or portray their lives themselves
in a flexible process that evolves contextually in response to the lived realities encountered in
the field. Ethnographic research is thus an intensive, ongoing, face to face involvement with
participants of the culture being studied in their natural setting and social over a long time. It
is a long time investigation of a group (often a culture).

Historical Research Design

o Historical research makes people aware of what happened in the past and enables them to learn
from past failures and successes. It enables people to learn what happened in the past and see
if the past experiences are still relevant to the present day problems and concerns. Historical
research is also useful as an aid to prediction. If an idea had been tried before, even in different
circumstances, the past results may offer insights into how the present result may turn out.
Historical research facilitates understanding the present from the past.

Phenomenological Research Designs

o Phenomenology research is based on the principle that people are active creators of their world
and that they have the consciousness that communicate to them everyday experiences and
knowledge. The purpose of phenomenological research is to accurately describe the lived
experiences of a people, rather than generation of theories or models of the phenomenon being
investigated.
o Phenomenology does not support the view that knowledge can be created through positivist
research. Rather, it seeks to understand group behaviour from the group’s point of view without
the researcher’s own assumptions and biases by examining human experiences through
detailed descriptions of the people being studied.
o Phenomenology involves studying a small number of subjects through extensive and prolonged
engagements to develop pattern and relationships of meaning.
o Phenomenologists examine phenomena without presuppositions and describe the deep
structures of a phenomenon based on observed internal themes.

Grounded Theory Research Designs

o Grounded theory is a research design which rather than prove theories, generates theories that
are grounded in the empirical data from which they are generated. In grounded theory, a theory
is developed from data rather than use data to prove a theory. It derives theories from the
immediate collected data rather than from prior research theories.
o In grounded theory designs;
Research Seminar
Oso Willis Yuko
50
 i. The researcher enters the field without prescriptions, strict designs, tale games, sample
determination, firm scripts, hypotheses or knowledge about the research object and without
theoretical definitions.
ii. The researcher carries out case studies and compares them with other studies to establish a
pattern or trend of events that could be similar, in order to generate tentative or provisional
hypotheses.

Research Seminar
Oso Willis Yuko
51
 ETHICAL ISSUES IN RESEARCH
Decision to undertake a research must rest upon a considered judgment about how best to
contribute to educational science and human welfare. Researchersare concerned not just with the
benefits of the research findings, but also with the rights of the subjects or participants in the
research process. Because research helps people to improve their lives and themselves – it remains
an integral part of human endeavor. However its value cannot be stretched beyond the dignity of
human society. The balance between search for knowledge and respect for the researched
community of people (or things) is what is called ethics.

Major Ethical Issues in Research

Informed Consent in Research
o Informed consent, as an ethical principle in research, demands that research subjects and
participants be fully informed of all aspects of the research that might reasonably be expected
to influence their willingness to participate in a study in which they have been asked to
participate; and that the researcher fully explains all aspects of the research about which the
participants may inquire. In the end, people should accept to participate in a study based on the
information provided.
o Prospective participants should be provided with information on:
i. The purpose of the research so that they can understand the nature of the study its likely
impact on them.
ii. The expected duration of participation in the study.
iii. The procedure to be followed in the study.
iv. Any unforeseen risks or discomforts to the participants that may arise from the study so that
participants can realistically expect what to anticipate in the study.
v. Any benefits to the subjects or participants that may accrue from the study.
vi. Alternative procedures or treatments, if any, that are available to the participants.
vii. The extent of privacy and confidentiality.
viii. That they can participate voluntarily and withdraw at any time.

Privacy and Confidentiality in Research

Privacy
Privacy refers to persons and defines the rights of the participants to keep from the public certain
information about themselves. Privacy is maintained when people can control who has information
about them or who may intrude in their lives. Privacy, as an ethical principle in research is the
requirement that participants’ territories be respected and entry to participants’ territories can only
be their permission. A territory can be as a small as an individual, or as large as an institution.

Research Seminar
Oso Willis Yuko
52
Confidentiality
Confidentiality has to do with data (or information) rather than with the participants. It is an
agreement between individuals that limits others access to privileged information, by defining who
will have access to data. Confidentiality as ethical principle in research is the requirement that
information provided by a participant, or that data collected from a study unit is not passed to a
third party unless with their express permission.

Anonymity in Research
Researchers are generally concerned with information provided rather that person who provided it
(though the source of information is also important). Researchers accept information without
bothering so much about personal details, or insisting the respondents provide their personal details
as well. In other words, respondents can provide information without giving personal details about
themselves, unless they wish tom provide such details. This is the idea of anonymity.Anonymity,
as ethical principle in research is the requirement that participants in an research undertaking have
a right that their individual identities are not made salient features of the research.

Researcher’s Responsibility in Research

The conduct of a researcher prior to, during and after the research undertaking is a major ethical
issue. Participants in a study have the right to expect that a researcher be sensitive to human dignity
and well meaning in his or her intentions throughout the investigation. Further, researchers are not
expected to allow their personal judgments to prejudice the collection, analysis and interpretation
of data; nor to give into fudging effect by directing results to conform to some preconceived
notions. Their integrity is foremost. Other responsibilities of a researcher include:
✓ Accuracy of information. Results must be based on data.
✓ Giving feedback to be researched.
✓ Admitting shortcomings and not reporting issues that are not based on the findings of the study.
✓ Respect for the researched and the research sites. A research cite should be left as undisturbed
as possible at the end of the study.

Other major issues in each of these stages are summarized in Table 8.1.

Table 8.1
Ethical Issues associated with Different aspects of Research Process
Stage of Research Ethical Concerns
Problem identification The problem must be beneficial to the individual cases being studied.
The problem should not marginalize or dis-empower the people being
investigated.

Research Seminar
Oso Willis Yuko
53
 Stage of Research Ethical Concerns
Purpose The problem must be cleanly specified and understood by the
statement/Research researcher and the participants in the same way. Deception occurs
questions when the researcher and the participants have different purposes in
mind.
Data Collection Participants should not be put at risk. Vulnerable groups and research
cites must also be respected. Informed consent is very crucial at this
stage as well gaining permissions to research cites and cases.
Data Analysis and The major ethical concerns at this stage are anonymity of respondents,
Interpretation keeping the data for some time (5-10 years) after the study, ownership
of the data, and accuracy of information.
Reporting The major concern includes use of unbiased and non-sensitive
language in any form. It is unethical to invent or to bend findings in
any way or for any purpose. This is regarded as a scientific misconduct.

Research Seminar
Oso Willis Yuko
54
 DATA COLLECTION
Classification of Datain Research
✓ Research data can be classified as qualitative and quantitative. Data is quantitative if it is
expressed in terms of amounts or magnitude (i.e. if it is numerical.) Quantitative data describes
how much, or many of the characteristics of interest are present. Qualitative data, on the other
hand, is expressed normatively or descriptively. Quantitative describes which characteristics
of interest are present or why are they present.
✓ Data can further be classified as primary and secondary. Primary data refers to first hand
information obtained from sources that were (or are) direct participants or witnesses of the
experiences being investigated. A researcher who interviews students who took part in a riot
about the causes of the riot collects primary data. A researcher who interviews the persons who
witnessed a riot about the nature of the riot also collects primary data.
✓ Secondary data, on the other hand, refers to information obtained from people who were not
direct participants or witnesses of the experiences being investigated, but who may have learnt
about them through other sources. More often than not, secondary sources report what they
have been told rather than what they experienced.

Data Collection Methods

Questionnaire Methods
✓ A questionnaire is a collection of precise pre-formulated written items designed to yield
specific information to meet particular needs for research, and to which research participants
are expected to respond usually in writing. A questionnaire may consist of one or more items
or sets of items depending on the aspects of data to be collected or the category of the
respondents. It can be used to determine the level of knowledge on an issue, opinions, attitudes,
beliefs, ideas, feelings, experiences and perceptions as well as to gather information about the
respondents. A questionnaire is the ideal tool for studying social phenomenon not amenable to
direct observation.

Types of Questionnaires
✓ The format of a questionnaire is determined largely from the types of items used in making it.
A researcher may choose to use close-ended, or open-ended items, or to mix both of these
items in a questionnaire. It is the way the items are framed that determines the structure of the
questionnaire.

Structured Questionnaire
✓ A structured questionnaire is formed solely from close-ended (or forced response) items.
Close-ended or forced response items restrict the respondents in the kind of information to
provide and how the information may be provided. They are constructed in such a way as to
require direct answers in particular prescribed formats.

Research Seminar
Oso Willis Yuko
55
✓ Structured questionnaires enable the researcher to collect a lot of data in a relatively short time.
Structured questionnaire are recommended when a researcher has some idea of potential
answers to the problem or when a researcher wants to compare different groups with regard to
an issue of interest. They also facilitate consistency of responses among respondents, though
they are difficult and time consuming to construct. Close-ended (forced Response) items may
take different forms. These include rank-order, multiple choice and rating scale items.

Unstructured Questionnaires
✓ These are like essay questions in an examination.Unstructured questionnaire are formed solely
from open-ended (free response) items which require participants to react in their own self-
constructed words.
✓ They enable the researcher to obtain information that would otherwise not be forthcoming if
structured questionnaires were used. They also require less effort to construct since the
researcher does not have to think and structure the possible responses. However, the responses
obtained by this method tend to be inconsistent in length and in content across the respondents.

Semi-Structured Questionnaires
✓ These are like examinations with structured and essay questions in the same paper. A semi-
structured questionnaire is formed from a blend of close-ended and free-response items in a
single instrument.
✓ It is used in areas where the researcher seems to have some idea of the possible answers while
in the same study there are areas where the researcher is not so well informed. Semi- structured
questionnaires enable researchers to balance between the quantity and quality of data collected.
They also make data analysis simpler than using entirely unstructured instruments would. On
the other hand, they provide more information than using entirely structured instruments would
yield. This delicate balance between the quality and quantity of information is useful for a
fuller explanation of phenomena under investigation.

The differences between the three types of questionnaires are summarized in Table 9.1.

Table 9.1
Summary of Types of Questionnaires
Questionnaire Meaning Strengths Weaknesses
Formed solely from close- i. Collect a lot of data i. Difficult and time
ended items that restrict the in short time. consuming to
responses to specific words ii. Easy to fill or score. construct.
Structured and format. Include, rating iii. Easy and simple data ii. Produces surface
ranking and multiple choice analysis. information.
items. iii. Does not produce
unexpected

Research Seminar
Oso Willis Yuko
56
Questionnaire Meaning Strengths Weaknesses
information.
Formed from a blend of items i. Collects a lot of i. The open sections
which restrict respondents in detailed information are rarely
words and formats, as well as in a short time. answered
allow respondents to react ii. Produces both surface correctly.
freely in their own self- and detailed
Semi-Structured constructed words. information.
iii. Fairly simple analysis
procedures.
iv. Produce balanced
quantitative and
qualitative data.
Formed solely from free i. Produces detailed i. Complex analysis
response items that allow information. procedures.
respondents react freely using ii. Easy to construct. ii. Responses
Unstructured
their own self-constructed iii. Produces unexpected inconsistent in
words and formats. information. length and
content.

Constructing a Questionnaire
Once a researcher has decided to use a questionnaire, the next task is to construct a uniformly
workable questionnaire. An ideal questionnaire for use in research has the qualities of a good law.
It is clear, unambiguous and uniformly workable, and its design must minimize potential errors for
respondents and coders. And since people’s participation in research is voluntary, a questionnaire
can help to engage their interests, encourage their co-operation and ensure answers that are as close
as possible to the truth. So constructing a questionnaire involves more than listing a few questions.

To develop an effective questionnaire, a researcher should:

1. Specify the information needed.
2. Design an appropriate response format.
3. Pre-test the questionnaire.
4. Design an administration plan.

The Interview Method

✓ An interview is a person-to-person verbal communication in which one person (or groups of
persons) asks the other person, or a group of persons, questions designed to elicit information
or opinions. In can generally be described as purposeful conversations. Interviews are used to
bring a researcher and respondents face to face, and to enable collection of first hand
information over what people think, know, or feel about the issue at hand.
✓ Interviews are used to collect wide range of information, and questions may be asked to obtain
information on:

Research Seminar
Oso Willis Yuko
57
i. Behavior such as what one is doing, or has done in the past. Such questions can elicit
descriptions of experience, behavior or activities that could have been observed but were not.
ii. Opinions or values such as what somebody thinks over an issue or a topic.
iii. Feelings like one’s attitude about an issue. These are directed more towards emotional
responses of a people to their experience.
iv. Knowledge in order to obtain factual information as contrasted to opinion or feelings.
v. Sensory perceptions of what one has heard, seen, tasted, touched or smelt, and
vi. Background/demographic issues such as age, education, religion.

Techniques of Interviews
The way people are asked questions, either individually, or in groups, give rise to the different
techniques of interviews. Interviews can be classified as key informants, survey or group
interviews. There are also protocol analysis and projective techniques, and Delphi technique.

Key Informants Interview

✓ In key informants’ (or key actors) interviews, data is collected from individuals who have
special information or perceptions that would otherwise be unavailable to the researcher if
people were interviewed at random.
✓ Key informants are people who are more informed about the issue at hand than the rest of the
group either because of their security or any other factor that puts them above others. Key
informants can offer more knowledge, communicate better and offer different perspectives
than other members of the study population.
✓ If the interviewees are not randomly selected, then a researcher should, while selecting the
respondents, consider:
i. Authority and Expertise of the interviewees. Always choose only people who are well
knowledgeable in the issue of interest, either because they have studied it in detail, or because
they practice it.
ii. Articulate: Key informants should be able to articulate their experiences and the experiences
of other people they represent, and explain in detail and describe in detail whatever it is that
the researcher is interested in.
iii. Personal Compatibility: It is important that a researcher gets on well with the people he or she
is interviewing. If the researcher and the interviewees are not compatible, they cannot do any
useful business.
iv. Availability and Willingness of the interviewee is important because interviews consume a lot
of time, and need to be kept as brief as possible. A researcher must be careful to select not only
respondents who are available, but who are also willing to respond to the questions.
v. Naiveté respondents are those whose knowledge has not been contaminated by external
influence. They are people who can explain the patterns of their culture the way they know it
from their life experience and not from outside influence or interaction.

Research Seminar
Oso Willis Yuko
58
Survey Interviews
Survey interviews are interviews conducted on individuals or groups that have been selected at
random. Survey interviews can be:
i. Confirmatory. These are structured to verify information from other sources.
ii. Participants construct. These are used to determine how informants structure their social and
physical world.
iii. Projective. These use ambiguous stimuli to elicit subconscious perceptions that cannot be
observed in the natural setting nor solicited through regular interviews.

Group Interviews
✓ These are interviews that address questions to homogenous groups of between 5 - 12 people
who have been assembled for this specific purpose, and who have been selected because they
are well informed on the issue at hand.
✓ Group interview can be focus group or nominal group. In a focus group discussion, decisions
are made by the researcher after synthesis and analysis of the data collected from the group. It
is the researcher who determines what counts as data based on the researcher’s own assessment
and judgment of the information collected.
✓ In a nominal group interviews, it is the group that determines what counts as data through
consensus.
In its ideal form, a focus group discussion is characterized by:
i. People’s involvement. Because of this of this, a focus needs to be small enough for everybody
to have opportunity to share insights, yet large enough for diversity of perceptions.
ii. Conducted in series: multiple groups with similar participants are needed to detect patterns and
trends across groups.
iii. Participants who are reasonably homogenous and unfamiliar with each other.
iv. Proviso for data: focus group pays attention to the perceptions of the respondents. It is not
intended to develop consensus, or arrive at a common plan.

Approaches to Interview
There are several approaches that an interview may take.
1. Informal conversation interviews refer to spontaneous generation of questions in a natural
setting. They resemble casual conversations that portray both the interest of the researcher and
the respondents in turn. They do not adopt any specific type of sequence of questions or any
particular form of questioning. Because the process is so natural, the interviewees may not
realize that they are being interviewed.
2. A general interview guide outlines the set of topics to be explored by the interviewer and the
order in which the topics are to be explored, but the general wordings are however not pre-
determined.

Research Seminar
Oso Willis Yuko
59
3. Standardized open-ended interviews have a pre-determined sequence of events and wording
of the set of questions to be asked to the respondents.
4. In a retrospectiveinterview, a researcher tries to get a respondent to recall and reconstructs
from memory something that happened in the past.

Questions in an Interview
There are several questions that a researcher may ask in an interview, and each question serves
specific purposes. A researcher must be careful at which point to ask which question. The opening
question is the robin question, usually factual, designed to be answered rather quickly to identify
the characteristics of the interviewee. Lead or introductory questions are not directly related to the
objective of the study but they are used to introduce a relevant subject area by preparing the
respondent to give more accurate and valid information. Such questions are usually not critical to
the analysis and are only intended to foster conversation and interaction.
Filter questions, on the other hand, attempt to identify whether a line of questioning would be
appropriate in the light of the respondent’s experience. Contingencyquestions are questions
intended only for those respondents who react to an item in a certain way. What is asked is
contingent on the response to the preceding question(s) in the series. Contingency questions are
intended to simplify respondents’ tasks in answering questions because they are not forced to
answer items that are irrelevant to them. There are also transitoryquestions that provide a bridge
from one topic to another so that there is no confusion about the purpose of the interview. They
help the interviewee to envision a topic in a broader scope. There should always be ending
questions that bring closure to the discussion, and which enables the respondents to reflect back
on previous comments. It may take the form of - all things considered, or summary, or the final
question. All these questions are intended to enable the researcher focus of key questions.

Table 9.4
Summary of Interview Techniques
Technique Main Features Strengths Weaknesses
- Questions emerge from - Interview can be matched - Less systematic and less
immediate context. to individual comprehensive.
Informal - No predetermined circumstances. - Data organization and
questions or topics. - Increases the relevance of analysis is complex.
questions.
- Topics and issues to be - More systematic and - Important and salient
discussed are specified in makes data collection topics may be in
General advance. somewhat systematic. advertently omitted.
Interview - The interviewer decides - Logical gaps in data can - Too much flexibility can
Guide the sequence and wording be anticipated and closed. result in different
of the questions in the responses.
course of the interview.
Standardized - The exact wording and - Increases comparability of - Standardized wording of
Open Ended sequence of questions are responses. questions may limit
Interview predetermined. - Reduces interviewee naturalness and relevance

Research Seminar
Oso Willis Yuko
60
Technique Main Features Strengths Weaknesses
- All interviewees are asked effects and biases when of questions and answers.
the same questions in several interviewees are - Little flexibility in relating
exactly the same order. used. the interview to particular
circumstances.

Observation Method
✓ Observation generally refers to seeing and noticing. It is the art of watching carefully to see
and notice features of interest. Even though people ‘observe’ everyday, all observations are
not always scientific.
✓ Scientific observation is a procedure by which a researcher relies on his or her ability to see
rather than communicate with another person in order to recognize and record what is occurring
in a situation. Scientific observations go beyond the mere use of eyes to see and notice, and
embraces the use of all the senses to perceive and understand experiences of interest.
✓ An observation becomes scientific if:
1. It serves a particular formulated purpose (i.e. it is related to some objectives); of it is
related to certain questions (i.e. it is intended to generate answers.)
2. It is planned methodically (i.e. it is done according to some approved format).
3. It is recorded systematically.
4. It is subjected to checks and balances of reliability and validity.
✓ Scientific observation allows researchers to:
i. See, verify and check informants’ contributions instead of relying on what the
respondents say about themselves.
ii. See what the informants may take for granted.
iii. See what the informants may be unwilling to reveal, because it is independent of a
person’s ability or willingness to respond.
iv. See what the informants may not articulate.
v. Observation makes it possible for researchers to record behavior as it occurs and avoids
reliance on memory.

Types of Observations
There are several ways of classifying observation; observation can be classified as direct or
indirect.
Direct and Indirect Observation
In a direct observation the researcher looks at the behavior (response) of the subjects he or she is
interested in as the behavior occurs. In an indirect observation, the researcher observes the effects
or results of the behavior rather than the behavior itself. The researcher may observe archives
(written records), or physical traces (erosion or accumulation or accretion) after the behavior has
already occurred.

Research Seminar
Oso Willis Yuko
61
Concealed and Unconcealed Observation
In a concealed observation, the researcher is disguised in some way so that the subjects do not
notice his or her presence. The researcher does not influence the behavior of the subjects in any
way. But in an unconcealed observation, the researcher is not disguised in any form and the
observant, if they have feelings, are aware of the researcher’s presence and are affected by it.

Structured and Unstructured Observation

An observation, whether direct or indirect, can be open-ended or close-ended. In open-ended
(unstructured) observation, the researcher attempts to obtain a complete record of everything that
happens within a particular time. In close-ended (structured) observation, the researcher limits his
or her observations to particular specific preselected issues.

❖ Observation, whether direct or indirect, concealed or unconcealed, structured or unstructured,

can be done through human or mechanical technique. In human observation, the researcher
observes the behavior as the main instrument of observation and of data collection. In
mechanical observation, the human observer is replaced with some form of static observing
device such as audio or audio visual recording.

Observation Settings
An observation setting refers to a place where a researcher actuality doers his or her observation.
Observation may take place in a laboratory setting or in a natural setting. In a laboratory setting,
observations are made from carefully arranged and controlled situations in which variables
(attributes) are selected and manipulated by the researcher. The researcher deals with selected
issues of interest in a controlled environment.In a natural setting, the behavior (responses) is
carefully watched as it occurs in a normal (natural) real life situation.

Observer Roles
Non-participant Observation
In non-participant observation, the researcher does not take part in the activity he or she is
observing; but takes a position in relation to the observants such that he or she does not disrupt the
normal functioning and behavior of the group. This is usually achieved through concealment and
there are many ways in which a researcher can conceal himself or herself.

Participant Observation
❖ In a participant observation, the researcher takes part in the activity he or she is observing,
though the researcher’s identity may not be known to the group being observed.
❖ Participant observation can be done through observer participant (OP), participant observer
(PO), and full participant (FP) techniques.
❖ In observer participant, observation role is a primary role and the researcher makes it known
to the observants. The researcher identifies himself or herself and makes no pretence of

Research Seminar
Oso Willis Yuko
62
 actively being a member of the group he or she is observing. The role of participation is
secondary and need not be revealed.
❖ In the participant observer technique,the participation role is primary and the observation is
subordinate.

Problems of Observation
Observer Effect
In almost all observations, the presence of the observer has a considerable effect on the behavior
of the participants and on the outcome of the observation. This is referred to as observer effect.
Firstly because the researcher may arouse curiosity which may result in lack of attention to the
task at hand and produce a behavior different from normal. Secondly, because the behavior of
those being observed may be influenced by the researcher’s purpose. If the observants are aware
of what the researcher is looking for, they may try to portray a behavior that pleases the researcher,
or decide to portray a contrary behavior.

Observer Bias
Observational data, to a great extent, reflect the biases and the viewpoints of the observer rather
than reality. This is referred to as the observer bias. Observer bias is the possibility that certain
characteristics or ideas of the observer may bias what he or she may see. Thus no matter how hard
an observer may try to be impartial, any observational report will posses some degree of bias. It is
also possible that two people can observer the same activity at the same time may differ in details.
This is referred to as observer bias.
❖ Closely related to observer bias is observer expectation. If a researcher is looking for a
particular behavior, then he or she may see or fail to see certain kind of behavior which may
not be how the subjects normally behave.

Document Analysis Method

❖ Virtually anything that has been written down in some form and virtually any object imaginable
is a potential form of information. However such materials can be grouped into documents,
numerical records, oral statements and records and relics.
❖ Documents are written or printed materials that have been produced in some form or another
and include annual reports, artwork, bills, books, cartons, circulars and so on. Broadly,
documents refer to any kind of information that exist in some type of written or printed form.
❖ Document analysis or content analysis is a method that enables researchers to study human
behavior in an indirect way through the analysis of their communications. As the name
indicates, document analysis is the analysis (or examination) of usually, but not necessarily,
written contents of a communication. Today there a tremendous number of communication of
one sort or another including magazines, editorials, advertisements, articles graffiti, films
which can be analyzed to reveal human behavior. Content of virtually any type of
communication can be analyzed. A person’s or a group’s conscious and unconscious beliefs,
attitudes, values and ideas are often revealed in their written communication.
Research Seminar
Oso Willis Yuko
63
❖ Many things produced by human beings such as pottery and weapons were not originally
intended as communication, but subsequently have been viewed as such. Document analysis
can be employed alone, but it can also be used as a useful adjunct to other types of data
collection methods. It can be used in any context in which a researcher desires a means of
systematizing and (often) quantifying information that is not previously organized to suit the
researcher’s purpose.

Research Seminar
Oso Willis Yuko
64
Table 9.6
Summary of Data Collection Techniques
Technique Meaning Advantages Disadvantages When to Use
The use of all senses to i. Eliminates reliance on memory. i. Researcher can be intrusive. When the researcher wants to:
perceive and understand ii. Allows firsthand experience ii. ‘Private’ information may be i. See for himself or for herself what
experiences of interest to without informants. observed that the researcher cannot participants actually do rather that
the researcher. iii. Information can be recorded as it report. what they say they do.
Observation occurs. iii. Researcher must have good ii. Bridge the gap between what
iv. Used to explore topics that may be attending/observing skills. participants say they do and what
uncomfortable to informants. they exactly do.
v. Unusual aspects can be notified
during observation.
Person to person verbal i. Can produce information that i. Provides ‘indirect’ information Used to:
communication in which cannot be directly observed. filtered through the views of the i. Obtain detailed information on
one person (or a group of ii. Can provide historical information. interviewees. specific issues.
persons) asks the other iii. Gives the researcher ‘control’ over ii. Provides information in a quasi-real ii. Collect information that cannot be
questions intended to the line of questioning. place rather than the natural field directly observed, or that are
Interview
elicit information or setting. difficult to put down in writing.
opinions. iii. Not all respondents are equally
available and authoritative.
iv. Researcher’s presence may bias
response.
Critical examination of Enable the researcher to: i. Entails search of information in When the researcher wants to obtain
public or private i. Obtain the language and words of hard to find places. unobtrusive information at his or her
recorded information the informants. ii. Requires transcribing and scanning pleasure, and without interrupting the
related to an issue of ii. Access data at researcher’s for computer entry. participants.
interest. convenient time. iii. The document may be incomplete
Document
iii. Obtain unobtrusive information. or inaccurate or not authentic.
Analysis
iv. Obtain data that are thoughtful in
that the informants have given
attention to compiling them.
v. To save time and expense in
transcribing.
A collection of items to i. High anonymity of respondents. i. High non-respondent rates. Suitable when:
which a respondent is ii. Collects a lot of data in a short ii. The population must be literate. i. A researcher wants to collect a lot

Questionnaire expected to react, time. iii. No probing of respondents’ of information in a short period of
usually in writing. iii. Uniformity of questions: each answers. time.
participant receives the same set of iv. Details and the circumstances ii. The population is literate.
Educational Research
65
Technique Meaning Advantages Disadvantages When to Use
questions phrased in exactly the under which the items were iii. Population is large and time is
same way. answered are unknown. limited.
iv. Yields more comparable data other v. Questions are fixed and cannot be iv. Information can be easily
methods of data collections. altered. transcribed.
v. Can be standardized if items are vi. Respondents’ motivation is v. A researcher wants to social
highly structured and conditions of difficult to assess. phenomenon not amenable to direct
administration controlled. vii. Unless random samples of returns observation.
are obtained, those returned
completed may represent biased

samples.

Educational Research
66
Educational Research
67