MEANING AND PURPOSE OF RESEARCH

An overview
In the third millennium, the role of research in development will be more important than ever
before. Environmental pollution, the quest for economic development, social and political
upheavals and the scarcity of natural resources have broadened the interest given to research.
Many governments have now recognized the direct link between national development and
research. In this chapter, we discuss the meaning and purpose of research, sources of
knowledge, and the relationships among research, science and theory.

Meaning of research
 To research is to carry out a diligent inquiry or a critical examination of a given
phenomenon. It implies exhaustive study, investigation or experimentation following
some logical sequence.
 Research also involves a critical analysis of existing conclusions or theories with
regard to newly discovered facts. This is very necessary in an ever changing world
with advances in technology continually creating new possibilities. Research,
therefore, means a continued search for new knowledge and understanding of the
world around us.
 Mouly1 defined research as a process of arriving at effective solutions to problems
through systematic collection, analysis and interpretation of data.

Purposes of research
1. The main purpose of research is to discover new knowledge. This involves the
discovery of new facts, their correct interpretation and practical application. Though
there are other sources of knowledge, research remains the most efficient and reliable
source of knowledge. Since research seeks demonstratable truth, it is also the most
accurate system of securing useful knowledge.
2. The second purpose of research is to describe a phenomenon. Accurate identification
of any event involves thorough description, for example, of size, shape, age, weight,
colour, height, change over time, etc. Description then provides knowledge that is the
basis for the other purposes of research stated below.
3. The third purpose of research is to enable prediction. Prediction is the ability to
estimate phenomenon A, given phenomenon B. For example, our knowledge about
the relative movement of the moon, earth and sun helps us predict with amazing
 accuracy the occurrence of lunar eclipses. We sometimes use a set of variables to
predict a given variable.
4. A fourth purpose of research is to enable control. In scientific research, control is
concerned with the ability to regulate the phenomenon under study. Many scientific
experiments are designed to achieve this objective. Usually one phenomenon is
manipulated in order to exert control over another. For example, in a laboratory
experiment, a drug that suppresses body growth is administered to a group of rats to
investigate its effect on weight gain in rats. Another group of rats is fed on normal
diet without the drug. All other conditions are controlled or held constant for both
groups of rats. After some time, the average weight of each group of rats is calculated.
If the average weight of the rats receiving the drug is significantly less than the
average weight of the group of rats, which did not receive the drug, then one can
confidently say that the drug controls weight gain in rats.
5. It should be recognized that the control and predictive functions of science are closely
related. In predictive inquiry, the researcher is interested in studying naturally
occurring associations between phenomena. In control, however, the researcher
manipulates one phenomenon to estimate its impact on another phenomenon. This is
experimental research and leads to more powerful statements about association
compared to predictive inquiry. However, the nature of some phenomena may not
permit experimental research; only predictive inquiry is possible. For example,
sociologists in studying socio-class structures, race, mob behaviour, personality, etc.
cannot manipulate these phenomena and hence must be content with predictive
inquiry. Similarly, research in astronomy largely involves description and prediction.
6. The fifth purpose of research is to enable explanation of phenomena. Explanation
involves accurate observation and measurement of a givenphenomenon. In order to
explain a phenomenon one should be able to describe it, predict its occurrence and
observe factors that cause its occurrence with certainty and accuracy
7. The sixth and last purpose of research is to enable theory development.
Theorydevelopment involves formulating concepts, laws and generalizations about a
given phenomenon. Research is also conducted in an attempt to confirm or validate
existing theories. This is sometimesreferred to as “falsification of theory”.
Sources of knowledge
1. As mentioned earlier, research is an important source of knowledge because it is
objective and involves systematic procedures. It however does not mean that there is
absolute objectivity in research. Pure objectivity in any research study is virtually
impossible. Different researcher are bound to apply research procedures relative to
their; own different realities. There is also an element of error in any research study
particularly because of the complexity and changeability of the phenomena we have
to deal with. Error is defined as anything that causes prediction to vary from what
actually occurs. The research process tries to identify and minimize the effects of
error; we can decrease error but can never eliminate it completely. Sources of error in
the research process include inaccurate observations during data collection, bias and
changes in conditions during the research process.
2. Experience is a common mode of obtaining knowledge. Human beings learn through
their experiences in life. For example, a farmer, after growing beans for several years,
may come to the conclusion that beans do best during the season of short rains. This
conclusion may be based on the fact that yields per acre are higher during the short
rains than during the long rains. However, no two human beings have exactly the
same experiences. For example, two people watching the same movie have different
interpretations and opinions about the movie.
3. Tradition is another source of knowledge. All human beings inherit a culture. Culture
is a reflection of an adopted system of rules, norms, standards and values. The socio-
cultural system embodies accepted knowledge of how things are or should be.
Acquisition of this knowledge for any member of the society is through indoctrination
and socialization.
Transmitters of cultural knowledge in the society include parents, peers, adults,
teachers, and the clergy. Most of us accept these ‘social truths’ without necessarily
testing them. This is the major weakness of tradition as a source of knowledge. If one
member of the society seeks a different understanding from what is traditionally
accepted, that person may become a social outcast.
4. Authority is yet another source of knowledge. This takes the form of an expert in a
specialized area, giving his opinion on a given issue. The acceptance of such an
opinion by others depends on the status of the person giving the opinion. For example,
we are likely to accept the fact that HIV/AIDS is not transmitted through handshakes
if the person stating the fact is a qualified doctor rather than an engineer. There is a
 danger in trusting the judgement of specially trained persons, particularly where there
are contradictory positions. Some dentists, for example, may argue that toothpaste that
contains fluoride is good for our teeth while others may argue to the contrary.
We may also be misled when we depend on the authority of persons speaking outside
their realm of expertise, for example, the popular athlete who states that a certain
brand of soap makes your skin soft and smooth in a television commercial. Authority
then may not be a legitimate source of knowledge, particularly because the experts
may err within their own special areas or overreach their domain with no one to
challenge them.
5. It has also been claimed that intuition is a source of new knowledge. Intuition is the
perception, or explanation or insight into phenomena by instinct. In other words, it is
the ability to gain new knowledge without conscious reasoning or rational process.
However, throughout history, intuition has remained a mystery.

Research and Science

The terms research and science or scientific method are often used interchangeably. In a strict
sense, the two terms have different connotations. Wehave defined research as a critical
examination or inquiry to discover facts.

A dictionary definition of science is “systematic knowledge of natural or physical

phenomena; truth ascertained by observation, experiment, and induction; ordered
arrangement of facts known under classes or heads; theoretical knowledge as distinguished
from practical; knowledge of principles and rules of invention, construction, mechanism, etc.
as distinguished from art”. Thus, research and science are both concerned with the discovery
of new facts. In research, facts are often collected to serve a useful purpose; the focus is on
application. Hence we talk of “applied research”. In science, however, facts are collected
with the sole purpose of testing or developing theory; the focus is on basic research.

Many research studies are not theory-guided and hence are conducted to solve specific
problems. In such studies, generalizable application or development of theory is not of
concern. Most research studies in education fall under this category; i.e. most studies are
done to solve specific problems. For example, studies may be done on new teaching
programmes in mathematics for secondary schools or on effective ways of dealing with
absenteeism in work places or in schools. This type of research is referred to as “action
research.”
In other cases, studies are conducted to test theoretical concepts in real situations in order to
develop generalizable applications. This is usually referred to as “applied research” which
includes research that may not necessarily be theory guided. In education, for example,
attempts are often made to develop generalizations about instructional, motivational or
learning processes.

Research that is theory-guided is termed “basic research”. Basic research or “scientific

inquiry” aims at developing theories and the existence of viable theories facilitates progress
in science and in all other areas of human endeavour.It can thus be argued that the basis of
science is research. Research is a necessary tool in the advancement of science. There could
never be advancement in science without proper identification, description, prediction,
control and explanation of events or phenomena-.

Research and scientific theory

Four purposes of researchto describe, to predict, to control and to explainare based on the
assumption of a single tangible reality that can be observed and measured. The scientific
method further assumes that phenomena are orderly and that their causes are not only
discoverable but can also be manipulated. These assumptions form the basis of scientific
theory.A theory is defined as a system of explaining phenomena by stating constructs and the
laws that interrelate these constructs to each other. A construct is a concept, abstraction or
idea drawn from the specific.

To explain this definition, let us consider Abraham Maslow’s Theory of Motivation.

Maslow’s Theory of Motivation is based on the assumption that human behaviour is goal-
directed and originates from unfulfilled needs. These needs produce tension. This tension
motivates human beings to act; tension is thus, released through certain behaviour or actions.
Maslow conceptualizes a hierarchy of human needs. These are: basic (food, drink, shelter,
etc.); safety (financial, protection, avoidance of pain); social (love and acceptance); self-
esteem (prestige, status, respect); and self -actualization (total development). Maslow’s
theory is a system in that it consists of constructs such as basic needs, social needs, self-
esteem, etc. This theoretical system is designed to explain a specific phenomenon: human
behaviour with respect to needs within a given environment.

Maslow proposed the law that these constructs are related to each other in an invariant
sequence. Each level of need must be fulfilled before the next level manifests itself. For
example, when starvation is the immediate problem, the need for status is non-existent. In
this case, the need that influences one’s behaviour is food. Once this need is fulfilled, the
need for safety will manifest itself. There will then be a different kind of behaviour to fulfill
this need.

Scientific theory serves several purposes. First, theoretical concepts 3 show commonalities in
phenomena that may seem isolated at a glance. We can then make sense of our experience
because theoretical concepts have; identified the universals of these experiences. Secondly,
the laws of a given theory help us to make predictions and to control events. Because of well-
developed theories in meteorology, we can make accurate predictions about f rainfall patterns
in different geographical areas. Thirdly, the concepts and laws of a theory help us to organize
isolated findings from different research studies into an explanatory framework. This helps
researchers maintain consistency in any field of study. When research findings do not fit the
expected general pattern or existing theory, further research in that particular area is
necessary. This requires researchers to collect more data to test the theory. From the results of
this further work, researchers are able to revise previously accepted theories and conclusions.

There is no limit to the amount of knowledge that could be created but we possess only a
portion of this knowledge. In any discipline, therefore the knowledge we possess will always
be cumulative with new knowledge being built on existing knowledge. The existing body of
knowledge theories and understanding are the best starting points for the creation of
knowledge. It is, therefore, necessary for researchers to base their research studies on existing
theories or known principles. A scientific theory explains some logical relationships that exist
among phenomena and research offers the means of observing whether those relationships
actually exist. A properly founded theory should be testable through research.

Review assignments
1. State the meaning and purpose of research.
2. Describe error in the research process and discuss how it arises.
3. State the relationship between:
i. Research and science
ii. Research and theory
4. Take one theory in your field of specialization and break it down into constructs and the
laws that relate these constructs to each other.
Notes
1. Mouly, 1978
2. Alee, 1980
BASIC TERMS IN RESEARCH
Overview
It is important to acquire a sound understanding of the basic terms used in research. A Clear
grasp of the meaning of terms is a prerequisite for effective communication. In this chapter,
terms commonly used in research will be defined with practical examples.

Population
Population refers to an entire group of individuals, events or objects haying a common
observable characteristic. In other words, population is the aggregate of all that conforms to a
given specification. Examples of populations in research might be:

 All standard eight pupils in the country

 All home economics extension workers in the district
 All office blocks over four stories high in urban areas of the country
 All indigenous trees in national forests
 All patients suffering from diabetes in the province.

A researcher first defines the population to which she or he wants to generalize the results.
Ideally, researchers would like to generalize results to the absolute population. This is
referred to as the “Target Population” or Universe.” It is often impractical to select a
representative sample from the target population because it may be difficult to identify
individual members. The number of members of a given population may be large or may be
scattered over a wide geographical area. A lot of time, money and personnel may be required
to locate all the members in order to select a representative sample. Researchers therefore
draw samples from an “accessible population”. This is a more narrowly defined and
manageable population. This however, creates a high likelihood of losing the generalizability
of the results and, therefore, researchers must try to demonstrate that the “accessible
population” is comparable to the target population in the characteristics that appear most
relevant to the study. This is referred to as “population validity” and it is a way of
establishing that the accessible population is in itself representative of the target population.
Hence generalizations to the accessible population can be j applied to the target population
with confidence. Examples of accessible populations might be:
 All standard eight pupils in the district
 All diabetic patients attending clinics in Nairobi hospitals
 Home economics extension workers in the Western Province
  All middle income residential housing units completed within the lastthree years in
Mombasa.

Sample
It should now be obvious that dealing with all the members even of the smaller accessible
population would still involve a tremendous amount of time and resources. Researchers
therefore further select a given number of members or cases from the accessible population.
This subgroup is carefully selected so as to be representative of the whole population with the
relevant characteristics. A sample is therefore a smaller group obtained from the accessible
population. Each member or case in the sample is referred to as a subject. Sometimes, the
terms “respondent” or “interviewees”are used.

The following are examples of samples that may be used in research:

 Two hundred standard eight pupils from a particular district
 One hundred diabetic patients from Nairobi clinics
 500 middle income residential housing units completed within the last three years in
Mombasa.

Sampling
Sampling is the process of selecting a number of individuals fora study in such a way that the
individuals selected represent the large group from which they were selected. The individuals
selected form the sample and the large group from which they were selected is the
population. The purpose of sampling is to secure a representative group which will enable the
researcher to gain information about a population. There are various methods that are used to
get a sample from a population. These methods are referred to as sampling techniques, and
are discussed in Chapter 7.

Variable
A variable is a measurable characteristic that assumes different values among the subjects. It
is therefore a logical way of expressing a particular attribute in a subject. Some variables are
attributes that are expressed quantitatively. For example, age is expressed in years, height is
expressed in metres, area is expressed in square metres and weight is expressed in
kilogrammes. Other variables are expressed in categories. For example, occupation may be
expressed as farmer, teacher, nurse, etc. Gender is expressed as male or female; and colour
may be expressed as green, yellow, white, etc.
There are different classifications of variables. There may be, for example, dependent,
independent, intervening, confounding and antecedent variables. A full discussion of these
types of variables is given in Chapter

Conceptual or theoretical definition of variables

Conceptual or theoretical definition of a variable is a way of specifying precisely what we
mean when we use a particular term to refer to a variable. Researchers cannot effectively
study a variable without first agreeing on a working definition of the variable. For example,
the “financial management” within a household is theoretically defined or conceptualized as
the process involved in administering income within a family, The variable “age” is
conceptualized as the period between the day one was born and the present date. It may be
reckoned in years or in years and months.

Sometimes researchers find it difficult to agree on a working definition of a particular

variable. Usually such terms or variables happen to be a combination of several other
concepts. Take for example the variable socio-economic status. This variable can only be
meaningfully defined by reference to other measurable concepts such as income, education
level, occupation, property owned etc. These are surrogates which are used to conceptualize
the variable socio-economic status. Depending on the type of study, researchers may include
only those surrogates that appear relevant to their studies. For the purposes of scientific
inquiry, the researcher should clearly specify the working definition of the main variables in
the study.

Operational definition of variables

Operational definition refers to the measurement of a variable. It is the description of the
operation that will be used in measuring the variable. Operational definition of a variable may
be stated quantitatively or categorically. For example: operational definition of the variable
“age” can be stated as the number of years or the date of birth. The operational definition of
the variable “type of instruction” may be stated categorically as: lecture, group discussion,
individual tutoring, independent study and so on.

Certain variables can only be operationalized by the use of indicators. Indicators are the
observable evidence of the degree of presence or absence of the variable one is interested in.
For example, the variable “financial management” within a family may be operationalized by
reference, to the | following indicators: budgeting, income and expenditure review, use of
credit cards and saving practices. The presence of most of these indicators within a family
would imply a high level of “financial management”. Examples of other, variables that are
operationalized by use of indicators are: intelligence, work performance, emotional attitude,
the, feeling of satisfaction with something and economic well-being.

Data
Data refers to all the information a researcher gathers for his or her study. Researchers
recognize two types of data: primary data and secondary data Primary data refers to the
information a researcher obtains from the field, i.e. from the subjects in the sample. These
data may be in the form of values obtained from the operational definition of the variables in
the study. These values are usually presented in the form of frequency distributions.
Secondary data refers to the information a researcher obtains from research articles, books,
casual interviews, etc. Data may also be classified as quantitative (numerical) or qualitative
(words, phrases).

Parameter
Like a variable, a parameter is a characteristic that is measurable and can assume different
values in the population. The difference between a variable and a parameter is that a
parameter refers to a population characteristic while a variable is related to a characteristic of
a sample drawn from the population. An example of a parameter is Kenya's per capitaincome
i.e. average income per person in Kenya

Statistics
Statistics is the science of organizing, describing and analyzing quantitative data. The term
statistics is also used to refer to indices which are derived from data through statistical
procedures. Examples of such indices include: means, standard deviation, correlation
coefficient, beta coefficient, etc. An index which is descriptive of a sample is called a
statistic; if the index is descriptive of a population, it is referred to as a parameter. Therefore
the mean of sample is a statistic while the mean of population is parameter. A sample statistic
is assumed to be an unbiased estimate of the population parameter

Descriptive statistics
Descriptive statistics are indices that describe a given sample. Examples of descriptive
statistics are: measures of central tendency (mean, mode, median), measures of dispersion
(range, standard deviation, variance), distributions (percentages, frequencies) and
relationships (correlations).
Inferential statistics
Inferential statistics are a branch of statistics which researchers use to draw inferences about a
given phenomenon in the population. Such inferences are based on the results from a
randomly selected sample. The purpose of inferential statistics is to test hypotheses and
enable the researcher generalize the results from the sample to the population.

Objectives
An objective is any kind of desired end or condition. Objectives in research Studies refer to
specific aspects of the phenomenon under study that the researcher desires to bring out at the
end of the research study. For example, a researcher may want to carry out a study on the
factors that contribute to malnutrition of children in Kenya. The objectives of such a study
would be to:

1. Establish the socioeconomic status of the parents of malnourished children

2. Explore how knowledgeable the parents are about nutrition in children
3. Identify the steps taken by parents to curb malnutrition in their families
4. Identify the common causes of malnutrition in Kenya
5. Explore the sources and availability of information on proper nutrition in Kenya

Literature review
Literature review involves locating, reading and evaluating reports of previous studies,
observations and opinions related to the planned study. It therefore leads to appreciating and
understanding the research that has already been done in one’s area of interest.

Problem statement
A problem statement is a specific statement that clearly conveys the purpose of the research
study. The statement focuses on the phenomenon that the researcher desires to describe,
predict, control, or explain. Some researchers express the problem statement in question
form. The following are examples of problem statements

1. What is the effect of design parameters on the maintenance cost of residential

buildings in Nairobi?
2. The purpose of this study is to establish the human and technical factors that
contribute to road accidents in Kenya.
3. The purpose of this study is to compare the effect of two teaching methods, lecturing
or group discussion, on the academic performance of first year university students.
Units of analysis
The Unit of analysis, also called the unit of statistical analysis, refers to those units that we
initially describe for the purpose of aggregating their characteristics in order to describe some
larger group or abstract phenomenon. Units of analysis are therefore the individual units
about which or whom descriptive or explanatory statements are to be made.

Individual human beings are the most typical units of analysis for social scientific research.
Examples of groups whose members may be units of analysis at the individual level are
students, residents, voters, parents, teachers, etc. Groups may also be units of analysis.
Families, for example might be the unit of analysis in a group. Other units of analysis at the
group level could be married couples, cities, or even geographical regions. Other examples of
units of analysis are organizations such as churches, colleges and supermarkets.

Social artifacts are also sometimes studied as units of analysis. Social artifacts are products of
social beings or their behaviour.Examplesocial artifacts are books, paintings, buildings,
songs, jokes, dances and sculptures.

Units of observation
A unit of observation is the subject, object, item or entity from which we measure the
characteristic or obtain the data required in the research study. This unit can be an individual
person, a house, all pupils in a class, an animal, etc.

In the majority of studies, the unit of observation is also the unit of analysis. For example, a
researcher may want to assess the effectiveness of a new drug in lowering the blood pressure
of hypertension patients. The researcher therefore randomly selects 100 hypertension patients
and administers the drug to them over a period of time. Each patient’s blood pressure is
recorded daily. In this study, each of the patients in the sample is a unit of observation
because the required data is obtained from individual patients. Each patient is also a unit of
analysis although the final results will relate to the whole sample.

However, in some studies, the unit of observation is not necessarily the same as the unit of
analysis. For example, in studies dealing with composite units such as schools, families,
industries, etc., the information or data required is obtained from an individual person but the
results obtained from the analysis of such information relate to the composite unit. The
results do not relate to the individual who provided the information. In studying small scale
industries, the unit of observation is the manager whoprovides the data but the unit of
analysis is the small scale industry. In studies dealing with families as units, the father or
mother (units of observation) may provide the information or data but the results from the
analysis of such data will relate to the family as a whole(unit of analysis).

Hypothesis
A hypothesis is a researcher’s anticipated explanation or opinion regarding the result, of the
study. For example, in a study on productivity of labourwithin manufacturing industries, a
researcher may hypothesize that more monetary incentives would lead to increases in
productivity of labour or provision of recreational facilities would raise productivity.

Theory
A theory is a set of concepts or constructs and the interrelations that are assumed to exist
among those concepts. A theory provides the basis for establishing the hypotheses to be
tested in the study. A good theory can logically be broken down into a set of hypotheses
which can then be verified through experiments or observations. Observations and
experiments hence provide the basis for empirical generalizations.

The process of developing hypotheses from theories and then testing these hypotheses
through observations or experiments is called deductive logic. Inductive logic, on the other
hand, is the process of constructing hypotheses and then theory from repeated observation.
Inductive logic has, however, been criticized as a scientific approach to theory building.

Review assignments
1. Differentiate between a variable and a constant.
2. Select four variables in your area of specialization and give:
a) The conceptual definition
b) The operational definition
3. Differentiate between a parameter and a statistic and a parameter and avariable.
4. State the differences between a sample and a population.
INTRODUCTION SECTION
An overview
The research process starts by formulating a research problem that can be investigated
through research procedures. Students as well as novice researchers often find it extremely
difficult to formulate a research problem. Sometimes it is a painful and laborious process
even for experienced researchers. In this chapter, we discuss the process of identifying a
problem, the characteristics of a good research problem, and stating the purpose, objective or
objectives and hypothesis or hypotheses of a research study.

Identifying a research problem

There is no reason why students or beginning researchers should be frustrated by the process
of selecting a research problem. It is important to bear in mind that there is no limit to the
number of good research problems that can be identified and studied. This is simply because
knowledge in any field is infinite and we process only a portion of it.

The first step in selecting a research problem is to identify the broad area that one is
interested in. Such an area should be related to the professional interests and goals of the
researcher. For example, one may select broad areas such as: low-cost housing, nutrition
status of Kenyan children, productivity of workers in industries, small-scale businesses, etc.

After selecting a broad area, the next step is to identify a specific problem within it that will
form the basis of the research study. This implies that the researcher should narrow down
from the broad area to a specific problem

In selecting specific problem, the researcher should consider the key factors that help in
identifying a researchable problem. First, the research problem should be an important one.
An important research problem is one that should:
 Lead to findings that have widespread implications in a particular area
 Challenge some commonly held truism
 Review the inadequacies of existing laws, views or policies.
 Cover a reasonable scope, e.g., not too narrow or too general.

Certain factors determine the scope of a research study. These include.

 The time available to carry it out
 The money available to carry it out
 The availability of equipment - if needed - to carry it out
  The availability of subjects or the units of study.

There are several ways of identifying a specific research problem from the broad area. These
are:

Existing theories
An existing theory in an area is a good source of a research problem. A theory contains
generalizations and hypothesized principles which can scientifically tested. Such tests can be
done through the research process.

This leads to validation or otherwise of existing generalizations principles. However, theory

is a source of research studies as; long; assuch studies are theory-based. There are many
studies that are not theory-based.

Existing literature
A systematic reading programme in the general area of interest is perhaps the best way of
locating specific research problems. Textbooksandarticles on the broad area give the
researcher a good background of basic information and an insight into various issues that
could be studied.

Discussions with experts

Discussions on general topics either in class or seminars are a source of current problems that
could be researched. Such discussions usually involve experienced and well-informed
researchers in the particular that one may be interested in.

Previous research studies

Previous research studies in the field of interest usually indicate areas of further research. A
review of such studies provides the researcher with researchable projects that would, when
carried out, add to the knowledge that has been built up by other researchers

Replication
Replicating a study involves carrying out a research project that has been done previously. In
this case, the problem and the procedures of the research are identical to a study that has been
previously done. Replicationis usually done to find out whether findings hold over time and
across regions.
The media
Issues which are frequently reported in the media can also form the basis of a research
problem. Such issues are usually discussed in public fora and are therefore important to the
majority of the people

Personal experiences
First-hand observations and reflection on intriguing experiences could be sources of research
problems. Such personal experiences usually lead to vivid images, a hunch or an intuition on
the part of a serious researcher. This helps the researcher to formulate a specific research
problem from a general area.

Stating the problem

A research stud usually starts with a brief introductory section. In this section, the researcher
introduces briefly the general area of study. The researcher then narrows down to the specific
problem to be studied. In general, a good problem statement has the following characteristics:
 It is written clearly and in such a way that the reader’s interest is captured
immediately
 The specific problem identified in the problem statement is objectively researchable
 The scope of the specific research problem is indicated
 The importance of the study in adding new knowledge is stated clearly
 Finally, the problem statement must give the purpose of the research.

Stating the purpose

As mentioned in the problem statement, the researcher introduces the general area and then
narrows down to a specific problem. In defining the specific problem, the researcher conveys
the focus of the research study in one or two sentences. This is referred to as the purpose of
the study. The purpose crystallizes the researcher’s inquiry into a particular area of
knowledge in a given field.

Careful thought must be given to formulating the purpose because it is the centroid of the
research study. If the purpose is accurately expressed, the research process will be carried out
with ease. Where the purpose is poorly formulated and stated, the researcher will not be able
to seek the desired answers.

Novice researchers often make the mistake of expressing the purpose in a very general
manner. Topic phrases such as the following should not be used in stating the purpose:
 a) Using incentives in primary schools.
b) Shortage of low-cost housing in Kenya.
c) Nutrition in children.
d) The lack of creativity in college students.

The above topic phrases would need to be narrowed down to specific purpose statements in
order to facilitate the research process. With sufficient planning, reading and inquiry into a
problem area, the researcher should be able to formulate the purpose of study to meet the
following criteria:
1. The purpose must be indicated clearly, unambiguously and in declarative manner.
2. The purpose should indicate the concepts or variables in the study.
3. Where possible, the relationships among the variables should be stated
4. The purpose should state the target population.
5. The variables and target population given in the purpose should be consistent with the
variables and target population operationalized in the methods’ section of the study.

The following are examples of purpose statements that might appear in j problem statement
section:
1. The purpose of this study is to investigate the resource management skills of women
in small-scale businesses in Kenya.
2. The purpose of this study is to compare the effects of three types of cattle feed on the
productivity of milk in cows.
3. The purpose of this study is to determine the effect of alcohol on reaction time of
heavy truck drivers in Kenya.
4. The purpose of this study is to examine the influence of high school curriculum on the
career choices of women in public and private universities in Kenya.
5. The purpose of this study is to find out the relationship between cattle grazing
practices and land degradation in marginal areas of Kenya.

In stating the purpose of the study, a researcher must choose the right words in order to
convey the focus of the study effectively. The purpose should not merely be implied by the
tone of the introduction, neither should the reader be left guessing which of the sentences in
the introductory section express the purpose. The researcher should also avoid the use of
value-laden, biased or subjective words or sentences in expressing the purpose. Particular
attention should be given to the verbs we use to state the purpose. The verbs given in column
A below reflect bias and subjectivity. Those in column B are neutral and are commonly used
in stating the purpose.

Table 1:Biased and neutral words in expressing the purpose of a study

Column A: Biased Column B: neutral
1. To show 1. To determine
2. To prove 2. To compare
3. To confirm 3. To investigate
4. To verify 4. To differentiate
5. To check 5. To explore
6. To demonstrate 6. To find out
7. To indicate 7. To examine
8. To validate 8. To inquire
9. To explain 9. To establish
10. To illustrate 10. To test

Stating the objectives

In chapter two, research objectives were defined as those specific issues within the scope of
the stated purpose that the researcher want to focus upon and examine in the study.
Objectives help the researcher to keep to the scope of the study by defining the area of
knowledge that the researcher is focusing on. In the example given below, the purpose is
stated and several objectives are formulated from it. It should be noted that the use of neutral
verbs or phrases as discussed under “stating the purpose” also applies to stating the
objectives.

Example
Purpose
The purpose of this study is to explore the influence of resource management skills on the
performance of small scale businesses owned by women in rural areas of Kenya.

Objectives
1. To identify the socio-economic status of small-scale businesswomen in rural areas of
Kenya.
2. To investigate the saving practices of women in rural areas ofKenya
3. To investigate the financial management practices of women in ruralareas of Kenya.
 4. To determine the level of knowledge of management processes among women in rural
areas in Kenya
5. To investigate the sources of information on the management processes in rural areas
of Kenya.

Some authors suggest that objectives could be stated in question form. For example, objective
number one above could read. What is the socioeconomic status of small-scale
businesswomen in rural areas of Kenya?”
Objectives are important in a study not only because they specifyparticular issues to be
examined, but also because they guide the researcher in formulating testable hypotheses. The
next section deals with hypotheses.

Formulating hypotheses
In chapter two, a hypothesis was defined as a researcher’s prediction regarding the outcome
of the study. A hypothesis states possible differences relationships or causes between two
variables or concepts. Hypotheses are derived from or based on existing theories, previous
research, personal observations or experiences. A study can have one hypothesis or where the
study includes several variables, multiple hypotheses can be formulated. Each hypothesis will
usually express a predicted relationship between one or more variables or concepts.

Some authors have recommended that hypotheses which do not havestrong support from
theory, previous research or professional experience should be avoided. Researchers usually
find it difficult to formulate hypotheses in an area of knowledge that has very little previous
research. In such an area, not much is known that can support a hypothesis Researchers also
find it difficult to formulate hypotheses in certain types of research. For example, where the
research is exploratory, researchers usually omit the hypotheses and the study is guided by
the stated objectives

It is very important for a researcher to give a lot of thought to his or her hypotheses because
the whole study revolves around the stated hypotheses. It is the hypotheses that are put to
empirical tests in a study. The test of a hypothesis involves collection and analysis of data
that may either support or fail to support the hypothesis. It is to be emphasized that testing a
hypothesis does not ‘prove’ or, ‘disapprove’ the hypothesis: The collected data is analyzed to
determine whether the hypothesized relationships exist. If the results fail to support a stated
hypothesis it does not mean that the study has failed. Such a situation actually implies that
existing theories or principles need to be revised or retested under various situations.
Purpose of hypotheses
1. Hypotheses provide direction. They bridge the gap between the problem and the
evidence needed for its solution.
2. Hypotheses ensure collection of the evidence necessary to answer the question posed
in the statement of the problem
3. Hypotheses enable the investigator to assess the information he or she has collected
from the standpoint of both relevance and organizations
4. Hypotheses sensitize the investigator to certain aspects of the situation that are
relevant regarding the problem at hand. Researchers should however, guard against
being led to acceptance of false data through sensitization.
5. Hypotheses permit the researcher to understand the problem with greater clarity and
use the data collected to find solutions to problems
6. Hypotheses guide the collection of data and provide the structure for their meaningful
interpretation in relation to the problem under investigation.
7. Hypotheses form the framework for the ultimate conclusions as solutions.
Researchers usually base their conclusions on the results of the tests of their
hypotheses.

The characteristics of good hypotheses

A sound review of literature or of existing theories often leads to good hypotheses. In
formulating good hypotheses, a researcher should generate as many ideas as possible and
then examine each statement critically before stating his or her hypotheses. The exercise of
formulating hypotheses is carried out after the review of literature but prior to data collection.
Formulating hypotheses after collecting and examining the data is a bad practice and should
be avoided. Good hypotheses should have the following characteristics:

1. They must state clearly and briefly the expected relationship between variables
2. They must be based on a sound rationale derived from theory, or previous research, or
professional experience.
3. They must be consistent with common sense or generally accept truths.
4. They must be testable. Data can be collected to support or fail to support hypotheses.
This also implies that the variables stated in the hypotheses can be operationalized.
5. They must be related to empirical phenomena. Words like “ought”, “should”, “bad”
reflect moral judgement and should be avoided.
 6. They should be testable within a reasonable time. For example, the hypothesis that
children who are breast fed for longer periods will have a longer life expectancy in
adulthood would take more than fifty years to test.
7. Variables stated in the hypotheses must be consistent with the purpose statement,
objectives and the operationalized variables in the method section.
8. A good hypothesis must be as simple and as concise as the complexity of the concepts
involved allows.
9. A good hypothesis must be stated in such a way that its implications can be deduced
in the form of empirical operations with respect to which relationship can be validated
or refuted.

The following are examples of hypotheses that meet the above criteria:
1. High alcohol content in the blood influences reaction time among drivers in Kenya
2. High mathematics anxiety influences students’ performance in statistics quizzes at
Kenyatta University
3. There is a positive relationship between level of education and income among civil
servants in Kenya.
4. The amount of rainfall and type of fertilizer used influence the yield of wheat per acre
in the Rift Valley Province of Kenya.
5. Promotion as part of an incentive programme, increases productivity of workers in
both public and private sectors.

In each hypothesis stated above, a relationship between two or more variables is clearly and
briefly indicated. In each case, a rationale can be derived from theory, research or
professional experience in the particular field. They are also consistent with common sense.
Because the variables in case, such a test would involve data collection and analysis, an
exercise that would not take an unreasonably long time.

Researchers should avoid the use of value-laden, biased or subjective hypotheses. For
example, the following hypotheses are biased and value-laden:

1. Jogging one mile a day will be shown to be a more effective method of weight control
compared to dieting.
2. The study will show that students from urban primary schools perform better in
national examinations compared to students from rural primary schools.
Types of Hypotheses
Researchers recognize three types of hypotheses: null hypotheses, alternative non-directional
hypotheses and alternative directional hypotheses.

Null Hypotheses
A null hypothesis is sometimes referred to as a statistical hypothesis. A null hypothesis
always states that no real relationship or difference exists: Any relationship between two
variables or difference between groups is merely due to chance or error. The following is an
example of a null hypothesis.

There is no difference in the performance of national examinations between standard eight

students from rural primary schools and standard eight students from urban primary schools
in Kenya.

We can test the above hypothesis by sampling several schools from each region and working
out an average score on the national examination for each region. We can then start by stating
that the two averages will differ. Statistically, this can be represented by the following
symbols:

H0: µ1 =µ2

This null hypothesis states that the population mean of children in rural primary schools is
equal to the population mean of children in urbanschools.

Alternative non-directional hypotheses

An alternative non-directional hypothesis is also referred to as a research hypothesis. This
type of hypothesis states that there is a relationship or differences but the researcher does not
know the nature of such a difference or relationship. An alternative non-directional
hypothesis is suitable where previous research findings are conflicting or where a strong
rationale to support a predicted relationship does not exist. Stating a hypothesis in a non-
directional form is a conservative approach; the researcher avoids commitment to a specific
outcome. Consider the following hypothesis:

1. High alcohol content in the blood affects reaction time among truck drivers in Kenya.
2. There is a difference in the performance of national examination between children
from rural primary schools and children from urban primary schools.
The first hypothesis states that a relationship exists between the variables but the researcher
does not indicate the direction of that relationship i.e. does alcohol reduce or increase reaction
time? The second hypothesis states that a difference between the two groups exists, but we
are not told which group performs better than the other. Using statistical symbols,we can
represent the second hypothesis as follows:

H1: µ1 =µ2

This means that the population mean, on national examinations children in rural primary
schools and the population mean, on national examinations, of children in urban primary
schools are not the same.

Alternative directional hypotheses

An alternative directional hypothesis specifies the nature of the relationship or difference
between variables. This means that a relationship may be stated as being greater than, less
than, increased, decreased, higher than lower than, etc.

Where the researcher is not sure of the form of the relationship, he or she should not use a
directional hypothesis especially where there is a high chance of obtaining an opposite
relationship from the one hypothesized Consider the following examples:

a. High alcohol content in the blood decreases reaction time of truck drivers in Kenya.
b. Irrigation and use of cow dung manure decreases the yield of maize per acre in arid
districts of Kenya.

From previous studies and experiences, it is highly likely that the first hypothesis will be
borne out by the results of a research study. Even from common sense or general observation,
drunk drivers take a longer time to react to dangerous situations compared to sober drivers.
However, it is highly unlikely that the second hypothesis will be supported by any data. In
statistical form, directional hypotheses make use of the signs greater than (>) or less than (<).
The hypothesis that, in national examinations, pupils in urban schools perform better than
pupils in rural schools isrepresented by:

H1: µ1>µ2
or
H1: µ1<µ2
The above alternative hypothesis implies that the population mean in the national
examination of children from urban schools is greater than the population mean, in the same
examination of children from rural schools. Therefore pupils from urban schools perform
better than pupils from rural schools in the national examinations.

It is important for researchers to note that the type of alternative hypothesis stated (directional
or non-directional) determines the type of statistical test applied. In a non-directional
hypothesis, a two tail test is used while in a directional hypothesis, a one-tail test is
applicable. It is easier to obtain statistical significance with a one-tail test than with a two-tail
test.

Assumptions and Limitations

Many researchers include in the introduction section any identified assumptions and
limitations that might affect their studies. An assumption is any fact that a researcher takes to
be true without actually verifying it. For example, in a study, a researcher may want to
investigate the effect of acid rain on different roofing materials. Here the researcher may
assume that the standard of both manufacture and installation of the roofing materials was the
same. If the materials were manufactured and installed using different standards, then the
assumption is questionable. In another study, the researcher may want to investigate the
difference between computer aided teaching of statistics and ordinary lecturing in statistics.
The researcher can assume here that none of the subjects in each group had received
instruction in either computer or statistics. This assumption is questionable if some of the
students graduated from high schools where computers were available. An assumption puts
some boundary around thestudy and provides the reader with vital information, which
influences the way the results of the study are interpreted.

A limitation, on the other hand, is an aspect of a research that may influence the results
negatively but over which the researcher has no control. A common limitation in social
science studies is the scope of the study which sometimes may not allow for generalizations.
Another common limitation is the sample size. Factors which can be controlled by the
researcher should not be included as limitations, unless a good rationale is provided.
Review assignments
1. Differentiate between problem statement and purpose in a research study.
2. State the characteristics of a good:
i. Purpose statement
ii. Hypothesis
3. In your area of specialization, choose a research article or a thesis and| identify the
following:
i. The problem statement
ii. The purpose statement
iii. Objectives
iv. Hypotheses.
4.
THE REVIEW OF LITERATURE
An overview
The review of literature involves the systematic identification, location and analysis of
documents containing information related to the research problem being investigated.
Literature review should be extensive and thorough because it is aimed at obtaining detailed
knowledge of the topic being studied.

The purpose of literature review

1. The main purpose of literature review is to determine what has been done already related
to the research problem being studied. A detailed knowledge of what has been done helps
the researcher to:
a) Avoid unnecessary and unintentional duplication
b) Form the framework within which the research findings are to be interpreted
c) Demonstrate his or her familiarity with the existing body of knowledge. This
increases the reader’s confidence in the researchers’ professional ability.
2. A review of the literature will reveal what strategies, procedures and measuring
instruments have been found useful in investigating the problem in question. This
information helps one to avoid mistakes that have been made by other researchers and
also helps one to benefit from other researchers’ experiences. The information may also
help to clarify how to use certain procedures which one may only have learned in theory.
3. In most cases, literature review will suggest other procedures and approaches. This is
very useful information because a researcher could try out suggested approaches,
especially if they will improve the research study.
4. Another purpose of literature review is to make the researcher familiar with previous
studies and thus facilitate interpretation of the results of the study. For example results
can be discussed in terms of whether they support or contrast previous findings. If there is
contradiction, literature review might provide rationale for the discrepancy.
5. In some cases, a researcher may not have narrowed down to a topic the start of a literature
review. In such cases, the literature review helps the researcher to limit the research
problem and to define it better. The review of literature will give the researcher the
knowledge needed to convert a tentative research problem into a detailed and concise
plan of work.
6. Literature review helps determine new approaches and stimulates new ideas. The
researcher may also be alerted to research possibilities which have been overlooked in the
past.
7. Approaches that have proved to be futile will be revealed through literature review. This
helps in the research because there is no point repeating a certain approach in a study if
that approach has been found to be consistently unproductive or unreliable.
8. In most cases, authors of research articles include specific suggestions and
recommendations for those planning further research. The suggestions are usually found
when reviewing literature and should be considered very carefully.
9. Literature review pulls together, integrates and summarizes what known in an area. A
review analyses and synthesizes different results revealing gaps in information and areas
where major questions remain.

The Scope of Literature Review

Some very common questions especially from graduate students are:
 How wide or narrow should the literature review be? In other words when should one
stop reviewing?
 What kind of literature is related to my topic? In other words, how do 1 know the
relevant literature?

These are appreciable questions and they get answered with experience. There is no formula
for arriving at a decision. However, the following are useful hints towards determining scope.

1. If the area of study has been studied for a long time and therefore thereis a huge body
of literature, one can read only those studies that are reasonably close to one’s
research topic. In these thoroughly explored areas, much greater depth is available and
the researcher can personally cover a narrower topic range to greater depth.
2. In new or little researched areas, where little depth is available aresearcher would
require to review any relevant material in orderto develop a logical framework for the
study and appropriate hypotheses
3. The researcher should avoid the temptation to include all available material.
Excessive material does not mean great research information. A smaller well
organized review is preferable to a review containing studies that are only remotely
related to the problem.
 4. Another indicator in abandon literature review is to constantly reencounter material
already reviewed

Steps in carrying out literature review

1. Be very familiar with the library before beginning the literature review
2. Make a list of key words or phrases to guide your literature search. For example, if the
study deals with family conflict, other phrases that could be used to search the
literature are “family violence” or “abuse”, “family dissolution” etc.
3. With the key words and phrases related to the study, one should go to the source of
literature. Library staff are generally very helpful in offering guidance
4. Summarize the references on cards for easy organization of the literature
5. Once collected, the literature should be analyzed, organized and reported in an orderly
manner. Such organization, analysis and reporting represents the hardest part of
literature review.
6. Make an outline of the main topics or themes in order of presentation. Decide on the
number of headlines and sub-headlines required, depending on how detailed the
review is.
7. Analyze each reference in terms of the outline made and establish where it will be
most relevant.
8. Studies contrary to received wisdom should not be ignored when reviewing literature.
Such studies should be analyzed and possible explanation for the differences given.
They should be analyzed with a view to accounting for differences of opinion.
9. The literature should be organized in such a way that the more general is covered first
before the researcher narrows down to that which is | more specific to the research
problem. Organizing the literature in this way leads to testable hypotheses.
10. Some researchers prefer to have a brief summary of the literature and: its
implications. This is however, optional depending on the length of the literature under
review.

Sources of literature
Sources of information can be classified into two broad categories:
1. Primary sources: A primary source is a direct description of any occurrence by an
individual who actually observed or witnessed the occurrence. In research, this is the
description of a research by a person who actually carried the research. The review of
 literature should be based on primary sources as much as possible because information
from secondary sources may be altered by the writers.
2. Secondary sources: These include any publication written by an author who was not a
direct observer or participant in the events described.

Examples of sources of information

1. Scholarly journals
These are crucial research documents. They are to be found in libraries. Recent ones will be
paper copies while older journals are retained microfilms. Today, owing to access to the
internet, many journalsthe web. Properly referenced journal articles will have the author’s
name, year of publication, title of the article, title of the journal and volume number to enable
one to locate it easily.

2. Theses end dissertations

All graduate students who wish to receive a Master’s degree or Doctor of Philosophy (Ph.D)
undertake original work which they write up as a thesis or dissertations. Such manuscripts
could prove of great value to the researcher

3. Government documents
These include policy papers, and research reports owned by governments, some of which are
normally sponsored by international agencies. In Kenya examples are the “Master Plan of
Education and Planning” (MPEP), “Kenya Demographic and Health Survey” (KDHS), and
the “Situation Analysis of Children and Women in Kenya”.

4. Papers presented at conferences

Research papers presented at conferences are also good sources of literature. Eventually such
papers get published in conference proceedings or referred journals. However, a reader is
able to make reference to them as presentations even before they are published.

5. Books
Subject bibliographies are available in most libraries. These give a list of books in general. If
one is interested in a particular book, then the catalogue which is alphabetically indexed by
author, subject and title can be checked to see if the book is available in the library.

6. References quoted in books

The references given at the back of a relevant book may reveal more relevant source e.g.
books or journals.
 7. International indices
These are important sources which list these and dissertations which have been written in a
particular area of specialization, for example in the USA, there is “Dissertation Abstracts
International” which lists dissertations with their authors, titles and universities.

8. Abstracts
Abstracts give a list of journal articles with summaries. Abstracts will give the name, volume
and issue number of the journal where the full article can be found. Examples of such
abstracts are:
 Nutritional abstracts
 Home economics abstracts
 Biological abstracts

9. Periodicals
As their name suggests these include journals, magazines or local newspapers which are
published periodically. In some libraries for example Moi Library at Kenyatta University,
there is periodical holdings list which lists all the available journals in the library classified
by subject. A very useful source of information.

10. The Africana section of the library

This is a special area in a library (libraries in Africa) where any material on Kenya by
Kenyans is kept. However, unique materials which am irreplaceable or which are very scarce
may be kept in Africana for safety. In most libraries in Kenya, all theses and dissertations by
Kenyans are kept in the Africana.

11. Reference section of the library

This is another special area in libraries where all books classified “reference books” are
catalogued in the same way as books on open shelves only that the inscription “REF’ is
added. Such books include encyclopedias, dictionaries, manuals, handbooks and biographies
w considered either rare or extremely expensive.

12. Grey literature

Grey literature refers to anything written but not published. Examples are notes from lectures,
papers presented at conferences or any other useful information collected by the library. Very
many researchers overlook this very important source of information. It is wise to always ask
the librarian for the grey literature available in one’s area of interest.
 13. Inter-library loan
An important source of information which is often overlooked. In Kenya, if a certain book is
needed and it is not in a particular library, staff can borrow that book from another library in
Kenya. The inter-library loan system also exists between public library university libraries
and research institutes in Kenya.

14. The British Lending Library

The British Council in Kenya has a library that is open to members of the public. Members
can borrow books from the library. If one requires a book which is not available in Kenya,
the librarian can use a coupon to get the book from British lending libraries. Articles from
unavailable journals can also be acquired.

15. Computer search

There are computer facilities to get required information through a computer search. In such
cases, all reading materials are computerized which makes the process of literature review
easier.If a computer is being used to search the information, key words are used. For example
in a study on ‘family conflicts’ the key words would be ‘family’ and conflict’. Related words
can also be used e.g. ‘violence’. Once e key words have been keyed into the computer, it
shows all the articles or books in the library which have titles bearing on that key word. The
computer also shows the call number of the article or book. One could also search by keying
in the subject of study. The computer will show the titles of books and articles dealing with
that subject some libraries in the Kenyan universities are already computerized.

16. Microfilm
This is where information, especially old information is stored on microfilm and retrieved
only when needed. There are microfilm machines which one can use to locate the article
needed. The article is then printed from the microfilm by special machines so that one can get
copies of microfilmed materials.

The purpose of storing old materials especially journals and magazines in microfilms is to
reduce the storage space. Such old materials are also not very frequently sued, yet retrieval is
easy to when the materials are needed.

17. The Internet

With the access to internet, it is very easy to browse the web and get literature from the latest
journal articles, conference papers and policy documents from various organizations. One is
also able to off load literature from a digital library via satellite. In most universities in
developed countries, the internet is accessible to students. In Africa, selected universities
including Kenyatta University (Kenya), Makerere University (Uganda), have access to a
digital library through the African Virtual University (AVU), a project funded by the Word
Bank. Some scholars are predicting very limited use of libraries in future, especially in
developed countries, owing to advances in information technology.'

Tips on good reviewing of literature

1. Do not conduct a hurried review for fear of overlooking important;; studies.
2. Do not rely too heavily on secondary sources.
3. Many people concentrate only on findings from journals when| reviewing literature.
One should also read about the methodology used and the measurement of variables.
4. It is important to check daily newspapers as they contain very educative current
information.
5. It is extremely important to copy references correctly in the first place so as to avoid
the frustration of trying to retrace a reference later.

Referencing within text

There are two very common methods of referencing within the text, refers to accrediting a
statement or finding to another author, to show that the statement or finding quoted is by that
author. The statement may be paraphrased and therefore need not be in quotation marks.

Method 1
The author’s last name and year of the document’s publication are put after a paraphrased
statement in a text. The name and year are put in brackets.

Example
Among the economic factors that affect satisfaction with quality of life income has been
found to be positively related to satisfaction with q of life (Berry and Williams, 1987).
In the above example, Berry and Williams are the authors, and the *** from which the
statement came was published in 1987.
Sometimes, the author’s name and year come at the beginning of sentence. In such a case,
only the publication year is put in brackets.
Example
Berry and Williams (1987) found a positive relationship between *** and satisfaction with
quality of life.

Method 2
In this method of referencing within the text, numbers representing names of authors are
used.

Example
Communication is an important planning tool that enables a household to realize goals (1,
1987).

In the above example, 1 refers to the number against the particular reference as listed in the
cited references. 1987 is the publishing year. A variation of the above method is when the
number is put after the statement but not in brackets.

Example
Demographic factors have been found to be related to quality of life

Footnotes
Footnotes refer to references which appear at the bottom of a journal paper or a book rather
than at the end of the text. Footnotes serve two purposes:
1. To provide additional information. For example, if the reader needs specific
information which if included would disrupt the continuity of the regular text, details
can be presented in a footnote.
2. To indicate the source of an idea. Footnotes are helpful to readers who wish to refer to
the sources of facts or opinions presented. For this purpose, accurate and complete
information is essential.

Styles of indicating footnotes

There are several styles acceptable for indicating footnotes. The more traditional one is to
number footnotes consecutively throughout a chapter and then put the references on each
page. If it is a second reference to the same work on the same page, the term ‘Ibid’ is used.
This term means ‘in the same place’. When other references have intervened, the term ‘op
cit’ is used meaning ‘in the work cited’. These terms are used following the author’s name.
A variation is to put the numbers indicating references consecutively throughout the chapter,
and then putting the references at the end of the chapter. Some scholars call this footnoting
while others refer to this style of references.

Quotations
A direct quotation may be used for several reasons. The most obvious useis when no change
is made to the text for the sake of accuracy. Also when a statement is unusually clear and
well put, its exact wording may be the most effective way to present a significant thought. A
third reason of using quotation isto add strength to one's work especially when the quotation
is from a recognized authority. Whenever a direct quotation is used, it must be reproduced
exactly from the original publication, and no corrections should be made. If a part of the
original quotation is omitted the break should be indicated by the ellipsis. If anything is added
to a direct quotation, this should be placed in brackets or be rendered in italics. Quotations
must be put in quotation marks and the name of the author, the year of publication and the
page number must be given.
When quotation marks are used, whatever additions that may be added should be outside the
quotation marks.

Example
“From the point of view of FAWE, all girls should have access t education at all levels of
Educational System.” (FAWE, 1998, p. 20).

Cited reference and bibliography

Cited references refer to a list of references which the researcher actually read and cited in the
text. Bibliography refers to the list of materials that were read whether they are cited or not.
Bibliography and references are sometimes used synonymously. A researcher could decide
to: -
1. Include the cited references only.
2. Have cited references and also a bibliography. In this case the cited references are
separate and do not need to be repeated in the bibliography
3. Have the bibliography only which will have all the references those cited and those
that were just read to gather information

There are various ways of writing references and bibliographies. The format will depend on
the university (in case of a graduate student) or of the journal publishing the research work.
The following examples are selected formats of referencing, taken from the American
Psychological Association (APA) manual, one of the most commonly used by social
scientists.

The following examples show how to reference different sources information using the APA
manual.

Examples
1. Journal Article- one author
Moon, M. (1990) Consumer issues and the elderly. Journal of consumer Affairs, 24
235 - 244
2. Journal Article – two authors
Hira, K. Tahira&Mugenda, O.M (1987) Families’ perception of the bankruptcy
process. Family Perspectives, 21, No. 1, May, 1987.
3. Journal Article – more than two authors
Palmore, F.B., Fillenbaum, S.S. & George, L.K. (1984).Consequences of
retirement.Journal of Gerontology, 39. 109 – 116
4. Paper presented at a conference
a. Hira, K. Tahira&Mugenda, O.M (1986). The determinant of money managers
satisfaction with level of living and level of assets. Paper presented at the building
with level of living and level of assets. Paper presented at the building family
strength national symposium, Lincoln, Nebraska, May 1986
b. Mugenda, O. (1999) Redefining and Actualizing the Research Mission in African
Universities. Paper presented at the BOLESWA educational Research symposium,
Maseru, Lesotho, July 1999.
5. Newspaper Article
a. Ng’weno, H.B. (1993, September). Multiply and fill the earth. The weekly
Review, pp 15-17
b. WatoroKamau(August 9th, 1999) New campaign on child labour. The Daily
Nationpp 17
6. A book
Haris, H.I. (1970) Why family estate planning? San Francisco, CA Bancroft-Whitney
Co
 7. Chapter, essay or section of an edited book
Helse, D.R. (1969) Problems in path Analysis and Causal Inference.In E.F. Borgata,
& G.W. Bohnstedt (Eds).Sociological Methodology (pp 38-730. San Francisco:
Jossey Bass.
8. Film or video
Maas, J.B. (Producer), & Gluck, D.H. (Director). 1979. Deeper into hypnosis (film).
Englewood Cliffs, NJ: Prentice Hall.
Reference: Hult Christine (1996) Researching and Writing in the social sciences.
Allyn and Baron, Boston.
9. Televiison Program (Hypothetical)
Njoroge, Arthur. (Producer). (1990, May 21). Career Choices Among Kenyans.
Nairobi: KBS-TV
10. Journal Article from the Web
Herz, J.C. (1995, April). Surfing on the Internat: A nethead’s adventures online.
(Online serial).Urban Desires, 1.3.
Available internet:www/desires.com/ud.html
For more details, the reader should refer to the American Psychological Association
Manual.

Review Assignments
1. Discuss the importance of literature review
2. Explain how one can draw tentative hypotheses from a review of literature
3. Choose a topic of interest and write a three page literature review
4. List four indicators of a good review of literature

Note
1. Neuman, 1994
2. Ibid
3. Ibid
THE METHOD SECTION
An overview
The method section of a research study describes the procedures that have been followed in
conducting the study. At this stage, techniques of obtaining data are developed and data is
actually collected to test hypotheses, if any. The steps involved in conducting the study
should be described in detail. This helps other researchers in understanding one’s study,
particularly where replication may be desired. In this chapter, we discuss populations,
samples, sampling methods, sampling bias, sampling error, sampling fluctuations, the normal
curve, types of variables and measurement scales.

Populations and samples

In Chapter 2, a population was defined as a complete set of individuals, cases or objects with
some common observable characteristics. A particular population has some characteristics
that differentiate it from other populations. For example, all three-bedroom residential houses
in Nairobi, all primary school children, etc. A subset of a particular population is called a
sample. As stated in Chapter 2, it is often impossible to study the whole of the target
population and therefore researchers identify and define an experimentally accessible
population; sometimes referred to as a survey population. The target population was defined
earlier as that population to which a researcher wants to generalize the results of a study. In
defining the accessible population, one should not be influenced by convenience. One should
define the population as consistently as possible with the purpose of the study. Consistency
supercedes convenience in research.

There must be a rationale for defining and identifying the accessible population from the
target population. This rationale must be based on some theory, previous studies or
professional experience. For example, a researcher wants to study the effect of different
teaching methods on the performance of standard eight pupils in mathematics. The target
population here may be defined as all standard eight pupils in the country. The accessible
population may be defined as all standard eight pupils in a particular district. The accessible
population must be the most representative of the target population. The accessible
population and the target population must be comparable on many characteristics, which are
important to the study. This was earlier referred to as population validity. The accessible
population must, therefore, be defined accurately so that others can determine how far one’s
results should be generalized.
Having identified and defined both the target and accessible populations, the next step is to
select a sample from the accessible population. The- sample must be large enough to
represent the salient characteristics of accessible population and hence the target population
assuming, of course that population validity does exist. Novice researchers are often worried
about the sample size. Generally the sample size depends on factors such as the number of
variables in the study, the type of research design, the method of data analysis and the size of
the accessible population. Gay suggests that for correlational research, 30 cases or more are
required; descriptive studies, ten percent of the accessible population is enough and for
experimental studies, at least 30 cases are required per group. A more accurate calculation of
the sample size is described below.

Sample size
Where time and resources allow, a researcher should take as big a sample as possible. With a
large sample, the researcher is confident that if another sample of the same size were to be
selected, findings from the two sample would be similar to a high degree. The danger with
small samples is that they do not reproduce the salient characteristics of the accessible
population to an acceptable degree.

The discrepancy between the sample characteristics and the population characteristics is
referred to as sampling error. Sampling error will be discussed in another section. The smaller
the sample, the bigger the sampling error. There are particular situations when large samples
are required; such cases include:
1. When many variables are held constant in the study.
2. When hypothesized relationships or differences between variables a| very small in real
life. That is, anticipated effect size is small.
3. When the study requires the sample to be broken into sub-groups. For example,
groups of rats under different diets, different income groups, etc.
4. When it is expected that many subjects or cases in the sample will not respond, drop
out or die in the case of animals or plants.
5. When a very high level of statistical significance is required (e.g. a significance level
of .001).
6. When the accessible population is highly heterogeneous on the variable under study
(if it were possible for the population to be wholly homogenous, studying one case
would be enough).
New researchers often have problems determining the sample size required for their studies.
The rule of thumb should be to obtain as big a sample as possible. However, resources and
time tend to be major constraints in deciding on the sample size to use.

In social science research, the following formula can be used to determine the sample size.

2
Z pq
n= 2
d

Where:

n = the desired sample size (if the target population is greater than 10,000).

Z = the standard normal deviate at the required confidence level.

P = the proportion in the target population estimated to have characteristics being

measured.

q = 1-p

d = the level of statistical significance set.

If there is no estimate available of the proportion in the target population assumed to have the
characteristics of interest, 50% should be used as recommended by Fisher et al.

For example, if the proportion of a target population with a certain characteristics is .50, the
z-statistic is 1.96, and we desire accuracy at thelevel, then the sample size is:

2
1.96 (.50 ) (.50)
n=
(.05)2

= 384

If the target population is less than 10,000, the required sample size will be smaller. In such
cases, calculate a final sample estimate (nƒ) using the following formula:
n
nƒ = /N ¿
1+ n ¿

Where:
nƒ = the desired sample size (when the population is less than 10,000)
n = the desired sample size (when the population is more than 10,000)
N = the estimate of the population size

Samples of procedures
After deciding on the sample size, the researcher formulates a procedure of selecting the
subjects or cases to be included in the sample. To select a representative sample, a researcher
must first have a sampling frame, sampling frame is a list, directory or index of cases from
which a sample can be selected. Subjects or cases selected from the sampling frame form the
units of observation in a study. Examples of sampling frames include: national census list,
lists of professionals in a particular field, students registers, lists of income tax payers, lists of
car owners from the Registrar of Motor Vehicles, lists of approved plans from the City
Commission telephone and post office directories, etc. It is important to note that degree of
generalization of a study depends on the accuracy of the sampling frame from which the
sample was selected. If a sample is drawn from an incomplete sampling frame, the findings
from that study may be generalized beyond the sample or the sampling frame from which the
sample was drawn.

In some situations, a sampling frame does not exist or it would require much time and
resources to compile one. In such cases, a researcher could decide on any other method of
selecting a sample provided that such procedure is described in detail in the research report.
At other times, target population is so small that selecting a sample would be meaningless.
Taking the whole population in such cases is advisable. The various techniques of selecting a
sample are described next.

Probability sampling
The goal of probability sampling is to select a reasonable number of subjects objects or cases
that represent the target population. Previous studies have indicated that probability sampling
can provide us with accurate information about groups that are too large to study in their
entity. It provides us with an efficient system of capturing, in a small group, the variations or
heterogeneity that exist in the target population. Random sampling is the key to obtaining a
representative sample. In random sampling, every sample of a given size in the accessible
population has an equal chance of being selected.

Random sampling allows generalizability to a larger population with a margin of error that is
statistically determinable. Random sampling also allows the use of inferential statistics;
statistical indices calculated on the sample can be evaluated to determine the degree to which
they accurately represent the population parameters. In the section that follows we discuss
four common methods of random sampling: simple random sampling systematic random
sampling, stratified random sampling and cluster sampling

Simple Random Sampling

This method of sampling involves giving a number to every subject or member of the
accessible population, placing the numbers in a container and then picking any number at
random. The subjects corresponding to the numbers picked are included in the sample.

Another strategy involves the use of a table of random numbers. Tables of random numbers
are usually included in statistics books. Alternatively, random numbers can be generated by
use of computer programmes. The following is an example of selecting a sample using
random numbers.

A researcher wants to select 500 first year students in our public universities to participate in
a research study. Supposing there are a total of 12,000 first year students in the national
universities:

1. The researcher compiles a list of 12,000 first year students. Each student is therefore
assigned a number from 00001 to 12,000. This list is the population frame for this
study.
2. The researcher decides on the desired sample size eg 500.
3. The researcher, blindly points at any number in the table of random numbers (see
Appendix II). Assume that you close your eyes and point at the number 05919 on the
second row, second column. Then student assigned the number 05919 is selected for
inclusion in the sample.
4. In our example, the researcher should consider all the five digits of each tabled
number as long as that number is within the range 0 through 12000 (i. e. the total
number of subjects in the population frame).
5. After selecting the number 05919, the researcher may then decide to progress through
the table in any way. One can go down or up the column or across to the left or right
along the row. For example, if the researcher decides to progress down the column,
the next two numbers are 84638 and 75277. These numbers should be ignored
because they lie outside the range 0-12,000.
6. The next tabled number is 05055 and since this number lies within our range, the
student assigned the number should be included in the sample.
 7. When the researcher finishes with one column, he or she should move on to the top of
the next column on the right. All cases within the required range are selected until the
sample size of 500 students achieved. If the sample size is very big and one goes
through all columns to the right, it is quite correct to start all over again at the** of the
first column.

If numbers already picked are selected a second time, they should be ignored. It is important
to note that a computer can generate an infinite number of random numbers and hence any
sample size can be selected using a table of random numbers.

Systematic random sampling

In systematic sampling, every Kth case in the population frame is selected for inclusion in the
sample. To obtain a truly random sample using method, the list of all the members in the
sampling frame ** randomized. A list that is arranged in alphabetical order or numerical
order is not random. Cases or objects arranged in numerical order follow a certain pattern, for
example size, colour, etc. and the picking every K th subject may eliminate cases of a certain
size entirely list of individuals is arranged in alphabetical order, a subgroup population may
be concentrated within one section of the alphabet some subgroups have names that often
start with a particularly alphabet. In such a case, a section of the population maybe
alternatively the majority of subjects in a particular sub-group may be included in the sample.
This introduces systematic error in sampling.
Another common example is where a sample of flats or maisonettes is drawn from a
population which is arranged numerically (e.g. 100, 101, 102, etc.). Systematic sampling in
this case may produce a sample of houses that tends to have one feature in common.

Once the population frame is randomized, the researcher then decides on the sampling
interval. A sampling interval is the distance between the cases that are selected for the
sample. The rule of thumb in deciding the sampling interval is to divide the total population
by the sample size. Using the example of first year students discussed under simple random
sampling, the following are the steps one would follow in selecting a sample using systematic
random sampling:
1. List all the first year students in the national universities in random order (e.g. 00001 -
12,000).
2. Determine the sample size, e.g., 500.
 3. Determine the sampling interval by dividing the total population by the size (e.g. 12,
000/500 = 24).
4. Blindly select from the table of random numbers the starting point. Because this
number has to be between 0 and 10, you only have to consider the first or the last digit
of the random number selected from the table. If the selected random number is say
05919, we should start from the student assigned the number 0 or 9 and then pick
every 24th student.
5. This process is continued until the required sample size is achieved. If the list of the
population is exhausted, start again at the top of the list, e.g. the student assigned the
number 00001.

Stratified random sampling

The goal of stratified random sampling is to achieve desired representation from various
subgroups in the population. In stratified random sampling, subjects are selected in such a
way that the existing subgroups in the population are more or less reproduced in the sample.
This means that the sample will consist of two or more sub-groups. It is important to
distinguish stratified sampling from random assignment. In random assignment, a random
sample is first selected by whatever method of sampling. Once a random sample is selected,
subjects within the sample are randomly assigned to two or more subgroups or treatment
groups. In random sampling, the population is divided into two or more groups using a given
criterion and then a given number of cases are randomly, selected from each population
subgroup.

To use stratified random sampling, one must first decide on the criteria under which the
population and hence the sample will be stratified. For example, the criterion or variable used
for stratification may be sex, income level, size, colour, etc. The next step is to decide on the
size of each stratum or subgroup in the sample. Using our first year university students
example, we could stratify the sample by the variable sex. If there are 8,000 male students
and 4,000 female students, we could select our sample in such a way that the ratio of males to
females in the population is reflected in the sample. For example, if a sample of 600 first year
students is desired, we could randomly select 400 male students and 200 female students. The
actual method of sampling from each subgroup of the population can be simple random
sampling or systematic random sampling as discussed above.
However, Selltiz, Weightsaman and Cook4 argue that, for greater accuracy in the findings,
the number in each stratum should be based on the relative variability of the characteristic
under study rather than proportionate to the relative size of each subgroup in the population.
For example, if there are 6000 male students and 6000 female students in our example above,
it would be advisable to sample more female students than male students if experience
suggests that, for the variable under stud| there is more variability among female students as
compared to male students. This argument is based on the fact that the more heterogenous the
population the bigger the sample required to capture the variation within the population. The
more homogenous the population, the smaller than the sample required. In fact, were it
possible to have a completely homogenous population, one subject or case would be enough
to give us the required information regarding the population as stated elsewhere.

The steps involved in stratified random sampling are the same as those involved in simple
random sampling. However, the sampling frame has to be divided into the defined strata.
Stratified random sampling involves the following steps:
1. Identify the population
2. Define the criterion for stratification.
3. List the population according to the defined strata or subgroups.
4. Determine the required sample size and the appropriate representationin each stratum.
This can be proportionate or equal.
5. Select, using random numbers, an appropriate number of subjects foileach stratum.

It is important to note that stratification can be done on more than one variable.
The obvious advantage in stratified random sampling is that it ensures inclusion, in the
sample, of subgroup, which otherwise would be omitted entirely by other sampling methods
because of their small numbers in the population.

Cluster sampling
Cluster sampling is used when it is not possible to obtain a sampling frame because the
population is either very large or scattered over a large geographical area. Cluster sampling
involves selection of an intact group. All the members of such an intact group are then
included in the sample and each member becomes a unit of observation.

For example, a doctor may want to study patients suffering from malaria in Kenya. It would
be time-consuming and expensive for the doctor to try and list all malaria patients who are
hospitalized in Kenya. The doctor may consider it logical to confine himself or herself to
district hospitals. A list of all district hospitals would be compiled and one or two district
hospitals would be randomly selected, depending on the required sample size. All malaria
patients hospitalized in the selected district hospital or hospitals would then be included in the
sample.

It is evident from the above example that in cluster sampling, it is the groups or clusters that
are randomly selected and not the individuals or cases. In this method of sampling, it is
assumed that clusters are similar in characteristics. Examples of clusters would include:
schools, towns, city estates, hospitals, large scale industries, Government ministries, etc.
Cluster sampling involves less time and is convenient. The degree of generalization of the
research findings depends on the degree to which the clusters within the population are
similar. Where the clusters are similar to a very high degree, cluster validity is said to exist.
Cluster sampling involves the following steps:
1. Identify the population.
2. Define the cluster forming the population.
3. Determine the required sample size.
4. List all the clusters in a random order.
5. Using the table of random numbers, select the required number of clusters according
to the sample size required.
6. All members in the selected clusters are included in the sample as units of
observation.

Cluster sampling is sometimes done in two or more stages or cycles. This method is referred
to as multi-stage cluster sampling. For example, a researcher wants to study the housing
conditions of Kenyans in rural areas. Since the population in the rural areas is large and
scattered, one can start by sampling, at random, one province from all the provinces in
Kenya. From the selected province, one can then sample two districts at random by first
listing all the districts in that particular province. From each of these two districts, one can
sample one division at random. From each division two locations can be randomly selected,
giving a total of four locations. One sub-location from each of the four locations is then
randomly selected and all the households in the selected sub-locations are included in the
final sample. The main problem with this technique is that achieving reasonable degree of
population validity between the selected clusters ant the target population is often difficult.
This limits generalizations.
Biased sampling or non-probability sampling
Non-probability sampling is used when a researcher is not interested ii selecting a sample that
is representative of the population. Most qualitative studies use non-probability samples
because the focus is on in-depth information and not making inferences or generalizations.

Purposive sampling
Purposive sampling is a sampling technique that allows a researcher to use cases that have the
required information with respect to the objectives of his or her study. Cases of subjects are
therefore hand-picked because they are informative or they process the required
characteristics. A researcher who proposes to use purposive sampling must specify the
criteria for choosing the particular cases. For example, a certain age range, religious sect, or
educational level. Some researchers’ use purposive sampling as part of multistage sampling
procedure. In such cases, purposive sampling ii applied to get the location or district in which
the units of observation have the required characteristics. Within the selected location or
district random sampling may then be applied to obtain the actual sample of caser Borg et al.
have outlined other methods of non-probability sampling. For example, extreme or deviate
case sampling, which involves selection of special cases or unusual cases? The disadvantage
of this method of sampling is that the results may easily be dismissed for being extreme
although the, may be representative of an extreme group, however small.

When researchers want to select cases that offer in-depth information and are less extreme,
intensity sampling is preferred. This is sampling that involves selection of cases that manifest
the phenomenon of interest intensely but not extremely

Maximum variation versus homogenous sampling

Borg et al. refers to maximum variation as a method of sampling where effort is made to get a
sample containing very varied cases. On the other hand, homogenous sampling is where a
sample of similar cases is selected to enable in-depth study of the group. Maximum variation
sampling enables the researcher to establish where characteristics of a certain phenomenon
cut across the units of observation with maximum variation. Selecting a homogenous sample
allows the researcher to select a smaller sample (since the’ cases are very similar) thus
enabling the researcher to collect more intensive data.

Snowball sampling
In this method, initial subjects with the desired characteristics are identified using purposeful
sampling technique. The few identified subjects name others that they know have the
required characteristics until the researcher gets the number of cases he or she requires. This
method is useful when the population that possesses the characteristics under study is not
well known and there is need to find subjects. For example, if a researcher wants to study
some aspects of Kenya’s nationalist movement, he or she may identify a few people who
participated in the Mau Mau uprising. The few thus identified may assist in identifying others
that they know of.

Quota sampling
This technique is similar to stratified random sampling and the objective is to include various
groups or quotas of the population in the study based on some criteria. For example, a
researcher may want to include a certain religion or social class in the sample and therefore
picks quotas of each. The researcher purposively selects subjects to fit the quotas identified.
The selection of actual participants is not random since subjects are picked as they fit into
identified quotas.

Convenient sampling or accidental sampling

This technique involves selecting cases or units of observation as they become available to
the researcher. It is also referred to as “volunteer sampling” or “accidental sampling”. For
example, a radio programme may ask a few questions to radio listeners and tell them to take
the answers to specified place. Such a sample could be referred to as convenient and biased
because only those people who have radios (and listen to them) will answer. Even among
those, many will ignore the questions. Another example of this sampling method is where a
researcher selects anyone who is convenient, such as a roommate or neighbour. The main
feature of this method is that subjects are easily and conveniently available, and are also
accessible. Researchers who use this method have little or no basis to argue that it is
representative.

Sampling error and sampling fluctuation

Sampling error has already been defined as the discrepancy between t characteristics of the
population and the characteristics of the sample although the sample is drawn from that
particular population.Sampling error is due to errors inherent in random sampling and never
entirely under the control of the researcher. This is simply because a random sample,
however accurately drawn, is never an exact replica the population.

Suppose we want to calculate the mean age of all first year students in our public universities.
We could draw a sample of 500 students a] calculate the sample mean. Supposing the sample
mean was computed as years. If the true mean age of all first year students in our public
universities is actually 19 years, the discrepancy between the true population mean and the
sample mean is one year. If we were to draw many equal samples with replacements and each
time calculates the sample mean, we would get different means. This phenomenon is called
sampling fluctuation and it is again due to sampling error.

As we draw more and more samples of equal size, it would be obvious that the majority of
the sample means would be distributed around population mean, e.g. 19 years. If infinite
samples were to be drawn random and their means calculated and graphed, the graph would
resent a normal distribution. This phenomenon is called the sampling distribution of the
mean. The mean of all such sample means would be equal to the population mean of 19
years. The standard deviation of all such possible sample means is a measure of the sampling
error and is called the standard error of the mean.

Any statistic computed on a sample has a sampling distribution and sampling error. To
minimize the standard error of any statistic, it’s necessary to take as large a sample as
possible. Small samples tend to have large standard errors.

The normal curve

The normal curve is really a family of curves often referred to as normal distributions. A
normal distribution is a mathematical abstraction based on chance or probability. The
mathematics involved in the derivation of a normal curve are complex and are therefore
beyond the scope of this book. However, it is important to discuss the major properties of the
normal curve because the mathematical properties of the normal curve facilitate
generalization of research findings from the sample to the population based on the
assumption of normal distribution.

Simply put, a normal curve is like a frequency polygon, the height of the curve at any point
on the base line indicates the proportion of cases at that particular point. As shown in Figure
1, the normal curve is a bell-shaped, symmetrical curve with the majority of the cases
concentrated in the middle. As we move further from the centre, along the base line or X-
axis, cases become fewer. The further we move, the less the number .of cases on either side
of the centre.

Theoretically, the curve does not touch the base line or X-axis; the curve is therefore said to
be asymptotic to the X-axis. When the units along the X-axis are in standard deviation units,
the curve is referred to as the standard normal curve. If a distribution of measures or cases
forms a standard normal curve, a fixed percentage of the cases or measures will fall between
given standard deviation units above or below the centre as shown in Figure 1.

Figure 1 the standard normal distribution

It has previously been indicated that the sampling distribution of a givenstatistic is in fact a
normal distribution. For example, if we calculate a correlation coefficient between the
variables height and weight among preschoolchildren and if we take a very large number of
equal samples with replacement, the correlation coefficients computed would form a normal
distribution in the population. In social sciences, for example, many complex human
characteristics and behaviour approximate normal distribution.

As discussed earlier, random sampling aims at reproducing in the sample, as accurately as

possible, the characteristics of the population. If the variable under study is normally
distributed in the population, the variable will also be normally distributed in a random
sample drawn from that population. If a variable is not normally distributed in the
population," random sampling does not ensure normal distribution of that variable in’ the
sample. Normality is an assumption often made in the use of many: inferential statistics.
Violation of the assumption of normality often leads to: inaccurate generalizations of the
findings.

With the use of computers, normality of any distribution can quickly be estimated. Deviation
from normality in a distribution is measured in terms'^ of skewness. A coefficient of
skewness can be computed. The smaller the] absolute value of a coefficient of skewness, the
smaller is the deviation from normality of that distribution. In a research study that uses
inferential] statistics to test hypotheses, the assumption of normality may be violated to a
certain extent without serious implications on the accuracy of generalizations of the findings.
Two types of skewness are recognized: a positive skew and a negative skew (Figures 2 and
3). In a positive skew, majority of the measures of i particular variable are low with very few
high measures. In a negative skew, the majority of the measures are very high with very few
low measures. For example, a very difficult test in statistics would produce very few high
scores and many low scores thus forming a positively skewed distribution. A very easy test in
statistics would produce very high scores with very few low scores thus forming a negatively
skewed distribution.

Figure 2 positive skew

Figure 3 negative skew

Another type of deviation from the standard normal curve is described as the Kurtosis.

Kurtosis refers to the ‘peakedness’ or the ‘flatness’ of a curve (see Figures 4 and 5). A curve
that is peeked is said to be leptokurtic. A relatively flat curve is said to be platykurtic. The
kurtosis of (he standard normal curve lies between the two extremes and is referred to as
mesokurtic. For our discussion so far, it is obvious that we can perform certain
transformations on the standard normal curve to produce a family of normal curves. We can
move the standard normal curve along the X- axis or change its spread. In fact, there is an
infinite number of curves that can be created from the standard normal curve. However,
certain characteristics still hold for all the curves so created.

Figure 4 leptokurtic

Figure 5 platykurtic

Classification of variables
The method section of a research study describes in sufficient detail the types of variables in
thestudy.In particular, the procedures used in measuring the variables in the stud should be
well explained. A variable was defined as a measurable characteristic that assumes different
values among the subjects. Obtaining these different values for each subject on a particular
variable was referred to as operational definition of the variable. Measurement of variables of
the operation definition of variables is a very critical step in the research process. Two similar
studies using exactly the same variables may ha; completely different results, depending on
the way each study operationalized the variables. In both the social sciences and the physical
sciences, researchers have, over time, developed standard measures for some variables. For
example, variables like time, volume, length, age, sex, blood pressure, heart beat, IQ,
temperatures, etc. have standard measure. However, there are very many variables in any
field where standard measures have not been developed. For example, there are no agreed
methods of measuring sadness, happiness, religiosity, satisfaction derived from owning a
house or a car, etc. Researchers therefore try to develop scales for measuring these abstract
concepts and such scales differ depending on the emphasis each researcher gives to a certain
concept. Construction of measurement instruments will be discussed in Chapter and
measurement scales will be discussed later in this chapter.

We shall first discuss the various types of variables that one is likely find in a research study.
There are eight common types or classifications variables. These are independent, dependent,
control, extraneous, intervening, antecedent, suppressor and distorter variables. It is important
to note that the types of variables used in a study influence the statistical method of analysis
that can be applied and the type of research design.

Independent variables
An independent variable is a variable that a researcher manipulates in order to determine its
effect or influence on another variable. Independent variables are also called predictor
variables because they predict the amount of variation that occurs in another variable.
Consider the following three examples:
1. The influence of alcohol on reaction time.
2. The influence of income on housing conditions.
3. The influence of gender on career choice.

The variables alcohol, income and gender are all independent variables. In example one, the
amount of alcohol in the blood may increase reaction time, i.e. one takes longer to react to an
emergency while driving. In example two, a high income level means one can afford a much
better house than those at lower income levels.

In example three, males may tend to choose certain types of careers while females choose
other types of careers. Independent variables can be divided into two: those in which the
researcher has manipulative control and those which have already occurred and hence a
researcher has no manipulative control. The former are referred to as experimental type of
independent variables and the latter are referred to as measurement type of independent
variables.
Experimental type of independent variables is common in biological - and physical sciences.
For example, a researcher may want to investigate the influence of various amounts of a
fertilizer on the yield of wheat. The independent variable is ‘amounts of fertilizer’. The
researcher has control over this variable because he or she is the one who decides the
different amounts of the fertilizer to apply to each experimental wheat field. Another example
would be: to determine the influence of alcohol on reaction time. The researcher decides on
the amount of alcohol (independent variable) to administer to each subject.

Measurement types of independent variables have already occurred; they have fixed
manipulative and uninfluenceable properties. Most of these variables are either
environmental or personalogical. Such variables include: age, gender, marital status, race,
colour, geographical location, altitude, soil type, nationality.For example, one may want to
determine the influence of nationality on choice of foods. The independent variable
‘nationality’ has already happened. We are just measuring and grouping the subjects on the
variable ‘nationality’ rather than experimentally manipulating it.

Dependent variables
A dependent variable, sometimes called the criterion variable, attempts to
indicate the total influence arising from the effects of the independent variable. A dependent
variable therefore varies as a function of the independent variable. Consider the following
examples:

1. The influence of hours studied on performance in a statistics test.

2. The influence of distance from the supply centre on cost of building materials.
3. The effect of altitude on daily mean temperatures.

The variables statistics test performance, cost of building materials and daily mean
temperatures are all dependent variables. In example one, the researcher may find that the
dependent variable, ‘performance in statistics test ’, varies as a function of the independent
variable. That is, the more the hours one spends studying for a test, the higher the test score is
likely to be.In example two, costs of building materials are likely to be higher, the further the
construction site is from the supply centre of the materials; and in example three, as altitude
increases, daily mean temperatures decrease holding other factors constant.
Control variables
If a researcher suspects that a certain variable is likely to influence the research results, he or
she should control for that variable in the study. There are several types of variables that may
influence results of a study without the researcher being aware of them. However, serious and
experienced researchers are able to identify variables whose omission from their studies may
influence the results. Previous studies and theory will usually indicate possible variables that
may influence the results if they are not considered in a study. Variables which influence the
results of a study when they are not controlled for are referred to as extraneous variables.
Several types of extraneous variables will be discussed later. These will include four very
special cases of extraneous variables, namely: intervening, antecedent, suppressor and
distorter variables.

When a possible extraneous variable is built into the study, it is referred to as a control
variable. Some researchers refer to control variables as concomitant, covariate or blocking
variables. The introduction of a control variable in a research study increases the validity of
the data and therefore it leads to more convincing generalizations.

However, absolute control of extraneous influence is not possible in any study. It is for this
reason that research results are interpreted on the basis of degrees of confidence rather than
certainty. The first thing in dealing with a possible extraneous variable is to assess the
relevance to your study of such a variable. If previous research, theory or professional
experience indicates that a certain variable may influence your research findings, then you are
well advised to control for it. This eliminates the need for controlling for many irrelevant
extraneous variables. Once the major extraneous variables are identified, the researcher
controls for them in three basic ways:
1. Build the extraneous variable into the study.
2. Hold the variable constant; consider only one level or category of the
variable.
3. Remove the effects of the extraneous variable by statistical procedures.

The first method of accounting for the effects of a possible extraneous variable is to include it
as an independent variable. For example, if previous studies indicate that gender may also
influence reaction timeamong drivers, then, in determining the effect of alcohol on reaction
time, we should add sex as an independent variable in our study. Using a statistical procedure
such as regression, we can then measure the effect of alcohol on reaction time, controlling for
sex

The second method of controlling extraneous variables is to include them in the study but
only at one level. In our example above, reaction time (dependent) is influenced by alcohol
levels (independent) and also by sex (extraneous). We can control for sex differences by
sampling only female or only males. Our subjects will therefore be homogenous on gender. If
age was also a possible extraneous variable, we could control for it by considering subjects
between 25 and 35 years old. The disadvantage with this method of controlling extraneous
variable is that generalizations are limited to a smaller population. That is, we can generalize
only among males or females persons aged 25-35 years.

Finally, we can control the effects of an extraneous variable by statistically syphoning its
effect on the dependent variable. The two main statistical procedures that are used to achieve
this are analysis of covariance and partial correlation. The mathematics involved in these
procedures are beyond the scope of this book.

Extraneous variables
As earlier stated, extraneous variables are those variables that affect the outcome of a
research study either because the researcher is not aware of their existence or, if the
researcher is aware, she or he does not control for them. In other words, an extraneous
variable will influence the results ofstudy unless it is controlled for through an appropriate
procedure as discussed under control variables. The validity and generalizability of research
findings really depend on the degree to which extraneous variables are accounted for in a
study. If extraneous variables are not considered, it is difficult to determine how much
influence on the dependent variable is due to an extraneous variable, and how much is due to
the independent variables. Extraneous variables are sometimes referred to as confounding
variables because they confound the effect of the independent variable on the dependent
variable.

Two major classifications of extraneous variable are recognized by researchers. These are
symmetrical extraneous variables and asymmetric extraneous variables. Symmetrical
extraneous variables are those variables which have a spurious relationship with both the
independent and dependent variables in the study.
In the first case stated above, a relationship between the independent the dependent variables
is found not to exist when the symmetrical extraneous variable is controlled for, but exists
when the symmetrical extraneous variable is not controlled for. This relationship is only
accidental and meaningful at all. For example, supposing a researcher finds that thereis a
higher proportion of one-year-old children in areas consistently inhabited by a certain species
of birds. It would be incorrect to interpret this finding as meaning that one-year-old children
attract a certain species' birds. The relationship between children and birds is purely
accidental fortuitous. The relationship is actually explained by a third variable; run urban
location. Most rural areas may have high birth rates as compared, urban areas. At the same
time, most species of birds are found in rural areas; their natural habitat. It is therefore very
important to examine critically, such an accidental relationship before making erroneous
conclusions.

In asymmetrical relationships, the effect on the dependent variable is d to both the

independent variable and an extraneous variable, which has been accounted for. This means
that a true relationship exists between the independent and the dependent variables and the
extraneous variable only magnifies such true effects. The difficult part would be to establish
how much of the total effect on the dependent variable is due to the independent variable and
how much is due to the extraneous variable that has not been controlled for.

For example, a researcher wants to study the effect of alcohol on reaction time. The sample
on this study includes subjects aged over twenty five years. The researcher finds that reaction
time increases with the level of alcohol in the blood. But suppose a big proportion of the
sample is over fifty five years old. The increase in reaction time may have something to do
with age as much as it has to do with the level of alcohol in the blood. Old people take a
longer period to react to emergencies while driving. Another possible extraneous variable in
this example would be gender. If previous studies indicate a relationship between gender and
reaction time, then age and sex, in addition to the independent variable (level of alcohol in the
blood) will influence reaction time. If sex and age are not accounted for in this study,
interpretations of the finding will be erroneous; the validity of the findings is questionable.
To a certain extent, accurate randomization controls possible influence of extraneous
variables in a study. Randomization ensures that no systematic difference or error on a given
characteristic exists among the subjects. It ensures equivalent representative groups that are
essentially similar on a major characteristic.
Intervening variables
An intervening variable may be considered as a very special case of an extraneous variable.
The distinction between the two is probably a theoretical issue rather than a statistical
one.The difference lies only in the assumed causal relationships among the variables. With an
extraneous variable, there is no causal link between the independent and the ‘dependent
variable. However, each of these two variables is independently associated with a third
variable; the extraneous variable. The logical status of an intervening variable is that it is
recognized as being caused by the independent variable and as being a determinant of the
dependent variable. This causal relationship can be represented schematically as follows;

Independent Variable Intervening Variable Dependent Variable

An intervening variable comes between the independent and dependent variables. When
intervening variables are used as control variables, one must establish a dominant direction of
influence. The independent variable influences the intervening variable and the intervening
variable influence the dependent variable.

Consider the following example borrowed from Zeisel. In a study on factory absenteeism
among female workers, married women were found to have a higher rate of absenteeism
compared to single women. One could therefore conclude that marital status is a determinant
of absenteeism,search for a possible intervening variable in this study would reveal that the
variable ‘amount of housework’ is indeed an intervening variable. One can easily argue that
were it not for the amount of housework, there would not be any difference in the rate of
absenteeism between married and single women. When the intervening variable ‘amount of
housework’ was controlled for, no different e in the rate of absenteeism was found. The
sample in this case comprised both married and single women who have different amounts of
housework.

Therefore, ‘amount of housework’ is a consequence of marital status and a determinant of

absenteeism. The total effect of an independent variable on a dependent variable can be
partitioned into direct and indirect effects. The indirect effects are those effects of an
independent variable, which are transmitted through an intervening variable. Direct effects
are not transmitted through another variable. The choice of the right intervening variables
therefore helps us not only determine accurately the total effect of an independent variable on
the dependent variable, but also partition the total effects into direct and indirect effects.
Antecedent variables
The logic of introducing an antecedent variable is probably the same as that of introducing an
intervening variable in a study. However, an antecedent variable comes before the
independent variable as shown below:

Antecedent Independent Dependent

An antecedent variable does not interfere with the established relationships between an
independent and a dependent variable. Rather, an antecedent variable clarifies the influence
that precedes such a relationship. There are three conditions that must hold for a variable to
be classified as antecedent variable. These are:
1. The variables, including the antecedent variable must be related in some logical
sequence.
2. When the antecedent variable is controlled for, the relationship between the
independent and the dependent variables should not disappear; rather, it should be
enhanced.
3. When the independent variable is controlled for or its influence removed, there should
not be any relationship between the antecedent variable and the dependent variable.

Probing into causal sequences is of great importance in understanding many phenomena. It is

common practice for serious researchers to reflect on which way a relationship exists, say,
between two variables. Many phenomena are subsystems of bigger systems. It is therefore
possible to carry a causal relationship as far back as is theoretically meaningful.

Thus, a researcher might hypothesize that political stability would attract many investors; that
high investments would lead to increased job opportunities; that increased job opportunities
would lead to a high standard of living, which would lead to a reduction on child mortality
rate, etc. The sequence could be endless.

Suppressor variables
Under extraneous variables, it was noted that a false relationship may exist between two
variables due to the intrusion of another uncontrolled variable. That is, one can find a positive
relationship between an independent and a dependent variable simply because they are both
spuriously related to a third variable. When the third extraneous variable is controlled for, the
original relationship between the independent and the dependent variable disappears.
In a sense, a suppressor variable is an extraneous variable, which, when not controlled for,
removes a relationship between the two variables. When a suppressor variable is not
controlled for in a study, it intercedes to cancel out, minimize or conceal a true relationship
between the independent and the dependent variables. In this case, it is the non-relationship
between the independent and the dependent variables that is said to be spurious. When a
suppressor variable is introduced in the study as a control variable, a true relationship
emerges.

Consider the following example: A researcher wants to test the effect of t e amount of alcohol
in the blood on the time it takes a driver to react to an emergency situation (i.e., reaction
time). The researcher hypothesizes t at sex and age are likely to affect the reaction time and
so controls for them by sampling only males between the ages of 25 years and 30 years.

After analyzing the data, the researcher finds that there is no relationship between the two
variables although the hypothesis was that the higher the amount of alcohol in the blood, the
longer it takes a driver to react to an emergency situation. Previous research also seems to
support this hypothesis and therefore the researcher is shocked to find no relationship. Let us
assume that in this study, subjects were given different amounts alcohol and after the alcohol
was absorbed into the blood, subjects were asked to press a button when a flash of light
appeared. Reaction time was measured in seconds; from the moment the flash of light
appeared to moment the subject pressed the button. Let us also assume that a red flash of light
was used in this study. Unknown to the researcher was the fact that the colour of the flash
light has a suppressing effect on the influence of alcohol. That is, a red flash affected the
subjects in such a way that even those who had taken large amounts of alcohol reacted much
faster than expected. As a consequence, the effect of alcohol on reaction time was suppressed
by the colour of the flash of light.

The variable colour of flash light is a suppressor variable in this study because it conceals a
true relationship. If we change the flash light to neutral colour, say white, a true relationship
is found to exist between the independent and the dependent variables.

Distorter variables
A distorter variable is a variable that converts what was thought of as positive relationship
into a negative relationship and vice versa. This effect leads a researcher into drawing
erroneous conclusions from the data. Where a distorter variable is controlled for, a true
relationship is obtained. Consideration of distorter variables in a study reduces the chances of
making a “type one” or “type two error”. Type one and type two error will be discussed in a
later chapter.

Exogenous and endogenous variables

Exogenous and Endogenous variables are commonly used in testing hypothesized causal
models. Path analysis is the statistical procedure often used in testing the validity of causal
relationships in a theory or model. It’s a procedure that tests causal links among several
variables.

In path analysis, one cannot strictly speak of independent and dependent variables. One can
only talk of exogenous and endogenous variables. Consider the causal model in Figure 6.

Figure 6 a hypothesized causal model

In this hypothesized causal model, variables X 3 and X4 are called endogenous variables. Each
endogenous variable is caused or explained by the variables that precede it. For example, the
endogenous variable X4is caused by the variables X1, X2and X3. Variables X1 and X2 are
called exogenous variables and are, in the context of the hypothesized model, left
unexplained. In other words exogenous variables lack hypothesized causes in our model.The
arrows or paths indicate the direction of causation. The curved arrows between any two
exogenous variables represent simple product- moment correlations.

Finally, the quantities labeled U and V are called residuals. In path analysis, residuals provide
us with a way of expressing the idea that numerous other variables enter into the explanations
of each endogenous variable but that such other variables, whether minor or major, are
uncorrelated with the variables in our hypothesized model.

Measurement scales
The operational definition of variables yields information or data which can then be analyzed.
In empirical studies, the data is usually quantitative and hence, can only be analyzed
quantitatively. In studies that yield such data, statistical procedures can be used to analyze the
data. The choice of the statistical procedure to be applied is largely determined by the type of
measurement scale used in the operationalization of the variable. There are four measurement
scales: nominal, ordinal, interval and ratio.

Nominal scale
A nominal scale is considered to be the lowest level of measurement. This type of scale
merely groups subjects or cases from the sample into categories. Subjects or cases in each
category have some common set of characteristics. Variables which can only be measured at
the nominal scale include: sex, race, marital status, employment status, language, roofing
materials, religion, shape of a building, colour etc.

For example, a researcher may include the variable ‘marital status’ in his or her study. The
researcher would then operationalize this variable as married, single, divorced or separated.
The researcher would then find out from her sample, the number of subjects that fall into
each category. This data would indicate the frequencies in each category. For example, if the
sample had 80 subjects, the frequencies might be as follows: 42 married, 23 single; 10
divorced; and 5 separated.

Numbers or numerals are often assigned to various categories for the purpose of
identification. We could assign the numeral 1 to married; numeral 2 to single; numeral 3 to
divorced; and numeral 4 to separated. It is important to note that the numerals assigned to
various categories do not represent quantities or order. Therefore performing mathematical
operations (e.g. addition, subtraction, multiplication, division) on these numerals would yield
meaningless values.

A special case of variables measured at the nominal scale is dichotomous variables. A

dichotomous variable has two categories only. Gender or sex is the best example. In reality,
the variable sex can only take two values male or female. This is referred to as true
dichotomy. Sometimes, variable measured at a higher scale (e.g. interval or ratio) can be
change into a dichotomous variable by establishing some arbitrary cutting point for dividing
the subjects into two groups. This is referred to as artificial dichotomy.For example, a sample
or buildings in downtown Nairobi could be divided into two groups according to the number
of storeys each building has. ‘Tall’ buildings may be defined as those buildings over 15
storeys high and ‘low” buildings would be those less than 15 storeys high.
Ordinal scale
Ordinal Scale is an improvement on the nominal scale. It is important and necessary to use
the ordinal scale when one can detect a deferring degree of the attribute being measured
among the subjects. An ordinal scale not only groups subjects into categories, but it also
ranks them into so order. This could be in an increasing order. Therefore in an ordinal scale
numerals are used to represent relative position or order among the values of the variables.
Examples of variables that can be measured at the ordinal scale include: social class, military
rank, etc.

However, the order indicated in an ordinal scale does not imply quantitative diction. For
example, a researcher may want to measure the hardness’ of some mineral. The hardness may
vary according to the graphical location of the mines from which the mineral was obtained. |
he researcher may define the criterion of measuring the different levels 0f ‘hardness’ and then
assign the numeral 3 to the hardest category; the numeral 2 to the category with medium
hardness; and the numeral 1 to the softest category of mineral.

There is no way of knowing whether the difference in hardness between the group labelled 3
and the group labelled 2 is the same as the difference in hardness between groups 2 and 1. In
other words, although the groups are ranked in increasing order of hardness, the intervals
between the groups are not necessarily equal. An ordinal scale does not give us the level of
precision required in a research study, particularly when strong statistical procedures are to be
applied. Mathematical operations on the numerals would yield meaningless values.

Interval scale
In an interval scale, the numerals assigned to each measure are ranked in order and the
intervals between numerals are equal. Hence, numerals used represent quantities and
mathematical operations on the numerals would yield meaningful values.

However, these mathematical operations are limited to additions and deductions;

multiplication and division are not applicable. Another salient characteristic of the interval
scale is that there is no true zero point. The minimum and maximum points of the scale are
only arbitrary.

For example, a lecturer gives a statistics test to a sample of students. If one student scores 60
points and another student scores 30 points, the student who has a score of 60 has twice as
many points as the student who scores 30. However, we cannot say that a student who scores
80 points knows exactly twice as much as a student who scores 40 points. On the other hand,
a score of 0 would only indicate the lowest level of performance in the statistics test. It would
not imply absolute absence of any knowledge in statistics. An example of interval scale
measures includes temperatures (Fahrenheit), test scores, etc.

Ratio scale
The ratio scale is the highest level of measurement. It is also the most precise method of
measuring variables. A ratio scale would generally have all the characteristics of the other
scales (nominal, ordinal and interval). The only additional characteristics in this type of scale
are that it has a true zero point and all mathematical operations can be applied to yield
Meaningful values.

Most physical objects can be measured at the ratio scale. For example height, weight,
distance, age, area, blood pressure, etc. In ratio scale intervals between numerals are equal
and mathematical operations yield meaningful values. For example, a building that is 60
metres tall is twice as tall as a building that is 30 metres. The difference between 80 kilogram
and 60 kilograms is exactly the same as the difference between kilograms and 40 kilograms.
We can also quite correctly state that zerometres indicate the absence of any height or ‘zero
kilograms’ indicate absence of any weight.

Table 2 summary of characteristics of measurement scales

Scale Characteristics Statistics Applicable
Nominal Subjects classified under a Measure of central tendency: mode
common characteristic
Ordinal Subjects classified in order. Measure of central tendency: mode
Numerals reflect increasing Median
amounts of the attribute but not
at intervals
Interval Numerals reflect increasing Measure of central tendency: mode,
amounts of the attribute with Median, mean
equal intervals. Measures of dispersion: range
A zero point does not exist, it is standard deviation, variance
only arbitrary
Ratio Numerals reflect increasing Measure of central tendency: mode,
amounts of the attribute with Median, mean
equal intervals Measure of dispersion: range standard
A true zero point exists. deviation, variance
When variables are measured at the ratio scale, powerful statistics procedures can be applied.
This improves the quality of findings obtained from such studies and generalizations can be
made to a wider population is important to note that statistical procedures appropriate to a
lower scale of measurement such-as nominal are also appropriate to a higher scale
measurement. However, the reverse is not necessarily true. For example regression analysis
can be used at the ratio scale but cannot be applied nominal measureswith the exception of
true dichotomous variables as gender.

With variables that are measurable at the ratio scale or interval scale distinction is made
between continuous and discreet variables. A variable said to be continuous when its
measures can theoretically occur at any point along a continuum. For example, if one person
is 1.80 metres tall and another person is 1.70 metres tall, theoretically, it is possible to have
many persons whose heights fall between the range 1.70 metres to 1.80 metres. On the other
hand, measurements of discreet variables can only take exact values. A common example
would be the number of persons i household. One household could have 5 persons and
another household could have 6 persons. It is impossible to have a household with a number
of persons less than 6 but more than 5. Measures of discreet variables can only be in whole
numbers, e.g. 0, 1, 2, 3…………………….n.
Review assignments
1. Discuss the following items:
Population validity, accessible population, target population, probability sampling,
sampling error and sampling fluctuation.
2. Discuss the properties of a standard normal curve.
3. Select a research study in your area of specialization and discuss the following:
a) The sample size.
b) Sampling method used.
c) The type of variables in the study.
d) Measurement scales used in the operationalization of these variables.
Notes
1. Gay, 1983
2. Fisher, Laing and Stoeckel, 1983
3. Ibid
4. Cook, 1976
5. Borg et al., 1997
6. Ibid
7. Ibid
8. Zeisel, 1975
DEVELOPING RESEARCH INSTRUMENTS
An overview
A researcher needs to develop instruments with which to collect the necessary information. In
social science research the most commonly used instruments are: questionnaires, interview
schedules, observational forms and standardized tests. This chapter discusses the process of
developing these research instruments.

Questionnaires
Questionnaires are commonly used to obtain important information about the population.
Each item in the questionnaire is developed to address a specific objective, research question
or hypothesis of the study. The ^researcher must also know how information obtained from
each questionnaire item will be analyzed. A questionnaire which is not well thought out has
the following disadvantages:
1. It may confuse respondents as to the nature of the information required
2. It may discourage respondents to the extent of discarding the questionnaire and
3. It may leave out important information required in the study.

Kinds of questions used in questionnaires

There are two broad categories of questions that are used in questionnaire^
1. Structured or closed-ended,
2. Unstructured or open-ended

1. Structured or closed-ended questions

These types of items refer to questions which are accompanied by a list of all possible
alternatives from which respondents select the answer that be describes their situation.In
many cases, it is impossible to exhaust all the categories since the researcher may not know
all possible answers. In such cases, it is customary to include a category called “Other” to
take care of all those responses which may not fit in the given categories.

Example
Question: What type of house do you live in?
1. Bungalow 2. Flat 3.Maisonette 4. Other
Advantages of closed-ended questions
a. Closed-ended questions are easier to analyze since they are in an immediate usable
form.
b. They are easier to administer because each item is followed by alternative answers.
c. They are economical to use in terms of time and money.

Disadvantages of closed-ended questions

a. They are more difficult to construct because categories must be well thought out.
b. Responses are limited, and the respondent is compelled to answer questions according
to the researcher’s choices.

2. Unstructured or open-ended questions

This refers to questions which give the respondent complete freedom of response. These free
response questions permit an individual to respond in his or her own words. The amount of
space provided is however a good indicator of whether a brief or lengthy answer is desired. In
some cases, only a phrase or a figure is needed

Example
Question
How do you keep record of your monthly expenditure in your household
___________________________________________________________________________
___________________________________________________________________________
___________________________________________________________________________

Advantages of open-ended questions

a) They permit a greater depth of response. When a respondent is allowed to give a
personal response, usually reasons for the response given may be directly or indirectly
included.
b) Open-ended questions are also simpler to formulate mainly because the researcher
does not have to labour to come up with appropriate response categories.
c) The respondent’s responses may give an insight into his feelings, background, hidden
motivation, interests and decisions.
d) Open-ended questions can stimulate a person to think about his feelings or motives
and to express what he considers to be most important.
Disadvantages of open-ended questions
a) When the respondent is free to give an individual response deemed to be proper, there
is a tendency to provide information which does not answer the stipulated research
questions or objectives.
b) The responses given may be difficult to categorize and hence difficult to analyze
quantitatively. Whenever information cannot be categorized, it is advisable to include
it in the study in narrative form. Some open-ended responses are very useful for
qualitative analysis.
c) Responding to open ended questions is time consuming. This may put off some
respondents.

3. Contingency questions
In particular cases, certain questions are applicable to certain groups of respondents. In such
cases, follow-up questions are needed to get further information from the relevant sub-group
only. These subsequent questions which are asked after the initial questions are called
“contingency questions” or “filter questions”. The purpose of these kinds of questions is to
probe for more information. They also simplify the respondents’ task in that they will not be
required to answer questions that are not relevant to them. There are various formats used to
formulate contingency questions. Two are given below:

Format One
Please put a tick (√) in the box next to the right response:
Question 1: Have you ever voted in a political election?
1. [ ] Yes 2. [ ] No
If yes, how many times have you voted?
1. [ ] Once 3. [ ] 2-4 times
2. [ ] 5-10 times 4. [ ] Over 10 times

Format Two
Question 2: Have you ever used birth control measures?
[ ] Yes (Please answer questions 3-5)
[ ] No (please skip questions 3-5 and go directly to question 6)
As can be seen from the above two examples, contingency questions can be closed ended as
in format one above, but could also be open ended, where a respondent is required to give
some explanation in a subsequent question
4. Matrix questions
These are questions which share the same set of response categories. T1 are common
whenever scales (e.g. Likert Type Scale) are being used.

Example
Beside each of the statements presented below, please indicate whether are extremely
satisfied, satisfied, dissatisfied or extremely dissatisfied.

Question: How satisfied are you with your

a. Financial status 1 2 3 4 5
b. Quality of life 1 2 3 4 5
c. Level of assets 1 2 3 4 5
d. Ability to save 1 2 3 4 5

Key:
1 = Extremely dissatisfied 2 = Dissatisfied 3 = Neutral
4 = Satisfied 5 = Extremely satisfied
Advantages of matrix questions
1. When questions or items are presented in Matrix form, they are easier to complete and
hence the respondent is unlikely not be put off.
2. Space is used efficiently.
3. It is easy to compare responses given to different items. This easy comparability is
advantageous to both the researcher and the respondent. The researcher can easily
detect a trend just by glancing at the responses.

Disadvantages of matrix questions

1. Because they are easier to construct, many researchers use them when in fact the kind
of information being sought could better be obtained in another format.
2. Some respondents, especially the ones that may not be too keen to give right
responses, might form a pattern of agreeing or disagreeing with statements.

Scales used in questionnaires

There are several scales that are used in questionnaires and interview schedules. The Likert
type scale is the most commonly used rating scale.
Rating scales
These types of scales are used to measure perception, attitude, values and behaviour. Rating
scales consist of numbers and descriptions which are used to rate or rank the subjective and
intangible components in research. The numerical scale helps to minimize the subjectivity
and makes it Possible to use quantitative analysis. The most commonly used rating scale is
the Likert scale. Likert scales are often used with matrix questions. The items that are used in
Likert scales are usually declarative in form. The Likert scale comprises 5-7 response
categories. The numbers in a Likert scale are ordered such that they indicate the presence or
absence of the characteristic being measured. The following is an example of a Likert scale:

Example
Direction: Please circle the number that best describes your feelings about family planning.
The numbers represent the following responses:
1 = Strongly disagree 2 = Disagree 3 = Neutral 4 = Agree 5 = Strongly agree.

Family planning improves the health of both mother and child.

12345

Both parents are responsible for planning the birth of their children
12345

Rules for constructing questionnaires and questionnaire items

1. Before constructing a questionnaire, list the objectives that you want the questionnaire
to accomplish. This will help in writing the items since each item must relate to a
certain objective.
2. Determine how information obtained from each questionnaire item will be analyzed.
3. Clarity is essential. Terms like “several” or “most” have no precise meaning and
should be avoided.
4. If a concept has several meanings and that concept must be used in question, the
intended meaning must be defined. This will avoid misinterpretation of the question.
5. Short questions are preferred to long ones. This is mainly because short items are
easier to understand.
6. Items should be stated positively as much as possible. Negatively stat' items are
usually misread by many respondents. In most cases negative word is overlooked.
7. Double-barrelled items should be avoided. These are items that have two separate
ideas in one question
 8. Leading or biased questions should be avoided. These are questions that suggest that
one answer may be better than the other. Hints as to preferable answers should be
avoided.
9. Very personal and sensitive questions should be avoided. The respondent may
naturally be dishonest in replying to them.
10. When constructing questions, simple words that are easily understandable should be
used. Difficult words or words that are not familiar will discourage the respondent.
11. Questions that assume facts with no evidence should be avoided. Such questions
offend and discourage the respondent. For example: “Have you stopped buying goods
and services on credit?”
12. Avoid psychologically threatening questions.
For example: “Are you worried about the consequences of a divorce?”
13. Include enough information in each item so that it is meaningful to the respondent

Ordering the items in a questionnaire

Items in a questionnaire should be arranged in a logical sequence. One advantage of a well-
organized questionnaire is that respondents can find their way around with ease. Well
organized questionnaires are also easy to code. The following are important tips on how to
organize your questionnaire:
1. Begin with non-threatening, interesting items. If very touchy, sensitive items are put
at the beginning, they will put off the respondent. For example, it may not be a good
idea to start with a question on respondent’s sexual behaviour.
2. It is not advisable to put important questions at the end of a long questionnaire. Such
questions might be overlooked or answered in a hurry.
3. Have some logical sequence when putting items together. For example group together
related items or items that will elicit a similar response.
4. Some researchers arrange their questions according to the themes being studied. This
is another acceptable sequence.
5. If the questionnaire is arranged into content sub-sections, each section should be
introduced with a short statement concerning its content and purpose. The purpose of
such a short introduction is to help the
Ways of administering questionnaires
Questionnaires are mainly administered by using three methods.

1. Self-administered questionnaires
Respondents are asked to complete the questionnaires themselves. In most cases
questionnaires are sent to them through the mail or hand-delivered to them.

2. Researcher administered questionnaires

In some cases, the researcher may decide to use the questionnaire to interview the
respondents. This is mostly done when the subjects may not have the ability to easily
interpret the questions probably because of their educational level. In such cases, the
researcher reads the items and thecategories to the subjects and the responses are written
down.

3. Use of internet
Due to advancement of information technology, a lot of researches especially in developed
countries are able to collect data through theinternet. The people sampled for the research
receive and respond to the questionnaires through their web sites/home pages or through their
e-mailaccounts. If the respondents do not have their personal e-mail accounts or home pages,
they can use someone else’s facility, for example, at institutional facility to access the
instruments from the researcher’s website or e-mail and respond to them. Advancement in
this technology makes it possible for the researcher to transfer the data straight from the
researcher’s web site or e-mail directly into a data management softwarefor example, SPSS
for quantitative data or SPSS text editor or NUDI*ST for qualitative data, ready for analysis.
This saves alot of time because the process of sending and receiving the instruments by mail
from the respondents is eliminated. The time taken to receive the instruments, codeand key in
the data is also eliminated.

In developing countries, such as Kenya, this technology is only available to a few individuals
and institutions. To open an e-mail account, own a website or have access to internet,
individuals in these countries have to pay monthly fee to a provider, which is very costly.
However, in developed countries, these facilities are accessible to people at a minimal cost.
The use of this technology in developing countriestherefore very limiting because the
researcher is not able to reach many respondents. However, a few international organizations,
for example, UN agencies and a few public universities have access to these facilities free or
at minimal cost. In such cases, the technology can be used for research purposes in those
institutions.

The Letter of transmittal

The letter of transmittal is also called a cover letter. A letter of transmittal accompanies every
questionnaire. The letter must be brief but must contain an adequate brief about the research.
A letter signed personally instead of a photocopied letter adds a personal touch that could
make a lot of difference. This will however depend on the number of questionnaires because
it might be too time consuming to sign them all.

Responding to a questionnaire and sending it back should be made as easy as possible for the
respondent. A stamped, addressed return envelope should be included. All the respondent
does is to complete the questionnaire and drop it at the post office. If a questionnaire was
taken by hand to the respondent, arrangements should be made to have it picked up once
completed.

A questionnaire can also be prepared in such a way that when folded, the return address and
the stamp are visible and the questionnaire doesnot therefore need a separate envelope to
mail. In such a case, the respondent ismerely asked to complete the questionnaire, fold it up
correctly seal it and drop it at the post office.What information should a cover letter contain?
1. The letter should explain the purpose of the study. Respondents need to be aware of
the purpose of the study for them to co-operate.
2. The lettershould also briefly explain the importance and significance of the study.
This should be done in simple language rather than technical language, which may not
be understood.
3. It is also advisable to commit yourself to haring the results when the study is
completed. Promising to share shows the respondents that the researcher is genuine
and there is nothing to hide. If sharing the information is not possible, the researcher
should not make promises that will not be kept.
4. A brief assurance of confidentiality should be included in the letter. If the topic is
very sensitive, it may be necessary to describe the steps that will be taken to ensure
confidentiality.
5. If the study is affiliated to a certain institution or organization, it is advisable to have
an endorsement from such an institution, or organization. This motivates and
reassures respondents.
 6. In a sensitive research, it may be necessary to assure the anonymity of respondents. If
the respondents are anonymous, this increases the chances of getting honest replies.
7. The letter should contain specific deadline dates by which the completed
questionnaire is to be returned. This should give respondents enough time to respond
without inconveniencing them. Two to three weeks should be sufficient.

Follow-up techniques
After the period allowed for sending back the questionnaires has expired, the researcher
should follow up by reminding non-respondents about the questionnaires. There are various
follow-up procedures.
1. Sending a follow-up letter. A letter with some encouragement is sent to non-
respondents. The letter should explain the importance of the study again. The follow-
up letter should assume that the individual wished to respond but because of
unforeseen circumstances it was not possible. The letter should never be rude or
appear to be putting pressure on non-respondents. The letter should be polite, kindly
asking the subject to respond.
2. A questionnaire and follow-up letter.Another technique is that c sending a new copy
of the questionnaire plus a follow up letter. Though expensive, this is a better method
because most questionnaires which have not been returned are usually misplaced.

In cases where the subjects were anonymous, the follow up letter is sent to all the subjects. A
follow up letter sent to all subjects should thank those who have already responded.

Response rate and non-respondents

Response rate
This refers to the percentage of subjects who respond to questionnaires. Many studies show
that the first mailing yields about a 70% response rate. If the response rate is low, then the
researcher must question to representatives of the sample. What is an acceptable response
rate? The authors feel that a response rate of 50% is adequate for analysis and reporting. A
response rate of 60% s good and a response rate of 70% and over is very good. These figures
have no statistical basis and should be taken only as guidelines. The researcher should use all
means available to increase the response rate in order to have a representative sample for
meaningful generalizations.
Non-respondents
Non-respondents are subjects who do not respond to the questionnaire. There should be
concern when 30% or more of respondents do not return questionnaires. The researcher
should be concerned because this could affect the results of the study especially if the non-
respondents are similar in characteristics that are critical to the study. Similarly, results could
be affected if the non-respondents are significantly different from those who have responded.

If non-respondents represent more than 30% of the sample, it is advisable to check a portion
of them by selecting a small; random number. The selected non-respondents can be
interviewed to determine whether they are significantly different from those who have
responded.If non-respondents are found to be different, this factor should he discussed in the
findings or acknowledged as a limitation. If the non-respondents are not different the
researcher should assume that the sample of those who responded is representative enough.
The authors would like to mention that it is not always possible to get non-respondents to
agree to an interview. Constraints of time and money make this exercise difficult. Unless the
response rate is very low (for example below 50%), it is usually safe to assume that the
sample is representative enough. However the response rate should always be reported. This
enables readers to interpret me results and conclusions considering the given response rate.

Interviews
An interview is an oral administration of a questionnaire or an interview schedule. Interviews
are therefore face-to-face encounters. To obtain accurate information through interviews, a
researcher needs to obtain the maximum cooperation from respondents. The researcher must
therefore establish a friendly relationship with the respondent prior to conducting the
interview.

Advantages of interviews
1. Interviews are advantageous in that they provide in-depth data which is not possible
to get using a questionnaire.
2. Interviews make it possible to obtain data required to meet specific objectives of the
study.
3. Interviews guard against confusing the questions since the interviewer can clarify the
questions thereby helping the respondent give relevant responses.
4. Interviews are more flexible than questionnaires because the interviewer can adapt to
the situation and get as much information possible.
 5. Very sensitive and personal information can be extracted from the respondent by
honest and personal interaction between the respondent and the Interviewer.
6. The interviewer can clarify and elaborate the purpose of the research and effectively
convince respondents about the importance of the research. This is advantageous
because respondents can then give complete and honest information.
7. Unlike questionnaires, the interviewer can get more information by using probing
questions.
8. Many respondents do not like exposing their negative side. Interviewer are able to get
such information through interaction and genuine conversation.
9. Interviews yield higher response rates mainly because it is difficult a subject to
completely refuse to answer questions or to ignore t interviewer

Disadvantages of interviews
1. Interviews are more expensive. Researchers have to travel to meet respondents.
2. Since it is easier to ask questions rather than to administer tests or, conduct
observations, interviews tend to be misused to get fact responses which could be
obtained more accurately through o methods.
3. Interviewing requires a high level of skill, perhaps yet to be acquired by the novice
researcher. It requires communication and interpersonal skills.
4. Interviewers need to be trained to avoid bias. Interviewer bias can interfere with the
responses originating from varying techniques in interviewing. For example, a
respondent might answer a question in certain way depending on how it is put.
5. Since interviews are adaptable, flexible and show concern for human interaction, they
often introduce bias and subjectivity into the study. The subject may be too eager to
please the interviewer or the interviewer may ask questions which tend to support his
or her perceived notions. In such cases, there exists a difference between a subject’s
response and the true answer. This is called “the response effect”.
6. Interviews generally involve smaller samples because they are time consuming. If a
researcher is interested in using a big sample, interviews become a constraint.
7. Responses may be influenced by the respondents’ reaction to the interviewer. For
example, an interviewer who appears very affluent going to a very poor community
may not be received very well. This may be reflected in the responses
Rules pertaining to interviews
1. The interviewer must be pleasant. Any kind of antagonism between the interviewer
and the respondent will affect the responses given.
2. Interviewers must show genuine interest in getting to know respondents without
appearing like spies.
3. In order to create good rapport, the interviewer must try to find out very early what
kind of a person the respondent would like to interact with.
4. Interviewer should be relaxed and friendly. However they should not appear to be too
casual or clinging. If interviewers try to be too friendly this may backfire by arousing
suspicion.
5. The interviewer should be very familiar with the questionnaire or the interview guide.
6. When interviewing, the interviewer should have a guide which indicates what
questions are to be asked and in what order.
7. The interviewer should interact with the respondent as an equal.
8. Respondents should not be talked down to or belittled. This will create antagonism
and will affect the responses given.
9. The interviewer should pretest the interview guide before using it to check for
vocabulary, language level and how well the questions will be understood.
10. The interviewer should inform the respondent about the confidentiality of the
information given. Depending on the situation, the interviewer should explain the
procedures that will be used to keep the information confidential.
11. Interviewers should not ask leading questions.
12. The interviewer should remain neutral in an interview situation in order to be as
objective as possible. Probing is commonly used to get deeper information. Examples
of probing questions are “How?”, “How did it happen?”, “anything else”?

An interview schedule
An interview schedule is a set of questions that the interviewer asks when interviewing. An
interview schedule makes it possible to obtain required to meet specific objectives of the
study.Interview schedules are also used to standardize the interview situation so that
interviewers can ask the same questions in the same manner.

Structured interview schedule

An interview schedule can have structured, semi-structured or unstructured questions.
Structured questions usually have categories with responses and the interviewer simply
checks the respondent’s response. In structured interview schedules, answers are not followed
up to obtain greater depth of information.

Unstructured interview
Unstructured interviews sometimes called interview guides have a general plan that the
interviewer follows. In this kind of interview, the interviewer asks questions or makes
comments intended to lead the respondent towards giving data to meet the study objectives.

Because of the open nature of unstructured interviews, probing is commonly used to get
deeper information. Probing is very time consuming and therefore a major disadvantage of
unstructured interviews. Another disadvantage is that the interviews are very subjective
because a lot of freedom is given to the interviewer.

Semi-structured interviews
In such interviews some structured questions are asked together with some open-ended ones.
Most interview schedules are semi-structured.

Note taking during interviews

Note taking refers to the method of recording in which the interviewer records the
respondent’s responses during the interview. The responses are recorded as the respondent
talks. The recording of answers should be done very carefully.

The interviewer should record the respondent’s answer exactly as expressed. Attempts should
not be made to summarize, paraphrase or correct bad grammar. It is also important to have
some comments on the margin regarding observed gestures. The interviewer should note the
gestures and interpret their meaning, for example to portray anger, uncertainty,
embarrassment or sadness.

Advantage of note taking

1. When responses are noted as the interview progresses, it facilitates data analysis since
the information is readily accessible and already classified into appropriate categories
by the interviewer. This is especially so if the interview guide is semi-structured or
structured.
2. If note taking is done as the interview progresses, no information will be left out
owing to forgetfulness or any other kind of omission.
Disadvantages of note taking
1. Note taking may interfere with the communication between the respondent and the
interviewer. This problem is more serious where the interview guide has a large
number of unstructured questions.
2. Sometimes it becomes impossible to maintain rapport with the respondent because of
the interruption as the interviewer takes notes.
3. Note taking might upset the respondent if the answers are personal and sensitive. The
respondent may not like the idea of the answers being recorded as the interview
progresses.
4. If note taking is delayed, important details may be forgotten.
5. Note taking makes the interview lengthy and boring.

Tape recording
In this method of recording, the interviewer’s questions and the respondent’s answers are
recorded either using a tape recorder or a video tape. However a tape recorder is more
commonly used.

Advantages of tape recording

1. Tape recording the interview reduces the tendency for the interviewer to make
unconscious selection of data in the course of the recording
2. The taped interview can be played back and studied more thoroughly than would be
the case if only the notes taken during the interview were available.
3. Tape recording makes it possible to reanalyze the data in order to test objectives or
hypotheses which may not have been there originally.
4. When the responses are recorded, a person other than the interviewed can evaluate
and categorize responses. This makes it possible to establish the reliability of the data.
5. A tape recorder speeds up the interview because there is no writing
involved. However the interviewer might find it necessary to jot down’ few comments
as the interview progresses.
6. Communication is not interrupted.

Disadvantages of Tape Recording

1. The presence of a tape recorder changes the interview situation considerably.
Respondents get nervous when they know that whateverthey are saying is being
 recorded. Some researchers, however, arguethat soon after the interview starts, most
respondents forget about presence of the tape.
2. If the interview involves sensitive information, respondents may be reluctant to give
information if they know they are being taped. If the interviewer notices hesitation or
fear, the importance of recording should be emphasized and the confidentiality of the
information should be stressed.
3. Transcribing the tapes and then analyzing the information is time consuming and
tedious.

Communication during interviews

Effective communication during interviews is crucial. Interviewers should be well trained
before the study begins.

Some rules of communication

1. The language used must ensure effective communication between the interviewer and
the respondent. The respondent must fully understand the language.
2. Technical terms must be avoided when talking to lay people because they will not
understand the question. Any technical words used should be simplified or clarified.
3. The interview should explain the purpose of the study and, when necessary, the
purpose of each question. The interviewer must be hoi with the respondents because if
the respondents suspicion is aroused responses will not be honest. Communication is
enhanced if the respondents appreciate the purpose of the study.
4. The interview should first establish some rapport to put the respondent at ease and
also to inspire confidence.
5. The interviewer should monitor the interview situation. For example, if the
interviewer recognizes a sensitive question, it can be withdrawn or revisited
differently.
6. When communicating, avoid harsh words and discriminatory remarks. A well
prepared interview guide will help one to avoid such negative approaches.
7. Sometimes, the respondent gets carried away when trying to explain a point. In such
cases, the interviewer should tactfully bring the respondents back to the point without
embarrassing them.
Training of interviewers
Interviewers must be trained if objective and reliable information is to be maintained.
Interviewers should study the interview schedule in detail. They should familiarize
themselves with the interview conditions, logistics, controls, safeguards and variables being
studied. This will help them conduct the interviews fluently without hesitating, back tracking
or rereading the interview guide. If the interviewer is not very familiar with the interview
guide, the confusion will be apparent and the respondent will lose confidence in him.

In addition, interviewers should practice interviewing and receive constructive feedback until
their performance becomes standardized and reaches the desired objectivity and reliability.

Tips on interviewer training

1. Training interviewers should be done in groups rather than with individuals. This
makes the training uniform and helps to standardize the procedure.
2. The trainer should explain the study, its purpose, general guidelines and procedures to
the interviewer.
3. The group should go through the interview schedules, question by question to see how
each will be asked and to point out possible probing questions.
4. Prepare specifications to accompany the interview guide. Specifications are
explanatory comments on the various ways interviewers should handle difficult
situations during the interview.
5. The trainer should demonstrate to the interviewers how to carry out interviews
properly.
6. After the demonstration, interviewers should be paired off to practice on each other.
7. Pretest the interviews by interviewing people under similar conditions to the actual
interview schedule. People in the actual sample should not be involved in pretesting
the interview schedule.
8. During fieldwork, the interviewers should check with the researchevery once in a
while to ascertain that they are doing the right thing.
9. Interviewers should be given enough latitude to deal with unforeseen circumstances
during fieldwork.

Probing questions during an interview

It is usually necessary to ask probing questions to get information especially for open-ended
questions. Examples of probing questions are “how is that?”; “anything else?”; “would you
like to elaborate on that?, in what ways?”; “how did that happen?”, etc. Silence is sometimes
a good way of probing. When there is silence the respondent will realize that meinformation
is needed.

Rules on probing questions

1. Probing questions must be neutral. The probing questions asked should not affect the
nature of the subsequent response. To satisfy this requirement, the interviewer should
avoid being biased or subjective
2. Although some probing questions are prompted by responses, as much as possible

Advantages of telephone interviews

1. Sometimes it is cheaper to call, than to travel to the subjects.
2. When no one answers there is little cost incurred and this makes
frequent call backs possible.
3. In countries where telephone services are accessible to most households,
it may be easier to interview subjects on the telephone than through
home visits.

Disadvantages of telephone interviews

1. If the subjects are sampled from a large area, calling could be very expensive. It
should be remembered that interviewing on the phone will take time and this can be
very costly.
2. The respondent may not be very honest with responses, since it is not a face to face
situation.
3. Respondents talking on the phone tend to be impatient and this affects the accuracy of
the responses given.

Standardized and non-standardized tests

Standardized tests
A standardized test is one that has consistency and uniform procedures for administering,
scoring and interpreting the behavior of subjects. Standardized tests can be divided into norm
referenced and criterion referenced tests:
Norm referenced tests
Norm reference tests compare a subject’s performance to that of others who have taken the
same test. The subjects to whom the norm test is compared must have similar demographic
characteristics like age, educational level, geographical region, etc.

Criterion referenced tests

These tests describe a subject’s performance without reference to the performance of others.
Rather, the comparison is made to some criterion of performance instead of a set of norms.
For example, students could be given a criterion referenced test on typing. The criterion of
performance in this case would be typing a specific number of words per minute.

There are various standardized tests available in Kenya depending on one’s area of interest.
Examples of some internationally accepted standardized tests in psychology and education
are the Stanford and Binet Intelligence Scale (SBIS) and Cognitive Scales. Others are the
Graduate Record Examination (GRE) which is usually taken by students wishing to pursue
further studies abroad.Before labouring to construct a questionnaire, it is a good idea to chef
on available standardized tests especially for those researchers in educative** psychology or
sociology

Non-standardized tests
These are tests that have not established a considerable procedure nor gone through the
construction process to minimize errors. Examples of non-standardized tests are classroom
tests. Such tests are not available for use in replicating studies.

Observation forms, schedules or checklists

A researcher utilizes an observation checklist to record what he or she observes during data
collection. First, the researcher must define the behaviours to be observed and then develop a
detailed list of behaviours. During data collection, the researcher checks off each as it occurs.
This permits the observer to spend time thinking about what is occurring rather than on how
to record it and this enhances the accuracy of the study.

In some cases, some studies require rating scales to be used. These require that the observer
not only observes the behaviour but also evaluate that behaviour on a rating scale. The most
commonly used rating scales are Likert type scales with 3 or 5 response categories. This type
of procedure requires a higher level of inference on the part of the observer since involves
observation and evaluation.
Points to bear in mind when using observation forms
1. Once the researcher has developed an observation form, it must be tested in situations
similar to those which are expected during the d collection. Pretesting the observation
form helps to correct any mistake that may be discovered.
2. Not too many behaviours should be observed. About ten behaviors are satisfactory. If
more behaviors are to observed, much training is required in order to ensure reliable
data.
3. The number of subjects must be controlled. It is not possible to study many subjects
as one would in a survey research since observing 10 or more behaviours is very time
consuming. If too many subjects observed, the results tend to be unreliable.
4. The behaviour to be observed should be defined in sufficient detail. This makes it
easier for the researcher to determine whether a particular behaviour has occurred or
not.
5. Where more than one observer is used in a study, training is necessaryin order to: -
a. Master the observation form
b. Standardize observation procedures.

The first step in training observers is to discuss observation forms. Each item is described
thoroughly so that observers are aware of the behavior to be observed.

The second step is to set up practice observations in which all observertrainees participate.
During this time, clarifications should be made, andrules should be set to help observers to be
as uniform as possible whileobserving. Videotaping the particular sessions helps improve the
observers’skills.

Review assignments

1. Compare and contrast questionnaires and interviews as research instruments

2. What is the importance of a letter of transmittal?
3. Define the term response rate. Why is it important to have a high response rate in a
research study?

Notes
1. Borg and Gall, 1984
2. Borg and Gall, 1997.
3. Borg and Gall, 1984.
4. Ibid