VARIABLES & SAMPLING IN RESEARCH

Variables in research
What is a variable?
A variable is defined as: a characteristic, attribute or property that can take on different
categories, levels or numeric values among people, events or objects. We revisit variables
and concepts here below:
Variables are developed from concepts or constructs. A concept ‘expresses an abstraction
formed by generalization from particulars’ (Kerlinger 1986).For example: ‘Temperature’
is a concept which expresses numerous observations of things that are more or less ‘cold’
or ‘hot’ likewise, ‘achievement ‘ in mathematics is an abstraction formed from the
observable behaviour of children or students. The behaviour is associated with
acquisition of knowledge and skills in mathematics which can be interpreted through
activities like: Solving arithmetic problems or obtaining a given score.

Activity
Find out the various behaviour or activities are put together and expressed in these words
(concepts) or constructs.

 Intelligence
 Aggressiveness
 Honesty
 Conformity

The major types of variables

Variables are classified in several ways. In this text, two important classification criteria
are used:
a) Classification by status of the variable in a given relationship.
b) Classification of variables by suitable measurement techniques.

The purpose of the first classification is to enable the researcher to explain the ‘problem’
under investigation. The second classification is to enable the researcher to collect the
right data, which will be analyzed using correct techniques.

Classification of variables by status in a relationship

In this classification, the assumption here is that any problem that a researcher wishes to
investigate represents a variable or a set of variables. As noted earlier, defining the
problem will specify certain concepts or constructs, some of which are scientifically
known from theory, principles, and models and so on. Every variable (or problem) does
not exist in isolation. Whenever something occurs or changes it will trigger subsequent
changes of related factors. Its occurrence or change is also triggered by some precedent
factor(s), which relate to it. However, not all variables have a relationship. It is, therefore,
necessary to be selective in your choice.
A construct is a concept, however, it has added meaning of being deliberately and
consciously invented or adopted for special scientific purposes. For example: terms like
intelligence and motivation have been used for scientific purposes.

Variables provide more observable and measurable indicators of concepts and constructs.
Concepts and constructs are defined in operational terms by specifying activities or
operations necessary to measure them. These activities, events or observations provide
concrete meanings to the concept or constructs.

Activity
Describe the activities, behaviour, or events that can be used to describe the following
concepts:
 Gender
 Education
 Social class
 Organizational productivity
 Occupational Mobility
 Verbal aptitude
 Anxiety
 Religious Affiliation
 Political preference.

It can be argued that: a research problem represents variables. The variables are things
that occur, change or are still changing and need to be explained or controlled. It is our
inability to explain and control them, which constitutes the problem. Constants rarely
pose problems. It is necessary for the researcher to identify and determine only those
variables that are genuinely related to the problem ‘variable’. This information can be
obtained from relevant theories, previous research, and models, known principles, among
others. Information about the relationships provides a preliminary explanation of the
problem, which can be confirmed through data collection and analysis.

In classifying variables in a relationship, you will use the following types:

1) Independent variables
2) Dependent variables
3) Extraneous variables
4) Intervening variables

Independent variables: these are variables that influence or cause other variables to
change. In this case, they are known as: ‘CAUSAL FACTORS’ in a given relationship.
They are also referred to as the ‘precedent’ factors because they logically change or occur
before other related variables can change.

Dependent variables: represent the variable that change or occur in response to previous
changes or occurrences of independent variables. The dependent variables represent the
effects in a given relationship. They are also known as the resultant factors in a
relationship.
In research, explanations about phenomena represent the cause – effect relationship
which is illustrated in the figure below:

Figure: Illustration of the cause- effect relationship

CAUSE EFFECT
Relationship
Independent
Variable Dependent
Variable
(X) (Y)

You will notice that, this is a one-way process. The assumptions are that:

 If X caused Y then the two variables must have changed together. In this way
they have a relationship.
 If Y changes or has changed we should observe the causal change in X
 If X changes, we should observe the resulting change in Y.
 If X is supposed to cause Y, then changes in X must precede changes in Y
 If X causes Y, then no other factor could have reasonably caused the change in Y
at the moment. (This can only be determined if we hold other variables constant).

Activity

Examine the following sets of variables in the table here below. Determine if they have a
Relationship with the other.

Variable A Independent or Variable B Independent

Dependent? or
Dependent?
1 Intelligence Achievement
2 Gender Verbal aptitude
3 Honesty Social Status
4 Aggressiveness Religious Affiliation
5 Achievement Educational Level
6 Age Social Class
7 Anxiety Stress
8 Political Preference Religious Affiliation

Extraneous variables: these are variables, which may influence the independent, and
dependent variables yet do not constitute a major part of the research study. Extraneous
variables are neither measured nor manipulated by the researcher. Unless they are
controlled, by holding them constant, they may affect the validity of research findings.
 Figure: Extraneous variables

Independent variable Dependent variable

A B

C D
Extraneous Variables

For example, in a study to determine the effects of teaching methods (A) on the
performance (B) of students, in technical education subjects, the research may
concentrate on data from the two variables only. It may happen that the performance (B),
at the same time, is influenced by other factors such as Resources (C) and student
Abilities (D). If these extraneous factors are not measured, then the result will be
exaggerated or underestimated, thus affecting the validity of findings.

Intervening variables: sometimes variables exist in a cause – effect chain. For example,
the relationship between A (an independent variable) and B (dependent variable) may
not be obvious or direct unless some other variable, C (intervening) changes.

Figure: Intervening variables

Independent Variable Dependent variable

A B

C
Intervening variable

For instance, we can introduce a method of teaching mathematics (A) to improve

students’ performance (B). However; unless students are motivated (C) when the new
methods are employed, no significant change in B) will be realized. Majority of
intervening variables represent psychological factors or constructs, which are difficult to
observe or measure directly. Common examples include perceptions, attitude,
motivation, morale and intelligence.
Classification of variables by suitable measurement techniques

Measurement is a process of collecting data. Variables are represented by numeric values,

levels or categories with reference to a group of people, objects or events, according to
their specific characteristics, attributes or property. Measurement is a means of
comparing people, objects or events by assigning them numeric values, levels or
categories. Measurement enables us to get data about their characteristics, attributes or
properties. This gives these characteristics some meaning to enable us to describe people
or explain their behaviour.

In terms of classification by measurement techniques, four types of variables are

identified:
a) Nominal variables
b) Ordinal variables
c) Interval Variables
d) Ratio Variables

We must recognize, however, that some problems relating to social phenomena can be
expressed in textual form. These are mainly developed in qualitative research. In this
case, specific variables are not emphasized because of the complex nature of social
phenomenon.

Nominal variables: these are variables which have categories that are arbitrary and
sometimes mutually exclusive. The categories are arbitrary because there is no continuity
or order between them. Also, they can be mutually exclusive because, in any given
population, every unit belongs to one and only one category at any given point in time.

The variable that is measured in a nominal scale generates nominal data.

Activity
Specify categories of the subsequent variables.

a) Religious Affiliation
b) Political Preference
c) Occupation

Example:
Variable Categories
a) Gender 1. MALE
2. FEMALE

Ordinal variables: nominal variables comprise categories or levels that can be arranged
in a rank order sequence. In this way, it is possible to tell which level is higher or lower
than the other. It is, however, not possible to determine the unit differences between the
levels. We cannot tell, by what quantity one level is greater than another. This variable
is measured on an ordinal scale.
Example:

Variable

1. Educational attainment Levels

1. Primary Education
2. Secondary education
3. University Education
Note that primary education is lower than secondary education, which in turn, is lower
than university education. We cannot, however tell by how much secondary education is
higher than primary education.

Interval variables: the interval variables comprise categories whose levels, can be
arranged in an ordinal scale. At the same time, it is possible to establish unit differences
between the levels. They are considered to be continuous variable. Each level of the
variable contains a range of numeric values, from a specific minimum to a given
maximum value.

Example:

Variable Levels Value range

Age – Group 1 Infant 0 – 2 Years
2 Children 3 – 12 Years
Adolescents 13 – 17 Years
Young Adults 18 – 35 Years
Adults 36 – 60 Years
Aged Over 60 Years
Note that the levels obey the order system. Each level has a range of numeric values.

Ratio variables: the ratio variable represents the highest level of all the variable types.
In this sense, each variable comprises levels, which obey the rank - order sequence. It is
also possible to determine unit differences between the levels of the variable. The
variables are continuous, which means that they can take on any value along a continuous
scale. Also, they generate absolute or exact values at the time of data collection.

Example:

Variable

 Age
 Exam score
Activity
Examine the following variables, indicate whether and why they could be nominal, Ordinal,
interval and ratio.
 Honesty
 Aggressiveness
 Education
 Social Class
 Anxiety
 Religious Affiliation
 Political Preference

Applications of variables research study

As noted earlier, the first step in research is to identify and define a problem. As the problem is
defined certain abstract concepts emerge. These concepts must be defined in operational terms as
variables. The purpose of this is to facilitate the data collection process. Evolving suitable
measurement techniques for each variable develops their operational definitions. In this process,
we must determine whether or not the variable is nominal, ordinal, interval or ratio. The observable
and measurable indicators must be specified in terms of activities, behaviour or events that
represent the variable or concept. These events, behaviour or objects can then be classified into
relevant categories, levels or numeric values, to define each variable.

Use of variables in formulating a conceptual framework

In developing a research proposal, it is necessary to specify key variables in the study. Prior
knowledge of the variable is necessary to enable the researcher formulate viable objectives,
research questions or research hypothesis.

In majority of research studies, the dependent variable represents the research problem. We could,
therefore, identify or determine possible causes (independent Variables) to the problem by
examining relevant theories, models or principles as well as previous research. Theory is defined
as: “A set of interrelated constructs (concepts), definitions and propositions that present a
systematic view of phenomena by specifying relations among variables with the purpose of
explaining and predicting the phenomena” (Kerlinger 1986: 9)

Miles and Huberman (1994) state that a conceptual framework explains the main things to be
studied in a narrative form or sometimes graphically. The main things to be studied include
concepts, constructs or variables. The explanation provides the presumed relationships among
them.

Both theory and a conceptual framework can be used to specify the independent variables for a
given research study. Once the problem is known, we then need to build a framework that relates
it to independent variables. A conceptual framework can be rudimentary, elaborate, theory –
driven, descriptive or casual. Conceptual frameworks are best done graphically rather than in text.
This enables the researcher to map the likely relationship, to identify variables that are
conceptually distinct and work with all information at once.
In studies relating educational programmes, a conceptual framework can be developed using a
logical system of the relationship among variables. For instance, this could be built on a systems
model.

In any system, four key components can be identified, namely:

 Inputs
 Process
 Output
 Feedback

For example, if we were to take Guidance and Counselling Programme as a sub-system in a school.

Figure: A system model

INPUTS PROCESS OUTPUTS

FEEDBACK

The inputs component represents the factors that should be put in place initially to enable the
system to function. For example: Human resources, physical facilities, policy guidelines, and
financial resources, among others.

The process component represents the techniques and methods that must be employed to reinforce
the inputs towards the achievement of the desired outcome in the system. For example, methods
of training, motivating students/ employees, performance measures among others.

The outputs represent the desired outcomes or results that should derive from the system. For
example: Achievement of institutional objectives, good student performance and other positive
results.

The feedback component provides checks and balances in the system by revealing possible
successes, achievements, weaknesses and limitations in the system. For example: Observed poor
performance among students, indiscipline cases and others.
In relation to your research study, the feedback represents the indicators of the research problem,
which in turn represents the dependent variables.

The problem can then be explained in terms of the possible causal factors within the inputs and
Process Components of the system. These factors, in turn, represent the independent Variables.

A schematic illustration can be developed showing the independent and dependent variables as
follows:
Figure: A schematic presentation of relationships among variables

Independent Variables Dependent Variables

It is possible to explain the relationship between these variables conceptually. Objectives of the
research study and research questions would then be formulated by evaluating each independent
variable against the dependent variables

Conclusion
The first step in any research study is to identify and define a research problem. Before identifying
a suitable problem for your study, explore suitable source of information relating to your area of
interest. Literature review is one of the major sources. After defining the research problem, it is
necessary to evaluate its viability before you begin the research process.
SAMPLING AND SAMPLING TECHNIQUES
Introduction

Various procedures are used in determining the sample of respondents in any research. The
procedures are presented in this section. The term sample is defined and different types of samples
discussed. The term sample is differentiated from the term population, with it is often confused.
Knowledge of sampling is useful since one cannot study the entire population that would give the
requisite data.

LEARNING OUTCOMES:
By the end of this section, you will be able to:

 Define the term “sampling”.

 Distinguish between a sample and population.
 Explain the criteria for a good sample.
 State advantages of sampling.
 Select a suitable sampling method for your study

What is sampling?
Sampling is a procedure used in the selection of sample units from the population. A sample
represents a small section of the population that has been selected for observation and analysis.
The purpose of sampling is to study a population.
A population represents all cases of people, objects or events that posses certain common
characteristics relevant to the purpose of a study. The target population for a research study
depends on what kind of information is required and who can provide it. For example, a study on
“effects of Management styles on school performance in Examinations” would target a population
of:
 Head teachers
 Deputy head teachers
 Assistant teachers
 Students
 Sub – ordinate staff.

The various categories of people mentioned here have certain common characteristics. They are
all members of the school environment where the various management styles’ prevail. Also, their
behaviour or actins contribute directly or indirectly to school performance.

Activity
Describe the target population for the following research study: Determinants of Effective
Financial Management Practices in Secondary schools in Nairobi.
The target population represents the universe in which the results of the study are generalized. This
implies that the sample measures (or statistics) are used to interpret or estimate the population
parameters (or characteristics).

The purpose of sampling is to study some population by selecting a representative sample. In this
case: “A sample is a set of cases drawn from the specified population for purposes of generalizing
or making inferences about that population” (Kohout, 1974:247). Look at the figure below which
illustrates a sample population.

Used to estimate or
Interpret the population Parameters
Example: pop. Mean: Pop. S.D: 

In this case the sample mean, X, is used to estimate the population mean, . Likewise, the sample
standard deviation, s, is used to estimate the population standard deviation, .

A parameter is a measure that summarizes or describes a characteristic of a population. A statistic

is a corresponding measure that describes a sample characteristic. The sample statistic can provide
a close or accurate estimate of the true population parameter. However, the two can be widely
varied. This difference represents the sampling error.

Sampling error: Is the difference between the true population measure (Parameter) and the statistic
used to estimate it. A good sample must fulfil the following criteria in order to minimize sampling
error.
Criteria for a good sample
a) Adequacy of the sample size
The sample size should be large enough to adequately represent the true population characteristics.
The smaller the sample size the greater the sampling error and vice versa.

b) Reliability of the sample

The selected sample should be reliable to minimize errors or bias in interpretation of research
results.

c) Independence of sample units

The sample units should be independent of each other. This means that the selection of one
population unit should not influence the selection of another unit. Each population unit
should be given equal chance of being drawn for the sample.

d) Homogeneity of sample units

The sample units should be homogeneous. This is to say that they must posses’ similar
characteristics which reflect the true population characteristics.

Reasons for sampling

What are the advantages if using a sample in a research study?

a) Cost factor
Sampling reduces the costs of the study, what would otherwise be high if the whole population is
used.

b) Time factor
Less time is spent in the study with a sample than that a total population. This enables the
researcher to meet specific time for research projects.

c) Accuracy
Handling fewer population cases enables the researcher to conduct a through evaluation and
analysis of data. This improves reliability of results.

d) Efficiency
It is possible to carry our several studies simultaneously in a single population when samples are
used. This increases efficiency in allocation of research resources.
Activity
Explain why sampling is preferred to a census study.
Designing a sample

The following steps are suggested in sample design:

Step 1: Re-examine the objectives of the study. The objectives indicate the relevant population
characteristics required in the study.

Step 2: Define the population characteristics, that is, categories of people, institutions, or
organizations.

Step 3: Determine the population size (N) and obtain a suitable sampling frame: A list of all
population units from which the sample will be drawn.

Step 4: Determine an adequate sample size (n). This could be at least 25% of the population
size (N).

Step 5: Select the sample units, using a suitable sampling method.

Sampling methods
Sampling methods can be classified into two major categories, namely:

a) Random/ Probability sampling methods

b) Non – Random (Non – Probability) sampling methods.
Random sampling: Refers to a procedure for selecting cases from a population in such a way that
every member of the population has an equal chance of being drawn for the sample.
Non – random sampling: Refers to a procedure for selecting cases from a population in such a
way that members of the population have an unequal chance of being drawn for the sample.

The differences between the two methods are summarized in Table 7.1.
Table 7.1: The differences between random sampling and non-random sampling
methods
NON – RANDOM SAMPLING
RANDOM SAMPLING METHODS METHODS
a) Selection is based on chance or a) Selection is judgmental or purposive
probability
b) Generates an accurate sample b) Generates a biased sample
c) Mainly used in quantitative research c) Mainly used in qualitative research

Random methods of sampling

Random sampling techniques include:
1. Simple random sampling
2. Systematic random sampling
3. Stratified random sampling
4. Cluster or area sampling.
The table below summaries the contexts in which one can use each of the methods of sampling

Table: How and when to use random sampling methods in research

Sampling When Used Selection Procedure

Method
1. Simple random  Target population 1. The population is
Sampling must be defined, that is,
Homogeneous. (homogeneous).
 Population size is 2. A sampling frame
known is obtained.
 A sampling frame 3. An adequate
is available. sample size is the
 Population is determined.
relatively small. 4. Population size is
determined
5. Sample units are
selected using
random numbers or
the lottery
technique.
2. Systematic Random  Target population 1. Define the target
Sampling. is Homogeneous. population
 Population size is 2. Obtain a suitable
known. sample.
 Sampling frame is 3. Determine the
available. population size.
 Population size is 4. Determine an
large. adequate sample
size.
5. Divide the
population into a
number of equal –
interval, where
each Interval =
Population size
over Sample size.
6. Select the first unit
from the first
interval at random,
and note its position
in the interval ( that
is, r th units from the
rest of the
intervals).
3. Stratified Random  The target 1. Define the target
Sampling. population is population.
heterogeneous 2. Obtain a sampling
 Population size is frame.
known. 3. Determine the
 Sampling frame is population size.
available. 4. Divide the
population into
logical components
or strata.
5. Determine the
number of units in
each stratum.
6. Select the sample
units each stratum
at random and
ensure that the
sample components
reflect the
population strata,
proportionately.

4. Cluster or Area  Population size is 1. Define the target

Sampling known. population.
 Population is 2. Determine the
distributed into population size.
wide geographic 3. Determine the
areas/locations. sample size
 Population units are required
grouped into 4. Determine the
clusters. number of logical
cluster in the
population and the
size of the smallest
cluster.
5. Select one or more
clusters at random
to represent the
sample.
Methods of sampling are non – random / non – probability.
Non –random sampling techniques include:

 Quota sampling
 Chain sampling (snowball)
 Maximum variation sampling
 Critical case sampling
 Extreme/deviant sampling
 Typical case sampling
 Intensity sampling
 Criterion sampling
 Homogeneous sampling

When and for what purpose the non – random sampling methods used
The table below explains the contexts in which the non-random sampling methods are used.

Table: Use of non-random sampling methods

Sampling Method Purpose and Use Selection Procedure

1. Quota Sampling Used to illustrate sub groups  Identify the

(Stratified purposive and facilitate comparisons. population strata.
sampling) - Used when the population  Estimate relative
heterogeneous but the sizes of population
population size is not strata.
known.  Design a sample by
allocating quotas to
the population strata
by size or
importance to the
study.

2. Chain sampling  Used to identify  The target

(snowball method) cases of interest from population is
the people who can defined.
identify others that  A few population
are familiar with units are identified.
population cases that  A referral network
are information rich. system is used to
identify other sample
units until an
adequate sample size
is achieved.
Activity
a) Describe the kind of data you will need for your research report
b) Indicate the method of sampling you will use to collect the data

Conclusion
The reliability of the research results will depend on the careful choice of a relevant sampling
method. It is important to ensure that the researcher selects a sample that fulfils the major criteria
for a good sample.