20

VARIABLES
CHAPTER PREVIEW
.Meaning
And Types of Variables
Dependent Variables And Independent Variables
Qualitative Variables And Quantitative Variables
Continuous Variables And Discrete Variables
Moderator Variables And Intervening Variables
Active Variables And Attribute Variables
.Methods of Measuring Dependent Variables
.laportant Considerations in Selection of Variables
.Important Approaches to Manipulating Independent Variables
.Techniques of Controlling Extraneous Variables
Technique of Elimination
Constancy of Conditions
Balancing
Counterbalancing
Randomization
Controlling Demand Characteristics

MEANING AND TYPES OF VARIABLES

nvaable, às the name
implies, is something which varies. This is the simplest and the broadest
way of defining a variable. However, a behavioural scientist attempts to define avariabiemore
ECIsely and specifically. From his point of view, variables may be defined as those attributes of
and which can be measured. In other
words, variables are ne
cis, events, things beings,
icteristics or conditions that are manipulated, controlled or observed by the expermenc
elC., ae
authoritarianism, achievemen,
eva anxiety, aptitude, income, education,
s O variables commonly emploved in psychology, sociology and education.
Can be classified in several Some of the commonly accepted es
dles below.
are presented
ways.

ependent Variables And Independent Variables borrowed

from the
field of

Theterms "dependent variable' and 'independent variable have been

into dependent
and

mathematics
atics in behavioural researches. The classification 0dependent variabiee
variable(V
research. Ihe aependent
defined as one requently employed in experimentala prediction.
The ine perimenter for the
t
about which the experi
erimenter makes the
is defined as onewhich is manipulated, measured and
selected by the ther words,
made.
measure lor the DV ISis mae
the DV
pose producing observable
of rvable changes in the
behavioural
the Deti h the
about

of which the predictioause isnd the

prediction
of
dependent
The occasio ariable is the variabl o n the basis
is
s t i m u l u s variables ane

Trdi a
whichthe IV
synonym of IV is controlled variable, calls
wth control variable. Underv.ood (1966:12)

433
 Sciences
Behavioural
Methods in
Measurements
and Research
434 Tests,
between the IV a n d .
the distinction
illustrate the
An example may upon the classro
effect of teaching methods
variables.
the response the
DV as
to study Ba
teaching, say, A,
DV. Suppose the experimenter
wants
employ three methods of
and subsequent!y. th
he may
achievement of pupils. For this purpose, these three
methods
of pupils by the example of the,
C and may teach
the same group methods constitute
measured or predicted.
Teaching if the experiment
achievement may be
the exampie of the DV. Similarly,
achievement constitutes he may tak
and the classroom attitude towards family planning,
wants to study the
effect of a religious group upon
groups and study
their attitudee
the Parsi as the four religious
the Hindu, the Muslim, the Sikh and which religious group has
Subsequently, he may be able to predict
towards family planning.
unfavourable attitude towards family planning
In
this example, the
favourable attitude or attitude towards family planning
constitute example
the of the IV and the
religious groups
constitutes the example of the DV.
selection and measurement
of dependent variable is that
One important thing related to the where a null result occurs
effect. A floor effect is said to nave occurred
of floor and ceiling Such an effect usually
bottom end of the scale.
because majority of subjects score at the very
too difficult for the subjects.
For example, if in any experiment the
emerges when the task is
to lights of different colours
but all the given colours are
purpose is to compare people's sensitivity
This illustrates floor effect. Ceiling effects
below thresholds, no meaningful data will be obtained.
are the opposite of floor effects and are
said to have occurred when subjects score too close to the
with the same example, if all the
top of the scale simply because the task was too easy. Continuing
coloured lights are clearly visible because they are bright, the task would be too easy for the
can be obtained with
subjects and therefore, ceiling effect will occur. These two types of effects
any task whose dependent variables cannot track
the full range of independent variable. These
effects must be prevented which can readily be done through pilot study that can check the
appropriateness of the subject pool and variables before carrying out the experiment proper.
in addition, a dependent variable must be sensitive, valid and reliable. To be sensitive, it
must change clearly when the independent variable changes. To be valid, it must measure what it
is intended to measure. To be reliable, a dependent variable must act consistently in relationto
the independent variable. If subjects respond a certain way on one occasion but in a wholly
different way on another occasion when the independent variables remain the same, the
dependent variable would be unreliable.
As it has been said above, the IV is manipulated by the experimenter and its eftect is
examined upon the DV. Some experts, depending upon the mode of manipulation have tried to
divide the IV into 7ype-E independent variable and
1970:11). Type-E independent variable is one which is
Type-S independent variable (D' Amato,
the experimenter, and Type-S independent variable is one
directly or experimentally manipulated by
which is manipulated through the
process of selection only. Such variables are difficult to
manipulate experimentally or directiy. A
example may illustrate the distinction between the Type-E and Type-S
Suppose the experimenter wants to study the effect of temperature independent variable
an industry. Here the IV is the
upon the rate of productionm
temperature and the DV is the rate of production. He may
manipulate the temperature by dividing it into three
examine its effect upon the rate of categories-high, medium and
production. Here the temperature is being directly low--a
by the experimenter and hence, it constitutes the manipulae
Suppose, for the time being, that the experimenter is
example of Type-E independent variad
interested in answering the the
rate of
production dependent upon the age of the workers? Age is here the question: is
For investigating this problem, the experimenter will have to select independent varia
of
get an appropriate representation from workers
of their ages in a way by which he can groups on the

ranging from, say, 16 to 55 years. Subsequently, he will different age 8 ined

by each age group and tinally, conclude whether or not compare the rate of production ola she
age is a factor in
experimenter by means of enhancement
production. The is
age being manipulated by the
selection in or
 ariables 435

te
eftects upon
determine its ettecr the rate of
prodction. Hence, this (onstitutes the exanple of Typee-5
n d e n t variable. A research or investigation whi h
inde
involves the manipulation of thee
ant variable is called experimetation, no Type-E
indeper
matter whether it 1s cone in a lateofatory or in
a research which involves the
a natral setting. Likewise, manipulatiorn
of the
l correlation research. Sometimes, a research is done in which there are no indetenedent
Typ5 variatble is

ca Sch
variables. Such research is termed as observation, e.g., when the re'searcher Wants to study the
underly ing the mating behaviour of dogs, he looks at this behaviour from a distanee for
n e time in order to have an idea regarding the factors underlying the mating behaviour. in this
evample the researcher or the investigator is not manipulating any variable, rather he is sirnply
obser ing the mating behaviour to know the underlying factors. Tnis illustrates what we
call obsen ation.

The independent variable (or the stimulus variable as Underwood calls it) may also be
classitied on the basis of the nature of the variables. Depending upon the nature of variables, the
independent variables may be classified into three categories: task variables, environmental
variebles and subject variables.
Task Variables
The task variables refer to those characteristics which are associated with a behavioural task
presented to the subject. It includes the physical characteristics of the apparatus as well as many
features of the task procedures. There are different types of apparatus used in an experiment.
Some are simple and some are complex. The simplicity or the complexity of the apparatus is
likely to produce a change in the behavioural measure. When, for example, the number of wrong
paths (blind alleys) in a particular maze is increased, the process of learning or traversing the
maze will be a difficult task for the subject. Likewise, when several more or less similar coloured
lights are added to a study of complex reaction time, the reaction time is likely to be longer

Environmental Variables
Environmental variables refer to those characteristics of the environment, which are not the
physical parts of the task as such, but tend to produce changes in the behavioural measures
Noise, temperature, levels of illumination, and tinme of the day are examples of environmental
variables. Suppose, for example, that the investigator wants to study how reading speed is
influenced by the degree of vagueness in handwriting. It is likely that apart from the degree of
vagueness, the reading speed may be influenced by the levels of illumination and noise occurring
the time of reading the materials. The intensity of the light and noise constitutes the example
of
at
the enviornmental variables because the variations in the intensity of the light and noise also tend
and that of sound are not the
to produce changes in the reading speed. Heretointensity light in the behavioural measures
of

physical parts of the task (handwriting) but tend produce changes

(relating speed).
Subject Variables
ubject variables refer to those ofthe subjects which are likely to produce changes
characteristics

measures. Sex, age, weight, anxiety,

intelligence, etC., are the characteristicsof
the behavioural which may be convenienty termed as subject
variables. Subject
ne
subjects (animals or human), induced subject
the natural subject variable and the
afiables can be divided into two types: wthin
are those variables,
which the subjects carry
variables
ariable. The natural subject sex, intelligence, and
anxiety are the enamples
before the start of the experiment. Age, instructional variables,
mselves variables, also known as the
natural subject variables. Induced subject i n s t r u c t i o n . Suppose
the
are induced by
the experimenter's
which
re those subject variables, the effect of two methods (A and B) upon the problem-solving another group
Vestigator wants to study of subjects with o n e set of
instructions and
to
instruct one group i n s t r u c t i o n s are ikely
naviour.He may The two difterenm sets
of
instructions.
set of
Subjects with another
 Secicnces
Behau ioural
Methodls in
and Research
Measurements
436 Tes/s,
produce changes in problem-solving behaviour. Likewise, the investigalor may want to.
upo
instructions to one grostudy
He may, for example, give that the tac
of ego-involvement. oI Subjects
earning as
a lunction
of his ability and to another group I n e s e two types
Subjects that a given
task is the test
situation.
followed in the
experimental
used as standard practice of the task.
Deing differences in períormance
instructions are likely to produce variable only for the
of
sake controlling
manipulates a
Sometimes, the experimenter is not an independent variahi.
results. Such a variable the reaction tim
unwanted eifect upon the
experimental whether o r not
interested in studying
the experimenter is to c o n d u c t the individh
for example, suppose time to tone. If he has decided
to light is shorter than the
reaction Now suppose, for the tim
presentations of
each stimulus.
200 randon random presentation, In
experiment, he may give simulus rather than a
presentation of the 100 trails he presents light and
being, he makes systematic
a
and in the remaining
the first 100 trials he presents tone is shorter than the reaction
conclusion that the reaction time to light
Subsequently, comes to the
not due to the ditference in stimule
time to tone. One may, here, argue that
the obtained result is
cumulative experiences with 100
have gained some
but to the practice effect. The subject might
reaction to light in the remaining 100
presentations of tone, which helped. in making quick he would be manipulatino
200 random presentations
presentations. Obviously, then, by making
the 'sequence of stimulus presentations'. This
variable is manipulated to control or eliminate its
unwanted effect upon the dependent variable and not for determining its influence upon the
variable. Thus, for a variable to be called an
dependent variable. Hence, this is not an indepedent
independent variable, it must be manipulated for the sole purpose of producing its effect upon the
dependent variable, otherwise it will not be an independent variable. In experimental situations,
those variables which are controlled by the experimenter because they are not of direct interest
but are likely to produce changes in the behavioural measures are known as control variablesor
relevant variables. Think of the difference between control variable and controlled variable
Control variables are also known as extraneous variables. For example, if the investigator wants to
study the relationship between the height and weight of some children, it is very likely that age
will tend to affect the relationship because with advancement in
age both height and weight of
the children will increase. Age, here, is the of
example the relevant variable because it tends to
influence the relationship between height and
weight of the children. Likewise, if the
investigator
wai.ts to
study the effect of two teaching methods-A and B--upon classroom achievement
intelligence may be one factor which, if uncontrolled, may produce additional
classrom achievement. It is differences in the
methoa of teaching and
expected that the intelligent children will
benefit more from eitne
may improve their classroom achievement.
example of the relevant variable. There are some Thus intelligence is,
variables, which have no discernible here, a
upon the dependent variable. These are ene
known as irrelevant variables.
extraneous variables, are of the Relevant variables,
any experiment. fo!lowing three types, which must be controllea u
"
1. Subject relevant variables:
constitute the characteristics of Subject relevant variables are those vhich

does not want to the subject are controlled by the

and variables
study their effect upon
Type-S independent variables which the behavioural measure (or the experimenter becaurhe
manipulation through selection
are the
characteristics dependent
of subjects and lend variadi to
Thus age,
intelligence, may be
conveniently grouped as the subject relevant ariablethemsee
subject relevant variables.race, aptitude,
The students personality classification, etc., are the a f the
independent variable in one should note examp
carefully that the samevariable may acta
an

experimental situation. A variable experimental

situation and as a relevan
v

ant variable in
n o t h e r

experimenter is called the which is

variable because he independent manipulated and whose effects are ed by the

does not want variable but when the stuu same

dependent variable, it becomes an to experimenter con

study its effect upon the o r the

example of the
relevant variable. behavioural med
 Variables 437

tianal relevant variables: The situational relevant variables refer

2.
and task variables whose effects (on the DV) are controlled by the to those
e n v i r o n m e n t a l

experimenter
are to produce
unwanted changes in the DV. The
likely temperature, levels of
nination, complexity of task, etC., are examples of situational relevant variables. All thos
independent bles whose effects are unwanted and which can be directly manipulated by the
herimenter (that is, Type-E independent variables) are included in situational relevant variables.
exper
3. Sequence relevant variables: The sequence relevant variables are those variables which
arisefrom the different ordinal positions that the conditions of the experiment occupy in a
sequence. For example, when the same subjects are exposed to two or more than than two
conditions of the experiment (that appear in a particular sequence), factors like practice, fatigue,
adaptation are likely to intluence the behavioural measure (or the DV). In the tirst condition the
Subjects' performance will not be allected by pratice biut in the ist condition it may be.
Similarly, fatigue may not be intluencing the periormarce in irst condition but it may
influence the performance in the subse quent conditions. Practice and fatigue are examples of
sequence relevant variables, which are usually controlled by the counterbalancing design
adopted by the experimenter (D' Amato, 1970).
Qualitative Variables And Quantitative Variables
The qualitative variables refer to those variables which consist of categories that cannot be
ordered inmagnitude. We cannot make such a statement regarding the qualitative variables, e.g,
'category X possesses higher (or lower) magnitude of the variable than category Y.' Thus the
qualitative variables comprise the categories, which do not have a quantitative relationship
among themselves. Sex, race and religion are examples of qualitative variables because they
cannot be ordered in magnitude. Since the qualitative variables cannot be ordered in the
magnitude, precise and accurate measurements are not possible. As a consequence, they are
least preferred in any scientific investigation. The quantitative variables refer to those variabies
which are composed of categories that can be ordered in magnitude. We can, for example, say
that category A possesses greater magnitude of the variable than category B. Intelligence, age,
levels of illumination, intensity of sound, etc., are examples of quantitative variables. We can
that group A possesses a higher magnitude of intelligence than group B and older people say
Tatigued sooner than younger and adult ones. Thus the variables can be ordered in terms get of
magnitude. With the quantitative variables, precise and accurate measurements are
Decause they can easily be ordered in terms of increasing or decreasing magnitude. In possible
anc education, fortunately, most of the variables belong to the category of psychology
insociology, qualitative variables are more common. quantitative variables.
Continuous Variables And Discrete Variables
dantitative variables are further divided into two categories, namely, continuous variables and
ete variables. A continuous variable is which is capable of being measured
one in any
rary degree of fineness or exactness. Of
course, the measurement is subject to the
limitations
d b l e tools. Age, height, intelligence, reaction time, etc., are some of the examples of a
u e s
variable. The age of the person can be measured in
years, months and days. To 8o to
alc dler unit of measurement, age be
ewse, reaction time can be measured inexpressed
can
hours, minutes in terms of and
seconds
seconds and milliseconds. If it
required, can also
S alred
est
in terms of microseconds.
Thus, all
such variables which can be measured in the
nyowarh degree of fineness, are
egorical variables) are those variables which variables.
of continuous are not capable of beingvariables
measured n
examples The discrete (also

eXampla gree of
fineness or exactness because the variables contain a clear
gap. or
membr rs nder of members in a family constitutes the example of a discrete variable. Ine
d n i y may be any number like 5, 6, 7, and so on. No amount of refinement in the
 Bebavioural Sciences
438 Tesls, Measuremenis and Research Methods in

As a
5.7,6.7or6members. general
measuring instrument can produce a value
of determined by countin
whose values can be ng
rule, discrete variables are those variables
family, the number of females in a particlar state tr
(D'Amoto, 1970). The number ofchildren in a
of discrete
are some of the examples
district, the number of books in the library, and so on,
variables. In psychology and education most of the quantitative variables belong to the categor
of continuous variables. It should, however, be noted that the independent variables may be

qualitative, quantitave, continuous or discrete. The same holds

true for the dependent variables
However, in most of the psychological and educational researches the independent variables as
well as dependent variables are quantitative and continuous.

Moderator Variables And Intervening Variables

The moderator variables are special types of independent variables (also called the secondary
independent variables) which are selected by the experimenter because he suspects that these
variables may alter or moderate the relationship between the primary (or main) independent
variable and the dependent variable. Thus, the moderate variables may be defined as those
variables which are manipulated or selected by the experimenter because they are suspected to
moderate the relationship of the independent variable with the dependent variable. For example,
suppose the investigator wants to study the relative effectiveness of the lecture method and the
demonstration method upon classroom achievement. He may take two groups of students of the
same class, one to be taught by the lecture method and another to be taught by the demonstration
method. But he suspects that the level of intelligence may be a factor, which can moderate the
relationship between the two methods (the main independent variables) and the classroom
achievement (the dependent variable). He may, therefore, manipulate intelligence by dividing
bothgroups into three subgroups having high intelligence, average intelligence and low
intelligence. Thus a total of six subgroups would be formed out of which the three subgroups
would be taught by the lecture method and the remaining three subgroups would be taught by the
demonstration method. Subsequently, the classroom achievement of these six subgroups may be
compared with each other. In the above example, the intelligence is manipulated by the
experimenter because he suspects that this factor may alter or moderate the relationship between
the independent variables and the dependent variable. The
intervening variables are the variables
which theoretically exist and tend to influence the behavioural measure.
Such variables cannot
be seen and/or manipulated by the experimenter and their effect can
be inferred from the effects
of the independent variables as well as the moderator variables
upon the variables.
Often such variables are not named in the proposed research and hence, littledependent
attention is given to
them. Suppose the investigator wants to test the
following hypothesis: A person who is allowed to
reach his goal displays less aggressive acts than the
person who is not allowed to reach his goal. n
this hypothesis the DV is the
aggressive act, the IV is being allowed or not being allowed to reacn
the goal, and the
intervening variable is the frustration. Thus the intervening variable is the
variable which is influenced by the independent
variable, the relevant variable and
moderator variable and in turn, affects the dependent variable. The intervening variable can tne
easily identified by carefully examining the hypothesis and asking the
the independent variable that will cause the question, "What is it abou
predicted outcome?" (Tuckman, 178).
Active Variables And Attribute Variables
Avariable which is manipulated by the
IS not
experimenter is the active variable and the variable wn n
manipulated but measured by the experimenter is the attribute variable or
organis
variable. Examples of active variables are reward, punishment, methods of teaching, etc.
the
examples of attribute variables are age, sex, intelligence, race, oles
are human anxiety, etc. These vaed
characteristics, which have already been determined. They cannot be directuy a
 riai's 39

t by the enperimentr. Whe the mvCstiaito Wanls to know whetheet Or

or isare
manipula
stuperior to ten-yoar-old boys in nuett alatility, it serves as an e.mple of
ten-ye-oldegiri ganismic variable on attibute v.un,ll tses p o n the dependent variable (dhe
the study
numericalability).
between the lctive vaiable an the attil tute variable i
The above distinction general and is
Sometimes contusing as well. Ther are some variables, whicihan be calegorized as attribute
variables as well as active ables. Forenample, anxiety is one sturh variable. Anxiety can be
ahipulated by giving a set instruction tothe Subjects. In ihis Case, it becomes an active variable.
Aniety can also be measured with the help ot a scale or lest. In this case, it constitutes the
n
example of an attribute variable. Thus the variable of anxiety can be studied either as an active
variable or as an attribute or organismic variable. Several other examples can be cited in this way.

METHODS OF MEASURING DEPENDENT VARIABLES

tn iact, the researcher selects a dependent variable in the study and attempts to provide an
Qperational definition of it. The dependent variable he will select and the operational definition
he will provide, depends upon the hypothesis and behaviour he is investigating, the subject
population and the perspective that he seeks. For measurement of dependent variables, the
following approaches are generally uncdertaken:
() Direct observation of the targeted behaviour The
direct observation and measurement of the interested behaviour.dependent
variable may involve
For
behaviour, Schachter (1968) measured the amount of food eaten by example,
in a study of eating
the subjects as a function of
diferent eating cues present. Likewise, Cunningham et al., (1990) conducted one study in which
they studied how mood influences our willingness to
cheerful or sad help others. They presented subjects with
statements, then observed whether subjects would
i) Indirect measures of unseen internal
help a confederate.
observed is more an indirect indication of unseenprocesses: Sometimes the behaviour to be
internal The dependent variable
measures an observable response that the researcher thinks is processes.
correlated with the unseen
Theretore, changes in the response are made to make inferences about
process.
processes. For example, several researches have been conducted in which thechanges in the unseen

physiological measurements of heart rate, breathing, blood pressure and researchers take
draw inferences about the subject's anxiety or stress level. perspiration level to
Ihe use of indirect measures is
especially common in studies of young children because they
cannot directly tell us about their experiences. Tronick
measured the length of time a child stared at a stimulus(1989) conducted one study in which he
or noted when he smiled at
posiTIVe emotional state produced by a given stimulus. In it, to infer a
memory research the number of words
correctly or incorrectly recalled or recognized under different conditions is used to inter
processes of retention and forgetting (Craik & Lockhart, 1972). cognitive
AVery common indirect measure is reaction time, the time a
US.On the basis of reaction time to stimuli that differssubject
takes in responding to a
Sion, the researcher uses differences in reaction time to infer alongunderlying
a
physicalcognitive
or mental
or
Onal processes involved therein. For example, Sternberg (1969) measured subjects
a 'probe word' that either was or was not a part of a reactiOn
earch through the previous list, for intering how
sube
informationtheir memory.
in
ahlo gements about a stimulus: In this approach to the measurement of dependent
le the subjects are asked to make judgements about a stimulus and then, it is observed n
how the gements
judgeme change as a function of the conditions. Another such approach utilize
In

the field-Choice technique

Oof perception
perd
in which the subjects select from the possible choices provided.
must
researchers often manipulate characteristics of a visual illusion andmes
Subjects tot Wnether or not they still perceive the illusion (Stuart & Day, 1991). Soe