Page 1
1 1
PHILOSOPHY AND ETHICS OF
PSYCHOLOGICAL RESEARCH
Unit Structure
1.1 Introduction
1.2 Epistemological positions in psychological research: scientific
realism, logical positivism; Ockham's razor
1.2.1 Epistemology
1.2.2 Scientific realism
1.2.3 Logical positivism
1.2.4 Ockham's razor
1.3 Popper and Kuhn‟s contribution: theory dependence of observation;
understanding theory: components and connections – concepts,
constructs, variables and hypothesis; Duhem – Quine thesis; Quine‟s
critique of empiricism
1.3.1 Theory dependence of observation
1.3.2 Understanding theory: components and connections – concepts,
constructs, variables and hypothesis
1.3.3 Duhem –Quine thesis
1.3.4 Quine‟s critique of empiricism
1.4 Ethical standards of psychological research: planning, conduction and
reporting research
1.5 Proposing and reporting quantitative research
1.6 References
1.1 INTRODUCTION The main objective of research in psychology is to find scientific answers
to questions r elated to individual‟s behavior and thought process. As
research in psychology can greatly influence not only the individuals but
also entire society, it is imperative for psychologists to have a skeptical
attitude while planning the research and using emp irical methods to do
research in scientific way.
1.2 EPISTEMOLOGICAL POSITIONS IN PSYCHOLOGICAL RESEARCH: SCIENTIFIC
REALISM, LOGICAL POSITIVISM; OCKHAM'S
RAZOR There is a close connection between philosophy and science. Science
gives scientific answers with evidence, to questions raised by philosophy
and helps to differentiate between knowledge and beliefs. For instance,
one may have a strong belief that air pollution is responsible for lung munotes.in
Page 2
2 Research Methodology for Psychology
2 diseases, but to accept it as a fact, one needs to follow scie ntific processes
and give strong evidences for that, then only it will gain respectability and
become a scientific fact instead of just a philosophical explanation.
1.2.1 Epistemology :
Epistemology is a branch of philosophy that looks into knowledge and
justification. It deals with questions such as how do we define knowledge,
what is the difference between knowledge and mere beliefs? How do we
know that what we have got is actually knowledge? One can say, knowing
about knowledge is of paramount importan ce in epistemology. All of us
have lots of beliefs that we might consider either as true or false. If we
believe any particular piece of information to be false, then it will not be
part of our knowledge. Logically, what it means is that if we say that we
know a particular thing, it means that we believe it to be true. For instance,
if I say Delhi is the capital of India, it means I believe this information to
be true, I cannot believe it to be false and then stating it as fact too.
Suppose a person believe s and states that height of Himalaya mountain is
12000 ft. but on checking, we find this information to be incorrect/ false,
then we can say that this person believed that he knew the height of
Himalaya but in reality, he did not know it. So, knowledge can be equated
with true beliefs.
However, in epistemology, true beliefs need to be justified too before they
become part of knowledge. Knowledge can be defined as justified true
beliefs. For instance, suppose, while playing the „Tambola Housie‟ game,
I believe that next number that will b e announced will help me to win the
number round. After several rounds, finally a number is announced that
helps me to win the game. In such a situation, I had a true belief that a
number will be announced that I need to win „full house‟, but this true
belief cannot be called knowledge, because there was no basis or reason or
justification for my true belief. Many rounds passed when my required
number was not announced. If my true belief was based on any particular
justification, then I would have been able to predict that in which round
my number will be announced. The epistemology believes that for a belief
to be labelled as knowledge, it must fulfill both the conditions – it must be
true as well as there must be sufficient reason or justification for the belief.
If scientific methods are followed to test or to generate the beliefs, then
these scientific methods can become the base for justification of the
beliefs. After all the word science comes from the Latin word scientia,
which means knowledge.
1.2.2 Scientific realism:
There are many things in this world that modern scientists assume to exist
but we cannot see them or sense them without the help of some other aids,
or instance, we cannot see atoms, electromagnetic radiation, etc. With the
help of sc ientific methods, scientists explain that there is reality behind the
way things appear. Scientific realism propagates that people should
believe in things that are not visible but have been assumed by the
scientific theories. Though, the critics ofscienti fic knowledge agree with munotes.in
Page 3
3 Philosophy and Ethics of Psychological Research realists that science is based on paradigm of rational inquiry and it has
progressively contributed to the growth of empirical knowledge, but at the
same time they also believe that scientific knowledge cannot be applied to
each an d every reality and it can be applied only in a limited degree to
certain areas. The critics of scientific realism also argue that whatever is
not visible becomes visible with advanced technology, and many of the
theoretical concepts that were part of past best scientific theories do not
exist anymore, then why there should be a distinction between appearance
and reality.
1.2.3 Logical positivism:
August Comte (1798 -1857) was a French philosopher , he invented the
term „positivism‟ and propagated the ide a that all societies go through
three stages – the theological, the metaphysical and the scientific stage. In
theological stage, people believe that adverse natural events such as
thunder, rain, drought, disease, etc. are caused due to wrath of the God or
spirits or due to magic. In metaphysical stage too, people believe that
adverse events are caused by unobservable elements but in the scientific
stage, people do not try to explain these phenomena, they do not assume
that they know what causes these events , instead, they adopt an inquisitive
attitude and try to understand the scientific causes behind these events, so
thatthey can predict them accurately. Comte propagated that European
society and social relations within it should be scientific studied. He asked
people to discard traditional calendar of Saint‟s Day as well as the
festivals based on religion. Instead he encouraged people to celebrate the
science and the scientists. Hume, another philosopher of that time,
separated the meaningful from meaningl ess and thus positivism originated
from empiricism.
Positivists in general lay great emphasis on:
(a) verification/falsification;
(b) observation/experience as the only source of knowledge (empiricism);
(c) not looking at the causation;
(d) not looking at theoretical entities;
(e) not looking at the explanation;
(f) being anti -metaphysics.
Originally, in 1920s, logical positivism revolved around a group, called
Vienna Circle, of Jewish scientists, mathematicians and philosophers,
some of whom were also so cialists. Later on, this group disbanded and
members migrated to either America or other places as the fascism
became stronger in Nazi Germany. At the same time, their ideas of logical
positivism had a great impact on the development of science as well as
philosophy. munotes.in
Page 4
4 Research Methodology for Psychology
4 Academicians and scientists have been pondering and arguing, for a long
time, to determine the difference between logical positivism and logical
empiricism. There are many famous positivists such as Moritz Schlick
(1882 –1936), Carl Hempel (1905 –1997), Carnap, Reichenbach and Ayer,
who adopted Hume‟s empiricism and Comte‟s dream of having fully
scientific intellectual culture. They also adopted mathematical logic,
developed by GottlobFrege (1848 –1925) and Russell. They wanted to
establish a simpl e connection between ideas and relevant experiences, so
that confusing metaphysical explanations could be avoided.
A fundamental principle of simplicity is Ockham‟s razor. The principle of
Ockam‟s razor states that generate only those assumptions of the fa cts or
entities that are extremely necessary to carry forward the reasoning or
discussion. This type of argument about simplicity is called ontological
parsimony. According to Ockham‟s razor, while comparing two
hypotheses, if everything else is equal, the n we should prefer the simple
hypotheses out of the two hypotheses. This is in sync with Hume‟s
empiricism. He also argued that if two hypotheses are related to things that
we can observe then it can be safely assumed that both of these two
hypotheses are equal.
1.2.4 Ockham's razor
Ockham was not actually the originator of the principle of simplicity. It
was used much before Ockham actually used it. For instance, Durandus of
Saint -Pourçain, a French Dominican theologian and philosopher used it to
discard abstraction as active intellect and considered it to be absolutely not
necessary. Later on, Galileo and other scientists also used the principle of
simplicity in their work.
Ockham was the first one to use the principle of simplicity constantly and
with such precision and intensity that it came to be known Ockham‟s
razor. He used the principle of simplicity to refute relations between things
as he thought that relations are part of foundation of things. He also
discarded the concept of causality as h e considered it to be merely a
succession of things or events. He believed that motion is nothing but
things appearing again at different places at different times. Similarly, he
believed that each sense organ has different psychological power and
when we speak about ideas in the mind of the person who has created
those ideas, we are talking about the person himself which is not distinct
from his ideas.
Scientists use Ockham‟s razor as a rule of a thumb to develop theoretical
model instead of using it to evaluate already published models. For
instance, Albert Einstein used the principle of simplicity as a guiding light
or heuristic, to explain his theory of special relativity. Similarly, Pierre
Louis Maupertuis and Leonhard Euler also developed and applied the
principle of least action based on the simplicity principle, and Max Planck,
Werner Heisenberg and Louis de Broglie used parsimony as a guiding
principle to develop quantum mechanics.
munotes.in
Page 5
5 Philosophy and Ethics of Psychological Research The concept of parsimony in scientific research is adopted to indicate that
there can be only one interpretation of the results and that too in a under
specific conditions. While using parsimony, lot of presumptions ar e made
while planning the study so that no extraneous variable can give any other
alternative explanation for the findings. No two research studies will share
the same tenability of parsimony, as there is no single universal principle
that will cover varie ty of subject matters.
1.3 POPPER AND KUHN’S CONTRIBUTION: THEORY DEPENDENCE OF OBSERVATION; UNDERSTANDING
THEORY: COMPONENTS AND CONNECTIONS –
CONCEPTS, CONSTRUCTS, VARIABLES AND
HYPOTHESIS; DUHEM –QUINE THESIS; QUINE’S
CRITIQUE OF EMPIRICISM In twenti eth century, Karl Popper was an extremely popular influencer,
having many followers, in the field of philosophy of science. He was
offered and he accepted to be a member of a prestigious scientific
association such as Royal Society of London. To begin with , he became
interested in the philosophy of science as he was trying to find the
difference between science and pseudo -science. He appreciated the
theories of physics but felt that theories used in fields such as psychology
and sociology were not scientifi c. He believed that people who mistook
the pseudo -scientific fields such as psychology and sociology as scientific
did not know what exactly made physics as scientific field. He propagated
the concept of falsification and it was whole hearted accepted by scientists
rather than by philosophers. He was instrumental in providing intellectual
criticism of Marxism. Two books authored by him ThePoverty of
Historicism and The Open Society and Its Enemies are popular among
political theorists even in present times .
Popper was very much against induction method and rejected all forms of
induction as a proper method to verify science. Induction method is used
to make generalizations from specific incidents. Popper argued that
science does not need induction. There i s a logical contradiction between
confirmation of universal generalization and falsification.When a
researcher gives many examples that favor generalization from a specific
event, one cannot deny that there is always a possibility of an event
coming up tha t refutes or falsifies that generalization. For example, if a
researcher makes a statement that all birds fly, then just one instance of a
creature that comes under bird category but does not fly is enough to
falsify his hypothesis.
Thus, Popper firmly b elieved that basically the job of science is to falsify
the theories and not to merely confirm them. He believed that theories can
be falsified by using deductive logic and not inductive logic. Therefore,
his theory of scientific method is known as falsifi cationism.
Another contemporary of Popper, Kuhn was a physicist and was interested
in the Copernican revolution and the history of science. Kuhn noticed that munotes.in
Page 6
6 Research Methodology for Psychology
6 most of the books on philosophical and historical work covered
Copernican revolution as a norm. T hese books were strewn with the
difference of opinion between Galileo‟s reason and experiment on one
hand and Catholic Church‟s superstitions and religious orthodoxy. There
was a strong argument in those books that the experimental data that
Galileo and hi s followers have found is totally against the Aristotelian
view of Cosmos.
Kuhn was of the opinion that these arguments in the text book were too
simplistic and Copernican revolution as well as other revolutions in
science were not in sync with either the induction principle nor with
falsification view of scientific method. In his book The Structure of
Scientific Revolutions (1962), he presented a totally different thought
about what is meant by knowledge and what should be considered as
scientific methodol ogy. This brought a paradigm shift in the history of
science.
His brand of philosophy of science became very popular among
academicians belonging to different streams such as literature and
management science. He was also instrumental in popularizing the word
„paradigm‟. He believed that theories can not be evaluated on the basis of
just induction or deductive principle, they should be judged keeping in
mind the local historical occurrences that might have influenced the
development of the theories. Simila rly, he argued that no data that is
collected through observation is totally objective data. A researcher‟s
observations and consequently the data is always influenced by theories.
Consequently, the extent to which an experiment confirms the hypothesis
is also not totally objective. There cannot be any one logical way of
deciding which evidence -based theory should be judged as true. He
explained that all research scientists develop new theories based on their
personal values and the entire scientific commun ity also judges those
theories on the basis of prevalent values of the society at that time. So, it is
not possible to judge theories objectively.
1.3.1 Theory dependence of observation :
As mentioned above, there are some scientists who believe that to
determine how the real world is, one needs to just look at the facts, i.e.,
observe or perform experiments. This is known as theory -based
observation. However, the critics of this idea point out that there is no
such neutral point from where a researcher can make observations about
the real world. Whatever observations are made or conclusions are drawn
from the data, they all are explained in the light of researcher‟s
background such as his prejudices, expectations, beliefs and the same facts
can be observed and interpreted in a substantially different way by
different researchers.
Historians of science have given many examples of instances where
promoters of antagonist theories have given very different interpretations
of the same empirical evidences, in acc ordance with their theoretical
commitments. An interesting illustrative case can be found in a popular munotes.in
Page 7
7 Philosophy and Ethics of Psychological Research drawing called the „duck -rabbit‟, a sketch which can be interpreted as
either a drawing of a duck or of a rabbit, depending on the „theory‟ one
applies i n interpreting the pattern of lines. While in this particular case
both interpretations are equally „correct‟, in many cases scientific and
philosophical disputes, however, many times it is not clear whether one,
both, or none of these different interpretati ons of the relevant facts are
right or no.
One more problem with observation method is that there is lot of observed
factual data and a researcher does not know how to choose and take into
consideration only relevant data. To choose only relevant facts, h e will
have to depend on some theory and there are possibilities that the chosen
theory may be under criticism. For example, a researcher may think that it
is not necessary for him to count every drop of rain water in different
geographical areas. But, if another researcher believes that every drop of
rain water is a blessing from God and mother nature is communicating a
message through these rain drops, then the significance and whole
meaning of these rain water changes. Both researchers would look at the
facts in a different way depending upon which theory they believe in.
How to resolve such problems has been the subject of considerable
philosophical attention, and remains an ongoing problem for any attempt
to provide a comprehensive philosophical underpi nning for scientific
inquiry.
1.3.2 Understanding theory: components and connections – concepts,
constructs, variables and hypothesis :
A theory can be defined as a method that helps a researcher and others to
understand given phenomenon. The basic goal of any theory is to give an
answer to the question „why?‟. For any researcher to learn and develop, it
is essential for him to ask the q uestion „why?‟. This helps him to not only
increase his knowledge of a given subject area, but also helps him to
reorganize his thoughts and opinions.
You must have observed that children are constantly asking the question
„Why‟. For example, children are often asking questions such as
“Why I can‟t eat full box of cookies?”
“Why is cow ruminating?”
“Why lemons are sour?”
“Why does it rain ?”
It can be very tiring for an adult to keep answering such questions and
some times the adult person may not have plausible answer to children‟s
question and may get irritated and ask the child to stop asking such
questions. But one must realize that asking the question „why‟ helps
children to learn, to understand the world and develops their own theories
about why things are the way they are. munotes.in
Page 8
8 Research Methodology for Psychology
8 However, the definition of theory will not be complete by saying that it
answers the question „why‟. Theory includes much more than that, for
instance, it performs following functions too:
It does not merely provide ans wer to the question „why‟ but gives
explanation to enhance the understanding.
A theory is not a general explanation – it is formulated on the basis of
consensus among large number of people about certain ideas and their
relationships.
A theory may n ot be based on facts – how a researcher understands
and explains the given facts depends on his cultural background and
the way he sees the world.
Components:
In order to generate new ideas and new discoveries, a theory must be
testable, coherent, economi cal, generalizable, and must explain the known
findings. If a theory has all these characteristics, then it will be considered
a „good‟ theory.
A theory has two components – The well -defined concepts and principles.
A concept can be defined as a symbolic representation of an actual thing,
for example, train, mountains, rivers, distance, etc. It expresses or
verbalizes an abstraction that is formed through generalizations from
specifics, for example, weight, achievement.
When a concept expresses an abstr action that has no physical referent such
as democracy, learning, happiness, etc. it is called Construct. Since
concept is an expression of an abstraction, it is always in the form of
words, so we can say that language helps in forming the concepts.
Howe ver, a construct is expressed in the form of a word that has been it
has an additional Construct has the added meaning of having been
purposely created or adopted for a special scientific purpose
On the other hand, a principle can be defined as an express ion of the
relationship between two or more concepts or constructs.
While developing a theory, a researcher extracts principles on the basis of
his research about how things or concepts are related.
There are two important functions of Concepts and pr inciples:
1) It is through concepts and principles that we understand what is going
on around us.
2) It is through concepts and principles that we can predict future events
(These predictions can be causal or correlational)
A Problem can be defined as a probing or inquisitive statement about the
relationship between two or more variables.Do teachers‟ comments cause
improvement in student performance? munotes.in
Page 9
9 Philosophy and Ethics of Psychological Research Now let us look at what is a research problem? Research problem can be
defined as any specific issue, difficulty, contradiction, or gap in
knowledge that a researcher wants to investigate in his research. The
research problem can be either practical problems or a theoretical
problem. The goal of a practical research problem is to help in ushering in
the change, and a theoretical problem helps in expanding the knowledge.
Though, generally a research study has either a practical or theoretical
research problem but there is no strict compartmentalization. It can have
both practical and theoretical research problem also. What kind of
research problem a researcher will chose depends on the broad topic of his
interest and the type of research he wants to do.
Now let us look at the importance of a research problem. A researcher will
not be able remain focused if his research problem is not well defined and
he will find his research becoming difficult to complete. He may waste
time and effort in repeating what has been already said by other or he may
be doing research without any goal an d justification. A well -defined
research problem, on the other hand, will guide him about what to pay
attention to and what to ignore related with his broad area of interest. This
sharp focus will help him to complete his research work more efficiently,
adequately and in given time frame. Furthermore, a well -defined research
problem will help a researcher to get new and appropriate insights about
the topic that he wanted to investigate.
So, we can say that having a well -defined research problem is the fir st step
in preparing a research proposal or a research paper or even a thesis. A
well-planned research problem will help a researcher in being very clear
about what he exactly he wants to do and why.
In research, a variable is any characteristics or thin g that varies, i.e., it can
have more than one value. For instance, gender can have two values -
male and female. Some of other examples of variables can be height, age,
temperature, or test scores on a psychological test.
In a research where the research er is trying to find cause -and-effect
relationship, there are two types of variables, independent variable and
dependent variable. An independent variable is the one that is controlled
and manipulated by the researcher. A dependent variable is the one tha t
varies due to the changes taking place in independent variable. We can
say, an independent variable is the cause and the dependent variable is the
effect.
For Example:
Suppose a researcher wants to test whether background music in a room
will have an effect on the math test scores.
His independent variable will be the background music in the room. He
can choose to have two groups of subjects and vary the loudness and the
type of music that he wants to introduce in the room in each trial. munotes.in
Page 10
10 Research Methodology for Psychology
10 Your dependen t variable is math test scores. You measure the math skills
of all participants using a standardized test and check whether they differ
based on back ground music in the room.
A Hypothesis is a speculative or theoretical declarative sentence that states
the relationship between two or more variables. Teachers‟ reinforcement
would have significant impact on students performance.
It is speculative because a researcher will not know what will be the result,
before he/she conducts the study. He will be making a speculative
statement on the basis of curiosity or certain premonitions about the
outcomes of the study. To test out whether his assumptions are true or no,
he will conduct study to gather data and analyze it. Apart from this general
function, there are many other functions p erformed by hypothesis, such as -
(a) It increases the objectivity and clarifies the purpose of a research
work;
(b) It keeps the researcher focused by highlighting what is the specific
scope of his research, i.e., what is to be included and not included in
his research.
(c) It is through hypothesis that a researcher will know from whom what
data to collect dependin g upon the scope and focus of the study
(d) Finally, it helps a researcher to develop a theory so that he can
determine what is what is true and what is not.
Initially, researchers used to use only research hypothesis to carry out their
research but now researcher scientists use two types of hypothesis –
research or null hypothesis and alternate hypothesis. While research or
null hypothesis assumes that there is no relationship or difference between
two variables or constructs, alternate hypothesis clear ly particularizes the
relationship between two variables or constructs. If the analysis of
gathered data indicates that research or null hypothesis cannot be accepted
as true, then alternate hypothesis is considered to be true. One can say that
alternate hypothesis and null hypothesis are opposite of each other.
1.3.3 Duhem –Quine thesis :
Duhem :
Quine thesis is also known as Duhem -Quine problem. This thesis has a
significant place in philosophy of science as it contradicts the most
prevalent views of positi vists. While positivists propagate the idea of
falsification to confirm or reject the hypotheses, Duhem -Quine thesis
propagates the idea of undetermination of theory by evidence. The thesis
theorized that researchers do not experimentally test any single h ypothesis
alone, they always test a bunch of hypotheses. In other words, when a
researcher makes any hypothesis, he has certain beliefs or presumptions
and hypothesis which is also an assumption is based on those
presumptions. For example, if a researcher comes up with the hypothesis
that dark clouds lead to rain, this assumption has a background munotes.in
Page 11
11 Philosophy and Ethics of Psychological Research presumption that clouds become dark in color when they are full of water.
So while empirical test is testing the hypothesis, it is also testing the
presumptions or supplementary hypothesis behind that main hypothesis.
If the observed data does not support any of these hypotheses, the only
thing that they can conclude is that one of these hypotheses is not
supported by the data, they cannot pin point which hypothes is needs to be
changed. In other words, he believed that a single hypothesis cannot be
decisively falsified or confirmed or dropped completely on the basis of
observed data. In other words, the cluster of hypotheses stand or fall
together and cannot be tes ted individually.
This highlighted two things :
a) He shifted the emphasis from testing hypotheses to testing theories.
b) He also emphasized that observed or experimental evidence do not
always lead to generation of unique novel theories.
Duhem held that many different theories can be generated about the world
and its ways from the same observed data. This is especially true for
abstract concepts about the world. In other words, Duhem argued that a
hypothesis cannot be accepted or rejected just on th e basis of some given
experiment or observed data as it may be very restrictive. However,
Duhem was applying his concept of holism predominantly to the field of
physics, and to some of the similar fields such as chemistry, as they have
similar logical stru cture, but he did not include a priori disciplines such as
logic and mathematics, as he believed that they cannot be tested. His idea
of holism and analytic -synthetic distinction makes it very difficult to
assess the match between theory and the real world .
In 1950, W.V.O. Quine presented a paper titled, “Two Dogmas of
Empiricism” in which he openly questioned the idea of analytic -synthetic
distinction. He argued that the thesis can be interpreted in a more
progressive epistemic holism manner.
Quine beli eved that the entire body of human knowledge (he called it „web
of beliefs‟) is one field that is bound by human experiences. He included a
priori disciplines like logic and mathematics too under empirical
investigation. For Quine, holism was a general the ory of meanings. He
looked at the relation between evidence and theory through semantic
prism.
He argued that if empirical evidence are there then even fields like logic
and mathematics can also be revised. He took support of quantum logic to
substantiate this logic. However, later on he disowned this idea as later on
he believed that quantum logic is not based on true values. He further
argued that if evidence do not confirm the given bunch of hypotheses,
then either core beliefs or supplementary belie fs or both can change. He
also pointed out that an empirical evidence can merely confirm or support
a theory, but it cannot say whether the theory is correct or no. munotes.in
Page 12
12 Research Methodology for Psychology
12 1.3.4 Quine’s critique of empiricism :
While presenting his ground -breaking paper “Two Dogma s of
Empiricism” (1951), Quine, first of all, dismissed the idea that logic and
empirical science differ significantly. This was similar to Wittgenstein‟s
idea that logical structure of a language can changes if appropriate
empirical evidences are presente d. Any change in human knowledge has
the potential to change human being‟s most basic and deep -rooted
inferential habits.
The second principle of empiricism, according to Quine, is that science is
nothing but a web of interconnected scientific or empiri cal statements
about given situations and their veracity is tested by the observer‟s
experiences or observations. However, critics pointed out that science
being nothing but a web of scientific sentences is not correct. The
observation often depends upon i nstruments and we cannot say that there
is relationship between the instrument and the phenomenon being
measured. For example, suppose we want to know the weight of a piece of
gold and for that we use an instrument. The instrument may be faulty and
give di fferent readings but we cannot say there is any relationship between
the instrument and the actual status of the piece of gold. The actual
weight of the gold does not change when it is placed on the weighing
scale.
Similarly, Wilfrid Sellars (1912 –89), an American philosopher, also
refuted the “myth of the given”. He argued that there is no inbuilt obvious
facts in our observation of either the world or the mind. The same idea
was propagated by the French philosopher and literary theorist Jacques
Derrida (1930 –2004) when he analysed the “metaphysics of presence”.
They believed that all human knowledge is nothing but the impressions
that we form on the basis of information received through our sense
organs.
Quine was of the opinion that language has no d istinct meaning as it has
no clearly established logical attributes and no direct relationship to
experience.
He went on to argue that He argued that, since there are no a priori
standards to find out whether two words have same meaning or no. In
fact, in philosophy, the very idea of meaning is doubted.
To prove his point, he described a thought experiment rela ted with
“radical translation”. He said suppose a linguist has to translate a
completely unknown language without taking any help from bilinguals or
other informants. Then to understand the vocabulary, structure, grammar
of an alien language, the only meth od that a translator can use is to
constantly look out for links between the events that are taking place in an
unknown environment and what the people in that environment are saying,
till he finds the pattern and can decipher the vocabulary and grammar of
that language. If this method is used by two translators, there is a
possibility of them developing two altogether different translation manuals
for that language, based on equally strong but different evidences. munotes.in
Page 13
13 Philosophy and Ethics of Psychological Research This will happen due to the fact that wor ds do not have well defined
meanings, that is a unique or specific content belonging to each word.
Quine said that in language, there are no well -defined “fact of the matter”
words.
1.4 ETHICAL STANDARDS OF PSYCHOLOGICAL RESEARCH: PLANNING, CONDUCTION AND
REPORTING RESEARCH Research in any stream of knowledge is not free from its answerability.
While on one hand, the credibility of research in psychology depends on
the scientific methods used, it is equally important for a researcher to
follow ethical gu idelines established by the world recognized bodies in
their field. For instance, Institutional Review Board (IRB) has laid down
ethical guidelines to be followed for psychological research about the
rights and welfare of human participants, Institutional Animal Care and
Use Committees (IACUCs) checks the research conducted with animals to
ensure that animals are treated humanely during the research.
The researcher must keep these ethical guidelines in mind even at the
planning stage of the research. There can be serious repercussions for not
following the ethical guidelines, e.g., it may hinder the advancement of
knowledge, gradually destroy the credibility and respect for scientists and
academicians, there can be legal and financial repercussions too.
Risk/benefit ratio:
A researcher needs to make a judgment about the possible risks and
benefits of a research project. If potential benefits of contributing to
scientific knowledge are more than the risk, then the research can be
conducted with appropriate constraints.
Some of the examples of potential risks in psychological research are
risk of physical harm, social harm, and mental or emotional stress.
To avoid or minimize social harm to the participants, participants can
be asked to give information anonymously, or if it is impossible to
have it anonymously, then the confidentiality of their information
should be maintained.
Informed Consent:
Informed consent procedure is a social contract between researchers
and participants that takes place befor e the actual study begins.
One of the ethical obligations of researchers is that they must clearly
describe the research procedures, and answer any queries or doubts
that potential participants may have about the research.
The potential research part icipants must be informed beforehand that
they can withdraw their consent at any time without any
repercussions. munotes.in
Page 14
14 Research Methodology for Psychology
14 There should be no direct or indirect compulsion on the potential
participants to participate in research.
Deception in psychological rese arch:
In psychological research, deception takes place when researchers
intentionally do not give full information or misinform participants
about the research. Naturally, such an act of a researcher is against the
ethical principle of informed consent.
Generally, deception is undesirable but in some of the research studies
in psychology, it becomes an imperative research strategy.
Debriefing:
Immediately after the research study, it is imperative for the
researcher to give detailed information to th e participants about what
was the research about and what was their role in the study and also to
tell them about the research process. The main objective of debriefing
is to make sure that individuals feel good about their participation in
the research.
In case researchers have used deception, they are ethically bound to
inform the participants, that the deception was used and why it was
used, immediately after the research study is over or as soon as
possible.
Debriefing not only informs the particip ants about the real nature of
the research study, but also gives a chance to researchers to learn
about participants‟ views about the research procedures, and gives
them potential insights into the nature of the research findings and
gives them ideas for f uture research.
Research with animal:
In some of the psychological research studies, animals are used to
gain knowledge that will benefit humans, for instance, a research
procedure may involve giving shock or investigating the effect of a
new drug.
In such studies, researchers are ethically bound to get animals legally,
to care for them and use them humanely, and later on dispose them
according to the local laws and regulations of the land as well as
according to the professional standards.
Wheth er it is ethical to use animals for gaining knowledge that will
benefit humans but causes lot of agony and even death to animals is
matter of heated debate and there is no straightforward answer to that.
It is a complex issue, having both pros and cons.
munotes.in
Page 15
15 Philosophy and Ethics of Psychological Research Reporting of psychological research:
The APA Code of Ethics gives guidelines for researchers to
communicate their research findings in peer reviewed scientific
journals.
Based on the scholarly importance of the contribution, there are clear
guidelines about who should get the credit for publication.
If researchers are using others‟ research, it is necessary for them to
acknowledge it in their research by using proper citations and
references, otherwise it will be considered a case of plagiarism, and
that will be violation of ethical codes. Plagiarism will involve legal
difficulties too apart from losing the credibility of the research.
1.5 PROPOSING AND REPORTING QUANTITATIVE RESEARCH Before starting a research study, the researcher has to prese nt a research
proposal to the research guide or supervisor, and to the research committee
at the university for their scrutiny anddecide whether that proposed
research is good enough to allot the degree, whether that proposed
research can be done with the given time and resources, as well as whether
it will meet scientific guidelines.
In a way, the researcher has to sell his research idea to the concerned
authorities of the university. In case, he is also seeking funds for his
research, then he needs to co nvince the funding agency too. If the research
committee does not approve the research proposal, then either the
researcher will have to submit a fresh proposal or rework and again submit
the rejected proposal on the basis of comments given by the committe e.
This of course will be time consuming and costly and may even dampen
the spirits of the researcher. Therefore, it is necessary to know what are
the salient features of the research proposal. A good research proposal
should include the proposed answer t o the following questions :
WHAT – Your research topic :
The first thing that you need to clearly and specifically mention in your
research proposal is what is the research topic. In other words, it should
clearly state what exactly you plan to do in your research and what is its
context.
For instance, suppose the research topic is the factors that may lead to
suicidal ideation in adolescents in India. Here the context is India
What‟s being investigated – factors that may lead to suicidal ideation
Who it involves – adolescents
In what context – Indian munotes.in
Page 16
16 Research Methodology for Psychology
16 The research proposal should give an exhaustive detail of the research
topic.
WHY – Your justification :
Research proposal should not only have a clear research topic but should
also include the rationale for doing that research or significance of that
topic. The researcher must give justification for choosing that particular
topic. He must be able to convinc e that this topic is not researched before,
that it is original. On the basis of review of literature, he should be able to
identify the gaps in the existing literature about that topic and explain how
his research is going to fill that gap. In other word s, he should be able to
explain how his research will provide significant inputs to the existing
knowledge or solve some unsolved problem and therefore will be value
creating.
For example, if a researcher is looking the possible factors contributing to
suicidal ideation in adolescent population and he can identify some of the
factors that were previously unknown or not related with suicidal ideation,
then his research will be making a significant contribution to the existing
body of knowledge related to su icidal ideation. This research may help in
framing government policies too. This theoretical and practical
contribution of the study becomes the justification for the research.
HOW – Your methodology :
Apart from convincing the research authorities about t he originality and
importance of your topic you also need to briefly explain the tentative
methodology that you will use to do that research. In other words, you
need to answer the questions such as –
What will be your research design to investigate your topic? Will you be
doing quantitative or qualitative research, longitudinal or cross -sectional
research?
Which sampling technique you will use and why? What will be the sample
size?
What will be the scope of your research study?
What will be method of collecting the data – survey method, interview
method, observation or any other?
You will also need to specify, how will analyse the data after it is
collected? Which statistical analysis you propose to use to test your
hypotheses?
Will you be adhering to all scientific and ethical guidelines?
Do you have the expertise and other resources to do that research and how
will you complete it in given time frame? munotes.in
Page 17
17 Philosophy and Ethics of Psychological Research It is obvious that to answer such questions, the researcher needs to be
aware of different types of research methodologies that are available. He
must also have a good knowledge of statistical tools, if he is planning to
do quantitative research.
Reporting quantitative research :
A quantitative analysis of the gathered data provides the information not
only about whether the researcher‟s assumptions were statistically
significant or no but it also provides enough information that helps in
making decisions about policy and planning for a program or organization.
It is very important for a researcher t o know how to write a good
quantitative analysis,irrespective of whether he is doing research for
publishing a research paper or for research thesis. A good qualitative
analysis ensures that data gathered is of good quality and the conclusions
drawn are ba sed on scientific principles. To ensure the quality of the data,
it is important that it is collected by using one of the well -established
methods, such as survey method, and to draw authoritative conclusions
from the data, the researcher must have good kn owledge of statistics.
Step 1 :
In the beginning of the report, first of all the researcher needs to highlight
why the report is being written. He must indicate what was the lacuna or
gap in the previous studies, that his study is filling up and what more can
be done in future beyond his study. He must clearly mention who are the
targeted readers of his report,that is, for whom it is prepared.
Step 2 :
Secondly, he needs to describe how the data was collected, that is, which
method was used and how that met hod was designed. Describe in detail all
the steps taken for data collection. For example, did he use survey method
or observation or interview method or anything else. If he has used survey
method, then whether he used a standardized questionnaire or made one
himself. If he has made it, what was the procedure adopted for
determining the reliability and validity. Had he conducted any pilot study
before collecting the actual data? What was the population from which
sample was taken? Which sampling technique was used and what was the
sample size? Depending upon the hypotheses, data should be subdivided
into relevant categories such as age, gender, socioeconomic status, etc. to
test the hypotheses. If there are any biases in gathering the data, or if there
have been some extreme responses, or certain data is missing, it should be
reported in this section.
Step 3 :
The next step is to report results by means of visual representation. It can
be in the form of tables and graphs. One can use histogram, pie charts, etc.
for describing the data as well as for reporting the conclusions from the
data. In result section, it is advisable to report only the most important
table and graphs that the researcher wants to high light. Rest of the results munotes.in
Page 18
18 Research Methodology for Psychology
18 can be put in appendices . For the sake of convenience of the reader, the
appendices should also include a blank form of the questionnaire used to
collect the data as well as the raw data.
Step 4 :
In the next section of the report, the researcher needs to write what
conclusions have been drawn from the results reported in previous section.
On the basis of results, the researcher may even suggest if further research
should be done to have more autho ritative conclusions. While writing
conclusions, the researcher should restrict himself to only those
conclusions that can be made on the basis of the findings.
Step 5 :
Lastly, make an executive summary of two pages. In this executive
summary very briefly repeat what the report contains, for example, write
concisely what was there in introduction, research design, sample, data
collection and analyzing methods, findings and their interpretations.
Executive summaries give a glimpse of what is there in the res earch
report.
1.6 REFERENCES Shaughnessy, J. J., Zechmeister, E. B. &Zechmeister, J. (2012).
Research methods in psychology. (9th ed..). NY: McGraw Hill.
Elmes, D. G. (2011). Research Methods in Psychology (9thed.).
Wadsworth Publishing.
Goodw in, J. (2009). Research in Psychology: Methods in Design
(6thed.). Wiley.
McBurney, D. H. (2009). Research methods. (8th Ed.). Wadsworth
Publishing.
***** munotes.in
Page 19
19 2
RESEARCH SETTINGS AND ME THODS
OF DATA COLLECTION
Unit Structure
2.1 Introduction
2. 2 Observation and Interview method
2.2.1 Observation
2.2.2 Interviews
2.3 Questionnaire
2.4 Survey research
2.5 Other non -experimental methods
2.6 References
2.1 INTRODUCTION To test his research hypotheses or questions, a researcher needs to collect
data with the help of an instrument. We can define data collection as a
method of collecting, measuring and analysing information by using
standardized validated tec hniques to get precise sagacity for research. The
main goal of data collection is to secure reliable data that can undergo
statistical analysis and yet is rich in information.
There are various different methods of collecting data and different
researcher s use different methods to collect data on the basis of their
research topics, the kind of information needed forhis work, sample (
children, adults or old people, patients or healthy people, etc.) the
instrument or measurement to be used (e.g., questionna ire, test,
observation, interview, case study, etc.), for data collection. It will also
depend upon the time line, resources available, the expertise or skill of the
researcher and the ethical requirements of the study. A researcher may
decide to collect data through online means such as an online survey via
Qualtrics or survey monkey, or he may decide to conduct experiments
online using Inquisit or Open Sesame. No matter which method is used to
collect data, one cannot deny that data collection is an imp ortant part of
any kind of research, quantitative or qualitative.
The researcher needs to pay attention to not only the method of data
collection but also to the research setting or environment in which the
study is carried out. This research environment i ncludes physical, social,
and cultural aspects and can significantly impact the data collection and
the interpretation of the data. For instance, a qualitative research will be
carried out in the natural environment of the participant as the researcher
is more interested in finding out the environmental factors that make
meaning for the participant, while quantitative study can be carried out in
either natural or artificial environment, e.g., field work or experimental munotes.in
Page 20
20 Research Methodology for Psychology
20 laboratory. The validity and the gener alizability of the study gets impacted
by where the research study was conducted.
Some of the common data collection methods are observation (direct and
participant), interviews, surveys, archival data, and tests. Each of these
methods have both advantages and disadvantages. To improve the
accuracy and veracity of theresults and their interpretations, it is advisable
to use a combination of many different methods of data collection. This
combination of different data collection methods is called triangulati on.
The next question that is often asked is how many times the data should be
collected.
How often the data should be collected:
How often to collect the data will depend upon the research topic and the
frequency of its occurrence in the participants’ lives. For instance, if a
researcher is interested to know the work profile of a teacher and if is
using observation method, then he needs to observe different teachers
throughout the year and still may miss on some of the details. For a topic
like this, it is better for a researcher to use survey method instead of an
observation method. To use survey method, he will need to have an
accurate relevant standardized instrument and he will need to ensure that
he administers this instrument to a representative sample of that
population.
One can get a representative sample through:
Time Sampling: It is an observational technique. It is used to evaluate how
many times and for how much time a particular behavior occurs in
different groups or individuals, in a spec ified time period. For instance,
how many times and for how long the violent behavior occurs in 5th
standard students in a single day and are there any gender differences in
their violent behavior.
Situation Sampling:
Situation sampling is used to enhance the external validity or
generalizability of a study. The researcher observes people under different
situations and under different geographical locations. This type of
sampling is not concerned about the fixed or predetermined time interval.
In fact, the particular behavior under study may be occurring infrequently
and randomly.
2.2 OBSERVATION AND INTERVIEW METHOD 2.2.1 Observation :
The observational method comes under the umbrella of descriptive
research and it allows a researcher to watch and record th e specific
targeted behavior of participants. It is different from survey method as
experimenter does not administer any questionnaire, and it is different
from experimental method, as the researcher cannot control any of the munotes.in
Page 21
21 Research Settings and Methods of Data Collection variables. Usually, observatio n method comes under the preview of
qualitative research.
Observation method can be used in various ways :
Naturalistic observation and structured observation, covert or disguised
and undisguised or open observation.
Naturalistic Observation:
Naturalistic observation can be of two types - observation without
intervention and observation with intervention.
Observation without Intervention :
The naturalistic observation is used with the aim of just describing the
natural behavior as it normally occurs.
The ext ernal validity of the research studies using naturalistic
observation is higher compare to those studies that are conducted in
laboratory settings.
In some of the research areas, it is ethically not possible to control or
manipulate certain variables. In s uch cases, using observation to study
naturally occurring behavior is the best way to do the research. For
example, areas like juvenile delinquency, child abuse, etc. Another
example can be the work of Jane Goodall, a primatologist, who
observed and record ed the behavior of Chimpanzees in a national park
in Tanzania. She observed their social behavior, family relationships,
tool making skills, etc. This study became a milestone in the field of
anthropology.
Observation with Intervention :
However, there are many areas of research where researchers can use
observation with intervention.
The observation with intervention can be done in three different
ways :
a) participant observation,
b) structured observation,
c) the field experiments.
Participative observa tion:
A researcher using participative observation method will completely
merge himself with the members of the targeted sample. He will adopt the
life style, culture and may be profession too. This type of observation
takes place in natural setting and th e researcher has no control and cannot
manipulate any variable. The behavior of the participants will be their
natural behavior, taking place in natural context, and they may or may not munotes.in
Page 22
22 Research Methodology for Psychology
22 be aware of being observed. This research method is also known as cove rt
or disguised observation method. The researcher observes and records
their behavior in a nonobtrusive way. The recording of the behavior takes
place after the behavior has already occurred or whenever the researcher
gets a chance to record it. In such c ases, the threat of personal bias or
forgetting some of the details, is very high but the threat of reactivit y in
participants’ behavior (i. e., participants’ natural behavior changing as they
become aware of being under observation and have desire to prese nt
themselves in socially desirable manner) is very low.
Structured Observation :
The structured observation method is generally used by clinical and
developmental psychologists. It is also known as systematic observation
method.
It is a non -participativ e method :
The researcher does not join the sample group as a participant. The
participants are aware that they are being observed and yet there is no fear
of reactivity, that is, their natural behavior does not change. The researcher
identifies the specif ic behavior(s) that he is going to observe and the
behavior can be recorded while it is naturally occurring. This also cuts
down, to some extent, the problem of researcher’s own biases or memory
interfering with the recording of the behavior. This method i s especially
useful with sample who cannot articulate their thoughts or emotions. He
has some control over the setting and the event that he is observing,
though the degree of control is not as much as in case of field experiments.
He records the occurren ce of the behavior in terms of either the frequency
of the occurrence of that behavior or when that behavior occurs.
For example, the researcher is a doctor who is studying the influence of
different dosages of a particular drug. The person on whom the dr ug is
administered is the subject of the experiment and is aware that his
behavior is under observation. Since the researcher is actively controlling
and manipulating the dosage of the drug and observing the changes taking
place due to this intervention, w e can say that researcher structures the
situation to observe and record behavior more effectively.
However, structured interview can also take place without the knowledge
of the subject. For example, Asch studied the impact of group pressure on
conformit y behavior. He used confederates to build the unspoken group
pressure on the actual research subject and observed his conformity
behavior. This experiment was not conducted in natural setting. It was in a
laboratory setting.
The Field Experiment:
In a fi eld experiment, researchers can manoeuvre independent variables in
a natural setting and observe its’ effect on behavior. The experiment takes
place in natural setting and the participants are not aware of being part of
research study and under observation . As there is no fear of reactivity, munotes.in
Page 23
23 Research Settings and Methods of Data Collection such experiments have higher external validity. The Stanford Prison
experiment in 1971 is a classic example of the field experiment.
Observation method has many advantages and disadvantages both. Let us
see them in det ail:
Advantages of Observational Studies :
1. The researcher gets the first -hand information about people’s specific,
naturally occurring behavior. He can see if any patterns are emerging
in their natural behavior, that will help him to make his research
questions or hypotheses more specific. As we have already discussed
before, the clearer the hypotheses are, more they will help in deciding
what data to collect and how to interpret. So, observation method
can be used as a part of exploratory study.
2. Since this method can be used at initial stage of the research study, it
also allows a researcher to make on the spot changes in the objectives
of the study, if needs.
3. Unlike in other research methods, it is possible to study the natural
behavior of th e subjects without plagued with any biases or reactivity
of the subjects. The researcher has to be merely conscious of his own
biases and ensure that they don’t contaminate his observations and
interpretations.
4. Since data is collected from natural set tings, the external validity or
generalizability of the research study will be robust.
5. There are certain areas of interest to psychologists that they can study
ethically only by using observation method in natural settings and not
in lab settings. If they use any other method, they will be violating the
ethical guidelines of doing research and their study will be considered
invalid. For example, one cannot study domestic violence in lab
settings.
6. Observation method is very useful to study cross -cultural differences.
Disadvantages of Observational Studies :
1. Compare to other methods, observation method is very time
consuming and costly and it requires more man power. If there
searcher decides to observe during a specific time period, there is no
guarantee that targeted behavior will occur during that time period. If
the researcher does not predetermine the time period of observation
and observe the behavior as and when it occurs, it may take place
after a long gap.
2. Generally, large sample siz e cannot be taken while using observation
method. As large quantitative data cannot be generated by using this
method, it is more suitable for qualitative or exploratory study only. munotes.in
Page 24
24 Research Methodology for Psychology
24 3. Since, experimenter is not controlling any variable, one cannot
establ ish cause and effect relationship by using observation method.
4. If the participants are aware that they are being observed, they may
not behave in their natural manner and reactivity or Hawthorne effect
may take place. To avoid the chances of Hawthorne effect occurring,
the researcher should use observation method to collect the data in
natural setting, in an unobtrusive manner, where participants are not
aware that they are being observed. For example, observing children
while playing in a garden.
Howev er, the disadvantage of observing in natural setting is that the
researcher has no control over the environment or any of the variables. For
example, suppose a researcher goes to a garden in the evening to watch
children’s prosocial behavior, but he finds that most of the children have
already left the garden for their homes as they all have to appear for class
tests the very next day. This extraneous variable ( class test on the very
next day)will affect his data as he will be able to observe only those
children who do not have class test next day. These children may be from
younger age group compare to those whom he wanted to observe.
5. Another major problem can be of observer’s personal biases. The
chances of observer’s bias influencing what will be obs erved and
recorded and how it will be interpreted are very high. Some people
suggest the use of multiple observers to overcome this problem. But
in such cases too, subjectivity cannot be eliminated. In fact, different
observers may give different interpret ation for the same piece of
behavior of a person.
6. In case of disguised participative observation in natural setting raises
the concern about the ethics of the research. It is not ethical to record
anyone’s behavior without taking his/her consent.
To overcome this problem, if one decides to do overt observation where
the participants are aware of being observed, the possibility of reactivity
cannot be dismissed.
2.2.2 Interviews :
Interview is a type of qualitative data collection method in which the
researcher asks questions to find out either the factual information or the
thoughts, feelings, values, experiences, meanings, etc. from the
interviewee. This method can be used either as a substitute or as a
supplementary to other data collection methods. Though, usually,
interview is done face to face, now technology facilitates it over phone or
through video conferencing too. There are various types of interviews.
Some of them are discussed here -
Structured Interview:
It is more like an oral questionnai re, where same questions, in same
sequence, with same multiple -choice answers to choose from, are asked to munotes.in
Page 25
25 Research Settings and Methods of Data Collection all respondents. The interviewer cannot change the sequence of the
questions asked, omit or change the wordings of any question, if he finds
that the interviewee is not comfortable with any question. It is possible for
an untrained person also to conduct this type of interview.
Semi -structured Interview:
The interviewer has a list of predetermined areas and few related
questions, to be covered during the interview but he can add or delete any
question depending upon the answers given to previous questions. The
respondent is not given any multiple -choice answers to choose from. The
interviewer does not ask questions in any predetermined sequence. In fa ct,
depending upon the comfort level of the respondent, the interviewer
decides impromptu what question should be asked and which area of
interest should be covered first. He has to only ensure that no area of
interest is left out for any of the respondent s, but the sequence of areas to
be covered can vary depending upon the respondent’s comfort. This
adaptability of the interviewer ensures that he gets in depth and complete
information from the respondent.
Unstructured Interview:
The interviewer does no t have any list of predetermined questions to be
asked. He has only a general idea of the areas to be covered and not the
specific detailed areas to be covered. The interviewer goes with the flow
and allows the interviewee to talk about whatever he wants t o talk. The
interview is in the form of a conversation, where both the interviewee and
the interviewer are free to ask as many questions to each other as they
want.
Technology Assisted Interviews :
Now days due to advanced communication technology and inter net it is
possible to conduct interviews either on phone or through video
conferencing sites such as zoom, google meet, skype, etc. While interview
through landline phone is not so satisfying and has its limitations,
interview through video conferencing ha s become very popular in last one
decade or so. It has the advantage that both interviewer and interviewee
need not spend time and money in travelling to meet each other face to
face. The interviewer need not incur the cost of finding a quiet place to
cond uct the interview. Such type of interviews can be recorded so that
supervision of such interviews as well as data analysis becomes easier. It
is also possible to access respondents from remote areas or places which
are not easily reachable, e.g., remand ho mes, jails, brothels, etc.
However, this advantage also becomes the disadvantage in the form of
serious sampling error. One cannot use random sampling method to
conduct this type of interviews. One can conduct such interviews only
with those who have acce ss to internet. Especially in poor and developing
countries, internet penetration is very poor and availability of electricity at
all times is also not guaranteed. So, a large population can not be accessed
for this type of interviews. munotes.in
Page 26
26 Research Methodology for Psychology
26 In this type of int erview, establishing rapport with introvert respondents is
also a challenge and one cannot observe the entire body language. There is
also the time pressure and such interviews cannot be as relaxed as face to
face interviews.
Advantages of Interview Method :
1. Compared to other methods of survey, this method is more flexible.
2. It is ideal to get much more information than what the researcher had
originally planned for and it can give lot of insights and context into
the problems, attitudes, values and l ives in general of the respondents.
3. Interview method provides the opportunity to clarify, during the
interview session itself, any doubts, misunderstandings, queries, etc.
that either the interviewer or interviewee may have.
4. It is possible for a s killed interviewer to establish rapport and put the
interviewee at ease, and talk about themselves. At the same time, the
interviewer can ensure that focus of the interview session remains on
the main objectives with which the interview session was initia ted.
5. In other survey methods, the chances of respondents not responding at
all or returning half -filled questionnaires, etc. is very high. In
interview method, the response rate is very high. So, if the researcher
needs to collect vast in -depth data from a small group of people,
interview method is the best.
6. It is also useful when the respondents have language barrier, have
limited reading writing ability or have limited capacity to articulate
written answers to open ended questions.
7. The inte rviewer can collect both verbal as well as nonverbal data in
interview, which is not possible in other methods. Through non -verbal
language, the researcher can also gauge whether the interviewee is
giving accurate information or falsifying. If the intervie wer feels that
the respondent is trying to avoid the question or giving false or
contradictory information, then he can probe further in that area and
get to know the real issues of the problem. This will also ensure the
accuracy of the data.
Disadvantages of Interview Method:
1. Compare to other methods of conducting survey, interview method is
more time consuming and costly. It requires a lot more time in
preparing for the interview as it requires in conducting the interview.
2. This method cannot be u sed if the sample size required is very large.
3. The effectiveness of interview method depends upon the skills of the
interviewer. Personal biases of an interviewer can influence what
information will be sought and how it will be sought from the
responde nts. In case of panel interview, where there is more than one munotes.in
Page 27
27 Research Settings and Methods of Data Collection interviewer, the problem gets further aggravated as each interviewer
will ask questions and interpret the answers in light of his own
personal biases. One can circumvent this problem by using in ter-
interviewer reliability, but it reduces the flexibility of the interview.
4. If an interviewer is not trained and is using unstructured form of
interview, then inadvertently, he may not cover all requisite areas of
interest that he was supposed to co ver, with all the respondents.
Another problem that an untrained interviewer may face is that he
may develop an empathy for the interviewee during the interview and
will not be able to do objective analysis of the data. Moreover,
interviewer must have good communication skills.
2.3 QUESTIONNAIRE The questionnaire is a scientific tool to get data from a large sample, and
not merely a list of questions. A researcher needs to do lot of planning
and pretesting to make an effective questionnaire, as faulty ins trument will
lead to faulty data and that will compromise the validity of the study. A
questionnaire can be in verbal or pictorial form and it can contain both
open ended or close ended questions. The respondent can answer in short
paragraph answer to open ended questions and close ended questions can
be answered by selecting an option from either binary form (e.g. yes/No)
or from multiple choices (e.g. Likert type Scale) or fill in the blanks. This
list is not exhaustive, there can be many more options tha t a researcher can
offer to the respondent. The questionnaire can have both open ended and
close ended type of questions. It can be univariate, bivariate or
multivariate type of questionnaire. A questionnaire collects various types
of data, e.g. :
Factual data, that is mostly demographic data that can be verified by other
means too.
Data related to cognitive factors, i.e., what people think and how they
make decisions, their attitudes and opinions, etc.
Data related to affective factors, i.e., the feelin gs and preferences of the
people
Data related to behavior, i.e., what people do or intend to do in a given
situation.
It can be administered either individually or in a group, either face to face
or electronically or by post. If it is a self -administered q uestionnaire,
which generally the questionnaires are, then it can be administered by
either the researcher himself or by his assistants or by data collection
agency. A well planned questionnaire should ensure that the questions that
are easy to answer and do not tax memory too much, are interesting, are
crucial for the study should be asked first. The questions that require lot of
thinking and taxes the memory, are embarrassing or difficult to answer,
are boring should be asked at the end. The questionnaire and the length of
the questions should not be too long. The questions should be prepared munotes.in
Page 28
28 Research Methodology for Psychology
28 keeping in mind the age, literacy rate and culture of the respondents in
mind.
Advantages of Questionnaire:
1. It can be used to collect data from a large sample at the same time, so
it is less time consuming and less expensive.
2. Since it is not necessary to administer it face to face and it can be sent
by post or can be uploaded on social media sites or sent by email, data
can be collected even from those who ar e geographically at far off
places. In fact, cross national studies also can be done by using the
questionnaire as a tool for data collection.
3. It is possible to maintain anonymity of the participants, if they so
desire.
4. A respond can take his ow n time to understand the question and think
about the answer before answering it, which is not possible in
interview method.
5. A well planned and constructed questionnaire helps us to code and
statistically analyze the data easily and scientifically.
Disadvantages of Questionnaire:
1. If the questionnaire is not well planned, it may contain personal biases
of the researcher. A faulty questionnaire may contain :
a) Leading questions, where he inadvertently gives a cue to the
respondent about the answer t hat the researcher wants him to give.
b) Ambiguous questions that can be interpreted in more than one way,
leading to either confusion in the mind of the respondent or
respondent may interpret it differently then what the researcher
intended.
c) Loaded questions where a respondent finds choosing any option as an
answer is embarrassing. For example, have you stopped stealing?
Choosing either yes or no both can be embarrassing for a person.
d) Hypothetical questions, ego boosting questions, double barre l
questions can all put a question mark on the useful ness of the data.
e) There is no guarantee that respondents will give honest answers and
their answers will not be contaminated by social desirability factor or
demand characteristic. Social desirabil ity takes place when the
respondent wants to present himself in socially acceptable way to the
researcher. He wants researcher to consider him to be good. The
problem of ‘demand characteristic’ takes place when the respondent
gives the answer that he think s the researcher wants to hear, instead of
expressing his true opinions, attitudes, or feelings, etc. munotes.in
Page 29
29 Research Settings and Methods of Data Collection f) When questionnaires are administered in group or by post or
electronically, the chances of either low response rate or getting half -
filled forms becom e very high. The researcher has to discard these
half-filled forms and then using questionnaire method can become
costly and time consuming. He will have to either get new
respondents or contact the same respondents and request them to
complete the forms. Very low response rate creates another question
in mind. The researcher has to think, do people who filled up the form
and sent back are different from those who did not bother to fill up or
filled it half and sent. In other words, there can be sampling er ror and
may reduce the generalizability of the results.
Guidelines for writing good Questions:
By now you must have realized that lot of care needs to be taken to write
effective questions for a questionnaire. Let us see some of the guidelines
for making g ood questionnaire.
1. Pay attention to the language:
You need to pay attention to the vocabulary, grammar and connotation of
the words.
One should use simple words that are understood by majority of the
people and avoid using any technical jargons, acron yms or culturally alien
words for the respondents. Avoid using ambiguous words that can be
interpreted in multiple ways. connotation of the words can also change
the meaning of the question and influence the way people respond to
them.
Words having simil ar meaning but different connotations in questions can
change the flavor of the question and impact the choice that a respondent
makes for answering it.
For instance, “Do you think Ravimust give test?”, “Do you think
Ravicangive test?” or “Do you think Rav imay givetest?”.
As you can see here, the word must indicates a compulsion, can indicates
the capacity of the management but not the possibility and may indicates
the possibility. This slight change can lead to different responses. Similar
result differ ences were observed when strong words such as prohibit was
used in the questions. Such strong words give the impression of control.
Avoid Ambiguous questions?
Suppose a researcher wants to study the popularity of certain types of
food. In the questionnaire , a question is asked “Do you like South Indian
food?” This is an ambiguous or vague question. The respondent will be
confused and wondering on what parameter I should be judging and
replying to this question. The questions do not specify South Indian food
from which South Indian state he should consider and whether he should munotes.in
Page 30
30 Research Methodology for Psychology
30 respond keeping in mind dishes served at meal time or he should consider
the snacks. He should indicate his liking on the basis of taste, smell,
texture or monetary value.
Avoid leadin g questions :
Avoid framing a question that puts mental pressure on the respondent or
gives him a cue about what the researcher wants the answer to be. For
example, “ All students like mathematics, do you ?”
Provide Context to the questions :
Wherever necessa ry, the context or reference should be provided to the
questions. One can use the technique of filtering and branching to give
context to the questions.
Multiple choice questions:
If you are asking multiple choice questions make sure that answers are in
mutually exclusive categories.
Avoid asking loaded questions:
Don’t ask questions that can encroach upon the privacy of the person (e.g.
questions related to their salary, religion, caste, etc.) or can be
embarrassing and emotionally inconvenient for them to answer (e.g., have
you stopped stealing?)
Unbalanced answer options in scales :
Depending upon the topic to be covered in the question, one should decide
whether to use Gutman Scale, Likert Scaleor Semantic Scale or any other
type of scale. If the scale type is not chosen properly it may force a
respondent to choose an option that does not reflect his true answer. For
example, suppose a researcher wants to study the eating habits of obese vs.
non-obese people on a Likert type of scale. He provides a five point scale
where one extreme option indicates that people starve themselves and the
other extreme option shows that people consume abnormal amount of
food, with a neutral middle indicating that they neither starve nor consume
abnormal amount of food. It has been found that people generally choose
the neutral option since they can’t find the right response option that truly
represents them. Unbalancing in the scale takes place when there are two
negative and one positive and one neutral option is given to choose from.
Avoid Double Barrel Questions :
Avoid asking about two variables in one question. For example, a manager
of a restaurant asking a customer, “ did you like the food and the service of
the restaurant?” It becomes difficult for a customer to answe r that
questions as he may have liked the food but not the service.
munotes.in
Page 31
31 Research Settings and Methods of Data Collection Avoid the use of long questions :
Long questions require more effort to understand and answer the question.
That demotivates the respondents. They may leave such questions or pick
up an o ption from response category without comprehending the question.
2.4 SURVEY RESEARCH One of the methods used for data collection in descriptive research is the
survey method. It is a very popular research method among social
scientists. In most cases, to conduct a survey, the researcher uses a self -
administered questionnaire to collect the data. As explained above, the
questionnaire can have both open ended and close ended questions with
various types of options to respond to those questions. Survey metho d can
be structured or non -structured. The structured survey method uses
questionnaire having predominantly close ended questions where the
answers are to be given on a forced choice type of scale. On the other
hand, unstructured or non -structured survey c an be the one that uses
questionnaire with lots of open ended questions. It is better to use
structured survey rather than the non -structured survey. In a structured
survey, it is easier to code every answer and statistically analyze the data.
In case of o pen-ended survey, though, it is possible to get lots of rich data
that gives insight into respondent’s way of thinking, behavior etc. , but this
data is not easy to code and statistically analyze. Open ended type of
questions are more suitable for qualita tive research rather than quantitative
research.
Survey method can be used in experimental research also. For example,
suppose a researcher wants to study the impact of mood on prosocial
behavior. He conducts the experiment in laboratory by taking a rando m
sample of college students. He divides them into two groups. One group is
experimental group and the other is control group. Students in the
experimental group are asked to recall all the negative things that have
happened to them during the past one wee k while the control group is just
asked to read a book.
After one hour, both the groups are asked to take a questionnaire and
indicate how likely they are to help another person who needs help. In this
experiment, the mood of the experimental group is man ipulated by asking
them to think of negative things and thus creating a negative mood in
them. While no such manipulations are done in control group. And yet
their prosocial behavior which is a dependent variable is measured through
survey method.
There a re various other types of survey method. Some of them are
discussed here.
Cross -Sectional Surveys :
In the cross -sectional survey, the questionnaire is administered only once
across various types of the sample from a given population. For example,
suppose I want to find out the mathematical ability of class fifth students. I munotes.in
Page 32
32 Research Methodology for Psychology
32 will administer themathematical ability test to the entire fifth class
students, having both boys and girls. After wards I can compare boys’
proficiency in mathematical ability with girl ’s mathematical ability. Thus,
a cross sectional study can find the difference between two segments of
the same population or it can be correlational study, that tries to find the
relationship between two variables. However, it is important to underline
once again that it collects the data only once. So, the biggest disadvantage
of this method is that by the time the results are published, the targeted
population’s attitudes, values, preferences might have changed due to
some sudden changes in the society.
Longitudinal Survey :
Longitudinal survey overcomes the problem of collecting data only
once. In longitudinal study, same set of respondents are studied over a
long period of time, sometimes lasting over years, and the changes in
their behavior are noted do wn.
There are various types of longitudinal studies, e.g., trend survey,
panel survey and cohort survey. In trend survey, the researcher tries to
find out whether there are any changes in the values or preferences of
the people. However, in trend survey, some of the original
respondents may drop out and same people may not be aware year
after year. Contrary to that, in panel survey, same people participate
in the survey year after year till the survey lasts. So, panel survey is
costlier and more difficult to carry out than trend survey, as the
researcher has to keep track of his sample over the years. In cohort
study, the criteria for taking sample is that people born in same year
or same generation people are taken as sample. For example, people
born imme diately after world war II. It is not necessary that researcher
has to take same people year after year, he has to merely ensure that
they should be born in the same year.
Mail Surveys :
In mail survey, the questionnaire is sent by post with a self -addresse d
envelope, so that the respondent can send back the filled up questionnaire.
Advantage of mail survey is that one can reach out to people even in
remote areas.
Disadvantage: The disadvantage is that it is very costly and time
consuming method and the resp onse rate is also very low. The potential
respondents either do not reply at all or they may send half -filled forms.
Phone Surveys :
Advantages: Compare to mail survey, this method is quicker and less
costly. It also generates higher response rate. The pr oblem of half filled
forms is also avoided as the researcher is asking questions on phone.
Disadvantages: Though it is better than mail survey method, but certain
problems still remain with this method too. First of all this method can be munotes.in
Page 33
33 Research Settings and Methods of Data Collection used only if a re spondent has a phone. Even if he has a phone, fixing up an
appointment when he will be willing to answer the questions becomes a
problem. The respondent may have difficulty in understanding the
questions spoken on phone and the researcher may have to repea t it many
times. The researcher’s voice or tone may reveal his personal biases. So
this method becomes costly and time consuming. The researcher can’t
take a large sample and will have to limit the number of questions that can
be asked. The respondent may refuse to answer embarrassing questions.
Web/Online Surveys :
Web or online survey can be done either by sending questionnaire in email
or through video conferencing devices such as zoom or google meet.
Advantages:
This method is still better than phone survey. Sending questionnaire
through email is faster and less costly than sending through mail.
There is no need for the researcher and the respondent to coordinate their
timings to talk to each other unless they are using video conferencing.
The respon dent can read the questions in email and take his own time to
understand the question and decide on his answer. It is less troublesome
for him also to send the filled questionnaire back to the researcher.
If the researcher is using google forms, then the a dded advantage is that he
can download it in excel sheet and analysing the data becomes easier.
In case of video conferencing, the respondent can understand the questions
much easily as he can see the researcher and his facial expressions. The
researcher can clarify any doubts or confusions that a respondent might
have while answering the questions.
Disadvantages:
Similar to phone survey difficulty, in case of web or online survey too, the
first condition is that the respondents must have either a laptop , desktop or
internet enabled mobile phone. Moreover, they must have good internet
connectivity. In some of the geographical areas, having electricity, good
internet connectivity might be a challenge. People belonging to poor class
may not have laptop or e ven internet enabled mobiles. In such cases, the
possibility of sampling error cannot be ruled out.
The questionnaire sent on email may go in spam and the respondent may
not be aware of it. The researcher needs to do the follow up and that may
become time consuming and push up the cost of doing research.
Multi -Mode Surveys :
When different modes of Survey method are used to collect the data and
the respondents’ responses are combined together to analyse the data, it is
called multi -mode or mixed mode survey method. This method helps in
reducing the sampling error as those who do not have access to internet or munotes.in
Page 34
34 Research Methodology for Psychology
34 do not respond can be contacted by using a different mode of survey
method. However, the impact of using different mode of collecting data
cannot be i gnored.
Advantages of Survey Method:
Since survey is mostly conducted by using the self -reported
questionnaires, it is possible to collect the data quickly from a large
sample. So it is less time consuming and less costly than experimental
method. Compared to experimental method, use of survey method is much
easier.
Surveys method can be used to study variety of topics, evens those topics
that cannot be studied through experimental method either due to ethical
constraints or due to lack of resources.
It is easy to analyze the data collected through survey method than many
other methods.
Surveys method is reliable as it follows the principles of scientific
research. The tool to be sued, the sample, the objectives and hypotheses
are well thought of .
Disadva ntages of Survey Method:
Since surveys are conducted through self -reported questionnaires, we do
not know whether the respondents have answered the questions accurately
or no. Some people may not give honest answers or may not answer all
the questions, be cause of their social desirability factor or due to demand
characteristic. They may want to project themselves in good light in front
of the researcher, or they find certain questions very embarrassing to
answer, or due to lack of knowledge or capacity to answer those questions
or they may be unable to recall the information at that moment when they
are filling up the form. Their responses can also get affected by their level
of motivation.
The data collected through survey method may be faulty if the tool used to
collect the data is faulty. For instance, if the researcher has constructed a
questionnaire and if that questionnaire has faulty format, wordings of the
questions, wrong scaling, wrong placement of the questions in the
questionnaire, not conducted the pilot study, reliability and validity has not
been established, then obviously the data collected through the use of this
instrument will be faulty. "Reliability" of an instrument indicates how
much there will be consistency in the answers given by t he respondents, if
the same questionnaire is administered again and again.
Validity of an instrument indicates whether it is measuring what it
intended to measure. Both reliability and validity are in the form of
degrees and not in absolute terms.
It is observed that questionnaires having more of close ended questions
have lower reliability than those which have balanced mixture of open
ended and close ended questions. Especially if the questionnaire is used to munotes.in
Page 35
35 Research Settings and Methods of Data Collection measure emotions or feelings. It has also be en observed that survey
method proves to be inadequate for studying complex social issues.
One cannot establish cause and effect relationship through survey method
as the researcher is not controlling any variable.
Other non -experimental methods :
Sometimes in social sciences it is not possible, either due to ethical
considerations or due to lack of resources, to use the principles of
randomization and control of variables. In such cases researchers use non -
experimental methods. Some of the non -experimental methods are survey
method, and observation method, case study method, etc. Data collected
through these non -experimental methods or quasi experimental methods
are more difficult to analyze and interpret than the data from experimental
method.
Case study M ethod:
Case study method is a systematic and scientific procedure for observing
or examining a phenomenon related to any specific event or person or
organization within its real -life context. Case study method can be used
for a single person as a subject ( e.g. in clinical settings) or it can study a
group of people or events or organization (e.g. success story of Maruti
car). Case study method can rely upon administration of standardized
scales, observation, interviews, etc. In other words, it combines mult iple
methods to collect the data. The data can be numerical as well as
qualitative.
A researcher can gain lot of insights and understand in a better way why
certain event took place or why a person behaves in a specific manner. On
the basis of this enhance d understanding, he can judge what should be the
future course of action to do research in that particular field.
Case studies can be instrumental in both generating and testing of
hypotheses. As mentioned before, initially case study method was
predominan tly used by clinical psychology but now it is used by other
branches of psychology too. Clinical psychologists firmly believe that to
understand a person’s physical and mental health, it is very important to
know his past and present history of health as w ell as about his past and
present social, physical and economic environment.
Apart from primary data, secondary data can also be a rich source of
information and insight for a researcher. One can collect secondary data
from books, personal sources, journa ls, newspapers, websites, government
records etc. There are many fields of social sciences that depend either
entirely on secondary data or secondary data plays a major part in research
in those fields, e.g., research in history, politics, economics, etc. It is
easier, less expensive and quicker to collect secondary data than primary
data. The researcher either does not need or requires minimal help from
others to collect such data. One of the sources of secondary data are
archival records. Let us see it in some detail. munotes.in
Page 36
36 Research Methodology for Psychology
36 Archival Records :
Archival records are generally the running records of the specific events
that have taken place or are taking place in public domain and these
records have some permanent value. For example, they may be documents
having info rmation about historical events, information about introduction
of some new laws or change of laws of the land. It may be record of
criminals and their past history of conviction, academic records of
students, etc. These records can be in the form of news paper articles,
government files, on official websites, in micro films, etc.
Archival records can be used to test hypotheses about human behavior in
natural setting. Though observation method also aims to study human
behavior in its natural setting, but ar chival records have an advantage that
it can give us information about natural human behavior that may have
taken place long back and in any part of the world. The researcher need
not be present in that era and at that place to study that specific human
behavior. Since behavior is studied in natural settings and unobtrusively, it
has high ecological validity. Apart from testing the hypotheses, it can also
help in generating the new hypotheses.
The advantage of this method is that it is quicker and cost ef fective. The
researcher can access the data of many people at the click of his mouse.
The disadvantage is that the data recorded was according to the
hypotheses of the researcher. It was collected by someone else for
different purposes. The researcher wil l have to see which data has
maximum relevance to his hypotheses.
Secondly, if there is already some inherent fault in the archival records,
then the researcher has no way of correcting it or even knowing it and will
have to use those faulty archival reco rds only. That may reduce the
reliability of his study. For example, suppose a researcher aims to use
archival records to find out the relationship between age of the participants
and their preferences for various political parties. The researcher decides
to use data available on Facebook. There is no guarantee that people have
put their real age or political leanings on the social media site. Similarly,
if a person wants to know the financial health of an organization and looks
at the records available on that organization’s website, there is no
guarantee that figures put there are the true ones. In such cases, the
researcher will have to seek same information from many different sources
to see whether there are any contradictions in the information given in
different records. If the contradictions are glaring, he cannot use that
information.
2.6 REFERENCES Shaughnessy, J. J., Zechmeister, E. B. &Zechmeister, J. (2012).
Research methods in psychology. (9th ed..). NY: McGraw Hill.
Elmes, D. G. (2011). Re search Methods in Psychology (9thed.).
Wadsworth Publishing. munotes.in
Page 37
37 Research Settings and Methods of Data Collection Goodwin, J. (2009). Research in Psychology: Methods in Design
(6thed.). Wiley.
McBurney, D. H. (2009). Research methods. (8th Ed.). Wadsworth
Publishing.
Forrester, M. A. (2010). Doing Qualita tive Research in Psychology: A
Practical Guide. Sage.
*****
munotes.in
Page 38
38 3
EXPERIMENTAL AND QUASI -
EXPERIMENTAL METHODS
Unit Structure
3.1 Introduction
3.1.1 Why Psychologists Conduct Experiments
3.1.2 Logic of experimental research
3.2 Independent groups design
3.2.1 Random groups design
3.2.2 Block Randomization
3.2.3 Threat s to Internal Validity
3.2.4 The Role of Data Analysis in Experiments
3.2.5 Establishing the external validity of experimental findings
3.3 Repeated measures designs
3.3.1 The role of practice effects in repeated measures designs
3.3.2 Balancing Pract ice Effects in the Complete Design
3.3.3 Balancing Practice Effects in the Incomplete Design
3.3.4 Data analysis of repeated measures designs: Describing the
results
3.3.5 The problem of differential transfer
3.4 Complex designs
3.4.1 Complex Designs with Three Independent Variables
3.4.2 Describing effects in a complex design: Main Effects and
Interaction Effects
3.4.3 Analysis of Complex Designs
3.5 Interpreting Interaction Effects
3.5.1 Interaction Effects and Theory Testing
3.5.2 Interaction Effe cts and External Validity
3.5.3 Interaction Effects and Ceiling and Floor Effects
3.5.4 Interaction Effects and the Natural Groups Design
3.6 True experiments
3.6.1 Characteristics of true experiments
3.6.2 Obstacles to conducting true experiments in natu ral settings
3.6.3 Threats to internal validity controlled by true experiments
3.7 Quasi -experiments design and program evaluation
3.7.1 The non -equivalent control group design
3.7.2 Sources of invalidity in the non -equivalent control Group
design munotes.in
Page 39
39 Experimental and Quasi-Experimental Methods 3.7.3 The issue of external validity Interrupted time -series designs
and Time series with non -equivalent control group
3.7.4 Program evaluation
3.8 Summary
3.9 Questions
3.10 References
3.1 INTRODUCTION we have seen that survey method can be used to describe p eople’s
attitudes and behavior. However, it does not establish cause and effect
relationship. It does not tell us why people have specific attitudes that they
have. Getting an answer to the ‘why’ of behavior is essential to make
predictions about the futur e behavior. Both description and prediction are
two of the goals of psychology. To determine causal relationship,
psychologists need to use experimental designs. In this unit, we will study
how do psychologists use experimental method. How groups are forme d
in independent group design research. We will also discuss how to
establish external validity of the experiment.
We will discuss the experiments that involve more than one independent
variable in one experiment only. Such designs are called complex des igns
or factorial designs. This design allows us to find out the main effect, that
is effect of each independent variable as well as the interaction or
combined effect of the independent variables. In last unit we touched upon
the difficulties in establish ing cause and effect relationships through
experimental method. In this unit we will elaborate further on those
difficulties, especially when experiments are conducted in natural settings.
We will also discuss how to overcome these difficulties by using qu asi
experiments. Researchers also try to determine the effectiveness of
changes made by government agencies and other organizations. This is
called program evaluation. We will briefly discuss the procedure and
limitations of program evaluation.
3.1.1 Why Psychologists Conduct Experiments :
One of the main reasons for conducting experiments is to establish the
cause and effect relationship between two or more variables. Researchers
first of all make hypotheses from existing psychological theories and then
empirically test those hypotheses to validate the assumed cause and effect
relationships between variables under study. For example, Pennebaker
et.al. (1989) developed a hypothesis that suppressed feelings about a
painful experience can lead to physical tol l. They derived this hypothesis
from ‘inhibition theory’. To empirically test this hypothesis, they used
experimental method, in which all participants were divided into two
groups. One group was asked to write down about personal emotional
events while th e other group was asked to write down about superficial
topics. The results showed that the group who wrote about personal
emotional events had better health later on than the other group that had
written about superficial events. However, in another versi on of the same munotes.in
Page 40
40 Research Methodology for Psychology
40 experiment, researchers divided the group into two groups – one group
was asked to dance expressively about an emotional experience while
another group was asked to write as well as dance about their personal
emotional experiences. It was fo und that the group that wrote and danced
both, had better health results then the first group that had only danced.
These two versions of the experiment led them to believe that there is a
cause and effect relationship between expressing one’s emotions and the
health outcomes of a person. If the results validate the hypotheses, the
theory is accepted, otherwise new hypothesis needs to be formed and
again tested through another experiment.
Apart from validation of the psychological theories, experimental me thod
is used to find out the effectiveness of treatments in various areas of
medicine and psychology.
3.1.2 Logic of experimental research :
As mentioned above, experimental method allows a researcher to firmly
infer causal relationship between independent variable and dependent
variable. This is made possible due to the use of experimental control. An
experimenter exerts experimental control through either manipulation or
holding conditions constant, or through balancing. Three conditions are
required to m ake a causal inference. These are – covariation, time -order
relationship, and elimination of plausible alternative causes.
Covariation :
As the name suggests, covariation means there is relationship between
independent and dependent variables. They change or vary
simultaneously.
Time order relationship :
A time order relationship takes place when independent variable is
manipulated first and then the subsequent changes in behavior are
observed. In other words, we can say that the change in behavior is
contingent on the manipulation of independent variable.
Elimination of plausible alternative causes :
Means applying the control procedures to ensure that no other factor than
independent variable is the cause of change in dependent variable. This
control ca n be achieved through holding conditions constant and
balancing.
If these three conditions are met, experiment will have high internal
validity and we will be able to say firmly that independent variable caused
the changes in dependent variable.
3.2 INDE PENDENT GROUPS DESIGNS In experimental method, to determine the cause and effect relationship,
two groups of participants are taken. In one group independent variable is munotes.in
Page 41
41 Experimental and Quasi-Experimental Methods introduced or manipulated while in another group, independent variable is
not introdu ced. Then the impact of this manipulation is measured or
observed on behavior. The measure used to record the change in behavior
is called dependent variable. An independent variable must have at least
two levels or conditions, e.g., exists/ does not exist , etc. One level will be
considered ‘treatment’ condition while the other level will be considered
as ‘control condition’. An independent variable can have more than two
levels too for additional comparisons between groups. This is known as
independent g roup design
3.2.1 Random Group Design:
Random group design is part of independent group design. In independent
group design, each group of participants is exposed to only one condition
or level of independent variable. If there are two groups of participa nts, it
is necessary to ensure that they are comparable. This can be done by
balancing or averaging the characteristics of participants across the
groups. This averaging can be done by randomly assigning participants to
any of the groups. This will make al l groups comparable or similar on all
important characteristics before the experiment begins. This is known as
random group design. Once the comparable groups are formed, then
independent variable is introduced in one of the groups and it is assumed
that a ny difference between the groups on the dependent variable must be
due to the introduction of independent variable.
Between Subject Design is another name for independent group design. In
either case, the basic principle remains same. Either two or more g roups of
participants are compared. Groups are similar or comparable but none of
the participant will be in more than one group.
Manipulation of an independent variable satisfies three conditions that are
necessary to firmly establish cause and effect rel ationships. These are :
1. Difference in measures of dependent variable covaries with the
changes or manipulation of independent variable.
2. The change in dependent variable takes place after the change in
independent variable condition (time order relatio nship)
3. By holding conditions constant and balancing, alternative
explanations for changes in dependent variables are eliminated.
Holding conditions constant ensures that the only factor that changes
systematically is the independent variable and nothing else. If
independent variable under study and a potential independent variable
are allowed to covary, confounding condition takes place and that
threatens the internal validity of the experiment.
However, it is important to keep in mind that researcher c annot hold
constant all possible covariants of independent variables. He will keep
constant only those factors that he thinks can be the plausible alternative
causes. But an experimenter should constantly keep looking for such munotes.in
Page 42
42 Research Methodology for Psychology
42 alternative possible factors that he had not considered or anticipated and
that may influence the outcome of his experiment.
Balancing :
The experimenter needs to use balancing as a control technique before
introducing the independent variable. Very often if the groups formed are
not equivalent groups, individual differences can become the confounding
variable that undermines the internal validity of the experiment. The
groups can be balanced by using random group design.
3.2.2 Block Randomization :
Block randomization is also known as randomized block design. It helps
the experimenter to balance the participants’ characteristics and other
confounding factors that may occur during experimentation. It also helps
to ensure a balance in sample size across groups over time. Block
randomizat ion is better than simple randomization because simple
randomization cannot rule out the possibility of known or unknown
confounders bringing severe imbalances in sample allocation. This method
is especially useful when sample size is small. Now let us se e how block
randomization is done.
In block randomization we form groups where participants are similar, so
that they can be compared with each other. Suppose we want to study the
difference in the effectiveness of online teaching and traditional method o f
teaching on class fifth students. We first create two homogeneous blocks
of students. In both the blocks we have students from fifth standard, both
male and female students and having same level of intelligence. Let us say
that our sample size is 60. So there are 30 students in each block. Now in
each block we randomly expose 15 students to online teaching and 15
students to traditional method of teaching. Let us take another example,
suppose there are five levels of independent variable, viz., A, B, C, D, E.
If the researcher wants to have 10 participants for each condition. Totally,
there will be 50 participants. There will be 10 blocks and each block will
consist of a random assignment of the five conditions.
Advantages of block randomization :
1. It produces groups of equal sizes. Having groups of equal size is very
important because the number of observations in each group affects
the reliability of the descriptive statistics for each group
2. It controls for time related variables. If an experiment ta kes a long
time to complete, the chances are very high that subjects may get
influenced by events that may take place while the experiment is still
going on. In block randomization, since every level or condition of
independent variable is tested in each b lock, these time related
variables are balanced across the conditions of the experiment. Time
related factors can be a traumatic event, change in the experimenter or
even change in the population from which the sample was taken. munotes.in
Page 43
43 Experimental and Quasi-Experimental Methods Block randomization will av erage out any characteristics of
participants.
3. It increases internal validity by balancing extraneous variables across
conditions of the independent variable.
3.2.3 Threats to Internal Validity :
By definition, internal validity refers to the degree to which difference in
performance on a dependent variable can be clearly attributed to the
effects of independent variable and not to the uncontrolled variables.
Uncontrolled variables are the alternative explanations for the results
obtained for a study. The uncontrolled variables are the threats to internal
validity of an experiment. Let us see what are these uncontrolled factors
and how to control them.
a) Testing Intact Groups :
Sometimes, in spite of using random assignment, comparable groups are
not formed. This problem comes when intact groups are formed at the start
of an experiment and randomly assigned to various conditions of an
experiment. In noncomparable groups, the confounding takes place when
individuals differ systematically across the intac t groups. For example, in
third year BA, while choosing courses, sometimes students choose
subjects on the basis of who will be the teacher, subjects taken by friends,
how easy it appears, how scoring it is, etc. Consequently, they will be put
in different divisions of the same class. For example, Div. A will have all
psychology students, Div. B will have all economics students, and so on.
If an experimenter randomly assigns different divisions to different levels
of independent variable, a confounding due to testing intact groups may
take place. Students may systematically differ across the divisions or
intact groups.
b) Balancing Extraneous Variables :
Potential variables that experimenter has not planned to study but these
variables can still influence th e outcome of the experiment are called
extraneous or confounding variables. For example, if all the participants in
the experimental group are tested by one experimenter and by another
experimenter in the control group, the levels of the intended independe nt
variable would become confounded with the two experimenters. Let us
take another example, Evans and Donnerstein (1974) found that students
who are willing to participate in the experiment at the beginning of the
academic term are the ones who are more academically oriented and have
internal locus of control, while those students who volunteered to
participate in the experiment in the later part of the academic term, were
those who were not academically oriented and had external locus of
control. The dif ference in participants’ characteristics would be a
confounding variable. munotes.in
Page 44
44 Research Methodology for Psychology
44 Psychologists suggested that block randomization method can be used to
balance extraneous variables across groups.
c) Subject Loss :
Internal validity of an experiment gets compromis ed when participants in
an experiment take part in the beginning of the experiment but do not
remain in the experiment till the end of the experiment. The subject loss
can be of two types – mechanical subject loss and selective subject loss.
d) Mechanical subject loss :
When a subject fails to complete an experiment due to an equipment
failure (and here we consider experimenter too as part of the instrument) it
is called mechanical subject loss. For example, malfunctioning of an
instrument, computer crash, experimenter giving wrong instructions or
someone inadvertently disrupting an experimental session, etc.
Mechanical loss is not as grave as selective subject loss, as it does not lead
to systematic differences between the characteristics of the subjects wh o
successfully complete the experiment. However, whenever mechanical
subject loss takes place, it should be documented along with the reason for
the loss. Then the lost subjects should be replaced with other tested
subjects.
e) Subjective loss takes place when :
(1) subjects are lost differentially across the conditions of the experiment;
(2) any particular characteristics of the subject is responsible for the loss;
(3) this specific characteristic of the subject is related to the dependent
variable of the study.
If the subject loss is selective, the groups can’t be compared. But the basic
reason for using random group design in experiments is to have
comparable groups. In such a situation it is not possible to have any
reliable results from the experim ent. Let us take an example to understand
selective subject loss. Suppose a gym instructor wants to test the
effectiveness of a one -month fitness training program. He gets total 80
volunteers for this experiment. He randomly divides them into two groups
of 40 each. He made sure that characteristics of the participants like
weight, fitness level, age, gender, motivation are same in both the groups.
Thus, both the groups are comparable at the initial stage of the experiment.
Participants in the experimental group start with the one -month fitness
training program, while the participants in control group continue with
their normal fitness routine work. At the end of the month, suppose only
38 participants in control group and 25 participants in experimental gr oup
remain in the experiment. On comparison, the experimenter finds that
average fitness score of 25 participants in experimental group is much
higher than the average fitness score of 38 participants in control group. It
will be wrong on the part of gym i nstructor to claim that his one -month
fitness training program has been effective. The selective subject loss has munotes.in
Page 45
45 Experimental and Quasi-Experimental Methods occurred, especially in experimental group. Out of 40 participants in
experimental group, only 25 continued till the end of the month. The
remaining 15 participants may have dropped out because they were less
fit, compared to the other 25 in that group, even before the program began.
Another potential possibility of difference in the fitness scores of
experimental and control group can be that 2 5 experimental participants
might have been more fit than control group and may have scored more
than them without training program too. Thus, the selective loss of
participants in the experimental group has most likely destroyed the
comparable groups that were formed by random assignment at the
beginning of the experiment. In other words, the characteristic of the
participants, i.e., their original fitness, became a confounding variable. To
prevent the possibility of subjective loss, the experimenter shoul d have
screened the participants through pretest, kept only those who were similar
in fitness and then randomly assigned them to either experimental group
or control group.
f) Placebo Control and Double -Blind Experiments :
Both experimenter and participant s come to the experiment with certain
expectations and these expectations can lead to certain biases that can be a
threat to internal validity of the experiment.
Demand characteristic is one such bias that originates from participants’
expectations. Parti cipants look out for cues or any other information to get
an idea about what kind of behavior is expected from them during the
experiment. For example, if a participant is given chocolate and he thinks
that experimenter expects him to feel happy about it, then he will behave
consistent with those expectations, irrespective of his real feelings.
g) Experimenter effect is another such bias that may compromise the
internal validity of the experiment. An experimenter may
unknowingly treat subjects differently in different groups. He may
treat subjects in experimental group in a biased manner to get
response that validates his hypotheses. For example, in an
experimenter wanted to test the effect of alcohol on cognitive and
motor functions of the subject. He div ided subjects in two comparable
groups. One group (experimental group) was given alcohol and the
other group (control group) was given plain water. The experimenter
read the instructions to the experimental group more slowly than to
the other group, thus c reating an experimenter bias. Furthermore,
experimenter effect took place again when the experimenter keenly
observed the subjects in experimental group for any unusual motor
movement or slurred speech as he was expecting that kind of behavior
from them,
Though it is not possible to completely eliminate demand characteristics
and experimenter bias but they can be controlled by using certain
techniques. The demand characteristic can be eliminated by using placebo
control group. The subjects can be divided into experimental and control
group. An independent variable such as alcohol or some drug can be given munotes.in
Page 46
46 Research Methodology for Psychology
46 to them while the control group can be given something which looks like
drug or alcohol but is actually inactive or plain substance. Subjects in both
groups have same awareness of taking alcohol or drug and similar
expectations or demand characteristics. Any difference in the behavior of
these groups can be attributed to independent variable. However, there can
be an ethical issue while using placebo contr ol group, especially if the
experiment is about testing the effect of any drug to treat an illness. If
patients benefit from the new drug that was introduced as independent
variable than those in placebo control group also would be expecting
similar benefi t. To overcome this problem, experimenters take written
informed consent from the subjects that they might get either new drug or
placebo. If the new drug proves to be effective, then those who got
placebo also will be given that new drug.
To overcome e xperimenter bias, double blind procedure can be used. In
this technique, both experimenter and the participants are unaware about
which group has been exposed to independent variable. In double blind
procedure, there will be two experimenters conducting th e experiment.
One experimenter will code the independent variable and control variable
separately and administer it to the groups. The other experimenter will be
the observer of the behavior and will not know in which group
independent variable has been in troduced, Therefore, the observer
experimenter will treat both the groups in the same manner and there will
be no experimenter effect.
3.2.4 The Role of Data Analysis in Experiments :
Data analysis is a very crucial part of any experiment. Without proper d ata
analysis, an experimenter cannot establish cause and effect relationship
between independent and dependent variable with surety. Robert Abelson
(1995) said that the basic purpose of data analysis is to determine whether
obtained data supports the assum ptions made in the hypotheses. One of the
best ways to find out the reliability of our results is to replicate the
experiment. Replication means repeating the experiment with same
variables, similar sample, using the same procedures and under similar
condi tions. If we get the same results in replicated study too, that indicates
that our previous results are reliable. However, it is nearly impossible to
have exact replication of the original experiment as the subjects and
external conditions of the experimen t will be different. Another problem
with replication is that it will be very cumbersome, costly and time
consuming to establish the reliability of each and every experiment
through replication method. As an alternative to replication, researchers
can use data analysis and statistics for determining whether the results of a
single experiment are reliable and can be used to state that independent
variable does have an impact on behavior.
Describing the Results :
Data analysis is done in three stages:
(1) get ting to know the data : this involves scoring, coding, inspecting the
data, removing errors and cleaning the data munotes.in
Page 47
47 Experimental and Quasi-Experimental Methods (2) summarizing the data : this includes using descriptive statistics such
as mean and standard deviation, percentages
(3) confirming what the data reveal : this includes testing the
hypotheses through inferential statistics.
One of the major concerns the researcher has is how much or significant
the effect of independent variable is on dependent variable. This question
can be answered by measure s of effect size. Measure of effect size is not
influenced by the sample size and indicates the strength of relationship
between the independent and dependent variables. One of the common
measures of effect size is Cohen’s d. He suggested that d values of .20,
.50, and .80 to indicate small, medium, and large effects of the
independent variable, respectively. Measure of effect size is also used in
another statistical technique called meta -analysis.
Meta -analysis is a statistical technique that helps to summ ate the effect
sizes from various independent experiments studying the same
independent or dependent variable. Meta -analysis is used to get answers to
questions like: Are there gender differences in ….? What is the effects of
class size on XYZ…...?Is solut ion-based therapy effective in the treatment
of …...? Meta analyses gives an efficient and effective way to summarize
the results of large numbers of experiments using effect -size measures.
Confirming What the Results Reveal :
As mentioned above, inferenti al statistics is used to confirm whether
independent variable has significant effect on dependent variable. There
are two methods of making these inferences – null hypothesis testing and
confidence interval.
A statistically significant difference between t he two groups indicates that
null hypothesis is not true and independent variable does have an impact
on dependent variable. In other words, it indicates that the difference
between the two groups is more than the difference expected due to error
variation . Error variation refers to non -systematic or random variation or
chance factor variation due to differences among subjects within each
group. The objective is to have as less error variation as possible. Though
it can’t be completely eliminated. When we d o null hypothesis testing
through inferential statistics, we are indirectly trying to determine what is
the likelihood or probability of our results occurring if null hypothesis was
true. It is well accepted in research that results with probabilities (p) of
less than 5 times out of 100 are judged to be statistically significant. The
probability value chosen by the researcher to determine that an outcome is
statistically significant is called the level of significance and is denoted by
Greek letter alpha (α ). The obtained value at alpha level or higher than that
is considered significant.
What Data Analysis Can’t Tell Us :
As already mentioned above, our data analysis cannot tell for sure whether
our independent variable has an effect on dependent variable. when an munotes.in
Page 48
48 Research Methodology for Psychology
48 outcome is not statistically significant, we cannot conclude with certainty
that the independent variable did not have an effect. We can only conclude
is there is not sufficient evidence in the experiment to claim that the
independent variable prod uces an effect. Secondly, data analysis cannot
tell us whether our experiment was meaningful or had any practical value.
There are two types of errors that a researcher can commit while using
inferential statistics. Type I error takes place when null hypot hesis is
wrongly rejected while it is true, and type II error takes place when null
hypothesis is accepted as true even when it is false
3.2.5 Establishing the external validity of experimental findings :
External validity refers to the extent to which find ings from a research
study can be generalized to individuals, settings, and conditions beyond
the scope of the specific study. Compared to experiments conducted in
laboratory, field experiments will have higher external validity as they will
be closer to r eal world setting. The researchers want to and should strive
for higher external validity of their studies as higher external validity
means higher generalizability. However, whether the researcher will
emphasize on internal validity or external validity d epends on the topic of
research. If the researcher is testing any existing theory, he may emphasize
on internal validity and if he is conducting a research in real world setting
to find solutions to a problem. Though, to increase external validity,
psychol ogists try to mimic real life situations while designing an
experiment in laboratory conditions, it is not always possible or ethically
permissible to mimic the real -life situation. For example, Ceci (1993)
conducted an experiment to test the factors that may affect the reliability
of eyewitness testimony in an assault case, but he cannot create actual
assault in laboratory as it is ethically not correct. Another factor that may
raise doubts about the external validity is the sample. Most of the
experiments in laboratory conditions are conducted with college students
as subjects. The question arises, do students represent the general
population, and if yes, then how much is this representation a true
representation.
One of the ways, psychologists can determ ine external validity is through
partial replication of the experiment. Partial replication establishes
external validity by showing that a similar experimental result occurs
when slightly different experimental procedures are used. Another method
of estab lishing external validity can be to use conceptual replication. For
example, Anderson and Bushman (1997) suggested that suppose we wish
to establish the external validity of the idea that insults lead to aggressive
behavior. In that case, we need to use di fferent words that may appear to
be insulting to different population. The words that children may find
insulting may be different from the words that adults find insulting.
Matched Group Design:
Matched group design is another design involving independe nt groups.
Random group design can be used when the sample size is large,
especially if the researcher is working with heterogeneous population. munotes.in
Page 49
49 Experimental and Quasi-Experimental Methods Random group design functions on the assumption that individual
differences get averaged out across groups. Wh en there are very few
people available that can be divided into two random groups or when it is
not possible to use repeated measure design, then matched group design is
used. By matching the subjects on dependent variable task and then
dividing them into two groups, matched groups become as good as
random groups. However, the matched group design will be effective only
if the two groups are pre tested on a good matchable dependent variable.
For example, if the researcher wants to study the impact of a trai ning
program on mathematical ability of the class 5th students. Then it is
important to match the students based on their mathematical ability and
then dividing them into two groups and introducing special training
program. Both the groups should have on average equivalent
mathematical ability. In such cases, pre test -post test can be used.
Ideally, the groups should be matched by pre testing them on dependent
variable. But some times the dependent variable may be such that it gets
affected by previous ex posure. For example, suppose we want to study the
amount of time taken by 5th classstudents to learn spellings of certain
English words. If we use the same list of English words for pre test and
post test to measure their spelling ability, we may not be su re that less
time taken in post test is due to the independent variable or due to the
familiarity or practice effect with the words. In such situations, it is advice
able to use dependent variable task which is similar to the dependent
variable that you ar e going to use in experiment but not exactly the same
one. This similar dependent variable task which is similar but not the
same as dependent variable task in experiment is called matching task.
There should be high correlation between the matching task and actual
dependent variable task.
Even after the subjects are matched on a particular task, one should ensure
that they should be assigned to the two or more groups in random manner.
This will help in averaging out any other factor, that is beyond the
matching factor. That may influence their performance.
Natural Group Design :
Natural group design is usually used in correlational studies. Researcher
tries to find correlation between the subject variables and dependent
variables. As you already know, in c orrelational studies, we cannot
establish cause and effect relationship. But psychologists are interested to
see the influence of individual differences or subject variables on specified
dependent variables. They cannot experimentally manipulate these subj ect
variables to see their impact on dependent variable, they can manipulate
these variables by selection method only. For example, experimenter
cannot increase or decrease variables like gender, age, height, race, etc., to
see its impact on dependent var iable ( e.g., performance in sports), He will
have to manipulate it by selecting participants naturally belonging to
different height, weight and form different natural groups and then
compare their performance on the dependent variable. Natural groups are
those groups where independent variable’s levels are selected instead of munotes.in
Page 50
50 Research Methodology for Psychology
50 being manipulated to form different groups. Natural group design is used
to describe and predict the behavior and not to establish the cause and
effect relationship.
3.3 REPEATED ME ASURES DESIGNS Why researchers use repeated measures designs :
Sometimes researchers use repeated measure design even when sufficient
number of subjects are available, because they find it convenient and
efficient in answering their questions. For example, Ludwig et.al. (1993)
were interested to find out how both hemispheres of the brain
communicate with each other. They briefly presented two letters having
either similar or different name to the participants. Either one or both
letters were presented to ea ch participant in either unilateral condition (one
hemisphere) or bilateral condition (both hemisphere). Since each
participant was tested in both unilateral and bilateral conditions, it is called
repeated measure design. This design was far better than if they had to use
two separate groups to test unilateral and bilateral condition separately.
Repeated measure design is more sensitive than an independent groups
design. It means that repeated measure design can detect even the smallest
of the effect of in dependent variable and thus have lower error variance.
There is less error variance in repeated measure design because there is
usually more variation between people than there is within people.
Psychologists find repeated measure design more suitable for longitudinal
studies, where they would like to record the changes taking place in
dependent variable over time, for example in learning experiments. This
design is very useful in psychophysics studies, where psychologists want
participants to compare two or more stimuli relative to one another, for
example, in Hawthorne experiment, researchers wanted to know how
much light will lead to optimum productivity, so they had to keep
measuring the subjects’ performance in different intensity of light.
Another ad vantage of this method is that individual differences cannot
confound the findings as same individuals are used in different conditions.
3.3.1 The role of practice effects in repeated measures designs :
In spite of repeated measure design having many adva ntages, it is not free
from shortcomings. These shortcomings can threaten the internal validity
of the experiment.
One of the factors that can threaten the internal validity of the repeated
measure design experiment is that due to repeated testing the par ticipants’
performance may change due to practice effect rather than due to
independent variable manipulations. The changes that participants undergo
with repeated testing in the repeated measures designs are called practice
effects. The participants may p erform better and better at the task as they
learn more about the task, or they may get worse at the task due to factors
like fatigue and boredom. munotes.in
Page 51
51 Experimental and Quasi-Experimental Methods Another threat to internal validity can come from participants’ natural
maturation process. Their performanc e may improve with age as they
become more mature.
3.3.2 Balancing Practice Effects in the Complete Design :
There are two types of repeated measure design – the complete and
incomplete design. The main goal of repeated measure design is to control
practice effect.
Practice effect can be balanced by using a technique called
counterbalancing technique. But counterbalancing technique is used
differently in complete and incomplete repeated measure design.
In the complete design, practice effect is balanced by using block
randomization or ABBA counterbalancing.
Each participant is administered each condition many times to control the
practice effects using different sequence each time. These conditions are
administered in different order on each trial. Each part icipant can thus be
considered a “complete” experiment.
When all the conditions of an experiment or block are presented in a
random order in each trial, it is called block randomization.
When a random sequence of all conditions is presented, followed by the
opposite of the sequence, it is called ABBA counterbalancing. It is better
to use block randomization rather than ABBA counterbalancing if :
a) practice effect is non -linear
b) when the performance of the subject can get affected by anticipation
effect
c) It is possible to have large number of trials and number of conditions
are also sufficiently large.
Anticipation effect takes place if the participants develop an expectation
about which condition will take place next in the sequence, and this
expectati on influences their responses to the task.
Generally, the number of blocks in an experiment are equal to the number
of times each condition is administered and the size of each block is same
as the number of conditions in an experiment. To balance out pra ctice
effect in block randomization, it is necessary to repeat each condition
many times.
It is better to use ABBA counterbalancing technique if :
a) the number of conditions is less,
b) it is not possible to repeat each condition many times
c) practice e ffect is linear. munotes.in
Page 52
52 Research Methodology for Psychology
52 d) Anticipation effect will not take place
In ABBA counterbalancing, as the name suggests, the condition is
presented in one sequence ( A then B) and then represented in opposite of
the first sequence (B then A). However, in ABBA counterb alancing, there
must be an even number of repetitions of each condition.
3.3.3 Balancing Practice Effects in the Incomplete Design :
In the incomplete design, a researcher will administer each condition to
each participant only once. The order of administe ring the conditions
differs across participants rather than for each participant. The general rule
for incomplete design is that each condition of the experiment must appear
in each ordinal position (1st, 2nd, 3rd, etc.) equally often.
Practice effects in the incomplete design are balanced out across subjects
rather than across each participant. There are two techniques, in
incomplete design, to choose the order :
a) All Possible Orders
b) Selected Orders
All Possible Orders of the Conditions :
When there are four or less than four conditions in an experiment, it is
better to use all possible orders of the conditions. Each participant is
randomly assigned to one of the orders. It is advisable to use it for not
more than 4 conditions because the number of or ders increase dramatically
as the number of conditions increase. For example, if there are 3
conditions then number of possible orders will be 6, for 4 conditions, there
will be 24 possible orders. The number of possible orders will go up to
120 if there are 5 conditions and there will be 720 possible orders if there
are 6 conditions.
This technique will be effective if at least one participant is tested with
each of the possible orders of the conditions. So, if there are 4 conditions,
at least 24 or mult iples of 24 participants will be needed. Consequently,
this technique can be used only if large number of participants are
available according to the number of conditions in the experiment.
Selected Orders :
When large number of participants are not avail able and the number of
levels of independent variables is more, it is better to use selected order
method. It is possible to balance practice effect with some of the selected
orders out of all possible orders. There are two techniques that can be used
to do balancing with selected orders. These are -
a) Latin square
b) Random starting order with rotation.
munotes.in
Page 53
53 Experimental and Quasi-Experimental Methods a) Latin square :
In Latin s quare each and every condition appears at each ordinal position
at least once, and each condition precedes and follows each o ther
condition exactly once.
b) Random starting order with rotation :
This technique begins with a random order of the conditions and rotates
their sequence systematically with each condition moving one position to
the left each time. This ensures that each condition always follows and
always precedes the same other conditions. The advantage of this
technique is that it is very simple to apply and it can be used with more
than 4 conditions.
3.3.4 Data analysis of repeated measures designs:
Describing the r esults :
The first step in analyzing the data is to check for any errors and outliers.
The second step will be to use descriptive statistics (e.g., mean, standard
deviation) to summarize the data for each condition of the independent
variable. It is easy to summarize data in random group design. In
incomplete repeated measure design, while summarizing, the researcher
needs to make sure that participants’ scores are listed with the correct
condition. In complete repeated measure design, the researcher needs t o
first compute a score for each participant in each condition before he begin
to summarize and describe the results, because each and every participant
is tested in each condition more than once.
Confirming what the results reveal :
Similar to random group design, in repeated measure design too,
researchers test whether independent variable produces an effect on
dependent variable by testing null hypotheses against set confidence
limits. However, repeated measure differs from random group design in
estimati ng the error variance. In random group design, error variance is
estimated by finding the individual differences among participants within
the groups, while in repeated measure design, differences among
participants are eliminated from the analysis. Repeat ed measure design is
considered more sensitive than the other design due to its ability to
eliminate systematic variation. Error variance occurs in repeated measures
due to the differences in the ways the conditions affect different
participants.
3.3.5 The Problem of Differential Transfer :
When performance in one condition differs depending on the condition
that precedes it, it is called differential transfer. Differential transfer is a
threat to internal validity of repeated measures designs. It also under mines
the external validity of the results by underestimating the differences in
the conditions. So, if there is differential transfer, the researcher should
use independent groups design. munotes.in
Page 54
54 Research Methodology for Psychology
54 3.4 COMPLEX DESIGNS 3.4.1 Complex Designs with Three Independent Variables :
The simplest complex design is 2X2 design. But we can have more than
two independent variables having more than two levels. As the number of
independent variables increase the power, the complexity and the
efficiency of the complex design also in creases. A two -factor design can
compute only one interaction effect while a three factor design can
compute three main effects and four interaction effects. For example, if we
take three independent variables – A, B and C. Then apart from the main
effect of each of these variables, we will get 4 different interaction effects
such as AXB, AXC, BXC, AXBXC.
In a three -factor complex design, a three -way interaction effect occurs
when the interaction of two of the independent variables differs depending
on the level of the third independent variable. Therefore, while describing
the results, all three independent variables must be considered.
Just as in case of 2X2 design, in complex design experiment too, the data
is first checked for any errors or outliers and then data is analyzed to check
for three potential sources of variations such as the main effects of each
independent variable and the interaction effect between independent
variables.
Descriptive statistics such as mean, standard deviation, and measures of
effect size is used to describe the results.
Inferential statistics such as null hypothesis testing and confidence
intervals are then used to determine whether any of the effects are
statistically reliable.
Both descriptive and inferential statistics a re used to interpret main effect
and interaction effect of an experiment. An effect is considered to be
statistically significant if the probability of observed effect of an
independent variable is less likely to occur by chance factor or it is the
probabi lity under the null hypothesis that is less than significance level of
.05.
3.4.2 Describing effects in a complex design :
When researchers study the effects of two or more independent variables
in one experiment, it is called complex design. These are al so known as
factorial designs. There are many types of complex designs. The simplest
complex design has two independent variables with each independent
variable having two levels. It is denoted as 2X2 design. In other words,
complex designs are denoted by the number of levels of each of the
independent variables in the experiment. As the number of independent
variables increase or the number of levels of each independent variable
increase, the design also becomes more complex and more powerful too,
e.g., we can have 2X2X3 design, 3 X4 design, 3X3X4X2 design, etc.
Though theoretically, there can be unlimited number of complex designs munotes.in
Page 55
55 Experimental and Quasi-Experimental Methods but generally, experiments do not involve more than five independent
variables having two or three levels. Moreover, these comp lex designs can
have either independent groups variables or repeated measures variables.
If a complex design has both an independent groups variable and a
repeated measures variable, it is called a mixed design
No matter how many independent variables and how many levels of
independent variables are there in a complex design, the nature of main
effect and interaction effect remains the same.
A simple main effect is defined as the effect of one independent variable
at one level of a second independent variab le. In complex design, one can
test the overall effect of each independent variable while ignoring the
effect of the other independent variables.
Interaction Effects :
An interaction effect refers to the effect of one independent variable
depending upon the level of second independent variable. In other words,
the effect of one independent variable differs depending on the levels of a
second independent variable. The order of the independent variables is
immaterial.
Main Effects and Interaction Effects :
The gross general effect of each independent variable in a complex design
is called a main effect. It shows the differences in the average
performances for each level of an independent variable colligated across
the levels of the other independent variable.
An interaction effect between independent variables takes place when the
effect of one independent variable differs according to the levels of the
second independent variable
A researcher can easily identify interaction effect by merely seeing the
graphic al representation (descriptive study) of the means or averages of
conditions under study. He can confirm the presence of interaction effect
by using statistical analysis. He can also choose the results of which
interaction effect to emphasize. Compared to experiments with only one
independent variable, the study of interaction effect in complex
experiment allows researchers to have better understanding.
As mentioned before, results of complex experiments can be summarized
in descriptive statistics and fin dings can be explained scientifically
through inferential statistics. Three main ways to summarize results are
tables, bar graphs, and line graphs. Tables are used to show the exact
values for each condition in the experiment. Bar graphs and line graphs
are used to show the patterns of the results without emphasizing the exact
values. Especially, interaction effect can be clearly seen in line graph.
Nonparallel lines in the graph indicate an interaction effect, while parallel
lines indicate no interaction e ffect. munotes.in
Page 56
56 Research Methodology for Psychology
56 Another way of finding interaction effect is the use of ‘subtraction method
especially in 2X2 design. In subtraction method, the differences between
the means in each row (or column) of the table is compared. If these
differences are different, an i nteraction effect is likely. While using this
method, it is important to calculate the differences in the same direction.
This method can be used only when one of the independent variables has
two levels. If both independent variables have three or more le vels, then it
is better to use graph to see the interaction effect.
3.4.3 Analysis of Complex Designs :
Just as in case of 2X2 design, in complex design experiment too, the data
is first checked for any errors or outliers and then data is analyzed to check
for three potential sources of variations such as the main effects of each
independent variable and the interaction effect between independent
variables.
Descriptive statistics such as mean, standard deviation, and measures of
effect size is used to descr ibe the results.
Inferential statistics such as null hypothesis testing and confidence
intervals are then used to determine whether any of the effects are
statistically reliable.
Both descriptive and inferential statistics are used to interpret main effect
and interaction effect of an experiment. An effect is considered to be
statistically significant if the probability of observed effect of an
independent variable is less likely to occur by chance factor or it is the
probability under the null hypothesis t hat is less than significance level of
.05.
Analysis Plan with an Interaction Effect :
Inferential statistics tests are used in conjunction with descriptive statistics
to determine whether an interaction effect has, in fact, occurred. If
significant inter action effect is present, then the source of the interaction
effect is identified using simple main effects analyses and comparisons of
two means. When three or more means are tested in a simple main effect,
comparisons of means testing two at a time can b e done to identify the
source of the simple main effect. Generally, researchers do not pay much
attention to main effects of each independent variable if interaction effect
is present.
Analysis Plan with No Interaction Effect :
If no statistically signific ant interaction effect is present, then the next step
is to see whether the main effects of the variables are statistically
significant. Once again for finding the statistical significance of main
effects, comparisons of two means or using confidence inter vals to
compare means two at a time are used.
Studies have shown that complex design can provide lot of information
even when there is no statistically significant interaction effect munotes.in
Page 57
57 Experimental and Quasi-Experimental Methods 3.5 INTERPRETING INTERACTION EFFECTS 3.5.1 Interaction Effects and Theor y Testing :
Very often, theories predict that interaction of two or more independent
variables influences the given behavior. These theoretical assumptions can
be tested through complex designs. Most of theories in psychology are of
complex nature. On being tested, they may produce contradictory findings.
In such cases, finding out the interaction effect helps in resolving the
conflicting findings. The complex designs are laborious but, very useful in
finding out the reasons for seemingly contradictory findi ngs when theories
are tested. Complex designs enhance the researchers’ ability to test
theories because they can test for both main effects and interaction effects
3.5.2 Interaction Effects and External Validity :
If no interaction effect shows up in a comp lex design, the main effect of
each independent variable can be generalized across the levels of the other
independent variable. This increases the external validity of the
independent variable. When interaction effect is present, it specifies the
conditio ns in which an effect of an independent variable will occur. These
conditions indicate the boundaries for the external validity of a finding.
The interaction effect also identifies what those boundaries are. When a
question is asked whether a particular in dependent variable has an overall
effect across other independent variables, the typical answer will be “it
depends”. Independent variables that influence behavior directly or
produce an interaction effect are called relevant independent variables.
Identif ying relevant independent variables is important for designing
effective interventions. The opposite of relevant independent variables is
irrelevant independent variable. There are many reasons that make it
crucial to identify irrelevant independent variab les too, such as :
1. If an independent variable has no effect in an experiment, it can’t be
assumed that this variable wouldn’t have an effect if different levels
of the independent variable had been tested.
2. If an independent variable has no effect in a single -factor experiment,
this doesn’t mean that it won’t interact with another independent
variable when used in a complex design.
3. If an independent variable does not have an effect in an experiment,
there is a possibility that an effect could have been seen with different
dependent variables.
4. The absence of a statistically significant effect may or may not mean
that the effect is not present.
Thus, presence or absence of interaction effect is important to determine
the external validity of the findings in a complex design. However, if there
is no statistically significant interaction effect, it does not mean that there
was no interaction between the independent variables. On of the reasons
for not finding statistically significant interaction ca n be that the munotes.in
Page 58
58 Research Methodology for Psychology
58 researcher may not have performed the experiment with sufficient
sensitivity.
3.5.3 Interaction Effects and Ceiling and Floor Effects :
When participants’ performance reaches a maximum in any condition of
an experiment, it is called ceiling e ffect. On the other hand, if the
performance reaches the minimum in one or more conditions of an
experiment, it is called floor effect and results for an interaction effect
become uninterpretable.
Researchers can avoid ceiling and floor effects by selectin g dependent
variables that allow sufficient chance for performance differences to be
measured across conditions
3.5.4 Interaction Effects and the Natural Groups Design :
When groups of people are formed by selecting individuals who differ on
some characteri stic such as gender, age, introversion – extraversion, or
aggressiveness, etc. they are called natural groups. The natural groups
design is efficacious for showing correlations between individuals’
characteristics and their performance. However, it is diffi cult to establish
cause and effect relationship through natural group design, as there can be
many other possible causes for difference in performance other than
individual differences. The problem of drawing causal inferences based on
the natural groups d esign can be dealt with by developing a theory about
the critical individual difference variable.
Three steps for making a causal inference involving a natural groups
variable are to state a theory for why group differences exist, manipulate
an independen t variable that should show how the theory was processed,
and test whether an interaction effect takes place between the manipulated
independent variable and natural groups variable.
Step 1: Develop a Theory
First of all, the researcher mustdevelop a theor y explaining why a
difference should occur in the performance of groups that have been
differentiated on the basis of an individual differences variable.
Step 2: Identify a Relevant Variable to Manipulate
Next the researcher needs to select an independent variable that can be
manipulated and that is presumed to influence the likelihood that this
theoretical process will occur.
Step 3: Test for an Interaction
Lastly, the researcher should try to produce an interaction effect between
the manipulated variable and the individual differences variable. This way,
the relevant manipulated independent variable will be applied to both
natural groups. munotes.in
Page 59
59 Experimental and Quasi-Experimental Methods If the analysis of a complex design shows that there is no statistically
significant interaction effect between indepe ndent variables, then we need
to determine whether the main effects of the variables are statistically
significant.
3.6 TRUE EXPERIMENT A true experiment can be defined as an experiment that leads to an
unambiguous result and clearly established cause and effect relationship.
3.6.1 Characteristics of True Experiments :
There are three important characteristics of true experiments :
1. True experiment will have some type of treatment in it.
2. In true experiments, the experimenter has high degree of control over
the assignment of the participants, experimental conditions,
systematic manipulation of independent variables and determining the
dependent variable. Random assignment of the subjects to different
groups is the hallmark of true experiments. However, experimenter of
true experiment in natural setting may not have same level of control
over either the assignment of the participants to different conditions or
even on the conditions of the experiment as he would have in
laboratory setting. But the experi ment conducted in laboratory setting
may not have same external validity as the one conducted in natural
setting.
3. True experiments always involve comparison, that is, finding out the
difference in the dependent variable due to different levels of
independent variable.
3.6.2 Obstacles to Conducting True Experiments in Natural Settings :
Some of the difficulties that an experimenter face while conducting
experiment in natural settings are :
Difficulty in getting permission from the authorities (such as sc hool
principals), to reach out to the potential participants and to conduct the
experiment in natural setting. Government officials may not financially
support a research if they think it is not useful. So financial crunch is
another obstacle to conduct r esearch.
Having access to potential participants becomes more crucial if the
research design requires randomly assigning the participants to more than
one group and compare them on a dependent variable.
If participants are divided into two groups on the basis of random
assignment, those in control group may feel discriminated as they are
denied to experience the independent variable. For example, if the
experimenter is testing the effectiveness of new teaching method
compared to the old teaching method, t hose who are randomly assigned to
control group will feel deprived of new method, especially if the new munotes.in
Page 60
60 Research Methodology for Psychology
60 method has proved to be better than the old method. On the other hand, if
the new method proves to be ineffective, then random assignment will
prove to be a boon for them as it would have protected them. To make
sure that researcher has potentially comparable group by using random
assignment and yet no participant is left out of either of the independent
variable conditions, researchers can use alternate treatments. That is both
the groups alternatively go through old and new method of teaching. Each
group will serve as control group for the other group.
However, there may be some experiments where random assignment
cannot be used. For example, if a rese archer wants to test the effectiveness
of a new drug, patients may not agree to get assigned to experimental
group where new medicine will be tested. In such cases, quasi -
experimental design can be used.
3.6.3 Threats to Internal Validity Controlled by Tr ue Experiments :
Internal validity of an experiment gets threatened when the results of an
experiment can be explained by alternative factors other than the
independent factors only.
Some of these confounding factors are history, maturation, testing,
instrumentation, regression, subject attrition, selection, and additive
effects with selection. Let us see how these factors can contaminate the
results.
History :
In true experiment participants in the experimental group and in the
control, group must have same history of experiences during the
experiment, except for the treatment. In natural settings, it may not be
possible for the researcher to have high degree of control, so internal
validity may be threatened due to confounding variable - history. For
exampl e, a teacher wants to test the effectiveness of an interactive
teaching program for fifth class students. She conducts a pretest,
introduces the interactive program and then conducts the posttest. She
finds the difference in the pretest and posttest scores . However, without a
comparison group, it is difficult to say whether the difference is only due
to the independent variable or some other factors also might have played
the part. For example, many students might be attending coaching class
also.
Matura tion:
Change associated with the passage of time per se is called maturation.
Participants in an experiment grow older and become more experienced as
the experiment progresses. This confounding factor influences especially
the longitudinal studies.
Testing :
Very often, it is observed that scores on the post test improve without any
intervention too. The reason is that exposure to pretest has an impact on munotes.in
Page 61
61 Experimental and Quasi-Experimental Methods post test that is similar to pre test. Participants become familiar with the
type of questions, instruc tions and experimenter’s expectation and this has
an impact on their subsequent performance. In pretest -posttest design, it is
difficult to separate the effect of independent variable and testing effect.
Instrumentation :
Apart from the participants, eve n instruments can change over time. For
example, if human observers are used to assess the change in behavior,
their judgment may suffer from observer bias such as fatigue,
expectations, and other characteristics of observers. It may be argued that
one can use mechanical observation instead of human observation to avoid
the problem of instrumentation. However, even mechanical instruments
may change with repeated use. For example, a machine used in pre -test
may become faulty by the time post -test is conducte d. In that case, change
in scores from pre -test to post -test may be due to faulty measurement
rather than the effect of independent variable.
Regression :
The error of statistical regression takes place when participants are
selected on the basis of their extreme scores on a specific test. There is no
guarantee that extremely poor or good scores on a particular test will lead
to similar scores on another test too. For example, if a student has faired
badly in one of the subjects in 12th standard exam, it d oes not mean that
he perform poorly in college exam too or vice versa . There can be various
reasons for his poor performance in 12th standard exam. If a student who
has scored very poorly in 12th standard exam gives the same exam again,
the chances are v ery high that his performance will be close to the average
of his overall scores. This is called regression to the mean. The chances of
statistical regression occurring are much more when a test or measure is
unreliable. An unreliable test will produce inc onsistent results over time.
Sometimes, researchers commit the error of assuming regression effect as
treatment effect.
Subject Attrition :
The internal validity of an experiment gets compromised if participants
drop out in the middle of the experiment. T he nature of the group that was
established before treatment changes if there is loss of participants. The
groups that were formed on the basis of random assignment may not
remain equivalent.
Selection :
If the groups are made on the basis of selection inst ead of randomization,
there might be inherent difference between the groups. This inherent
difference can threaten the internal validity of the experiment. The
chances of selection threat are more in natural setting experiments than lab
experiments.
munotes.in
Page 62
62 Research Methodology for Psychology
62 Addi tive Effects with Selection :
History and maturation combined with selection can be another cause for
threat to internal validity. These additive threats can be of three types –
a) Additive effects of selection and maturation :
As explained above maturatio n refers to natural increase in cognitive,
physical and emotional maturity as well as familiarity with the
environment. If a researcher compares fresh recruits in a company with
those who were recruited a year back, it will result in additive effect of
selection and maturation effect. Those who were recruited a year back
must be already familiar with the work culture and work routines while
new recruits will be still struggling to get adjusted. In such a scenario, any
difference in the behavior of two group s cannot be attributed solely to
variations in independent variable.
b) Additive effect of selection and history :
Additive effect of selection and history is more prominent when researcher
compares two intact groups. Both of these intact groups may not b e
equivalent as they may have different experiences, or they may experience
the same event differently depending on their specific characteristics. For
example, if a researcher wants to study the effectiveness of corona
prevention program among college stu dents. Students who have either
suffered an attack of corona or their near dear ones have suffered from
corona will pay more attention to this program rather than those who were
least affected.
c) Additive effect of selection and instrumentation :
An addit ive effect of selection and instrumentation will take place if the
instrument can detect or measure changes in one group but not in another
group. This becomes more prominent when there is floor or ceiling effect
is present in the groups. when a group scor es so low on an instrument
(floor effect) in the beginning that any further drop in scores cannot be
reliably measured, or so high (ceiling effect) that any more gain cannot be
measured. The floor or ceiling effect will endanger the internal validity, if
an experimental group shows relatively no change (due to floor or ceiling
effects) and a control group changes reliably because its average
performance was near the middle of the measurement scale right from the
beginning.
All these threats to internal vali dity can be controlled through true
experiments. But some of the threats may not be controlled through true
experiments too. Let us see some of such threats.
Problems That Even True Experiments May Not Control :
Although major threats to internal validity are removed by the true
experiment, there are some other threats that the researcher must guard
against, while working in natural settings. Some of these threats are : munotes.in
Page 63
63 Experimental and Quasi-Experimental Methods Contamination :
The term contamination refers to a general class of threats to internal
validity. Contamination takes place when information about the
experiment is communicated between groups of participants, which may
lead to resentment, rivalry, or diffusion of treatment. when an
experimenter unintentionally influences the results, true ex periments gets
affected by threats such as experimenter expectancy effects.
Observer bias takes place when researchers’ biases and expectancies lead
to systematic errors in observing, identifying, recording, and interpreting
behavior
i. Resentment :
Resent ment takes place when participants, randomly assigned to a control
group, come to know that they are receiving less desirable treatments or
that the other group is getting better treatment. It may dishearten the
control group participants or make them angr y, and they may give lower
performance due to resentment.But the experimenter may interpret this
lower performance compared to experimental group due to intervention
and not as a deliberate attempt by the control group due to resentment.
ii. Rivalry :
Anot her possible reaction of the control group, on knowing that
experimental group is receiving better treatment than them, is the spirit of
competition and rivalry. A control group might become motivated to
reduce the expected difference between itself and th e treatment group and
not look inferior to experimental group.
Diffusion of treatments :
Diffusion of treatments occurs when participants in a control group use
information given to participants in the treatment group to help them
change their own behavior. They may copy the behavior of participants in
experimental group. This will also reduce the differences between the
treated and untreated groups and affects the internal validity of the
experiment.
a) Novelty Effect :
Novelty effects occur when people’s be havior changes simply because
new element is introduced in their environment (e.g., an experimental
treatment). It produces excitement, energy, and enthusiasm. This
enthusiasm, rather than the intervention itself, may account for the
“success” of the inter vention. The opposite of a novelty effect is known as
a disruption effect. Disruption effect takes place when due to novelty in
work procedures, the routine work of employees gets disrupted to such an
extent that they cannot maintain their typical effectiv eness.
munotes.in
Page 64
64 Research Methodology for Psychology
64 b) Hawthorne Effect :
Hawthorne Effect refers to changes in people’s behavior brought about by
the interest that “significant others” show in them. The behavior changes
because participants are aware that someone is interested in them. It is a
kind of reactivity (i.e., an awareness that one is being observed). The
effect was named after such an effect showed in the experiment conducted
in the Hawthorne plant of the Western Electric Company. In Hawthorne
plant, the experiment was conducted to find out the whether variations in
amount of lighting in the plant will affect workers performance. Results
showed that experimental and control groups, both increased their
productivity during the study.
Cook and Campbell (1979) emphasized that judging the intern al validity
of a relationship is a deductive process. The researcher has to be his own
best critic, minutely examining all of the threats he can imagine. He must
systematically look at each of the internal validity threats and determine
how it may have inf luenced the data. Then, he must examine the data to
find out which relevant threats can be ruled out. He can make conclusions
about the causal relationship between two variables with confidence only
when all of the possible threats can be eliminated.
Apar t from threats to internal validity, the researcher must ensure to
eliminate the threats to external validity too.
External validity can get threatened if the sample is not representative of
the persons, settings, and times to which the researcher wants t o
generalize. Theoretically, representativeness can be achieved through
randomization, but in real life random sampling is not used often. If
complete randomization is not possible, then the next best alternative to
ensure external validity is repeating th e experiment with different types of
participants, in different settings, with different treatments, and at different
times. The experimenter can built partial replication into the experiment
itself such as selecting more than one group of participants and comparing
them.
3.7 QUASI -EXPERIMENTS The word quasi is a prefix and it means ‘resembling’. Quasi - experiment
means a procedure that resembles true experiment but is not true
experiment. Just like true experiments, quasi experiments too include
comparis on and some type of intervention, but they lack the
randomization and the control that is an essential part of true experiments.
So quasi experiments are used when it is not possible to have the rigour of
true experiments. Quasi experiments may be incapabl e of controlling all
confounding factors that may threaten internal validity of the experiment.
Shaughnessy and Zechmeister believe that first of all experimenters
should try to make quasi experiment as close to true experiment as
possible and they must id entify the specific shortcomings of the procedure
and give all possible evidence to overcome those shortcomings. munotes.in
Page 65
65 Experimental and Quasi-Experimental Methods One of the serious issues faced by quasi experiment is lack of opportunity
for randomization. Due to either practical considerations or admin istrative
decisions, the researcher may have to work with intact group, for example,
children in a particular class or employees of a particular organization. In
such situations, the researcher has to use one -group pretest -posttest design
that is also know n as pre -experimental design. Though it is assumed that
any difference between pre -test and post test scores is due to independent
variable, but it is possible that the difference in scores is due to
confounding factors (such as history, maturation, testin g, an d
instrumentation threats, etc. ) that threaten the internal validity. That is
why it is also known as bad experiment. The results of a bad experiment
are inconclusive about the effectiveness of a treatment. There are some
quasi experimental designs that can improve upon this pre -experimental
design. We will discuss here some of these designs.
3.7.1 The Nonequivalent Control Group Design :
There are two requirements of this design – there should be two
comparable groups and there should be a possibil ity of using pretest and
posttest on them.
As the researcher is forming comparison groups on the basis of non -
randomization, we cannot assume that participants in both the groups have
equivalent characteristics. The experiment suffers from selection threa t.
So, it is necessary to equalize them on the basis of pretest scores.
For example, we have two groups – a treatment group and a control group
and they are compared through pretest and post test measures. If the
pretest scores of both the groups were si milar, then we can say that both
groups are comparable, If post test scores of both the groups differ, it is
assumed that it is due to the effect of the treatment.
With this design, it is possible to control confounding factors such as
history, maturation , testing, instrumentation, and statistical regression. The
reason being that it is assumed that both the groups have similar
experiences (of confounding factors), except the treatment. Then
confounding factors cannot account for differences in post test s core, and
the researcher can safely claim the cause and effect relationship between
independent and dependent variable.
This experiment can be symbolized as follows :
O1XO2
----------
O1 O2
O1 refers to pre -test or first observation and O2 indicates the po st test
score or second observation. The dash line indicates that experimental and
control group were not formed on the basis of random assignment.
munotes.in
Page 66
66 Research Methodology for Psychology
66 Nonequivalent Control Group Design: The Langer and Rodin Study :
Quasi -experiments usually assess the overa ll effectiveness of a treatment
that has many components. A follow -up research may be required to
determine which components are crucial for achieving the treatment
effect. For example, Langer and Rodin (1976) conducted a study to test
the assumption that the lack of opportunity to make personal decisions
contributes to the psychological and even the physical debilitation. Their
argument was that environmental changes associated with old age partly
contribute to the feelings of loss, inadequacy, and low se lf-esteem among
the elderly, especially those living in a nursing home. They believed that
nursing homes provide a “virtually decision -free” environment. The
elderly are no longer allowed to make even the simplest decisions, such as
what time to get up, wh om to visit, what movie to watch, etc. They carried
out a quasi -experiment in a nursing home. The independent variable was
the type of responsibility given to two groups of elderly residents staying
in that nursing home. It was not possible and administrat ively undesirable
to randomly assign them to two different groups. So, researchers chose to
take residents of two floors. These floors were chosen on the basis of
similarity in the residents’ physical and psychological health and prior
socioeconomic status . It was decided to randomly assign any of the floors
to one of the two treatments. Residents of one floor were informed of the
many decisions they needed to make regarding how their rooms were
arranged, visiting, care of plants, movie selection, and so fo rth. Moreover,
these residents were given a small plant as a gift (if they decided to accept
it) and told to take care of it as they wished. This was the responsibility -
induced condition. The second group of residents, the comparison group,
was told that i t is staff’s responsibility to look after their needs. They were
also given a plant as a gift (irrespective of whether they chose to have one
or not) and were told the nurses would water and care for the plants for
them. Questionnaires having items related to “how much control they felt
over general events in their lives and how happy and active they felt” were
given to residents 1 week before and 3 weeks after the responsibility
instructions. Staff members on each floor were asked to rate the residents,
before and after the experimental manipulation, on traits such as alertness,
sociability, and activity. Differences between pretest and posttest measures
showed that the residents in the responsibility induced group were
generally happier, more active, and m ore alert following the treatment
than were residents in the comparison group. However, it is important to
know that the effectiveness of the overall treatment, not individual
components of the treatment, was assessed. So, we don’t necessarily know
whether the treatment would work with a smaller number of components
or whether one component is more crucial than the others. Generally,
research in natural settings is characterized by treatments with many
components and aims to assess the overall effect of the treatments.
Theoretically, however, it is important to determine whether components
of the treatment specified by a theory, as being critical, are really the
critical components.
munotes.in
Page 67
67 Experimental and Quasi-Experimental Methods 3.7.2 Sources of Invalidity in the Nonequivalent Control Group
Design :
Thou gh non -equivalent control group design generally controls for all
major classes of potential threats to internal validity yet for proper
interpretation of quasi experiments, the researchers check out the presence
of any threats to internal validity, such a s additive effects with - selection
and maturation, selection and history, selection and instrumentation, and
differential regression, observer bias, contamination, and novelty effects.
Let us look at each one of these threats.
Selection -Maturation Effect:
An additive effect of selection and maturation takes place when
participants in one group grow more experienced, more tired, or more
bored at a faster rate than participants in another group. This threat to
internal validity becomes more prominent when th e treatment group is
self-selected (the members deliberately sought out exposure to the
treatment) and when the comparison group comes from a different
population than the treatment group. We cannot say that both control and
experimental groups are equival ent, for various reasons, even if their
pretest scores are same on an average.
The first reason is that the natural growth rate of two groups from different
populations might be different, but the pretest may have been taken at a
time when both groups hap pened to be about the same. Both such groups
may show a difference at the posttest due to differential growth rate but it
could be mistaken for a treatment effect.
The second reason is that pretest is conducted to measure respondents on
only one measure o r only few measures. The mere fact that individuals do
not differ on one measure does not mean they don’t differ on other
measures that are relevant to their behavior in this situation.
Selection -History Effect :
Additive effect of selection and history ari ses when an event other than the
treatment affects one group and not the other. This is also known as the
problem of local history of effects. The more the settings of the
individuals in the treatment and comparison groups differ, more the
problems of loca l history becomes acute. For example, in above
mentioned Langer and Rodin study, suppose the results had to show that
the happiness and alertness of residents increased on one floor but not on
the other floor. Then these results could have been due to many possible
reasons such as change in nursing staff on one floor may have increased or
a decreased the morale of the residents’, depending on the nature of the
change and any differences between the behavior of a new nurse and that
of the previous one.
munotes.in
Page 68
68 Research Methodology for Psychology
68 Selection -Instrumentation Effect:
A threat of selection - instrumentation is more likely to take place when
changes in a measuring instrument are more likely to be detected in one
group than they are in another. Floor or ceiling effects may make it
difficult to detect changes in behavior from pretest to posttest. The threat
of selection -instrumentation effect is more likely to be prominent in
groups that are more non -equivalent and closer the group scores are to the
end of the scale.
Differential Statistica l Regression :
Regression toward the mean is likely to occur when individuals are
selected on the basis of extreme scores (e.g., the poorest readers, the
workers with the lowest productivity, the patients with the most severe
problems). Differential regres sion can takes place when regression is more
likely to be there in one group than in another. The changes from pretest to
posttest may be mistakenly interpreted as a treatment effect if regression is
more likely in the treatment group than in the control g roup.
Expectancy Effects, Contamination, and Novelty Effects :
This observer bias, or expectancy effect takes place if the observers are
aware of the objective of the study. They inadvertently try to prove the
hypothesis.
Possible contamination effect, tha t is participants getting demoralized after
knowing that other group is getting better treatment, can be controlled by
making sure that both the groups are geographically far apart or have
almost nil communication with each other. For example, in Langer an d
Rodin’s study, the residents of one floor had very little communication
with residents of the other floor.
3.7.3 The Issue of External Validityand Interrupted Time -Series
Designs :
As mentioned before that the best evidence for the external validity of
research findings is replication with different populations, settings, and
times. Same deductive process that was explained to determine internal
validity must also be used to examine external validity of the study. For
example, let us look at Langer and Ro din (1976) study once again. They
conducted the study in a nursing home that was rated one of the best
caring unit, having the best of the staff and facilities. Will the result be
different if the study was conducted in another care unit which was not so
highly rated? Another factor that can threaten the external validity is
whether the residents staying in this facility are comparable to elderly
inmates staying in other facilities. If inmates of different care units differ
in their background, then their r eactions to staying in the care unit also
might be different. Similarly, one needs to determine whether the staff of
this unit is comparable to the staff of other care units before any
generalizations can be made. munotes.in
Page 69
69 Experimental and Quasi-Experimental Methods Interrupted Time -Series Designs :
Gener ally, in simple time series experiments, observations are made
before the treatment and after the treatment. If abrupt changes
(discontinuities) in the time -series data occur when treatment is
introduced, it is safely concluded that the change is due to tr eatment.
However, the internal validity of time series experiments can be seriously
threatened from history effects and changes in measurement
(instrumentation) that occur at the same time as the treatment. To
overcome these threats, researchers can make many observations from
time to time to check the changes taking place in a dependent variable
before and after a treatment is introduced. This is called a simple
interrupted time -series design. This design can be outlines as
O1 O2 O3 O4 O5 T O6 O7 O8 O9 O 10
O refers to the observations and T refers to the treatment introduced. This
design can be effectively used when a new product has been launched, a
new social reform has been implemented etc.
One specific feature of time series experiments is that only abrupt changes
can be observed and not the gradual changes taking place. History,
instrumentation and seasonal variations are the major threats to internal
validity of time series design. Threats such as maturation, testing, and
regression can be controlle d in the simple interrupted time -series design.
However, simple interrupted time series design has serious problem with
external validity as it generally involves testing only a single group that
has not been randomly selected.
Time Series with Nonequival ent Control Group :
In a time series with Nonequivalent control group design, researchers
make a series of observations before and after treatment for a treatment
group and a comparable comparison group, both. This significantly
improves the internal validit y of the experiment. To implement time series
with Nonequivalent control group design, the researcher must have two
comparable groups that can be observed multiple times. He can make
multiple observations of dependent variable in both the groups and then
introduce treatment in experimental group, and again make multiple
observations of both the groups over a period of time.
This design is outlined as follows:
O1 O2 O3 O4 O5 T O6 O7 O8 O9 O10 - Experimental group
------------------------------------------- ----
O1 O2 O3 O4 O5 O6 O7 O8 O9 O10 - Control group
A dashed line indicates that the control group and the experimental group
were not randomly assigned. This design can eliminate the effect of
history. munotes.in
Page 70
70 Research Methodology for Psychology
70 3.7.4 Program Evaluation :
Posavac (2011) differentia ted between organizations in manufacturing
sector and in service sector. Program evaluation is more applicable to
service organizations such as hospitals, schools, government agencies, etc.
The efficiency of manufacturing organizations can be easily assess ed by
the profitability of the organization but that is not the case in service
organizations. Service organizations’ effectiveness is assessed through
program evaluation.
Posavac (2011) said that program evaluation is a methodology to find out
the depth and extent of need for a human service and also to find out
whether the service is likely to be used, whether the service is sufficiently
intensive to meet the unmet needs identified, and the degree to which the
service is offered as planned and actually d oes help people in need at a
reasonable cost without unacceptable side effects. (p. 1)
This definition of program evaluation highlights questions about four
areas – needs, process, outcome and efficiency. The process of answering
these four questions incl udes the entire process of conducting an
experiment. Let us see how it works. So for designing any program for the
people, the service organization has to first of all assess what are the needs
of the target population. For example, suppose the state gover nment
decides to built and maintain certain parks only for senior citizens and
names the scheme as ‘nana -nani park’. This scheme will be successful
only if senior citizens have the need or want to have parks exclusively for
themselves. The government agenc ies can use survey method to assess the
recreational needs of the senior citizen. This survey will help them to
decide whether exclusive parks are needed or some other form of
recreation is needed. On the basis of this survey, if exclusive parks are
provid ed for senior citizens, the next step will be to appoint program
evaluators to assess whether program or scheme is effectively
implemented or no and whether it is meeting its goals. If in spite of
providing exclusive parks, very few senior citizens are usi ng them, it will
indicate that either the program was not designed properly or not
implemented properly. Thus, program evaluator looks at how actually the
program or scheme is being carried out and what adjustments are needed
to make it more effective. Pro gram evaluators can be used to assess the
either existing schemes or new schemes to find the gap between needs and
what is provided for need satisfaction. Evaluation of outcome of the
schemes can involve either or both experimental and quasi -experimental
method, that is experiment in natural setting.
Efficiency of the program is also determined by the cost of the program.
The evaluator has to determine whether continuing the program is
economically viable or not. On the basis of evaluator’s reports, the
agencies can make a informed decision about whether to continue the
program, does it need improvement and how to improve it, or whether to
try an alternative program. If very few people are using the existing
facility, it may not be economically viable or it may require changes to be
made to make it more useful for the targeted population. Programmed munotes.in
Page 71
71 Experimental and Quasi-Experimental Methods evaluation is an example of applied research. Its goal is not to test or
formulate any theory, but practical goal. However, there is a reciprocal or
circular rel ationship between basic and applied research. Basic research
gives us certain scientific principles. When these principles are applied in
real world, new complexities are notices and new hypotheses are formed.
These new hypotheses are tested in controlled lab environment. That gives
rise to new theories or modifies existing theories and then these theories
are again tried in the real world. The cycle goes on.
Campbell (1969) emphasized that government agencies introduce and
implement many social reforms th at ultimately fail because most of these
programs are not based on hard scientific data, instead they are based on
certain assumptions and for political gains. Such ill -informed programs
lead to waste of public money. Public officials should not try to ap ply one
solution for all problems, instead they should use experimental method to
find out different solutions for different problems. Only then the solutions
will be effective.
3.8 SUMMARY The experimental method is used to establish cause and effect re lationship.
One way to determine the cause -effect relationship, it is important to
control as many confounding variables as possible. One of the
confounding variables can be sample itself in different groups. To avoid
this contamination, researchers use r andom group design, matched group
design, etc. In random group design, comparable groups are formed by
randomly assigning the subjects to different groups. This can be done
through block randomization. There are various factors that can undermine
the inter nal validity of an experiment. For example, it may be extraneous
variables like different physical settings, different experimenters, selective
subject loss across the conditions due to certain characteristics of the
subjects, etc. Some of these factors ca n be countered by using placebo
control and double -blind techniques. Data analysis is done by using both
descriptive and inferential statistics for analyzing and presenting the data.
Though confidence interval and null hypothesis are two powerful
technique s to test the hypotheses, but statistical analysis does not
necessarily lead to meaningful findings or be of practical use. One needs
to make sure that external validity of the experiment is also high. External
validity can be increased by conducting field experiments, by doing partial
replication and conceptual replication. Another technique for conducting
experiments is matched groups design. This method is used when only a
small number of participants are available and when experimenter needs
separate gr oups for each treatment.
Repeated measures designs are useful when the available number of
participants is small or independent variable can be tested over repeated
trials. This type of design is useful in fields like psychophysics. It is more
sensitive than other designs. In this design, each participant goes through
all conditions in an experiment. There are two types of repeated measure
designs – complete and incomplete repeated measures design. In an
incomplete repeated measure design, each participan t is exposed to each munotes.in
Page 72
72 Research Methodology for Psychology
72 treatment only once, and the practice effect is balanced across the
participants.
However, this method is susceptible to practice effect. Practice effect can
be balanced by using either block randomization or ABBA counter
balancing tec hnique. Block randomization should be used when practice
effect is expected to be nonlinear or if there is possibility of anticipation
effect taking place. In an incomplete repeated measure design practice
effect is balanced through either all possible ord ers or selected orders. In
selected orders, we can use either the Latin square technique or rotation of
a random starting order technique. Differential transfer is the biggest
problem in any repeated measures design.
When two or more independent variables are studied in the same
experiment, it is called a complex design. Complex designs can be used to
reveal the main effect as well as interaction effect between independent
variables. 2 X 2 design is the simplest possible complex design, in which
both indepe ndent variables are studied at two levels. Additional
independent variables can also be included to yield designs such as the 2
X 2X 2, the 2X 3X 3, etc. Complex designs can provide lot of information
to the researcher, irrespective of whether statistically significant
interaction effect is present or not present. Complex designs are also used
to resolve the contradictions arising from theories and to draw causal
inferences based on the natural groups design.
If a true experiment is not possible the researc her should use quasi -
experimental approach. A particularly strong quasi -experimental
procedure is the non -equivalent control group design. All major threats to
internal validity except those associated with additive effects of (1)
selection and history, (2 ) selection and maturation, (3) selection and
instrumentation, and (4) threats due to differential statistical regression are
controlled by non -equivalent control group design. An experimenter must
be sensitive to possible contamination resulting from comm unication
between groups of participants, problems of experimenter expectancy
effects (observer bias); questions of external validity; and novelty effects,
including the Hawthorne effect also.
In pretest -posttest design, simple interrupted time -series des ign can also be
used. In this design, the researcher needs to look for an abrupt change
(discontinuity) in the time series that coincides with the introduction of the
treatment. Some of the threats to internal validity in this design are history
and instru mentation. But instrumentation threat can be controlled by using
an equivalent control group and history threat can be controlled by using
non-equivalent control group.
Apart from psychologists, other professionals such as educators, political
scientists, and sociologists, are often involved in this process. Types of
program evaluation encompasses assessment of needs, process, outcome,
and efficiency.
munotes.in
Page 73
73 Experimental and Quasi-Experimental Methods 3.9 QUESTIONS a) Why psychologists conduct experiments?
b) Describe three conditions necessary for caus al inference.
c) What are the threats to internal validity and how external validity can
be established by the researchers?
d) Compare matched group design with random group design.
e) In which type of experiments, natural group design should be used?
f) What is practice effect and how it can be countered?
g) What is meant by complex designs. Discuss in detail the main effect
and interaction effect in complex design?
h) What is the difference between analysis plan with interaction effect
and without interaction effect in complex design?
i) Elaborate on the threats to internal and external validity of interaction
effect.
j) What are the characteristics of true experiments?
k) Discuss in detail the threats to internal and external validity of tru e
experiments.
l) What is meant by quasi experiments? Elaborate on the threats to
internal and external validity of quasi experiments.
m) What is program evaluation and why it should be done
3.10 REFERENCES Shaughnessy, J. J., Zechmeister, E. B. & Zechmeister, J. (2012).
Research methods in psychology. (9th ed..). NY: McGraw Hill.
Elmes, D. G. (2011). Research Methods in Psychology (9thed.).
Wadsworth Publishing.
Goodwin, J. (2009). Research in Psychology: Methods in Design
(6thed.). Wiley.
McBurney, D. H. (2009). Research methods. (8th Ed.). Wadsworth
Publishing.
Forrester, M. A. (2010). Doing Qualitative Research in Psychology: A
Practical Guide. Sage.
***** munotes.in
Page 74
74 4
QUALITATIVE RESEARCH
Unit Structure
4.1 Introduction
4.2 Philosophy and conceptual foundations; proposing and reporting
qualitative research
4.2.1 Approaches to building body of knowledge
4.2.2 Positivism
4.2.3 Relativist social constructionism
4.2.4 Attempts to move beyond the relativism -realism debate
4.2.5 Theoretical issues
4.3 Grounded theory
4.3.1 Background and definition
4.3.2 Analysis: Memo Writing & Coding
4.3.3 Writing up the analysis
4.4 Interpretive phenomenological analysis
4.4.1 Backg round, Understanding human experience
4.4.2 Double hermeneutic, Case study approach
4.4.3 Analysis
4.4.4 Writing descriptive summaries
4.5 Discourse analysis
4.5.1 Background
4.5.2 Designing your study and collecting data
4.5.3 Transcription & Coding
4.5.4 Analysis and interpretation of the data
4.5.5 Writing up the analysis
4.6 Narrative analysis
4.6.1 Background
4.6.2 Analysis and Interpretation of the Data
4.6.3 Writing up the Analysis
4.6.4 Other Models of Narrative Analysis
4.6.5 Critical Issue : Does Narrative Analysis Always Analyze Text?
4.7 Conversation analysis
4.7.1 Introduction
4.7.2 Taking turns in conversation: How people use a ‘locally
managed system’
4.7.3 Sequence & Structures in conversation
4.7.4 Structures in conversation
4.7.5 Transcription in Conversation Analysis munotes.in
Page 75
75 Qualitative Research 4.7.6 Analysis: Writing Up the Analysis
4.8 Summary
4.9 Questions
4.10 References
4.1 INTRODUCTION Psychologists have increasingly become concerned about the restrictive
nature of nomothetic approach (quantitat ive research) to research.
Psychologists have been acknowledging the fact that in psychology it is
the human beings doing research on other human beings and therefore
some other methods need to be used. Though various branches of
psychological research, su ch as clinical psychology, developmental
psychology, personality research, ergonomics, etc., had been using
qualitative research methods in the past but due to the dominance of
quantitative methods, it is only now that qualitative methods are getting
due r ecognition in psychological research. Some of the methods they have
found useful to understand and interpret the individualistic experiences are
interpretative phenomenological analysis, discourse analysis, narrative
analysis and conversational analysis. Q ualitative research methods have
become popular now due to specifically three reasons:
a) Growth in Theoretical Positions:
In the past, psychology was dominated by the philosophical idea of
positivism that favoured quantification. But in last few decades,
psychologists had been embracing social constructionism and
phenomenology and this has made them move towards qualitative
research methods. For example, since early 1980s, psychologists have
been paying more attention to language as they have started belie ving that
language is not just a tool to express our inner world but also a tool that
creates the reality of the world that a person lives in. This belief gave
impetus to discourse analysis.
b) Critique of Social -Cognitive Approaches to Psychology:
The se cond reason for qualitative research methods becoming popular is
that there is growing dissatisfaction of social -cognitive approach among
psychologists. Psychologists are of the opinion that research should study
people in context of their social worlds instead of conducting
individualistic, de -contextualised experiments.
c) Recognition of the Limitations of Quantitative Methods:
There has been an increasing criticism of psychology’s over dependence
on producing knowledge that is predominantly based on ex perimental
studies and quantitative measures. The argument was that quantitative
studies do not cover the entire richness of the human behavior and it is not
possible to generalize experimental conditions to other conditions. In other
words, quantitative s tudies are low in ecological validity. Critics believe munotes.in
Page 76
76 Research Methodology for Psychology
76 that naturalistic studies can give much more information that is related to
context and can be generalize to real world.
d) Influences from Outside Psychology:
Psychologists have been getting influenc ed by other social sciences that
are closest to psychology, e.g., sociology , social anthropology, etc. These
other social sciences have been predominantly using qualitative methods.
Psychologists felt that these qualitative methods can be gainfully used i n
psychology too. Apart from that, even funding bodies also gave a push to
psychology towards qualitative research methods, without undermining
the importance of quantitative methods.
Basically, quantitative and qualitative research, psychologists are
interested in finding out how people think, feel and behave, what
influences their thoughts, emotions and behavior, what are the meanings
that people attach to things, how ideas, events or things are represented in
language and how people make sense of them a nd what are their
consequences. To investigate these areas, researcher needs to raise
research questions and make predictions (hypotheses). Research questions
help us to explain what is happening, these explanations are called
theories.
However, quantitat ive and qualitative research differ in certain areas, such
as:
Research Questions:
In qualitative research, we don’t make predictions (or raise hypotheses),
we only raise research questions. These questions are much different from
the type of research que stions raised in quantitative research. The focus of
research questions in quantitative research is on statistical relationships or
differences, while in qualitative research focus is on participants’
experiences and making sense of those experiences.
Gathering evidence, in the form of data – in quantitative research it is
numerical while in qualitative research, it can be in the form of words,
pictures, etc.
Possible explanations or theories:
Some qualitative methodsaim to generate theories and examine how good
these theories are for explaining what is happening in the data.
Grounded theory is developed by analyzing qualitative data in an
inductive manner. The data analysis is influenced by researcher’s
theoretical sensitivity. In this theory, the analys is and reflection takes
place through memos. Memos are considered to be live entities and can be
in variety of forms depending on the data. Memos help the researcher in
raising new questions, comparing cases, developing concepts and
identifying their relat ionships. While writing report of grounded theory, to
gain credibility, it is necessary for the researcher to overtly explain how he
has adhered to principles of inductive logic. munotes.in
Page 77
77 Qualitative Research Now let us turn our attention to how qualitative research has moorings in
philosophical assumptions.
4.2 PHILOSOPHY AND CONCEPTUAL FOUNDATIONS; PROPOSING AND REPORTING
QUALITATIVE RESEARCH Philosophical issues are the bases on which psychologists conceptualize
and conduct research. Two branches of philosophy are especially relev ant
to research in psychology. They are epistemology and ontology.
Epistemology:
Epistemology refers to that branch of philosophy that asks questions about
knowledge, beliefs and truth.
It deals with questions like what is truth and how do we know whethe r
what we have got is truth or no, what is the difference between knowledge
and beliefs, what are facts, etc.
Ontology:
Ontology refers to a branch of philosophy that asks questions about what
things exist in the world. It is about what is in existence an d real. It
focuses on defining and cataloguing the things that exist. For example,
ontology will deal with question like does personality really exists?
Researchers’ beliefs (either explicitly or implicitly) about epistemological
questions, such as, how do we recognize knowledge, what can be
considered as evidence, what is truth and how do we recognize it, etc., will
determine how they do research and how they evaluate research done by
other people. In other words, we can say that there are many different
approaches and different assumptions about philosophical issues that a
researcher can adopt to do research in psychology. Let us look at some of
these approaches for better understanding,
4.2.1 Approaches to building body of knowledge :
Till 20th century, Po sitivism had been a dominant school of thought in
research in psychology, but now there are other approaches that are
entering in research in psychology, e.g., social constructionism. Let us
look at both of these schools of thoughts.
4.2.2 Positivism:
The main characteristics of positivism are:
Science states that objective knowledge or facts can only be gained
from direct experiences or observation. There is no place in science
for hypothetical or simply speculative things such as theories and
concepts. munotes.in
Page 78
78 Research Methodology for Psychology
78 If proper methods and tools are used then science can be value free
and objective process. Positivist believe that science gives us
objective tools that we can use to measure the world objectively, and
thus, we can bypass our own subjectivity to create objec tive
knowledge.
Science is based on the analysis of numerical (quantitative) data that
are collected through a strictly defined set of procedures. These
procedures are different from those that are used to gather ‘common
sense’ or lay knowledge.
The propos ition made within science are based on facts. Hypotheses
are tested to find out whether the facts are in congruence with the
propositions(theories) that have been put forward.
The main purpose of science is to create universal causal laws - that is,
overarc hing explanations of what things directly cause other things.
This is derived from the search for empirical regularities where two
things consistently occur together (this is also known as ‘constant
conjunction’).
According to positivism, cause is nothing more than constant
conjunction – and all that we need to demonstrate a causal
relationship is to observe (reliably and often - not just once) constant
conjunction.
We don’t need anything other than these types of general laws to
explain the world.
Psycholog ists can simply transfer the methods and assumptions of the
natural sciences to our discipline.
let us see how Positivistic approach influences the methodology of
research:
1. As positivists believes in directly observing the phenomena under
investigation , they will be skeptical about using participants’ accounts
and self -reports as useful data.
2. For positivists, it is very important to use tried and tested methods.
They believe that psychologists are not objective, they are not free of
their subjective experiences and biases. However, they can produce
objective knowledge with the help of proper use of standardized tools.
They can be trained to use these tools properly.
3. Positivists believe that quantitative data is better than qualitative data.
They have more faith in numerical data rather than non -numerical
data and its interpretation. However, this attitude is changing
gradually.
4. Positivists believe that experimentation is the most important method
to do research as it allows a researcher to hav e necessary control over munotes.in
Page 79
79 Qualitative Research the variables and they can conclusively establish the constant
conjunction between the things.
5. Consequently, positivists regard repetition as an important criterion to
determine the patterns of conjunction and to build the caus al laws.
Challenges to Positivistic Approaches:
Realism and Psychology:
Before we look at the challenges to positivistic approach, it is necessary to
understand that world can be divided into two parts –
Entities:
Entities mean anything that we consider actually exists in the world. This
can include things that we can physically touch and see as well as things
that consider that they exist but we can’t directly see or touch them, e.g.,
we can directly see people, things, places, etc. Things that we can’t
directly see or touch but believe that they exist can be personality,
intelligence, obedience, etc. These entities can be straightforward like a
person, door, table, etc., or they can be complex ones such as education,
justice system, etc.
Our Representa tion of entities:
This refers to the way we understand these things. It can include the way
we conceptualize and describe the entities that exist in the world. For
example, it can be our mental representation of things, which are basically
the products of our visual perception and cognitive representation. We
may describe these representations in the form of words or images.
Realism is the view that our representations of the things in the world are
relatively straight forward reflections of the way tho se things actually are.
This is also known as the ‘realistic ideology of representation’. Realism
believes that entities exist independently of being perceived, or
independently of our theories about them. They are represented on the
surface in the form o f behavior, language, knowledge, thoughts or
documents, etc.
A realist scientist tries to establish a link between surface representation
and underlying entities (reality). Realists believe that we can meaningfully
differentiate between entities and our r epresentations of them and thus can
judge the accuracy of the representations.
This realist approach is totally in contrast to positivistic approach,
especially if we positivism is applied to psychological entities like
memory. Personality, etc. Realism i s problematic for positivistic approach
because it believes that the link surface representation with underlying
entities (reality) is impossible, irrespective of whatever methods we use.
Our knowledge of the world is never a simple reflection of the way t he
world actually is, but is created and sustained through subjective social
processes – and particularly through language. It is difficult to munotes.in
Page 80
80 Research Methodology for Psychology
80 meaningfully separate surface representations from the reality of what
they represent. In such a situation, anoth er approach called relativism
gives another explanation. Relativism says that we cannot meaningfully
reach out to psychological reality, we can only access the representation of
it.
Relativism can be defined as the view that our representation of the thin gs
in the world are socially constructed and can’t be seen as simple
reflections of how those things actually are.
It is better to consider relativism and realism as two extremes of same
continuum, instead of considering them as two separate approaches.
Objectivity and the socio -political context of research:
In contrast to positivism, many scientists have argued that science and the
knowledge that it produces are not completely objective. They argue that
objectivity is not an automatic outcome of using c orrect research methods.
After all, research is being carried out and interpreted by human beings.
These researchers cannot be completely detached from their values and
biases. Secondly, like any other social activity, research is also carried out
within h istorical, political and social contexts and these all have an impact
upon the kind of research questions that are being asked and the methods
chosen to find answers to those questions. So, we can say, at least to some
extent, subjectivity is an inevitable part of research. It is important for
researchers to be reflexive – to reflect upon how their own views, attitudes
and experiences may influence their research activities.
Experimentation and Ecological Validity:
Positivism’s over emphasis on experimenta l methods and control is often
criticized for its lack of generalizability. The concern about how well we
can generalize the results from research situations to real world is known
as ‘ecological validity’. It is argued that positivist approach is very low on
ecological validity. It gives us lots of information about how people
behave in experiments but not in real world.
The Different views of causality: The importance of meaning
Some psychologists believe that positivism should not be used in
psychology, especially in social psychology, as it is not a suitable
approach. Positivism assumes that causation takes place because one
variable has causal properties that can impact another variable. But now
psychologists believe that to understand causation, it is important to pay
attention to the significance of ‘meaning’ while explaining the relationship
between two variables.
4.2.3 Relativist Social Constructionism :
Relativist social constructionism has its roots in other social sciences such
as sociology an d it became popular in 1970s. It has been responsible for
giving impetus to qualitative research methods in psychology. The main
characteristics of relativist social constructionism are: munotes.in
Page 81
81 Qualitative Research Science is just one way of looking at the world and there are many
other ways of looking at the world. Therefore, science and scientific
methods should not be considered superior to other approaches.
According to relativist social constructionism, it is not possible to
have a rational procedure to determine the truth or to determine which
forms of knowledge are better than others in a truly objective manner.
These decisions always get influenced by culture, morale, values, etc.
Our perceptions and understandings of reality are all we actually have
access to, so reality does not meaningfully exist as something separate
from our ways of understanding it.
Language is the most important means of representing and
understanding the world and should therefore be the main focus of our
research. If truth meaningfully exists in the f orm of our
representations of it, then we should study those representations
(means language) to get to truth.
To understand people, we must understand the context and meaning
in its full complexity.
Research gives us working hunches about the world, an d these
hunches are inevitably shifting and imperfect and do not give us fixed
facts.
Qualitative methods are more useful as they focus on language and
meaning.
Let us look at the methodologies used by relativist social constructionists:
Methodological i mplication 1:
Social constructionists argued that the veracity of academic attempts to
explain what is going on in the world can’t be objectively evaluated. We
can only check whether those academic explanations are feasible and
convincing. However, knowle dge can be evaluated by using another
criteria. For example, we may ask whether this academic explanation
helps me to find solutions to the problem or brings any desirable outcome.
Methodological implication 2:
The goal of psychology is not to find out pre-existing truth. It is more
relevant for psychologists to find out the consequences of believing certain
things to be true and other things to be false, or finding out the
implications of talking about things in a particular way, rather than finding
out whether things are actually true or not. Truth is something that we
create and derive through social interaction and through actively trying to
make sense of the world around us. Truth is not something lying
somewhere for psychologists to come and discove r.
munotes.in
Page 82
82 Research Methodology for Psychology
82 Methodological implication 3:
Many social constructionist researches use research methods that involves
the examination of language, e.g., Discourse Analysis.
Methodological implication 4:
Social constructionists view those research methods more us eful that
permit us to explore meaning. They value accounts of participants very
valuable while positivists consider them as problematic.
Challenges for relativist social constructionism:
1. Many social constructionists are not too happy about the more
relativistic form of constructionism as it prevents them from taking
any moral, ethical or political standpoints or to question any falsehood
and oppression. We cannot compare surface representations with
entities, so we can’t either support or refute any c laims.
2. Extreme relativism is more focused on language while ignoring many
other important aspects of the things under consideration. The
criticism for this is what came first - the language or reality? The
emphasis should be on reality more than the la nguage. Relativists
must find an approach that recognizes and accepts both the socially
constructed nature of the world and its material reality too.
4.2.4 Attempts to move beyond the relativism -realism debate:
Psychologists have been trying to move away from extremes of relativism
and realism and find other approaches that may help to find the balance
between these two approaches or find alternative ways of gathering
knowledge. There are two such approaches - Critical realism and
phenomenology. Let us see each one of them.
Critical Realism:
Many psychologists believe that both extreme realism of positivism and
extreme relativism of social constructionism are equally undesirable. The
moderate amount of both put together is called ‘critical realism’. Some
common characteristics of critical realism are –
It rejects the extreme realism of traditional positivistic approaches.
Knowledge is considered as historically and culturally specific. It is
believed that research methods are not totally objective from this
point of view and research is a social process that is always conducted
in the context of values.
Language is not simply a reflection of the ‘reality’ of the world but
also capable of shaping our thoughts and conceptions of the what is
real. Consequently , it influences which actions are seen as legitimate
and which are not. munotes.in
Page 83
83 Qualitative Research It is possible to gain access to a reality beyond discourse, even if this
access is not a perfect one.
Knowledge of this reality is always distorted to some extent by our
perspectives , power and culture.
Though knowledge and truth are social constructs to some extent but
truth claims can be evaluated against evidence.
However, research in psychology seldomly adopts an explicit critical
realist position.
Phenomenology:
Phenomenology is a philosophical school of thought as well as a popular
research method in psychological research. Edmund Husserl is known to
be the founder of phenomenology. In his phenomenological approach, the
research process begins with the ‘bracketing’ of the questi on whether
people’s experiences and their reporting of these experiences can be linked
to any kind of reality that is separate from those experiences. Bracketing
refers to an idea that we can leave aside the question of whether people’s
experiences are sep arate from reality. If it is agreed upon that
understanding of experiences is the main aim of psychology, then
scientific method, investigating the variables and their causal relationships
are of no use. It is also irrelevant to see whether the experiences of a
person match with some reality or not that is beyond that experience.
However, there are some forms of phenomenology that allow us to avoid
choosing between extreme realism and extreme relativism.
4.2.5 Theoretical issues :
There are two theoretic al issues that may influence the way researchers
plan and execute their qualitative research. These are:
i) the link between language, reality and thought
ii) the issue of experience and how we can explore it
Let us look at each one them.
The relationsh ip between language, reality and thought:
Psychologists believe that language is a set of symbols that we use to
share information about our inner states such as thoughts, feelings, etc.
But relativists believe that language is something that pre -exists an d
actually shapes our thoughts. Consequently, we can say that the way we
experience the world and even our internal states is only through pre
existing structures and forms. This also means that we can have thoughts
only through the concepts that pre -exist in the language and are given to
us by the language. It is due to this belief that some relativist social
constructionists say that there is ‘nothing beyond the text’. For them,
studying language is very important to do research in psychology, as it is
the basic requirement for making any sense or thoughts. Relativists munotes.in
Page 84
84 Research Methodology for Psychology
84 analyze people’s talk and interactions to find out how people make sense
and use language to achieve certain kind of things in interactions such as
to make claim, to lay blame, to defend the ir position or to work out their
identity. They believe that ‘talk’ is the medium through which the world
becomes real. Though there is no denying that there is reality beyond talks
too but it is not accessible to the researcher.
Experience and how we can explore it:
As we have already discussed that psychologists can gain insight into the
experiences of others through language, but it should be kept in mind that
this process gets influenced by their own views about what is language.
The question arises, c an we gain knowledge about the experiences and
perspectives of others by using relativist social constructionism approach.
To resolve the issue of experience and subjectivity, some psychologists
adopt psychoanalysis, while others adopt critical realist of social
constructionism. There are some similarities in phenomenological
approach and critical realism too. Proponents of each of these viewpoints
believe that research based on these viewpoints can help us in
understanding others’ experiences.
4.3 GROUNDE D THEORY Grounded theory is an extremely popular and powerful qualitative
approach to do research in psychology. The goal of grounded theory is to
understand the psycho -social phenomena that is grounded in the data. It is
based on the presumption that a ‘ theory’ we might have about a topic or
issue should be ‘grounded’ in the data we collect from people. Grounded
theory uses inductive approach to doing research. Generally, when we use
hypothetico -deductive model of research, we use deductive approach. We
begin with a theory, develop hypotheses to test whether that theory is valid
or no. On the other hand, grounded theory does not begin with review of
literature, but collects data right from the beginning. It looks at the details
of individual cases, uses in ductive logic and develops a theory that is true
for those cases.
4.3.1 Background and definition:
Charmaz defined grounded theory as a set of systematic inductive methods
for conducting qualitative research having the goal of theory development.
The term grounded theory highlights two things:
(a) a method having flexible methodological strategies
(b) the products or outcomes of this type of inquiry.
Grounded theory was originally developed by Barney Glaser and Anselm
Strauss. Glaser believed in standard hypothetico – deductive type of
research while Strauss was more interested in symbolic interaction. In
1965, both these researchers teamed up to conduct a study on the process
of dying in hospital. After two years , they formalised the grounded theory munotes.in
Page 85
85 Qualitative Research by publishing it in a seminal ‘Discovery of Grounded Theory : Strategies
for qualitative research (1967)’. In 1990s, Strauss joined hands with
Corbin and jointly they shifted from the concept of the natural emergence
of theory. They also believed that researc her using grounded theory should
not totally abstain from literature review. This led to split between Glaser
and Strauss.
It was argued that a study can be called grounded theory study only if it
produces a substantive theory, i.e., a theory that may not postulate
universal laws of human behavior, but has its own context. Apart from
developing theory, it is also necessary that researcher himself should have
theoretical sensitivity.
Theoretical sensitivity means a researcher should have characteristics a nd
skills that are important for developing codes and categories. Theoretical
sensitivity helps the researcher to become ‘more -in-time to the meanings
embedded in the data’. Charmaz (2009) explained that theorizing means
stopping, pondering and rethinking… To gain theoretical sensitivity, we
look at studied life from multiple vantage points, make comparisons,
follow leads and build on ideas…When you theorize, you reach down to
basics, go up to abstractions, and then probe into experience.
Researchers with t heoretical sensitivity can insightfully reflect on the
subtleties of a developing theory in a way that is creative and conceptual,
rather than merely descriptive.
Theoretical sensitivity determines the way a researcher chooses the sample
and collects the d ata. This is known as theoretical sampling that uses
gradual sampling strategy. In qualitative research, usually purposive
sampling is used as generalization from sample to population is not
emphasized. The participants can be either chosen a priori, that is, before
the start of the study, or they can be chosen gradual, i.e., participants are
chosen individually while analysis is still going on. Most of the grounded
theory studies start with a priori method of sampling and then use a
gradual strategy to sel ect further participants, after the analysis of initial
data has begun. The process of theoretical sampling helps the researcher to
look at all the possibilities of a theory that is developing from that study,
so this process goes on till the end of the st udy. When a researcher has
collected sufficient data to fully develop all his conceptual categories and
is satisfied with the theory that he has developed, he reaches a point called
theoretical saturation. Theoretical saturation is a point where adding n ew
data does not contribute new details or properties to the conceptual
categories already developed.
4.3.2 Analysis: Memo Writing & Coding:
Grounded theory develops when researcher reflects or analyses qualitative
data based on his theoretical sensitivit y. This analysis is an inductive
process that moves from the details of the participants’ lives to a theory
that explains the underlying process and dynamics in more general terms.
The analytical process goes on throughout the life of a grounded theory munotes.in
Page 86
86 Research Methodology for Psychology
86 project – it starts as soon as the researcher has collected some data and
continues until he has finished writing his report.
Memo Writing:
Analysis and reflection take place in the form of memos. A memo is a
written reflection of researcher’s analysis. It is through memos that a
researcher reflects on the definitions and properties of emerging concepts,
ask questions, compares one case with another, record insights, advances
tentative ideas about develop theories, etc. Memos can include diagrams
too, depictin g the relationship between concepts that the researchers is
trying to specify. Grounded theorists keep writing memos, sketching their
ideas, recording their ‘Aha’ moments, etc. Much of the analytical work
gets done by producing the memos. Memo -writing star ts at the very
beginning of the research process and continues until the final product of a
grounded theory project, i.e., the theory itself, is produced from memos
written at a late stage of this process. In the initial stage, these memos
may be just tho ughts noted down when the researcher is reading the
transcript for the first time. This may be like open codes and they are
tentative in nature. Later on, as the study progresses and the researcher
becomes surer of the concepts and categories, the memos be come more
formal. However, they still remain flexible and a researcher can change
his mind and consequently the memos too.
According to Corbin and Strauss (2008) ‘memos and diagrams are more
than just repositories of thought. They are working and living d ocuments.
Corbin and Strauss (2008) gave a list of 13 analytic tools. Some of the
common ones we will discuss here:
Constant comparison is a most popular analytic tool among grounded
theorists. It is used in all stages of the research. The researcher
constantly compares concepts to see how they might fit in the greater
scheme of things. This comparison shows the possible relationships
between concepts and that will help in structuring the conceptual
categories.
Asking Questions:
The grounded theorist cons tantly checks out his own analysis and
generates more questions, as he proceeds. This technique of constantly
questioning yourself will either provide the answers needed or it may
highlight to the researcher that the collected data does not have the answe r
to a specific question and the researcher needs to collect more data
through theoretical sampling.
Coding:
There are many techniques for analyzing the data, but the most obvious
one that is used in grounded theory is coding. Coding here refers to thre e
types of coding:
munotes.in
Page 87
87 Qualitative Research a) Initial coding/open coding:
When data is broken down into the conceptual components, it is known as
initial coding. It begins as soon as some data is collected to work on. It
basically means taking a large portion of text from the data and giving it a
label ( code title). While Glaser (1978) and Strauss and Corbin (1998),
called it open coding, Charmaz (2014). Referred to it as initial coding. The
size of the portion of text taken differs from one stage to another stage,
e.g., in th e beginning, a theorist may go through a transcript line by line or
phrase by phrase. Later on, as the theorist starts getting some ideas to
work on and develops certain questions, the portion size may increase to
one whole paragraph having one title. Thes e code titles are the first step
towards naming the concepts, and those concepts can be the building
stones of the theory. It is suggested that in these initial codes, it is better to
use verbs rather than topics or themes.
b) Intermediate coding / Axi al coding:
When these conceptual components are arranged in to categories it is
called intermediate coding. Glaser (1978) named this stage of coding as
‘selective coding’, while Strauss and Corbin (1998) called it as ‘axial
coding’ and Charmaz (2014) label led it as ‘focused coding’. In
intermediate coding we look for relationships that might indicate aspects
of a developing theory, such as causal connections or indications of basic
social processes. At this stage memos in the form of diagrams are very
usefu l. Sometimes the spatial arrangement of the categories can very
easily depict the nature of the theoretical relationship between them. A
grounded theorist can write an addition to these memos adding his
reflections about the idea that is developing. This p rocess of reflecting will
generate new ideas and new perspectives that also can be jotted down and
coded.
c) Advanced coding:
This is the final stage of coding in grounded theory. Strauss and Corbin
(1998) labelled this as ‘selective coding’, whilst Gla ser (1978) and
Charmaz (2014) both called it ‘theoretical coding’. At this stage, from
various categories, a core category is chosen and then all other categories
are organized around that core category. At this stage, the grounded
theorist needs to thorou ghly check all concepts that he has developed.
Corbin and Strauss (2008) gave some tips for the core category:
1. It must be abstract; i.e., one should be able to relate all other major
categories with it and placed under it.
2. It must show up in the d ata frequently.
3. It must be logical and congruent with the data. There should be no
forcing of the data. munotes.in
Page 88
88 Research Methodology for Psychology
88 4. It should be abstract enough, so that it can be used to do research in
other substantive areas, leading to the development of a more general
theory.
5. It should increase in depth and explanatory power as each of the other
categories are related to it.
Strauss and Corbin (1998) suggest that at this stage one should use ‘coding
paradigm’. Coding paradigm refers to a set of ways of thinking about
intermediate (axial) categories to make sure that the relationships between
and within the categories are fully explained. Elements of the coding
paradigm include:
Conditions:
The researcher must explain which conditions might have a causal
relationship be tween concepts or stages of a process, and which conditions
might show an important context or intervening function.
Actions and interactions:
How people (and organisations) deal with the situations as they arise, and
what habitual activities and rituals can be identified in these instances.
Consequences:
The ground theorist must also explain what are the outcomes of an event?
How extensively are these outcomes felt and how complex are their
effects?
However, Glaser has been very critical of this approach and said that this
approach is too prescriptive and it may force grounded theory analysts to
think about their data in a fixed manner and may deter the emergence of
the full detail of a theory from the data and the categories that arise from
the data. He suggested an alternative and said that the ground theory
analysts should use a wider set of theoretical coding families, which can
help them to ‘conceptualise how the substantive [intermediate] codes may
be related to each other’
Finalising Theory:
Finally , while writing the theory, the grounded theorist must keep in mind
that the theory should not be just the description of the data. It should be
strong enough to explain, in psycho -socio and cultural terms, the dynamics
of the contexts in which our partici pants carry out their lives. Though the
theory will be embedded in the specific situation, it should have more
extensive principles that can allow future researchers to test hypotheses
that might be generalization oriented.
One should also keep in mind th at collected data does not throw up only
one ‘right’ theory. Researchers’ own subjective experiences, beliefs, and
philosophical orientations may influence their analysis of the data which munotes.in
Page 89
89 Qualitative Research may be totally different from the theory already developed. There i s no
right or wrong theory, they just have different perspectives.
4.3.3 Writing up the analysis :
Grounded theory reports is generally written in a more or conventional
manner. Though it was explained in the beginning that literature review is
not essenti al in the beginning of the grounded theory study but in report it
is included in the beginning only just like other conventional reports.
It is very important for grounded theorists to be very clear and transparent
about the details of their research pro cess. The readers will be able to trust
a researcher’s analysis as a good example of grounded theory research
only if he gives enough detail to show them that he has actually
implemented the entire approach to full extent. If the researcher does not
provid e the evidence of his interpretations, it may lead readers to believe
that he might have just ‘borrowed’ the grounded theory coding process
and applied that to data collected from people selected with little
forethought to what the requirements of the stud y might have been. The
researcher must make sure that he gives evidence (such as interview
transcripts or field notes) from the data that he has analyzed to support his
claims.
The analysis section of grounded theory generally tends to be the largest
part of the report. It covers almost half of the total report. There is no fixed
rule about how to write the analysis section. But the most common way to
write analysis section is to write in detail about the core category and
about as many of the other categor ies as necessary to explain the grounded
theory that the researcher is proposing. The results section will mostly be
combining and clarifying some of the later memos. The reader should be
able to follow at least some of the major ideas right from transcri pt to
theory.
Lastly, in the discussion section the researcher must clearly show how his
grounded theory increases the readers’ understanding of the area under
investigation. Though it is not essential, but if the researcher wants he can
end his report by writing one or more testable hypotheses derived from his
data.
4.4 INTERPRETIVE PHENOMENOLOGICAL ANALYSIS (IPA) Interpretive phenomenological analysis is a qualitative method that is
influenced by phenomenology and hermeneutics. It aims to understand the
meaning of human experience. Phenomenology refers to the study of
human experience and the way in which things are perceived as they
appear to consciousness (Langridge,2007; extracted from Forrester
M,2010)
Hermeneutics refers to a theory of interpretation. In contrast to nomothetic
method, IPA adopts an idiographic method of inquiry. Nomothetic method
is used in quantitative research, which uses large data and gives munotes.in
Page 90
90 Research Methodology for Psychology
90 probabilistic conclusions. Idiographic analyses is done on a small scale. It
is based on the assumption that all individuals are unique, so the study is
done at the level of an individual case and conclusions are drawn on the
basis of individualistic studies.
4.4.1 Background, Understanding human experience:
The basic aim of IPA is to understa ndwhat personal and social experiences
mean to those people who experience them. The IPA researchers treat
experience as a thing as well as a process that people come across and are
active in. So, the unit of inquiry is ‘experiential account’.
Understandin g human experience:
IPA believes that reality exists but we cannot directly access it, We can
access that reality only through the particular perspective of the person
describing the event in a particular place and time. Therefore, instead of
just trying t o understand the experiences of the people, IPA researchers
tries to understand in which articular stage of life those experiences
occurred as well as what was the social, cultural, political and economic
context when those experiences took place. In other words, IPA researcher
wants to explore what it is like to empathize with another person and to
make analytical interpretations of his experiences as well as make
interpretation about the person as the ‘experiencer’. The researcher not
only pays attention to how the event is described by the participant but
also interprets how participant is doing his own sense making during the
interview.
4.4.2 Double hermeneutic, Case study approach:
During the interview, IPA researcher encourages a person to not only
describe his experiences but also to reflect himself on those experiences
and describe what meaning those experiences have for that person. At the
same time, the researcher is also trying to make sense of what the
participant is describing. Thus, dual hermene utic takes place where the
participant is trying to make sense of his own world and the researcher is
trying to interpret or make sense of how the participant is making sense of
his world.
Case study approach:
Since IPA is an idiographic method, for dat a collection it uses tools like
semi -structured interviews, participant’s diaries, case study method, etc. In
case of single case study method, data about one person is collected from
multiple sources. If there are more than one participant in the study,
analysis is done on a case -by-case basis. Fully worked up analysis is done
for case one before going to next case and then in the final stage of
analysis, these cases are compared. IPA is interested in getting an in -depth
information about each person as ge tting richness in data is very important
for IPA. Interpreting the data also requires lot of patience, time and
creativity of IPA researcher. munotes.in
Page 91
91 Qualitative Research 4.4.3 Analysis:
In IPA analysis, generally transcript is written in ‘playscript’ style, i.e.,
verbatim transcript is written down in sequence and only occasional
reference is made about non -verbal actions or events. Let us see in detail
how IPA analysis is done
Initial thoughts on reflection and quality:
Right from beginning of the research, an IPA researcher should keep a
reflexive journal and make notes of anything that comes to his mind
regarding the research. As transparency needs to maintained for research
to come across as trustworthy, it is advised that the researcher should also
keep copies of notes made, key extracts and themes identified at each
stage of the analysis. Apart from building up the trustworthiness,
maintaining notes helps the researcher in writing a clear detailed report
where a reader can understand the process the researcher went through to
produce that analysis.
Now let us look at all the stages that a researcher has to go through while
analyzing the data.
Familiarizing yourself with the data:
The first step in analysis of the data is that the researcher needs to
familiarize himself with the data. If the data was collected in digital audio -
visual format, then the researcher must read and re read the transcript and
must watch the video again and again. While familiarizing with the data, if
researcher comes across anything that he finds interest ing or important ,
worth noting, he must keep making notes in reflexive journal.
Coding and identifying Initial Themes:
The idea behind reading and rereading is that during re reading the
researcher must break down the transcript into small sections and d escribe
what is being said in each section. Then he should do ‘phenomenological
coding’ and ‘interpretative coding’.
In phenomenological coding, he needs to write down a summary of the
participant’s description of the experience or story. He should also m ake a
note of issues identified, events relayed and feelings expressed by the
participant This will give him an idea about what is important to his
participant, what were the topics of their conversation and who are the
people mentioned in his story.
Interpretative coding includes initial interpretations about what these
issues, events and feelings mean to the participant. This will indicate to
the researcher what contributes to participant’s making sense of his
experiences.
Breaking down transcript into small sections and describing these sections
ensures that analysis is data driven and not theory driven. Data driven
approach is also known as ‘bottoms up’ approach, while theory driven munotes.in
Page 92
92 Research Methodology for Psychology
92 approach refers to deductive logic where a researcher makes a decisio n
whether his analysis fits with existing theory or not. That is known as ‘top
down’ approach. Initial interpretations of these small sections ensure that
analytical process can be traced back to the raw data. Giving such audit
trial is important for the analyst to make sure that interpretations can be
traced back to the data.
4.4.4 Writing descriptive summaries:
As mentioned above, descriptive summaries are known as
phenomenological coding. It shows the phenomena under study from the
participant’s persp ective. In the early stage of reading the transcript, the
descriptive summaries merely sum up the content of what is said by the
participant. At this stage, the researcher should not try to make
interpretation of what is being said by the participant. If t he researcher
feels that he is getting lots of ideas while writing the descriptive
summaries, he should make a note of those ideas in reflexive journal.
Making initial interpretation:
Breaking down the transcript into small portions and writing down
descriptive summaries (phenomenological coding)of these portions helps
the researcher gets to know something about what is important to his
participant and about the things which make a difference to the participant
in terms of making sense of his experiences (interpretative coding)
Phenomenological coding is the building block for interpretative coding.
For doing interpretative coding, the researcher needs to re -read the entire
script once again right from the beginning, also read the summaries and
think abo ut what inferences can be drawn from the data and how it is
linked to the research question. It is important to keep in mind that
participant is describing an event in retrospect, an event that may have
taken place many years ago. Such description of a re trospective event will
include participant’s evaluation of the event too, as he also must have
thought about and assigned meaning to that event. It is important in IPA
analysis to identify the participant’s own reflection of the event and make
a note in r eflexive journal about the participant’s own sense making. This
will be useful when interpretations are made in the second stage. After all,
the main objective of IPA researcher is to understand experience from the
experiencer’s point of view and make infe rences that are specific to the
experiencer.
Clustering themes:
The themes are developed through several readings of the transcript
(familiarizing oneself with the data initially, reading for gist, dividing the
transcript into small chunks), writing descri ptive summaries, and a
comprehensive interpretative analysis of the transcript and the summaries
with regard to the reflexive journal. However, one should keep in mind
that themes don’t emerge from the data but develop from the data. If we
say themes emerg ed, it means they were pre -existing in the data, waiting
to be discovered, which is not correct. Themes develop as the researcher munotes.in
Page 93
93 Qualitative Research does close readings, summarizes the small portions, identifies the
important meanings from those summaries, finds out the recu rrent or
notable themes from initial interpretations. The researcher looks for
connections between these initial themes and creates a narrative account of
participant’s experiential account and that also reduces the data. Once the
cluster of themes is deve loped, the next step is to give a title for each new
theme.
Establishing the final themes:
A central overall theme is derived from the cluster of themes and this
becomes the final theme. This final theme is included in the research
report.
Continuing wi th other cases:
Though, ideally IPA studies can be single case studies, but it is possible to
have multiple participants in an IPA study. Once the analysis of the first
case is completed, the researcher can move to second person in his sample.
With the sec ond person as well as other subsequent cases, he needs to
repeat all the steps that he went through for the first case. For each case he
needs to make a final list of themes. There might be some cases, where in
the final list themes, some themes may be sam e, and in other cases, there
may be no similarity of themes between any two people. The goal of IPA
research is to look at similarities as well as uniqueness of the participants’
experiences that they share with the researcher.
Writing up the analysis:
The result of an IPA research is a narrative account of participant’s
account and the researcher’s interpretation of it. This is called final stage
of analysis. It must briefly define the theme and show the importance of
that theme for understanding the mea ning of participant’s experiences.
The report should be persuasive, defensible and must sufficiently
represents the participant’s story. It must also convey the interplay
between the description given by the participant, participant’s own
reflection on his experiences and the researcher’s interpretation of that
narration. Reporting in this manner makes the report dynamic, creative,
compelling and enlightening.
4.5 DISCOURSE ANALYSIS Discourse refers to any form of talk or text in any social interaction.
Discourse allows the researcher to understand the way language is used in
social interaction. This interaction may be in the form of chatting,
emailing, talking, etc. The word ‘discourse’ is used instead of ‘language’
or ‘communication’ to encapsulate the s ocial and constructive element of
interaction. If we use the word language, instead of discourse, it may focus
on grammar, punctuation or other technical aspects of the language
system, and the term communication may consider language as a means
for thoug hts to be transferred between people. munotes.in
Page 94
94 Research Methodology for Psychology
94 Discourse analysts (DA) are interested to know what is the function of
talk. They believe that it is through talk that we discuss, ask, complain,
flirt, console and deny, etc. In this process, it is through talk that w e claim
our identities, move a conversation forward or give details about our
version of events that may make it appear more believable. Discourse
analysts are interested to know how talk constructs a version of a reality
that has an interactional effect i n social situation. They are interested to
know what is being constructed or what is the effect of the talk at a given
point of time. Some of them are interested in finding answers to ‘how’
(e.g. how people’s talk exposes the shared meanings in our culture ),’
when’ and ‘why’ also. So, we can say that DA is ‘social constructionist’.
For DA analyst, talk is not just representing the emotions, thoughts or
reflecting the world as it is, instead, it constructs the world and the person
in interaction.
There are many different types of discourse analysis, each having their
own theory and methodology and emphasizing different aspects of what,
why, when and how questions. It is a flexible approach and can look at
variety of research questions.
4.5.1 Background:
Discursive psychology was developed in the 1980s, taking inspiration
from researchers like Austin and Wittgenstein and from theories within
ethnomethodology (the study of how people make meaning in interaction),
poststructuralism (a philosophical approach tha t does not accept that there
is a fixed meaning to language and considers meanings as relational),
conversation analysis and the sociology of scientific knowledge. There are
two approaches to the study of talk and text known as Discursive
psychology and Fo ucauldian -informed discursive psychology. Discursive
psychology offers a close -up analysis of talk, and Foucauldian -informed
discursive psychology (FiDP), looks at language in a broader way. We
will now check out each one of these.
Discursive psychology ( DP):
Discourse analysis is also known as discursive psychology. Discursive
psychology was developed in the 1980s. It was inspired by researchers
such as Austin and Wittgenstein and by theories within –
ethnomethodology (the study of how people make meanin g in interaction),
post structuralism (a philosophical approach that does not accept a fixed
meaning of language and considers meanings as relational), conversation
analysis and the sociology of scientific knowledge.
All these diverse influences had someth ing in common. They were all
concerned with the action -orientation of discourse (that words do things),
the empirical study of social interaction, and the development of methods
that enabled researchers to audio -record, transcribe and analyse real -life
conversations munotes.in
Page 95
95 Qualitative Research Derek Edwards and Jonathan Potter (1992) combined many of these
previous ideas and came out with a methodology that allowed researchers
to psychologically analyze every day interaction. It looks at the role of
psychological concepts such as ide ntities or emotions, to manage and
enable social actions within everyday interaction. DP does not try to
understand emotions as a mental or physiological state, rather it tries to
understand how emotions are produced in discourse and then used by the
participant to deal with psychological or social matters. It focuses on both
the turn -taking details of talk and the words used: on the sequence and
structure of conversations as well as the content of what is said. In other
words, discursive psychology conside rs discourse as something that can be
used to construct individuals’ versions of the world and answers questions
such as what, how and when psychological concepts are utilized to
perform particular actions in social interaction.
Foucauldian -informed discu rsive psychology (FiDP) was developed from
the late 1970s . It was also influenced by same influencers as discursive
psychology but mainly it is inspired by poststructuralist philosopher
Michel Foucault. FiDP examines how talk or text construct particular
versions of reality and the consequences in using one version and not
another version of what people can say, think or do. Often these
constructions are part of shared cultural ‘common sense’ and take the form
of relatively coherent ways of talking about o bjects and events in the
world. It focuses on the words people use in talk and writing (and
sometimes images) and considers how this helps us to understand
ourselves and our world in the wider social context that supports this talk.
In other words, FiDP is interested in the different ways that the issue is
constructed because it sees the society as consisting of different discourses
competing for how we understand an issue. People are thought to be
moving fluidly between different discourses.
There are many different versions of discourses, each one of them offers,
different ways to make sense of our world, either within a particular
sociohistorical moment or across time. This obviously leads to
subjectivity. FIDP is also interested to know what people can do with
those constructs. (subject positions) and what are the consequences of the
talk.
Steps in the research process :
Since DP and FiDP approaches are different, their methods are also differ
to some extent. We will discuss methods of each one of them. We first
begin with DP analysis.
Devising a research question :
In DP research, research questions are framed to get an answer to what are
the various ways in which people manage psychological matters in
everyday life, such as identities, accountabilities an d mental states and
how these psychological issues become relevant in social interaction.
munotes.in
Page 96
96 Research Methodology for Psychology
96 4.5.2 Designing your study and collecting data :
In DP, we collect ‘‘naturally occurring data’, means data taken from
everyday social interaction that would have taken place even if our
research project was not occurring. Alternatively, we can use other
methods also such a s interviews and focus groups.
Before you actually start collecting data, you need to plan your data
collection on the basis of your research questi on. Be very clear about what
you want to study and how to get access to this topic or area of social
interaction. Once you are clear about what information you need to
collect, then you need to identify and approach the potential participants
for individua l interviews. You need to get informed consent from the
participants and then ensure that interviews are video - recorded, while
maintaining all other ethical guidelines. Collect sufficient number of
interviews to enable you to identify any recurring patter ns or different
issues pertaining to your research question, emerging out of this data. But
do not collect so much data that it overburdens you and you find it
cumbersome to transcribe and analyze it. Next step is transcribing and
analyze the data.
4.5.3 Transcription and Coding :
Transcription :
After collecting the data in audio or video format, you need to convert that
data into written document. First the data is transcribed in ‘playscript
style’, where the focus is on the content of the talk and other fe atures of
the talk such as length of pause, intonation, etc. are ignored. In the second
step of transcription, the researcher identifies those sections of the data
that he wants to examine in more detail, and as a third step in transcription
DP researcher uses ‘Jeffersonian’ transcription system for only those
portions of data that were identified in the second step of transcription.
Jeffersonian system includes features of talk such as length and location of
pauses, changes in intonation, emphases and over lapping speech. This
helps the researcher to know, not only what was said, but also when and
how it was said.
Coding :
Coding of data in DP refers to organizing data into smaller portions on the
basis of relevance to research questions. These smaller portio ns are
clubbed together into a ‘corpus’ of data on the basis of emerged patterns in
the data. The process of transcription is iterative, i.e., a researcher needs
to keep examining the data, making tentative analyses and interpretations,
re-checking and ex amining the data, and focusing on small sections of
transcript each time. Consequently, to get final coded ‘corpus’, the
researcher needs to do coding many times. Discourse analysis is
considered as ‘data driven’, i.e., how analysis will be done is influe nced
by research question.
munotes.in
Page 97
97 Qualitative Research 4.5.4 Analysis and interpretat ion of the d ata:
The researcher should ensure that coding is as inclusive as possible so that
no potentially important extract is left out inadvertently. The researcher
must remain focused on what i s there in the data, instead of paying
attention to what he expects to find.
Analysis process in DP is not a linear process. There searcher needs to
work backward and forward through a series of stages while examining
the data. The researcher must read tr anscript line by line and must describe
in a simple manner what is happening on each line. He must note down
how, when and what is being said. He should not try to answer why it is
being said or what the participant meant.
Then comes the second stage of analysis, where the researcher starts
looking at internal states or psychological constructs. The pauses, quieter
talk and references are called internal states or psychological constructs.
He must begin to pay attention to things such as how one is ‘feeli ng and
thinking’ or ‘not liking’ something, as these influence social actions of a
person.
The second stage analysis will most likely highlight some features of the
talk where psychological issues are particularly noticeable. At this stage,
DP uses ‘disc ursive devices’: that is, identifiable features of discourse that
help to perform social actions. This helps in interpret the data and to
identify the way the discourse is constructed.
Next step in analysis is to focus on specific analytical issues.
Focus on Specific Analytical Issues :
It can be very frustrating task for a researcher to choose and focus only on
some topics out of a cluster of themes that he has identified. It will help
researcher if he goes back to research question and existing literature to
identify and focus on something that will be within the limits of his
research and also contribute to research in that area. But he should be as
specific as possible.
Re-code the Data for all Instances of this Issue :
Once he has decided what will be the focus of his research, he needs to go
back through the full data set to check for all instances where this occurs.
He needs to go through this step even if he feels that he was very thorough
when he did coding for the first time. He can check both the orig inal
recording as well as the transcript, since some matters are easier to spot in
visual form than in written form. This step will ensure that your research is
very rigorous and methodical.
Refine & Writing up the analysis :
Based on re -coding of the data, a new file is made. The researcher now
needs to go through each data extract in turn, comparing them with each
other and searching for any patterns or differences. The researcher can munotes.in
Page 98
98 Research Methodology for Psychology
98 now begin to write up his analyses. He can turn his analytical notes int o
sentences that can be used to provide some insight into the data and form
the basis of his report. He may not be able to use all his extracts in a
written report, so he needs to select only those extracts that represent his
analysis and give the clearest and most engaging examples of the issue
that he is focusing on.
Analysis Process in FiDP Research :
Devising a Research Question :
FiDP research questions are ‘what’ or ‘how’ questions. FiDP focuses on
the reality being constructed, including different and sometimes even
contradictory constructions, the consequences for what people can say,
think and do when drawing on these constructions; and the wider
discourses and socio -historic context that enables these constructions to
make sense. To begin with, the researcher may start with a general
question and later as the data is collected, research questio n starts getting
crystallized.
Data Collection :
For FiDP almost everything is a text that can be analysed. This is because
talk, writing, images, actions and everyday objects are part of symbolic
systems that can be read for the meanings they employ.
Transcription :
If the data is not already in textual form, the researcher will have to
transcribe it from his sound or video recordings. FiDP, generally, uses
audio recordings instead of videos as it wants to be less intrusive and also
because the focus is on language and not on visual cues. FiDP uses a
simplified and less detailed transcription notation than DP. FiDP
sometimes uses the tools of DP, so some more de tail might be given, such
as intonation, emphasis or rough length of pauses.
Data Coding and Analysis in FiDP :
Just as in DP, here also, the researcher has to read the entire data to
familiarize himself. Then he needs to re -read it to make notes, ensuring
that codes are as inclusive as possible. It’s important to be inclusive and to
write as many keywords as the researcher sees, because at this stage he
doesn’t know what will be the most relevant issues, and if he doesn’t code
them at this stage, he might miss them later.
Coding through a Discursive Lens: What, How, Why?
In the previous step we searched for FiDP questions: what are the issues,
how are they constructed and why? ‘what’ question can be answered by
summarizing what is being said. Researcher ca n summarize by Paying
attention to the exact words and phrases, summarizing what the participant
is saying in less words but staying close to the data. How question can be
answered by looking at how it is said. He can look for rhetorical devices munotes.in
Page 99
99 Qualitative Research (these are called discursive devices in FiDP). An example of rhetorical
device can be ‘hundreds of people’ in talk.
Researcher might also notice if there’s any ‘trouble’ in the talk. When
people find it hard to say something they often pause, change direction,
stop and start, or make ‘um’ and ‘er’ sounds. This tells him that something
difficult to say is being managed, so he needs to look for what that
something is. By seeing what’s not being said he might get hints at the
functions of what is being said. For examp le, avoiding being labelled as
unpopular. If the researcher gets an idea about the functions of the talk at
this stage, he should write it down. Thinking about ‘what’ and ‘how’
questions can leas the researcher towards ‘why’ questions. He might think,
why is this participant saying this, to think about the consequences of talk
for the speaker. Alternatively, he might be asking why it is in our society
that this talk would make sense. He might think about where else he might
have hear similar constructions o f reality such as in the media, government
policy or psychological discourse. The same text or section of a transcript
can be coded for many issues
Stop, Review, Consolidate and Conceptualise :
After coding through discursive devices, once again look at the keywords.
You will notice that many of the keywords refer to the same thing. For
example, the researcher may have coded different sections of the text as
contentment, pleasure, glad, joy, enjoyment etc., but they all can be
described by just one word, i.e ., happiness. This becomes researcher’s key
word and he needs to go back to the beginning of the data and again
analyze the text to further analyze the text in terms of this new
conceptualized key word. For example, the researcher may find that when
partic ipants talk about happiness in relationships, they talk about
‘reciprocity’. In this case, ‘reciprocity’ becomes the defining feature of
interpersonal relationships. The researcher may find that the data appears
differently with this new conceptualized wor d and gives an new insight
about the data. This is known as shifting from descriptive analysis to
conceptual analysis.
Collate, Confirm, Develop :
Next you need to collate all the extracts coded under each conceptual
keyword. This can be done either physi cally or by using a software such as
Transana and NVivo. Many discourse analysts prefer to use physical
sorting method as sorting printed extracts again and again helps to make
creative associations. Extracts may be coded with either single conceptual
keyw ord or may have multiple keywords. Next step is to look at each
extract and identify what is being constructed and then combine two or
more extracts together to make a bunch that is defined by the construction.
If extracts do not share same construct, just divide all extracts in two
bunches and label them. The researcher can look at the answer ’what’
question to facilitate him in creating these two bunches. If at this stage, the
researcher feels that these two piles should be coded differently, he will munotes.in
Page 100
100 Research Methodology for Psychology
100 have to start all over again by comparing the extracts till he again finds a
way to conceptually categorize them. So this is a cyclical process.
Identify Discourses to Focus on and Confirm the Research Question :
Discourse is like building up an object or bri nging an issue into existence.
The researcher needs identify the extracts that enables him to explore the
different ways your research topic is being brought into being. These
extracts are chosen on the basis of quantity (i.e., whether they were
regularly occurring in talk ) or on the basis of interest (i.e. an extract that
has not been said in other discourses). In this process, the researcher can
either stick to his original research question or if some other new
reoccurring topic emerges, he can decide t o focus on that issue and rewrite
his research question.
Analyzing Each Extract: What, How, Why?
Once again you need to analyze the extracts categorized under the same
discourse and check for each extract what is constructed, how it is
constructed and th e consequences for using these constructions.
4.5.5 Writing up Analysis :
After this repetitive analysis the researcher will havea set of discourses
that construct your research topic in different ways, and an analysis of the
extracts that articulate them. He needs to decide which extracts he wants
to use for his report. While choosing extracts, he has to keep in mind to
draw from his whole data instead of just picking up a few particularly
expressive and articulate participants. For each discourse, he need s to
explain how it constructs reality, how the talk makes this reality appear
credible, the consequences for constructing the reality in this way and the
wider discourse that supports this sense making.
4.6 NARRATIVE ANALYSIS A narrative can be defined as the telling of a series of connected events for
an intended audience. It is an answer to the question ‘and then what
happened?’
Very often the word ‘narrative’ is used interchangeably with the word
‘story’ but in real sense narration is the recountin g of a story. While
recounting an event, a narrator selects which aspects of that event to
emphasize or leave out. He includes those aspects of the event that he
thinks are important and links them in a manner that will hold the
audience’s attention. He co llaborates with the audience of his story to
decide what can be told and how it is to be told.
Psychologists realize that people share their experiences with others in the
form of stories. These stories are very useful to understand what sense and
meanin g people derive from the things that they come across and how
these meanings change when people experience life changing events. So
these stories can be used to understand the shifts in a person’s identity munotes.in
Page 101
101 Qualitative Research across changing situations in his life. To study h ow people subjectively
construct or reconstruct their identities across different developmental
milestones in their lives or across any significant events, narrative analysis
is the most apt method that psychologists can use.
Psychologists also recognize that people may describe the same event in
different ways, depending upon to whom they are narrating the story, in
what context and at what point of time in their lives they are telling the
story. For example, they might be telling the story immediately af ter
experiencing the incident or may be after many years. This time gap will
have the impact on the story that is being told. The context can include
family, historical, cultural or any other context that may have influenced
the story teller’s life. In na rrative analysis, the psychologist looks at not
only the content of the story but also the way the story was told. There are
various different types of narrative analysis models. Each one of them
focuses on one aspect of the story, such as, content, struct ure or function.
These stories are studied through semi -structured interviews using
predominantly open -ended questions. Other methods of analyzing stories
are visual, auditory and performance records, diary entries, letters, etc.
While analyzing the stor y, narrative analysts have to keep in mind that the
story should not be fragmented so that there is no risk of analysts
unintentionally changing the meaning of the stories while relaying them.
To systematically analyze stories different models of narrative analysis
can be used depending upon where they are situated, what they are about,
and how and why they are being told. At the same time, the analyst must
use reflexive practices to be aware of his own role in the construction and
interpretation of the sto ry.
4.6.1 Background :
It was in 1970s that narrative analysis became popular, as a method of
understanding the context, different perspectives on experiences or for
generating new theories using bottoms up approach to data analysis.
Narrative analysis is i nfluenced by both modernist and post modernist’s
concerns. It makes sure that participant’s representation and the agency
remain the main focus of the study. Narrative truth is not a record of
factual records, rather it is concerned with the constructed ac count of
experience.
Let us look at some of the technical words that we have used here.
Modernism:
In the late nineteenth and early twentieth centuries, a philosophical
movement focused on breaking from traditional forms of expression to
developing new a nd innovative ways of doing so. It views a person as
being a fixed entity with a whole and stable identity.
munotes.in
Page 102
102 Research Methodology for Psychology
102 Post-modernism:
It is a philosophical movement that came in the wake of the modernist
movement, concerned with challenging the modernist idea of a f ixed and
stable identity. It views a person as being multiple, fluctuating,
contradictory and fragmented, and is interested in how these processes
occur.
Agency:
Refers to the capacity of an individual to act freely and independently, and
to make their own choices in the world.
Bottom -up (or inductive) approach:
This involves identifying patterns or meanings through the analysis of the
data without trying to fit it into theoretical interests or the researcher’s
preconceptions, and so the findings and interp retations are strongly linked
to the data itself. This is in contrasts with a ‘top -down’ (or deductive)
approachwhere the analysis is driven by existing theoretical or analytical
concerns.
Subjectivity:
Subjectivity refers to the personal perspectives of t he individuals. It
influences the ways research participants and the researcher view their
world, such as through their perceptions, experiences, expectations and
desires, as well as their social, cultural and historical understanding.
As mentioned above, there are various ways in which narrative analysis
can reflect on above mentioned issues. In fact, narrative analysis is an
umbrella approach for the various ways in which stories can be told and
explored. Each model of narrative analysis uses different ty pes of
questions to find out the content, structure and function of the narrative.
For instance, one approach looks at story, as told by the participant to find
out what meaning it makes for that narrator. The story teller does not tell
everything relate d with the experience that he is narrating in the form of
the story, rather, he chooses what to include and what to exclude from the
story. He makes sense of the event by adding interpretive elements, which
allows him to construct a coherent narrative.
There is another approach, known as critical approach. This approach
proposes that coconstruction of the story takes place between the narrator
and the researcher (his audience) during the interview as well as during the
interpretation and representation. This coconstruction takes place through
researcher’s role, words and interventions in the story, through immediate
context such as interview setting, as well as through wider context like
socio -cultural setting. Let us now look at how exactly narrative analysi s
take place.
munotes.in
Page 103
103 Qualitative Research Steps in the Research Process :
Keeping in mind the subjectivity of the participants and the researcher, it
is necessary for the researcher to show how the narrative data was
obtained, analysed and interpreted and reported in the report. It is already
mentioned that there are various models of conducting narrative analysis.
We will be concentrating on Labov’s model of structural narrative
analysis. This model looks at the manner in which an event is described in
a story’s context and shows how the meaning and identity are constructed
in narratives about the experiences.
a) Devising a Research Question:
Narrative analysis research’s main focus is to investigate how individuals
reconstitute their identities following a life -changing event. The r esearcher
explores the identity of an individual by encouraging him to do personal
reflection and meaning making of the life changing event. He helps the
narrator to make sense of changes in the sense of self and in the
relationship with his surroundings. So, the questions asked by the
researcher will aim to understand the narrator’s self and identity, meaning
of experience and intimate and social relationships. For example, the
researcher may have research questions such as How do war veterans
describe the ir experiences of learning to surf following diagnosis of
PTSD?
What is the transition to second -time motherhood feel like for women
whose second child is labelled as having a disability?
How do the stories that couples narrate together construct their
relationship across the transition to second -time parenthood?
As you can see, these questions are open ended and broad questions. They
are concentrating on a topic as well as on how the narrative is told.
b) Eliciting and Collecting Data:
In narrative a nalysis method, generally the data is collected through semi -
structured interviews. These semi - structured interviews are considered as
collaborations between the storyteller (participant) and the listener
(researcher) in which the story teller is support ed and encouraged to speak
freely of what is significant to him about the topic under inquiry. He is
encouraged to reflect on personal experiences through the telling (and
sometimes retelling) of stories.
To elicit the stories from the participants during the semi - structured
interviews, it is necessary for the researcher to pay attention to the words
used in each question. The questions should be able to not only elicit the
stories from the participants but also get the details about the experiences
as muc h as possible in follow up questions.
Forrester, M. A. (2010) gave example of some of the narrative eliciting
questions such as : munotes.in
Page 104
104 Research Methodology for Psychology
104 “Can you tell me about a particular time when …?
Can you tell me about …?
Can you tell me what you do when …?
Can you rememb er your …/a time when …?
Can you tell me what you think about …?
Can you tell me what it means to …?
Can you tell me what it is like for you …?
Can you tell me the story of …?”
However, while initial questions should concentrate in getting story
related with the researcher’s research question, follow up questions will
depend upon the immediate context of the interaction.
c) Transcription :
Once the interview is over, the researcher needs to transcribe it verbatim.
The transcription needs to be done so t hat the researcher does not
unintendedly impose meanings of his own and also does not miss out on
the subtle aspects of participants’ story. To ensure that no such biases
creep into transcription, some narrative analysts prefer to have ‘rough’
transcriptio n. They carefully look at not only what is being said, but also
how it is said. To avoid the risk of unintentionally changing the meaning
that the narrator intended, they do not try to ‘clean up’ the speech and
keep the insertion of grammar and punctuatio ns to the minimum. If they
clean up the speech, it may run the risk of not only endangering the
important information but may also lead to risk of co -authoring the
narrative.
While narrative analysts transcribe every spoken word by the participant,
they m ake only side notes of laughs, pauses, hesitations, etc. The next
question that a analyst faces is whether to present this transcript as one
piece or it should be broken into lines. While breaking the data into lines
with line numbers makes it easy to iden tify units for analysis, but it
changes the flow of reading the data, and they may lead to making
undesirable assumptions and additional meanings. Rough transcription, on
the other hand, may allow to include different accents and the rhythmicity
of speech that can help in recognizing the subtle aspects of the interview.
Though breaking the data into line numbers may help the reader to know
where in the transcript the narrative has occurred to enable them to refer to
the topic that was being talked about or what has been asked before and
after it. This is also possible in rough transcription without breaking the
data into line numbers. The analyst can give this information through
explanation and inclusion of the interviewer’s question. It should also
show ho w far into the interview the narrative was recounted to allow
readers to understand the context in which it has been told. munotes.in
Page 105
105 Qualitative Research 4.6.2 Analysis and Interpretation of the Data:
After the process of transcription, the next step will be analysis of the data.
Thou gh there are many ways of analysing narrative data, here we will be
using Labov’s (1972) model of structural narrative analysis as a guiding
procedure.
Labov’s (1972) model of structural narrative analysis :
This model first of all identifies the differen t parts that make up the
structure of the narrative.
a) Identifying Narratives :
First of all, the analyst familiarizes himself with the text of narrative that
he is going to analyze. To familiarize himself, he reads and re -reads the
text, listens to the interview and transcribes it. Next, he tries to identify the
narratives in the text, by searching for the narration of events that follow
the conventional story form of having a beginning, a middle and an end. In
other words, he looks for sequences of even ts that can be temporally
ordered. For example, he may look for words such as ‘and’ ‘and then’.
The entire semi structured interview may form as a single narrative or it
may contain many narratives. This depends upon how the narrative
eliciting questions w ere asked.
b) Coding the Narratives into Component Elements :
Labov (1972) believed that stories that attract the attention of the audience
have identifiable elements. These elements are shown in
table 1 :
Elements in Labov’s model of structural narrati ve analysis
(Source: extracted from Forrester, M. A. (2010). Doing Qualitative
Research in Psychology: A Practical Guide. Sage.) munotes.in
Page 106
106 Research Methodology for Psychology
106 Researchers can easily understand the intended meaning in the stories if
they can recognize the structural elements in the n arratives. They can
understand the meaning of events and experiences of the participants by
questioning the function of each element and finding out the sequence of
these events. Entire data must be coded into different elements. The
elements may appear in different order in different stories. Single element
may get repeated many times or some elements may be missing within
each narrative. But complicating action cannot be missing. If complicating
action element is missing, it indicates that there is no eve nt, consequently,
there cannot be any narrative.
c) Writing Analytical Comments for Each Narrative :
Once the narrative is coded with elements, the next step will be to make
analytical notes about the function of each element in constructing the
story.
Keeping in mind the research question, analytical notes are also made for
note the presence, absence, repetition and interruption of the individual
elements in the narrative, as it is relevant to the structure of the narrative.
Such an in -depth analysis e nables the narrative analyst to make detailed
investigation of the functions of each elements, their interplay and
suggested overall structure of the narrative. The analyst can seek answers
to why the narrative has been told in the way that it has been an d what it
suggests about how the storyteller is constructing his experience, sense of
self and his relationships with others. He can also determine which
analytical notes are related to research question in hand.
Re-contextualising the Narratives :
Till th is point, narratives are analysed outside of their context. But at this
point, recontextualization is done. The analysts goes back to the entire
data and again reads it keeping in mind the analysis. To interpret the
narrative within the context of data, t he analyst may pick up some other
pieces of data that may come other narratives but support the analysis.
4.6.3 Writing Up the Analysis :
Writing of analytical notes and observations are loose first draft of the
report. The final step of analysis contains the story and its interpretation
that is relevant to research question. The final report may be restricted by
practical reasons such as word -number limit, etc, so the analyst will need
to select the most representative narratives of his analysis, to inclu de in the
report of his interpretation. Moreover, the report must carry a clear
example of narrative itself, and these examples should be linked to the
research question. Each narrative coded into its elements and selected to
be included in the final repor t, can be presented in table. In the table, each
elements’ content can be described, functions and interplay of the
elements can be discussed and overall structure can be described.
munotes.in
Page 107
107 Qualitative Research 4.6.4 Other Models of Narrative Analysis :
The structural models cannot answer the questions about why a story was
told or what message the teller intended to convey. Therefore, other
models of narrative analysis can be used instead of, or in addition to, a
structural model.
For example, an analysts may want to know more abo ut the content of
narratives, because they want to compare a data set of narratives collected
from different participants. This comparison can give insight to patterns
and meanings across and within narratives which can help to develop
conceptual categorie s of the self.
Researchers can choose the model of narrative analysis depending upon
their research question – whether they are interested in function of the
story, or its content, or the manner in which it was told, or finding out the
reason for why it w as told to the researcher. If the researcher is interested
in all of these questions or more than one question, then he can use a
pluralistic narrative analysis, that means, he can combine the analysis
models.
4.6.5 Critical Issue: Does Narrative Analysi s Always Analyse Text? :
Narrative analysts do not always use text to analyse the story. Apart from
using semi -structured interview, researchers can collect data through many
other ways. For example, while textual data can be collected through
collection of poems, diary entries, etc. The stories can be retold for an
audience through dance, or through the use of visual artefacts such as
drawings or photographs.
4.7 CONVERSATION ANALYSIS (CA) 4.7.1 Introduction :
Psychologists have been increasingly interested in studying naturalistic
everyday interactions between people such as talking to other people. In
psychology, the study of conversation is known as conversational analysis.
Background :
The origin of CA as a qualitative methodology is in ethnomethodology
which is part of research in sociology. Harold Garfinkel (1967), an
ethnomethodologists, explained that people spontaneously produce
sequences of activities that appear fairly random or even uncoordinated,
but in fact are very orderly. These are sense -makin g practices and are
methodical in ordinary conventional ways.
Ethnomethodology is defined as the study of the methods people use to
produce and interpret social interaction. Ethnomethodology concentrates
on providing a rational analysis of the structures, procedures and strategies
that people themselves use when they are making sense out of their own
everyday world and their interactions within it. munotes.in
Page 108
108 Research Methodology for Psychology
108 With the advent of portable sound recorders, conversation analysts started
recording conversations and analyz ing them to identify many different
kinds of structures within conversation. They found that when
conversations were converted into transcripts, one could find many regular
patterns in these conversations. For examining such data, conversation
analysts use d ethnomethodological approach that focuses on how people
themselves produce and recognize their own ‘sense -making’ practices as
they are going along.
Conversation is converted into Jeffersonian -style transcription as CA is
very much interested in the stru cture of talk, the sequence of the
interaction and the numerous things that people do when having an
everyday conversation.
While analyzing the extract, CA approach looks at :
1. The who, how and what people do as they are conversing.
2. How a person says something is plausibly significant and can give an
idea about something of what they are doing.
3. The transcript is organized so that researcher can analyze the various
structures that people use, such as turn -taking patterns, and
identifying when troubl e in the talk might occur.
4. The way a person says something (e.g. said something at a particular
point that was noticeably faster than the utterances around it) can
form part of the analytic rationale that the CA researcher employs.
5. Whatever has been said can be analyzed with a focus on the sequence
of the interaction. CA often asks: ‘Why that (utterance) now?’
6. By analyzing the fine detail of the structures of the conversation the
researcher can understand how, and what, people themselves orient to
as they produce their own conversations.
Steps followed in CA:
Step 1: Record conversations either in audio - or video -recorded form.
Step 2: Write down the conversations in full using a Jeffersonian style
transcription to describe all relevant actions.
Step 3: find out the elements and structures in the conversation.
Step 4: Stress on participant -oriented evidence for the methods that
people use for doing ‘talk -in-interaction’.
4.5.2 Taking turns in conversation: How people use a ‘locally
managed system’ :
In 1960s, Harvey Sacks et.al. (1974; extracted from Forrester,2010)
developed a model of conversational turn -taking that is called ‘Local
Management System’(LMS). This model explains what people themselves munotes.in
Page 109
109 Qualitative Research do while conducting a conversation either with each other or in a group;
that is, sorting out whose turn it is talk to talk now has the current turn at
talk, who might have the next turn and so on. The entire thing is conducted
‘locally’ on a ‘turn -by-turn’ basis in the immediate setting of the
interaction. This model represents a highly organized system and it is
called local because it always takes place in the immediate local context. It
is called ‘management’ because it is the people who are talking, they are
managing it as they proceed. The locally mana ged system is based on two
components, and a set of rules that operate on these components. The two
components are :
1. A turn -constructional element (TCU) : This refers to any kind of
utterance, gesture or sound (e.g. ‘ehm’) and it can be of any length i n
the conversation.
2. A turn -allocation element : According to turn -allocation element,
allocatio n of turns can work in two ways :
a) somebody choosing or selecting the next person to talk, or
b) the next person selecting themselves.
The point at which the turn -allocation occurs in the conversation is known
as transition -relevant place (or TRP). TRP can be in the form of short
pause and generally at TRP the speaker changes. For example, if you are
speaking to someone, the pitch and emphasis of your voic e will typically
change, even before you actually get to the end. This will indicates to the
other person to whom you’re talking to that you are just about to give up
your turn at talk and hand it over. If the other person is speaking to you,
you know prec isely when to enter the conversation and select yourself as
the next speaker.
In short, a turn is made up of a turn constructional unit (TCU), which can
be of any length or form. A transition -relevant place (TRP), on the other
hand, defines the gap betwee n one speaker and the next speaker.
Sacks et al. (1974), while identifying the turn -taking constructional and
allocation units (TCUs and TRPs), identified a set of rules as follows :
Turn -taking Rules :
Rule 1: This rule applied to the first transition -relevant place of any
turn.
(a) if the speaker who is speaking right now, selects the next speaker
during the current turn, then the current speaker must stop speaking
and the next speaker must speak next. And the current speaker must
speak next at the fir st transition -relevant place after this ‘next speaker’
selection.
For example, imagine how odd it weird it would be if, the minute you ask
somebody a question and they are about to answer, you start talking. munotes.in
Page 110
110 Research Methodology for Psychology
110 (b) If the speaker does not select a next spe aker during a current turn,
then anybody else present (after parties) can self -select and the first
person to do this will gain ‘speaker rights’ at the next turn.
For example, don’t you think, it is quite a skill to learn to recognize just
the right momen t to come into a conversation without appearing rude.
(c) If the current speaker has not yet selected the next speaker an d no
other speaker self -selects , then the current speaker can continue
(although this is not a requirement). In such a case, the curr ent
speaker gains a right to have a further turn -constructional unit (TCU).
Rule 2:
When rule 1 (c) has been applied by the current speaker, then, at the next
transition -relevant place, rules 1 (a) to 1 ( c) apply again, and will remain
enforce or re -applying until speaker change is accomplished.
For example, the whole system is ‘recursive’ going around and around as
the talk proceeds. Does this mean we sometimes feel fixated in a
conversation?
This system and all its elements incorporate an assemblin g of all
procedures, strategies and social conventions designed and used by
ordinary participants going about the ‘doing’ of everyday conversation.
Ethnomethodology calls it ‘member method’. All these practices and the
structures that make them up are prod uced to take care of at least one
inherent problem of interaction, that is, the taking of turns. If such rules
are not there, one can imagine, how messy and disorganized the turn
taking in conversation can be. One can see such chaotic situation in
children ’s conversation before they actually learn conversational skills.
Some of the features of conversation ensured by LMS are :
1. Speaker changes occur with relative ease.
2. Mostly only one speaker has a turn at talk.
3. Transitions (from one turn to the ne xt) are exceptionally sophisticated
and very often occur with no gaps or overlaps.
4. The length of any person’s turn and the order of turns is not normally
fixed in advance.,
5. The distribution of turns is not specified in advance, nor is what
people wi ll say during a turn.
6. Repair mechanisms exist for dealing with turn -taking errors and
violations (e.g., when two people start talking simultaneously, one of
them normally stops very quickly).
4.7.3 Sequence in conversation :
One regular element of intera ction and conversation is that one thing
always follows another in sequence. whenever we are in the presence of munotes.in
Page 111
111 Qualitative Research other human beings we are forever monitoring, either very subtly or most
of the time unnoticeably, our own and their behaviour. Not only we watc h
our own and others’ behavior, we present ourselves to others in such a
way that our own behavior can be monitored by whomever happens to be
around us. We are unconsciously very sensitive to a continually
‘unfolding’ sequence of ‘what happens next’ and ‘w hat’s meant to happen
next’, given what has just been said. We become conscious of it only
when this implicit attention to sequence seems to go wrong somehow or
somebody talking doesn’t seem to be paying attention to it.. As sequence
and ‘what happens next ’ is important, we realize that we are always
accountable in some way for what we are doing when in a conversation. In
CA this is called ‘sequential implicativeness’. It underlines the observation
that what you do ‘next’, following something that another p erson has
done, is automatically monitored by both parties. If any of the parties in
conversation, do not follow the rules of normal conversational
conventions, it is assumed that something is wrong or the conversation
was not understood by the other party . This may annoy the first party in
the conversation and he will look for reasons for breaking the rules of
normal conversational conventions. This indicates that no matter how
small or insignificant a behaviour might be, in the presence of other
people, w e are always accountable for our actions as a member of the
culture.
4.7.4 Structures in conversation :
While having conversation with another person, we unknowingly show a
sensitivity to various structural elements in talk, such as requests,
questions, gre etings, compliments, interruptions and many others. At the
same time, we also produce such structures ourselves, so that others can
recognise we have done so. To understand this procedure, we need to look
at some of the rules of turn -allocation. Here we wi ll look at three rules of
turn allocation. They are :
Adjacency pairs
Endings
Formulations
a) Adjacency Pairs in Conversation: The Talk Unfolded Two -by-Two
In conversation, lot of things come in two parts that are sequentially
organized. For example, question -answer, greetings -greetings, invitation -
acceptance, etc. Even ringing of telephone is considered as summons. It is
similar to someone tapping you on your shoulder to get your attention.
When a person picks up the phone and answers it in conversational
manner, it is a response to the summons. I n CA, these pairings are
delineated as adjacency pairs and they come in as ‘first and second’ parts
– a first -pair part (FPP) and a second pair part (SPP). We may not be
consciously aware of it but there is a very strong convention that when
someone produc es an FPP, then an SPP has to come somewhere later. It
may not come immediately, but later on it must come. munotes.in
Page 112
112 Research Methodology for Psychology
112 Some of the characteristics of adjacency pairs are :
1. They must be normally adjacent.
2. They must be produced by distinctly different speakers.
3. They are always sequenced as first -pair part/second -pair part
(FPP/SPP).
4. Both the pairs are conditionally relevant. The first pair initiates what
may take place as a second, and the second will be based upon what
has occurred as the first.
The adjac ency pair structure also follows LMS rule that once a first pair
part is produced, the current speaker must stop speaking and the next
speaker must produce at that specific time in the interchange, a second -
pair part to the same pair. This is the conventio nal practice or ‘members’
methods’ that we all produce and adhere to when we talk.
People unconsciously follow conversational conventions and become
conscious of it only when somebody tries to rectify, change or repair the
‘breaking of the rule’ that has j ust occurred. So, conversation analysts
strongly believe that interpretations, suggestions or claims made about the
data being analyzed (the actual conversations) should rest upon
identifiable evidence in the conversations themselves.
b) Endings: Closin g Sequences and How to End a Conversation
The sequential characteristic of talk -in-interaction indicates that one
person's turn will always followed by another's (your turn, my turn, your
turn, my turn, and on, and on).This can be never ending activity. Peo ple
may find it difficult to end the conversation without being rude. It will be
rather unconventional to simply walk away from the conversation after the
main topic is over. Both the parties in conversation must make the end of
the conversation possible i n a smooth and acceptable way. Whoever
decides to make the first move towards stopping the conversation must
produce an FPP (first -pair part) that indicates a move towards possibly
finishing the conversation. Where exactly that FPP is produced is
important because the orientation of the SPP (second -pair part) to this
special kind of FPP will show whether the ‘next speaker’ to the current
speaker has taken up this offer. Generally, a speaker who wants to end the
conversation may use pre -closing phrase or wor d such as ‘well….’,
‘right…..’, and may also change his intonation or may use a long pause.
Sometimes, the respondent party may not take up the offer of closing
down, i.e., they may decline an end move, instead the respondent may
introduce a new topic and the conversation may continue. For example,
the respondent party may continue by using phrases like –
Oh, there was one more thing …
hmm, by the way, I just wanted to say …
I just wanted to mention one other thing …
By the way, I meant to say … munotes.in
Page 113
113 Qualitative Research c) Formu lations in Talk :
Formulation is another structure that appears quite frequently in talk.
Formulation refers to a moment in the ongoing conversation when
somebody refers to, or spells out, what they have been saying. Phrases
such as ‘Look, what I’m getting a t …’ or ‘Oh I see, what you’re
suggesting is …’ or ‘The thing I’m saying is …’, It shows that while
talking people are also making sense of what is going on as it is happening
– in the here and now. Formulation indicates to us that the main aim of
talking is to show to each other our understanding of what is being said.
This demonstration of understanding becomes part of the ongoing
conversation. So, it can be said that, initially the first speaker will produce
a formulation, and then the person spoken to w ill produce a response to
the formulation that is either a confirmation or a dis -confirmation. Of
course, the preferred response is confirmation or agreement. If the
respondent responds with disagreement, disconfirmation or non -
commitment, it will be an in dicate that participants do not have a shared
sense of what is going on. We can say that in typical conversation
formulation has four function -
1. to show the accumulative importance of the talk that has preceded the
formulation itself;
2. summarizing to make an overall point, a form of ‘summing up’;
3. to work as a method to move towards ending a conversation;
4. to indicate the combined understandings of what is going on, that is,
what is being achieved by both parties.
4.7.5 Transcription in Conversat ion Analysis :
Conversation Analyst transcribes by playing back small sections of
conversation repeatedly, and gradually writing out the words and sounds
of the conversation, according to the orthographic conventions used in
Jeffersonian transcription. This gives the analyst access to ‘lived reality’
of the interaction that is not available in any other way. The analyst must
listen to the recordings with ‘unmotivated attention’, i.e., not having any
pre-specified goals. He should pay attention to even unremar kable features
of the talk or other conduct. He should listen to the talk with unmotivated
attention again and again, unless he is sure that he has noted down every
sound and can begin analysis. Throughout transcription process, he should
be asking, “ why that utterance now?”. He should pay close attention to the
dynamic and sequential nature of the conversation.
4.7.6 Analysis :
Since CA usually studies everyday conversation, the analyst can have a
huge body of material to compare talk -in-interaction. Th is conversation
can be in formal as well as in informal settings, e.g., at home or at work.
However, most of the research studies have focused on how ‘talk -in-munotes.in
Page 114
114 Research Methodology for Psychology
114 interaction’ occurs in interview settings. For example, they look at the
questions like :
What pa rticular procedures do people use so as to have extended turns -at-
talk?
How exactly does an interviewer indicate that they are listening attentively
without being overbearing?
What are the procedures employed at the beginning of an interview to
encourage a free-flowing easy conversation?
How people end an interview and what resources they use to close the
interaction itself. What kind of formulations and adjacency pairs are used
to end the conversation.
Writing Up the Analysis :
Doing the analysis and writi ng it up really go together. CA is basically
interested in ‘members’ methods’; that is, how people themselves use
resources in service of their everyday sense -making practices. Two
resources that people call upon in conversation are formulation and
adjacen cy pair structures. Formulations can be short or longer utterances
and can be presented in many forms. Formulations highlight the reflexive
nature of conversation as action – it indicates what the two parties are
doing or have just done, as the conversatio n is taking place. The sequential
structure is the adjacency pair that people produce and respond to, in two -
part formats. Formulations help people to overcome ‘turn -taking’ problem
in ending the conversation as formulations sum things up. formulations
also serve to indicate ‘cumulative understandings. People treat the
successful production of formulation – confirmation pairs as reflexion of
understandings
4.8 SUMMARY In this chapter, we have looked at how research gets influenced by
epistemological and t heoretical issues. The philosophy that a researcher
believes in influences the kind of research problems investigated, the
methods chosen and the interpretation of the findings. The qualitative
methods differ depending on the theoretical assumptions, they are based
upon. So, in contrast to quantitative methods, there are variety of
qualitative methods. Positivism ruled in the past and led psychologists to
go for quantification. Later on, social constructionism became dominant
and today psychologists prefer a balanced approach
Grounded theory is an analytic approach to qualitative research that uses
inductive approach. Some of the attributes of grounded theory are
theoretical sensitivity, theoretical sampling and theoretical saturation. The
grounded theory p rocess involves cycling back and forth between data
collection and analysis until a substantive theory develops. Grounded
theory develops through constant production of reflective documents
called memos. Through these memos, the researcher uses the constan t munotes.in
Page 115
115 Qualitative Research comparative method to develop and explain the conceptual content of the
developing grounded theory.
Memo writing is done through coding. Grounded theory commences with
early coding, which leads to the development of a structure of
intermediate, conceptua l categories. A principal category is selected from
these, and then a theory is developed that emerges from the relationship
between the core category and other major concepts.
IPA is an idiographic from of inquiry that is influenced by
phenomenology and h ermeneutics. It focuses on the individual level of a
person’s experience and involves ‘double hermeneutic’. It is important in
IPA to do phenomenological coding and interpretative coding as well as to
maintain reflexive journal. This helps in developing c lusters of themes and
then a final theme for the final report. Since IPA is data driven, it is
important to maintain audit trail.
DA is a term used to describe a range of approaches used for analysing
talk and text in all forms of social interaction. DA a ssumes that discourse
constructs reality instead of reflecting reality. There are different forms of
discourse analysis and each concentrate on different aspects of discourse.
For instance, DP focuses on the micro -management of fact, interest and
accountab ility, and the ways in which psychological terms (such as mental
states, identities or personality) are used to perform social actions. FiDP
focuses on how talk constructs facts about the world and the people in it;
how the talk makes this reality appear c redible; the consequences for
constructing reality in this way (for subjectivity and practice); and the
wider discourses that support this sense -making
Narrative analysis tries to understand more about how people make sense
of themselves and their lives th rough the stories they tell. Narrative
analysis views individual meaning -making as emplaced within contexts
that include their biography, history, societal and cultural influences.
There are variety of models of narrative analysis, and each one focuses on
a different aspect of the narrative, for example, its structure, its function or
its content. Labov’s model of structural narrative analysis looks at how an
event is told in a story context by closely analyzing how the different parts
in the structure of t he narrative function.
Conversation is not simply talk but an interaction between two people. It
is ethnomethodologically inspired. It is not important for CA to focus on
the content of the talk but he must pay attention to the methods people use
to make sense of their social world as they are producing talk. We need to
record, attend to and describe the interaction to our maximum personal
capacity. CA researcher should analyse the transcription with
‘unmotivated attention’. He must develop all his suggest ions or arguments
based on a careful ‘line -by-line’ sequential examination of how
participants themselves treat ‘what happens next’. If there is no evidence
or indication for what the analyst is suggesting, then the researcher must
be very sceptical about what might be said about the interaction.
munotes.in
Page 116
116 Research Methodology for Psychology
116 4.9 QUESTIONS 1. What is the difference between quantitative and qualitative research
and why qualitative research has become popular in recent times?
2. What is positivism and what are the challenges to positiv ism?
3. Write a detailed note on relativist social constructionism.
4. Discuss in detail the background, analysis and coding in grounded
theory.
5. Discuss in detail how IPA analysis is done.
6. What are the steps involved in the discourse analysis resea rch?
7. What is the analysis process in FiDP research?
8. Write in detail the steps involved in narrative analysis research?
9. Discuss Labov’s model of structural narrative analysis.
10. Elaborate on the background of conversation analysis research
method .
11. Write a short note on
a. taking turns in conversation
b. Sequence in conversation
12. Write a detailed note on structure in conversation.
4.10 REFERENCES Charmaz, Kathy. " Constructing Grounded Theory A Practical Guide
Through Qualitative Analysi s." SAGE Publications. 2006.
Charmaz, Kathy. "Grounded Theory." The SAGE Encyclopedia of
Social Science Research Methods. 2003. SAGE Publications. 24 May.
2009.
Corbin, J., & Strauss, A. (2008). Basics of Qualitative Research:
Techniques and Procedures fo r Developing Grounded Theory (3rd
ed.). Thousand Oaks, CA: Sage
Edwards, D., & Potter, J. (1992). Discursive psychology. Sage,
London.
Forrester, M. A. (2010). Doing Qualitative Research in Psychology: A
Practical Guide. Sage.
***** munotes.in