Académique Documents
Professionnel Documents
Culture Documents
Abhay S Nair
Definitions:
and effect. This helps to understand the present events and predict
future events
13)
Qualitative research: qualitative research is designed to reveal a
target audiences range of behavior and perceptions that drive it with
reference to specific topics or issues.
14)
Quantitative research: deals in number, logic
Importance of research:
1) To find the optimal solution to business related problems
2) To analyze and find out what is the position of the company in the
market in which it operates.
3) To forecast the demand of already launched product or a product to be
launched
4) Research helps the government to come up with an effective policy to
run the state administration.
5) Helps the government to adopt a technology that is beneficial for the
society.
6) Social control and performance
Steps involved in research
1) Defining research problem should be specific, uniformly and similarly
understandable to all
2) Identifying the research objectives
3) Research design
a. Selection of research approach
b. Selection of sampling plan
c. Selection of questionnaires, schedules, observational format
** schedule self filled by the interviewer
**questionnaires filled by the interviewee
d. Pilot study
4) Collection of data (primary or secondary)
5) Presentation of data
a. Tabular
b. Graphical
6) Analysis of data
7) Interpretation of result / outcomes
8) Validation of outcomes
Studies consumer behavior
Marketing research provides data about consumer behavior. It provides data
about age, incomes, likes, dislikes etc. of the consumers. It also finds out the
opinions of the consumers about a companys product. This data is used to
make production and marketing policies
Research design
Research design is a blueprint for the research to be conducted
Terms
1) Variables: are factors which might or might not affect an experiment
**Extraneous variables: these are those variables which interfere with the
research
**Confounded relationship: when the external factors are not under
homogeneity
**Treatment: special conditions given to research
**control group: the group to which treatment is not applied but is
homogeneous to the experimental group
**Experimental group: the group to which treatment is given.
Schedule
Lesser coverage
Personal touch
Higher cost
Questionnaire
Higher coverage
No personal touch
Lower cost
Interview technique
An interview is a conversation between two or more where questions are
asked by the interviewer and responded by the interviewee
Types of interviews
1) Structured interview
2) Unstructured interview
3) Focused interview
4) Non directive interview
5) Personal interview
6) Telephonic interview
A) Structured interview: standardized interview or research administered
interview
interviewer decides and plans in advance the nature and scope of
questions to be asked from the interviewee. Structured interviews are
the preferred means of collecting data for a statistical survey.
B) Unstructured Interview: interviewer changes or manipulates the
questions in order to meet respondents intelligence, understanding or
belief. It does not offer limited, pre-set range of answers for a
respondent to choose, but instead advocates listening to how each
individual person responds to the question. Mostly used for sociology
otherwise this technique is rarely used.
C) Focused interview: this technique is used to collect quantitative data
by setting up a situation (the interview) that allows a respondent the
time and scope to talk about their opinions on a particular subject. The
focus of the interview is decided by the researcher toward areas that
the researcher is interested in exploring. It uses open-ended questions.
D) Non directive interview: this is the interview which is unstructured
and flexible in nature.
Observation Techniques
I.
1) Disguised/Hidden: when the subject under observation is not aware
that he is being observed. Observers presence is not known
2) Undisguised / open / visible: when the subject under observation is
aware that he is being observed
II.
1) Structured: format is decided beforehand from where to start and
where to end. Systematic approach for scientific observation is
always structured
2) Unstructured: format is not pre-decided and is not systematic
III.
1) Overt observation: observers approach is active. The researcher
actively participates in the observation taking place (usually
designed). Doing the same activities as the participants
2) Covert observation: researcher or observers approach is passive
i.e. not an active participant in the observation process
IV.
1) Natural observation: when you look at the subject in the normal,
routine environment
2) Contrived observation: when looking at the behavior of the
subject in an artificially created environment
What are the advantage of observation technique?
Data is very much of current scenario accurate.
We dont need the cooperation of the subject.
When the subject is illiterate and not willing to cooperate.
Conditions for observation techniques:
1) Observations should be inferable / to be able to draw conclusion
2) Observation should be repetitive in nature to be able to draw
conclusions (repetitive frequencies)
Situation
Respondent
Type of Error
Incorrect
question,
inappropriate
analysis,
misinterpretation.
Wrong sampling technique, not being true representative of
population.
Biased attitude, misinterpretation, carelessness etc.
Inappropriate scale, ambiguous questionnaire, complex
words, inadequate space to reply, response choice
omission etc.
Lack of rapport, lack of assurance, condition that places
strain on interviewee
Casual attitude to reply, fatigue, may not admit ignorance,
boredom, etc.
to
accurately
Importance of scaling
Measurable
Precise
Variables are amenable to mathematical treatment
Methods of Scaling
Direct/undisguised
method:
prone
to
errors,
awareness in admissibily, self-incrimination, etc.
Indirect/disguised method: Reveals respondents
attitude, belief, etc. indirectly.
Scaling Techniques
Rating Scales
o Graphic- symbols, squares, circles, pictorial.
o Itemized
Ranking or comparative scales
a) Semantic
Limitation Unidimensions
Construction-tedious & complex
Not reliable- complex objects
Experimental research
It involves manipulation of at least one variable and control over the other
relevant variables so as to measure its effect on dependent variable
The variable(s) which is manipulated is also called independent variables, a
treatment, an experimental variables or the cause
Experimental research will always have two or more groups for comparison
on the dependent variables. Its objective is to explore and understand cause
& affect relationship.
Experimental research design a blueprint of the procedure that enables the
researcher to test his hypothesis by reaching valid conclusions, about
relationships between independent and dependent variables. It refers to the
conceptual framework within which the experiment is conducted.
Professor R. A. Fisher realized by dividing agricultural fields into blocks and
conducting experiments in each of these blocks individually results in more
reliable information/outcomes
Data Processing
Process of converting raw data into useful information or knowledge
It involves
1) Data editing
2) Data classification
3) Transcription
4) Coding
Data Editing
A process of locating and removing errors, incompleteness or inconsistency
in survey data so it ensures that survey data is accurate, complete and
consistent
Objectives of data editing are:
1) To ensure the accuracy of data
2) To establish the consistency of data
3) To determine whether or not the data are complete
4) To ensure the coherence of aggregated data; and
5) To obtain the best possible data available
Data Classification
Categorization of data for its most effective and efficient use is done by
arranging data in homogeneous groups based in similar characteristics. A
well-planned data classification system makes essential data easy to find.
This can be of particular importance in risk management, legal discovery,
and compliance with government regulations.
Transcription
The process of conversion of data from one medium to another is called
transcription. The data is transferred from interview schedule to a card such
that there is only once card corresponding to each unit in the survey
Transcription makes sorting of information easier; and the records remain
intact without any marking on them
It is recommended that two individuals make independent transcriptions to
eliminate the possibility of creeping transcription errors.
Coding:
Transferring the data from interview schedule to a card is in the form of
abbreviation, a number, an alphabet or symbol which is assigned by
researcher to every schedule item and response category. Such
representation of data in card is referred to as coding. Mock code book is
constructed for pivot study to show various codes that are temporarily
assigned to different response categories. Later raw data code book is
prepared that contains final codes to different response categories.
Guidelines of coding
1) One code should be specifically indicating one and only one kind of
response category. There should be no overlapping.
2) Each set of categories should be collectively exhaustive. For
unanticipated responses, a miscellaneous category should be created
3) Separate categories should be created for non-applicable, refusal or
ignorance
4) Inter-coding and intra-coding tests should be conducted
Inter-coder agreement test
Degree to which any coder will
classify and code the given response
in such a way that any other coder
will also classify and code the same
given response in similar way
Common tools
Discriminant analysis
A statistical analysis to predict a categorical dependent variable by one or
more binary independent variables.
Cut off score = n1*y1+n2*y2/n1+n2
What is discriminant score D?
D is a weighted linear combination (sum) of the discriminating variables.
If the discriminant score of data set is less than D, then include the member
of entity representing the data in below category; else in the above
category.
The purpose
To establish a linear combination if variables that discriminates between
categories of dependent variable in best possible manner
To investigate the difference between groups on the basis of attributes and
indicating as to which attribute contributes most to the group separation
s1
s2
s3
s4
w2
D2
c4
A3
B3
w3
B1
A2
D4
C2
w4
C1
D3
B4
A4
replacing all records repeatedly 52 times and realize that 65.4% of cross
validated group cases are correctly classified.