NATIONAL INSTITUTE OF MANAGEMENT

19th Senior Management Course

Research Methodology
Data Collection
Khalid Mahmood Lodhi
Chief Instructor
1
 Data
Basic/major sources of data collection are the
following:

1. Documents
2. Questionnaires
3. Interviews
4. Sampling
5. NET
6. Observation
(Useful source of data collection for
anthropological research, may not be very
relevant and useful here)
2
 Data
1. Documents

• Documents are primarily of two types - Primary and

Secondary

• Primary sources are preferred sources in research –

they are original, ‘straight from the horse’s mouth’

kind of material - anything recorded by the principal

actors involved in the events being analyzed

Contd… 3
 Data
Examples:

• Personal Records – Memoirs, Diaries, Letters,

Autobiographical/Eye-witness accounts

• Official (Public) Record – Government documents/

studies/reports/inquiries/notifications/minutes, etc

4
 Data
• Secondary Sources – material produced after

recalling the event/s

• The material tends to be descriptions, analyses,

interpretations or statements of those who were not

present during the event/s
• Mostly books, monographs, journals, magazines,

and newspapers

5
 Data

• Documents also include: photographs,

maps, pictures, audio and video

casette recordings, tape recorded

interviews, and computerized records

6
 Data

Major reasons for using Secondary Sources:

1. Collecting primary data is difficult, time-
consuming and expensive
2. Not enough primary data
3. Makes sense to use it because this is the data you
want and is also readily available
4. May shed light or complement the data you already
have from the primary sources
7
Contd…
 Data

5. Because it may confirm, modify, challenge

or contradict your argument/s
6. Because it allows you to focus your
attention on analysis and interpretation
7. Because you cannot conduct a research
study in isolation from what has already
been done
8
 Data
How to use Documents?
1. Never use documentary sources uncritically
2. Newspaper reports are at times, if not mostly,
inaccurate or biased. Don’t rely on one paper’s
views. Check against other newspapers to get the
balance
3. Finally, and most importantly, look at your own
prejudices. Are your searching for evidence to
support your own viewpoint and dismissing out of
hand information that seems to disagree with you?
9
 Data

2. Questionnaires (Mailed and Schedule)

• A good questionnaire grows out of a good hypothesis

• MORE care required for the selection and designing of

questions for mailed questionnaires

• Selection of the respondents should be done very

carefully

10
 Data
(1) Designing the questionnaire:
(a) Preparatory work
(b) Questionnaire format
(c) Questionnaire content
(d) Pre-testing
(e) Training the interviewers
(f) Coding the data
(g) Analysis of the data

11
 Data
(2) Kinds of Questions:
1. Open or Closed questions
Advantages of Open-ended questions:
i) Greater freedom to express opinion
ii) Respondents express their own opinions and will not be
influenced by a ready-made answer just because it is offered
iii) Bias reduced because response is not limited to certain
answers
Disadvantages:
1. Time-consuming – to analyze
2. Field workers need training as they are liable to misunderstand
the answers given by the respondents
12
 Data
Closed Questions (yes/no answers type)
Advantages:
1. Quick to administer
2. Easier to code and analyze
Disadvantages:
3. Misleading questions can be drawn because of the
limited range of options
4. Respondents unable to express personal opinions
and may choose the nearest answer to what they
really think
13
 Data
What to avoid in designing questions?

1. Problem question – a question that may not be

clear and unambiguous when read by some one
2. Double – barrelled question
A good question must seek answers to one
question, not two. (Should the government
spend more on education and less on the
infrastructure?)
Contd…
14
 Data
3. Leading question
Leading questions are often emotive (would you
say that animal welfare charities are a good
thing?)
Make sure that your questions don’t begin with
statements such as:
‘Do you agree that…
‘Do you think that…
‘Would you say that…,
and so on
Contd… 15
 Data
4. Hypothetical questions
Try to avoid asking questions where the
respondents have to imagine themselves in a
particular situation
Avoid starting your questions with,
‘if….
5. Memory questions
Questions that recall events, dates, or information
are likely to lead to inaccurate replies, which in
turn, will make your research meaningless

Contd… 16
 Data
6. Sensitive questions
Asking a sensitive question is best left towards the
end of your questionnaire ( especially Schedule)
7. Long questions
It is better to reduce the length of the long
question or make it two questions if it is necessary
to obtain that information
8. Questions requiring prior knowledge:
If the respondents have to look up information in
order to answer your question, they may abandon
it altogether
17
 Data
Layout of the Questionnaire
If it is badly designed, the interviewers and
respondents can lose their way and miss important
questions
Layout is as important as the wording of the
questions themselves:
• should be clearly printed
• should be presented logically
• should have plenty of space for replies
• all instructions should be easy to
understand and follow
18
 Data

Coding the Questionnaires

Code box on the right side of the questionnaire –

convenient

Coding before printing the questionnaire – simple

and most obvious questions at the beginning, more
complicated as the questionnaire proceeds

19
 Data
How to distribute the Questionnaire?
• Covering letter (unless a Schedule – face to
face)
• Ensure confidentiality
• Return Date – essential to give a date by which
the questionnaire should be returned (Mailed )
• Stamped self – addressed envelope
• Reminders – with intervals.
• Letter of thanks (finally)
20
 Data
3. Interviews

• Interview is a purposive conversation whose purpose may vary

widely
• Interview is a very useful technique for data collection

Why Interview? -
to collect information/data

1. based on emotions, feelings, expressions

2. based on sensitive issues
3. based on insider experience – privileged insight and
experience
21
 Data
How can the Interview be used?

1. Supplement information provided, for example, in

the Questionnaire
2. Help pilot a questionnaire - interview a few
people to test
3. To follow up a questionnaire - select some in-
depth interviews
4. To add variety to other methods - such as
documents, questionnaires, etc
22
 Data
Kinds of Interviews

1. Structured Interview
• The structured interview is used with a large
number of people – voting, shopping surveys, etc
• The questions asked – are usually to the point and
short. Responses are also brief.
• Easier to code and analyze
• Once the questions are formulated, more than one
person can ask the pre-arranged questions
23
 Data
2. Unstructured Interview
• Unstructured interviews are best left in the hands of
experienced researchers.
• They require a great deal of expertise to get the data
required
• They also require deft control of conversation
• Otherwise, one may have incomprehensible data -
rich in itself but not easy to draw conclusions in
relation to the subject
24
 Data
3. Semi-Structured Interview
• Semi-structured interviews enjoy the best position

as far as interviewing goes

• Much more flexible

• Along with the within - questions - they allow for

additional comments. This additional information

will be extremely useful

25
 Data

Who should be Interviewed?

1. People in position of authority - potentates

2. People with special knowledge - experts

3. People that form the subject of research -

sample of the population, for example

26
 Data
Techniques
1. Be non-judgmental
2. Choose the right vocabulary
3. Express an interest in what you are told
4. Be aware of non-verbal communication, body-
language
5. Be silent when the interviewee pauses (may
need time to gather thoughts)
27
 Data
What to avoid?
• Interesting questions should not be asked at the
expense of information relevant to your research
• Keep the interviewee focused on the information
you need
• Be careful to avoid leading questions
• Avoid questions that are too probing, especially in
the beginning
28
Contd…
 Data
• Don’t appear threatening

• Don’t ask excessively long questions,

multiple questions, technical or jargon-filled

questions or questions that will suggest

bias and prejudice

• Don’t use cliches or unfamiliar words

29
 Data
Shape of the Interview – Format

1. Introductions
2. Social comments about time and place
3. Background information to the interview, how it will
be conducted? What you are asking for, and so on?
4. The interview itself?
5. A few straightforward questions to finish the
interview off and some information about use and
contacts
6. Closing down and thanks

30
 Data
Tape-recording or note-taking?
Tape-recording:
• Using a tape-recorder may not be mostly welcome by
an interviewee – but you will have a verbatim record

Note-Taking:
• Note-taking gives you an instant record of the key
points of an interview – but note-taking can also be
distracting

Basically a researcher’s call! Advantages and

disadvantages with both methods
31
 Data
4. SAMPLING
• How to choose your participants?
• Choose a smaller, more manageable number of
people to take part in the research
Many different ways to choose Samples, but
Basically two types:
(1) Probability Samples
(2) Non-probability (Purposive) Samples

32
 Data
• In Probability Samples – all people within the
research population have a specifiable chance of
being selected
• Very useful – if the researcher wishes to explain,
predict or generalize the whole population
• In Non-probability samples – description rather
than generalization is the goal.
• Not possible to specify the possibility of one
person being included in the sample
33
 Data
SAMPLE SIZE
• Large-scale quantitative surveys require contacting
more people (than you would for a small, qualitative
piece of research)
• Will also depend on what you want to do with your
research – if large amount of cross tabulation
required, the more people you contact the better
• In quantitative research, the larger the sample the
more accurate your results
• In this type of research, it is best to use statistical
methods 34
 Data
Types of Sampling
(1) Random/Systematic Sampling

For large, relatively homogenous but geographically

concentrated populations
• Straightforward strategies like random or
systematic (for ex: every fifth or tenth
pedestrian) sampling are often adequate to
capture a representative sample
• Sometimes these may be combined into
systematic random sampling, where the starting
point is chosen randomly but subsequently the
sampling interval is systematic
35
 Data
(2) Stratified Sampling
However, if the idea is to capture a reasonably
representative sample of each category of the
population concerned (such as ethnicity, religion,
class, home language, income group, housing type,
residential area or political affiliation) – then the best
strategy will be stratified sampling.
This involves identifying the relevant population
categories (strata) and the approximate proportion
that each comprises of the total population – and
then - taking an appropriate random or systematic
sample within each
36
 Data
(3) Cluster Sampling

• However, for particularly complex and diverse

populations, multi-stage clustering may prove ideal
• It involves, as the name suggests, three or four
successive stages of sampling
• Generally, there is a trade-off between the number of
initial clusters and the number of people/households
to interview in each
37
 Data
Quota Sampling/Non-probability, Purposive Sampling

Quota Sampling:
• In this kind of sampling, the population is split up
into non-overlapping sub-groups, as for stratified
sampling. Quotas of desired number of sample
cases are then calculated proportionally to the
number of elements in these groups
• Major problem - with quota sampling is that attempts
to deal with one known source of bias may well
make matters worse for others not known, or at least
not known until after the data have been collected
and it is too late 38
 Data
Error, Sample Size and Non-response

• Two categories of errors in research – sampling

error and non-sampling error – refer to different
entities
• The difference between sampling and non-sampling
error is that the extent of the former can be
estimated from the sampling variation – whereas the
latter cannot

Contd… 39
 Data
• Major source of non-sampling error related to the
sampling process itself include - sampling - frame
defects, non-response, inaccurate or incomplete
response, defective measuring instruments (i.e,
questionnaires or interview schedules), and
defective data collection or management
• Many of these effects are, or can be, controlled by
proper randomization in sampling
40
 Data
Non-response
• Non-response is a major source of error in many
surveys
• Response rate is influenced by such design matters
as the appearance of a questionnaire, its layout,
length, and readability
• Increasing sample size to cover expected non-
response would, infact, be more likely to increase
than to decrease bias. More money would be spent
to no avail
41
 Data
5. Net
Internet or NET is a useful source of Data Collection
• Easy and fast access
• Search Engines, like Google, Yahoo, etc
• Remarkable range of information – enter key
words or phrases
• Mostly copyable – from the source document
(Plagiarism?)

42
 Data
Information at Net:
• Public/Government Records
• Books/Articles
• Newspaper reports
• Individual research projects
• University/Academic Professional sites, etc

43
 Data
Problems:
• Web information is limited
• Web is subject to copyright laws
• Credibility of the sites?
• How old or relevant is the information?

Proper citation required

44
 Data
Conclusion
Which method is the ‘best’ method?
• All methods are fine, with strengths and
weaknesses
• Choose the relevant method – some methods
are more suitable for certain fields, disciplines
• Combine methods, preferably
45