Data Collection—Types, Sources, and Instruments

Types of Data

The types of data could be different with respect to the following:

a) Sources

b) Nature

c) Dimension

d) Scale of measurement
a) On the basis of sources, data could be classified as primary and
secondary data.
 Primary data are collected directly from the entity.
 The entity may be individuals, households, organizations, firms, etc.
 The person collecting data has the first-hand information on the
variable relating to these entities.
 Secondary data refer to the data that have already been collected by
others and are available in published or unpublished forms.
b) On the basis of Nature, Data can be classified as qualitative and quantitative.

Data or information relating to the categorical variables are known as qualitative

data .

These are are quantified using counting numbers using a process. For example,
brand loyalty, efficient employees, and store ambience are qualitative variables.
These could be measured using specified processes.

Income, expenditure, weight, height are quantitative variables.

• The former are measured indirectly but the later can be measured with interval or
ratio scale of measurement.
c) On the basis of Dimension, data can be classified into time series,
cross-sectional, and panel data.

In time series data time varies & entity remains the same.

In cross section data the entities vary but the time is fixed.

In panel data have both time and entities vary.

Panel data could be balanced and unbalanced depending on the

equality of time series.

In empirical analysis, time series, cross-sectional, and panel data are
used depending on the purpose of the study.
Scale of Measurement

Data could be different with respect to the scale of measurement in which the
variables have been measured.

There are four types of scale of measurement, namely nominal, ordinal, interval and
ratio.

Data represented in the first two scales are known as non-metric data , whereas data
in interval and ratio scale of measurement are known as metric data .

Qualitative variables or the categorical variables are measured in non-metric scale of

measurement.

Quantitative variables are measured in metric scale.

 Instruments for Data Collection

• Time series, Cross Section and Panel Data can be collected from secondary sources.

• These are collected from various published sources or data banks, such as publications of the
World Bank, International Monetary Fund (IMF), the United Nations, and from several websites
at the international level.

• Statistical offices of international agencies e.g the United Nations, IMF, UNESCO, and
statistical organizations of the ASEAN countries, Eurostat, and OECD also publish statistical
data including population figures.

• Different countries across the globe have several organizations that collect and manage date
banks. There are various publications , statistical offices, web sites, data banks etc at the
national levels in different countries from where secondary data could be collected.
 Instruments for Data Collection

• Data from primary sources are basically cross-sectional data.

• In some cases, short time period data can be collected from primary sources using a
questionnaire or a schedule.

• Both qualitative and quantitative data can be collected from primary sources.
Instruments for Data Collection
Census and Sample Surveys with schedules and questionnaires
Observations
 Interviews
Focus Group Discussions (FGDs)
Experiments
Projective techniques
Warranty cards
Store audits
Pantry audits
Consumer panels
Surveys with Schedules and Questionnaire

• The most important and widely used instruments of primary data collection in census and
sample surveys with schedules and a structured questionnaire.

• Schedules incorporate a set of questions with respect to the topic and objective of the study,
which are asked and recorded by enumerators by personal contact.

• Questionnaires refer to a set of questions with a structured format that are filled in by the
respondent or used by other agents to collect the data.

• These two are almost the same in their structure and characteristics but differ with respect
to administration or in the ways they are used for data collection.
Questionnaire can be used with the following

Personal contact or personal interview

With the help of field investigators or enumerators

By postal methods

Electronic Mail

Telephonic interviews
Surveys with Schedules and Questionnaire-A Few Guidelines for
Questionnaire construction

The questionnaire or the schedules are normally used for primary data collection in
census or a sample survey. These are used interchangeably.

There are several guidelines given by several authors about construction of a

questionnaire/schedules. However, the guide-lines are subjective.
The following are some of the points which a researcher may take note of while framing
a questionnaire/schedule for a survey.

1. Questions and statements

The term 'questionnaire' suggests a set of questions to be asked to the

respondents using a format.

The researcher can also include statements/attributes to elicit responses in a Likert

scale to measure attitude of the respondents.

A Likert Scale may be of a 'seven' or a 'five" point scale.

Likert Scale Example
1. How important are the following characteristics for the performances of the
company?
Characteristics V.Imp Quite Imp Undecided Of little imp Not imp

Finance

Tie up for raw

materials

Sale/Marketing skills

Quality of products

Competitive pricing

Diverse customer
base
Using a continuum

Alternatively a continuum may also be used in a Likert Scale.

Two extremes are specified and the rest is specified with numbers only
 1-Strongly disagree & 7-Strongly agree

 1-Least Important & 7-Most important

Example: Rate your employees communication skills on the following scale by encircling the appropriate
number.

Very Open Secretive

1 2 3 4 5
Example 2
Likert Scale …Contd

• In a market research study for a consumer non - durable such as Premium Soap
the researcher wants to find the importance attached by the consumer to
specified attributes.

• He may use a Likert Scale to elicit responses from the consumers and identify
attributes/group of attributes which are more or less important.
Example: Premium Soap.
Please rate the following attributes in a scale of 1-7.

Attributes 1 2 3 4 5 6 7
Brand
Price
Freshness
Perfume
Size
Hygiene
Lather
Multipurpose use
Life
2. Open-ended and close-ended questions

Open-ended questions: The researcher provides space for the answer.

Close-ended questions: The respondent is asked to select answer(s) provided by the

researcher.
Two structural requirements for this type of questions.
 The response category should be exhaustive and it should end with an option
'others'.
 The answer category should be mutually exclusive except in case of multiple
choice questions.
3. Sequence of questions

The questions should be properly ordered to facilitate the respondents to answer

systematically and with ease.

• Properly ordered questions become interesting for the respondent.

• Very often, it is better to start a questionnaire with identification particulars and

demographic features.

• Some put these at the end of the questionnaire.

4. One should make items/questions clear.

Often the intent /perspectives are clear to the researcher since he identifies a research
topic, the objectives of the study and prepares the questionnaire to collect data.

 But it may not be clear to the respondent

Example:

a) What is your income ?.....................

(from all sources? HH Income ? Per year?)

b) The proposed policy change will have a positive effect on the sales volume of your
company. Yes/No
5. Questions Should Be Relevant and the Respondents Should Be
Competent to Answer

Asking on brand preference to a person who does not know about it is meaningless.

• Asking a respondent in a village:

1.How much do you spend in your children's' education per year?......................

2. What was your income in the last five years?.......................................

6. Questionnaire should include short items. One should avoid double
barrel question.

 Double-barreled question asks about more than one construct in a single survey question.

The word “and” is a hallmark of a double-barreled question

The respondent should be able to read the items quickly, understand the intent and select or provide
an answer without much difficulty.

 Short, clear and pinpointed questions help the respondent in understanding the intent of the
questions and the answers will be accurate to a large extent.
Consider the following question:

Q. The advanced countries should allocate fewer funds for production of arms and divert more funds for
helping the developing countries. Yes/No

When the researcher is using 'and', he should check whether he is asking a double-barrel question
involving two different aspects.

When some people would instantly agree, other may disagree and a few may not be able to answer.

Thus, the answer will not be comparable.

Q. Do you agree or disagree with the following statement: “Managers in my organization are helpful, but
the lack of diversity in the organization is disappointing.”
7.Avoid negative items:

Inclusion of negative items leads to the possibilities of misinterpretation. Often 'not' or

'no' in the question goes unnoticed by the respondents and the answers may lead to
wrong conclusion.

 Q1. Store loyalty is not same as brand loyalty. Yes/No

Q2. I do not buy other brand when my brand is not available. Yes/No

Q3. There is no reason why Kashmir problem should be considered as an

international issue. Yes/No

In all these questions, the respondent has to think carefully about his answer.
8.Questions Should Avoid Use of Biased Items or Format Leading to
Biased Answer
Questions encouraging/forcing the respondents to answer in a particular way is a
biased question.

1. Biased Question

Don't you feel that smoking is a bad habit? Yes/No

2. Inadequate/faulty response category:

Specifying yearly Income category.

< 1 Lakh 1-2 Lakh 2-5 Lakh 5-10 lakh 10-15 lakh >15Lakh

There may be problem while interpretation.

9. Do not use jargons or specialized language and avoid questions that require
calculations

The researcher should see that the specialized terms or languages are not used.

Specialized terms and expression could be made simple and included in the questionnaire so
that the respondent understands it quickly and is at ease to understand.

However, in case the survey includes a sample where the sample units are technical and can
understand the jargon and specialized languages, we can use these.

However, in case of a social/market survey, where a representative sample is drawn which

include people from different background, the use of jargons or specialised language may be
avoided.
10.A Request to the Respondent

In case of the questionnaire being sent by post or by electronic mail, it may be
necessary to write a forwarding letter intimating the respondent about the study, its
objective and requesting the respondent to fill up the questionnaire.

The questionnnaire should have

A general instruction to the respondents regarding giving answer or choices of codes

if codes are used.

A set of instructions to the investigators in cases where the data collection is done by
the investigators.
11. Pre-testing of Questionnaire

 A draft questionnaire requires pretesting in the field which gives an opportunity to

identify gaps/mistakes/omissions, if any, during the preparation of the questionnaire.

A limited number of respondents may be selected for such pretesting.

There is no set rules for fixing up a number for the pretesting, but has to be
subjectively decided by the researcher with respect to the objectives of the study the
sample size and its characteristics.