Sampling Design

-An Introduction

By
R. Venkatesakumar
 Sampling Design- an overview
• Population/ Universe
is defined as the totality of cases that
conform to some ‘designated’
specifications.
• The specification define the elements to
be in or out.
• Population includes not only individuals,
but also firms, channel members etc.,

Sampling-introduction 2
 Two Methods of Data Collection
• Census
– it is the complete enumeration of the
population.
– All the members who defined as
‘universe’ are met, and the opinion are
recorded.
• One must have complete knowledge about the
universe and its size.
• Even a large number such as 10,00,000 are
considered as ‘countably infinite’ population.

Sampling-introduction 3
 Sampling
• Sample is a portion of population, which
possess the characteristics of the
population, selected for study purpose
through an established procedure and from
which the universe characteristics are
inferred.

Sampling-introduction 4
 Why Sample?
• Lower Cost
• Greater Accuracy of Results
– Possibility of better interviewing (testing),
thorough investigation of missing data etc
• Greater speed of data collection
– Time between need for data collection &
availability of it
• Availability of population elements
Sampling-introduction 5
 Characteristics of a good sample
• Accuracy
– Degree to which bias is absent from the
sample
– No systematic variation
• a variation in the measures due to some known
or unknown influences that ‘cause’ the scores
to lean in one direction more often than other
• Precision
• Representation
Sampling-introduction 6
 Sampling Plan
Any good sampling plan passes through the
following steps.

Define the
Define the Unit / Sampling
Population Sampling Technique
Frame

Choosing an Sample Collection

appropriate size issues of the data
procedure
 Relevance of defining Population /
Universe
• Population is defined from the
Management problem / Research
problem we have chosen
• Whether the individual should be
considered / or household / firm / or its
managers?

Sampling-introduction 8
 Unit Selection

• A Subset from the Theoretical Population

• A List of elements from which the sample
is actually drawn – Sampling Frame
• Sampling unit is a singles element or
group of elements subject to selection in
the sample.

Sampling-introduction 9
 RANDOM SAMPLING ERROR AND
NONSAMPLING ERROR
• Random sampling error is the difference between
the sample result and the result of a census
conducted using identical procedures.
– Random sampling error is a function of sample size
– As sample size increases, random sampling error decreases.

• Systematic (non-sampling) errors result

from non-sampling factors
– primarily the nature of a study’s design and the
correctness of execution.

Sampling-introduction 10
 Sampling Techniques

Sampling can be divided into 2 broad categories,

viz.,
(1) Probability Sampling
(2) Non-Probability Sampling

Sampling-introduction 11
 Probability Sampling
In this method, the samples selected from the
universe, has a ‘Known- Non zero’ probability
of being included in the study.

Non- Probability Sampling

No elements in the population has a known
probability of being included in the study.

Sampling-introduction 12
 Types of Samplings
Probability Sampling
1. Simple Random Sampling
2. Stratified Random Sampling
– Proportionate Allocation
– Disproportionate Allocation
3. Systematic Sampling
4. Cluster Sampling
5. Area Sampling
– One Stage Area sampling
– Two stage Area Sampling

Sampling-introduction 13
 Non- Probability Sampling

1. Convenience Sampling
2. Purposive Sampling
(that conforms to certain criteria)
1. Judgment Sampling
2. Quota Sampling
3. Snow Ball Sampling.

Sampling-introduction 14
 Simple Random Sampling

• Every individual in the population having an

equal opportunity of being selected, and every
individual in the sample is having an equal
opportunity of being in the sample.

Sampling-introduction 15
 – Knowledge about the universe is must.
– Size of the Universe is known to the
researcher.
– Complete list of it is available with the
researcher.
– Random number generation
– Lottery methods

Sampling-introduction 16
 Stratified Random Sampling

• The entire population is divided into

mutually exclusive and collectively
exhaustive classes based on the
predetermined basis
(such as Age Group, Profession, Education, or
Socio- Economic Classification etc., )
• And sample is selected from each
classes.

Sampling-introduction 17
 – The classes are known as stratums.
– The samples are selected either on
proportionate size of the stratum or
disproportionate size of the stratum.
– With in the stratum, mostly simple
random sampling is adopted to select
the samples.
Sampling-introduction 18
 Systematic Random Sampling

• First the population is divided into required

group based on the sample size.
– For example, let there be 1000 dealers for the company in Tamil
Nadu
– the researcher wants to select a sample size of 100.
– The entire population is divided into groups of 10(1000/100=10)
– a number between 1-10 is picked at random, say 7.
– Then the researcher started choosing 7,17,27, etc as the samples
• the basic requirement is listing of all the
population elements
• hence a knowledge about the population size
is must.

Sampling-introduction 19
 Cluster Sampling
• First the population is divided into required group
based on the characteristics/parameters pre-
specified just as in the case of stratification.

• Here there is one difference - all the groups are

similar in nature and all the respondents are
identical in nature

• From the identical clusters, some of the clusters

are considered for study.

• That is either all the respondents in that clusters

or portion of it is studied.

Sampling-introduction 20
 Cluster Design Issues
• What extent the clusters or homogeneous?
– As far as possible, clusters should be
heterogeneous
• Shall we seek equal or unequal clusters?
– Try to develop equal clusters by breaking larger
into smaller and clubbing smaller into larger ones
• Size of the clusters
– No priori defined limits

Sampling-introduction 21
 Cluster Design Issues

• Use of single stage or multi-stage

clustering?
• Sample size to be taken for the study
purpose

Sampling-introduction 22
 Area Sampling
• First the population is divided into required group
based on the geographical basis
• such as Northern part of the town, southern
part of the town etc.,
• the groups are homogeneous in nature
• within the group, it is heterogeneous

– From this group, either one group or some of the

groups are considered for study.
– If we take one area - study it completely - it is one
stage area sampling
– if we apply sampling, I.e., choose a portion of the
area through a sampling technique, it is two-stage
area samplings.
Sampling-introduction 23
 Convenience Sampling
• The researcher chooses the sample according to
whims and fancies/ convenience to him/her.

• Cannot be classified into a proper technique

• Universe characteristics are not considered for

the study purpose

• There is no guarantee of the representation of the

population parameters.

• Nothing can be inferred about the population

Sampling-introduction 24
 Judgment Sampling

• Here the researcher uses ‘some’ judgments to

select the respondents
• It may be an expert opinion or an observation
which the researcher observed during the
initial phase of the study.

• It is better than a procedure just discussed.

• Again you cannot infer anything about the
population characteristics/ representativeness
of the population elements etc.,

Sampling-introduction 25
 Quota Sampling

• It is a procedure as similar to stratified random

sampling
• The stratification was made according to the
pre-specified characteristics.
• Then for each stratum/class/group a fixed size
is allocated
• The researcher/field staff who is collecting the
information just like that can collect the
samples
• (only goal is the sample size - no botheration about the
universal characteristics)

Sampling-introduction 26
 Snow Ball Sampling
• Here we start the survey with an initial lead.
• Then from the initial lead, the next respondent
is getting introduced
• The second respondent introducing the third
and so on.
– The researcher has no control over the universe
parameters are included in the study or not.
– May be very useful for the researcher who is new to a
geographical area
– Certain extend it is a better sampling than the
Convenience sampling.
– But any generalization is beyond the scope as it is
applicable to all the Non- Probability sampling

Sampling-introduction 27
 How can we overcome the issue of
randomization in non-probability
sampling?
• In the multi-stage sampling procedure employed; probability
sampling has been applied up to a certain stage.
• In subsequent stages, non-probability sampling has been
followed.
• Calder, Philips, and Tybout (1981) suggested that because
strict statistical sampling often is not feasible, other
procedures might be invoked to enhance the
representativeness of individuals in the research.
• Furthermore, using non-probability sampling on a relatively
homogenous population reduces the potential for random
errors when compared with a sample from the people
(Sproles and Kendal, 1986; Bakewell and Mitchell, 2004).

Sampling-introduction 28
A Case:

Rural consumers within a village often exhibit relatively

homogeneous pattern of socio-cultural profiles sometimes the
economic profiles are also very similar. In certain blocks of the
district, a majority of the farmers prefer to cultivate the same
crop.

Solution:

Mehta, Lalwani, and Han (2000) put forward that when the
convenience sampling method is used, a researcher could
randomize data collection to the extent of place and time of
collection.

For example, data may be collected from the same place at

different times of the day, i.e., morning, afternoon and evening.

Sampling-introduction 29
 Thank you

Sampling-introduction 30