Sampling Techniques

In a research it is not always required, possible or

desirable in view of the time and resources constraint
to consult every unit of the target group or observe
the phenomena in detail. Analyzing large masses of
data may be proved as a futile and unnecessary
voluminous exercise. In such circumstances
scientific use of sampling technique will be
meaningful and useful.
❑ All the units, elements or items in the field of inquiry
constitute universe. It is also known as population.
❑ If all the units are consulted and covered the method of
research is census method.
❑ It assures highest accuracy and concrete description of a
phenomena without any element of bias as all the
elements are taken into consideration without any chance
of being left.
❑ Certain studies always require this method. For example,
population census, literary digests, utility subscriptions,
voting registrations etc., where one cannot opt for
sampling.
❑ This type of inquiry involves large paraphernalia , time,
money and energy besides organizational abilities.
 ❑ Data should be carefully collected lest the
wrongful entry of the characteristics may result in
large bias.
❑ The universe may be finite or determinable
universe or may be infinite where the possibility of
covering all the elements will not be there.
❑ In order to overcome the difficulties in conducting
the census method a scientific sampling design is
advocated.

A good sample fairly and accurately fit into the description

of the whole. It facilitates inference about the phenomena
with unknown or large parameters and it serves as a
technique to statistically test the hypothesis.
➢ According to Goode and Hatt a sample is a smaller
representation of a larger whole:
➢ According to Kothari it is representative of the total
population ….. in order to produce a miniature cross-
section.
➢ According to Mohsin a sample is a portion of entire lot of
certain kind of objects. The entire lot is called the
universe or population.
➢ According to Wimmer and Dominick a sample is a sub
set of the population that is taken to be representative of
the entire population.

❑ The procedure adopted to draw the sample may be

termed as sampling technique and the definite plan for
obtaining the sample from the universe is termed as
sampling design
Universe

Sample
Advantages of sampling method:

➢ Saves the time and reduces the cost

➢ Provides quicker results
➢ Save human resources and will be within the
manageable limits
➢ Assured accuracy of results if properly executed the
technique
➢ Only method when census method could not be
adopted
➢ Amenable to statistical analysis and testing.
Limitations of Sampling method:
➢ Requires specialized knowledge of the sampling
techniques in order to execute the work
➢ Chances of systemic bias or sampling errors are more if
the adopted method is not suitable or defective.
➢ If it is a very complicated and complex phenomena
sampling technique may not facilitate representative
sample
➢ By very nature as noted require only census method
➢ If the universe itself is small representative sample may not
be possible.
Characteristics of good Sample:

✓ Truly representative in nature

✓ No systemic bias or sampling error
✓ Proper devices to control sampling error and systemic
bias
✓ Viable in the terms of both human and monetary
resources
✓ Must yield to precise estimates and lead to accurate
results
✓ Must be adequate in size for drawing proper inferences
 Sampling methods (representation basis)

Non-probability sampling Probability sampling

1. Simple Random
1. Haphazard Sampling sampling
or convenience sampling
or accidental sampling
2. Complex Random
Sampling (Cluster and area
sampling; systematic
2. Purposive sampling, sampling; stratified
judgment sampling, sampling; multi stage
quota sampling or snow- sampling; sub-sampling;
ball sampling replicated sampling) etc
In deciding whether to use one or the other type one has
to consider
1. Cost Vs Value
2. Time constraints
3. Purpose of the study and
4. amount of the error allowed.
In Probability Sampling sample is selected according to
according to guidelines whereby the chance of selection of each
unit is known. Each unit stands for a chance of selection. It is
based on the theory of probability. Hence it results in a
representative sample. Because of randomization the systemic
bias is reduced, and genralisation is accurate.

Where as in non probability sampling the mathematical

guidelines are not followed and does not give scope for
probability. The units do not stand for equal chance of being
selected. The representative aspect of the sample is not assured.
There is a scope for systemic bias and statistical errors.
Inferences thus drawn may thus become questionable.
However, as a matter of expediency for certain studies non
probability sampling method is adopted for its simplicity,
convenience, availability of time and low cost or the results are to
be localised. Practicability is the basis for adopting this
technique.
Types of Non-probability Samples:

Haphazard sampling, or convenience sampling or

accidental sampling techniques are available sample
collections from the readily accessible units of study. It may
be like hit and miss fashion. Such studies may be elementary
or exploratory in nature with the danger of unknown
quantity of errors and suffer from least reliability. The
proponents of this method argue that if every
unit represents whole and basic homogeneity is
assured even by this type of sampling the
characteristics of the phenomena can be
estimated. One has to accept that these types of
sampling are informally used by the researcher for
validation of a questionnaire, in solving the problems
faced in adopting the methodology etc.
Purposive sampling is a better type of non probability
sampling. Though it is deliberate selection of the
sample, it is based upon some pre-determined criteria.
It is also known as judgement sampling as it depends upon
judging the appropriate selection which according to the
researcher or an expert is most suitable. But still it is
subjective. It does not guarantee representativeness.
However, the generalizations drawn may be appropriate
with regard to particular unit. For example the study of
status in a particular industry, RTC etc.
Quota sampling is a method of selecting a particular pre
determined percentage of units for the purpose of study. This
is generally done when the universe consists of various
categories like, religion, race, caste, sex, geographical
distinctions etc. such percent is selected based upon
accessibility. Ie., it is a type of stratified sampling with
non-random selection within the strata.

Snow-ball sampling is based upon the selections within the

category through some informants. For example if the
students of a particular origin have to be contacted the list
provided by the authorities of the school may be relied upon
Probability Sampling methods:

The elementary type of probability sampling is the simple

random sampling. In this type each unit of the universe
has an equal chance of being selected. If the selected unit
is removed from the subsequent selection the procedure is
termed as simple random sampling without replacement.
If the selected item is again placed in the universe the
procedure is termed as simple random sampling with
replacement. Sample with replacement is used less
frequently.
Sampling at random may be drawn either from finite
population or infinite population.

The finite sampling gives each element in the population

an equal probability of getting into the sample and all
choices are independent of one another. It gives each
possible sample combination a probability of being
chosen.
Selection of each item in random sample from an infinite
population is controlled by the same probabilities and that
successive selection are independent of one another. It
goes with sample with replacement
❑ The selection of items in a random sampling technique can be
made by various devices like lottery in different modes.
❑ However such methods have limited utility. Some of the
statisticians like Tippett, Yates, Fisher have prepared random
number tables.
❑ Among them Tippett is popular which gave 10400 four figure
numbers. He selected 41600 digits from the census reports
and combined them into fours to give his random numbers
which may be used to obtain a random sample.
❑ Random sample in telephone surveys often use random digit
dialing where certain digit numbers are selected ether
generated by the computer, or number tables or through
directories.
Advantages of simple random sample:
1. specialized knowledge about the universe is not
required
2. external validity may be statistically inferred.
3. representative group is easily obtained.
4. errors are eliminated

Disadvantages:
1. Listing of the universe is at times impracticable
2. sampling errors are possible
3. procedure may be expensive and lengthy
Complex Random Sampling methods:

❖ In a Systematic sampling, an improvised simple

random sampling, every ‘n’th item is selected from the
universe at a regular sampling interval at a determined
sampling rate.
❖ For example if 50 or 5% sample is to be drawn from
thousand respondents the sample at random starts say
at 1 and goes on 20,40, 60 and so on. First unit is
selected at random and remaining items are selected at
regular intervals.
❖ Systematic sampling depends upon adequacy of the
sampling frame. One major limitation over this method
is that the procedure is vitiated by fixed periodicity.
Advantage:
1. selection is easy
2. selection is more accurate than in a simple random
sample
3. procedure is inexpensive

Disadvantages:
1. listing of the population is not always easy
2. periodicity may cause bias in the proicedure
※ Stratified Sampling is adopted when the chosen
universe is not homogeneous. Under this technique the
population is divided into sub-groups having common
characteristics known as strata. Then selection of items
from each stratum constitute sample. Here strata are
purposively formed based upon expertise, experience,
judgement of the researcher or experts or based upon
some pilot study. The sample, thus collected gives a
better estimation of the whole.
※ According to Babbie,
stratified sampling ensures the proper representation of the
stratification variables to enhance representation of other
variables related to them. Taken as a whole, then, a
stratified sample is likely to be more representative on a
number of variables than a simple random sample.
While attempting at this type the following procedures are to be
clearly laid.
► Procedure to form strata
► Procedure to select items from each stratum
► Procedure relating to the number items to be selected from each
stratum
► procedure relating to fixing sample size.
Advantages:
1. Selection can be made from the homogeneous sub-group -
Sampling error is reduced
2. representativeness of variables is assured
3. comparative analysis of other populations can be made
Disadvantages:
1. In order to form strata prior thorough knowledge of the universe is
required,
2. The procedure involved is time consuming and may prove costly
3. If the universe is limited finite and small it may be difficult to draw
the strata and thereby valid sample
Cluster Sampling method is used when the area under investigation
is a larger one. In such cases the area can be divided into smaller
non overlapping areas known as clusters. The clusters are selected
at random and sample consisting of all the units in the cluster is
collected.
Advantages:
1. If the clusters are well defined the cost will be reduced
2. Only a part of the population need to be enumerated
3. The estimates of the cluster parameters can be compared to the
population
Disadvantages:
1. Clusters may not represent the universe
2. Sampling errors may creep in and it is less precise than the
random sampling
3. Each unit has to be assigned to some specific cluster
If the clusters relate to a geographical area it is termed as Area
Sampling.
Multi stage sampling is an improvement over the above
cluster sampling and utilized in a big geographical areas.
If a country level investigation is attempted, initially cluster
sampling can be done at state level, then at districts,
blocks, panchayats or pin-code areas. The sampling
frame under multi-stage sampling is developed in partial
units thereby making it easier to administer in
considerably large geographical areas.

Sequential sampling technique usually applied in the

context of statistical quality control. The size of the
sample is not set in advance of the investigation but
adopted when survey progresses and is determined
based on appropriate mathematical rules.