SAMPLING

C. R. Darolia, Ph. D.

Dept. of
Psychology
Kurukshetra
University
General Concepts
Generally research is conducted on a sample
rather than on the whole population.
Population: is the aggregate of all the units/
elements that share some common set of
characteristics.
It comprises the universe for the research under
consideration.
Census: Before sampling, population needs to be
surveyed for various informations. Census
involves a complete enumeration/listing of the
elements of the population.
General Concepts
Sample: is a portion of the population selected
for participation in the study. On the basis of
which inference is drawn for the population.
General Steps
Define the population -> Determine the
sampling frame -> Select sampling techniques
-> Determine the sample size -> Execute the
sampling process
 Sampling Techniques

Probability Sampling Nonprobability Sampling

Simple Stratified Cluster

Random Random Random

Incidental Purposive Quota Snowball

Sampling Sampling Sampling Sampling
Probability/Random Sampling
Each element in the population has a known and
equal probability of selection in the sample.
It implies that every element is drawn
independently of every other element.
1. Simple Random Sampling
Each possible sample of a given size has a known
and equal probability of selection.
Firstly, the researcher compiles a sampling frame.
Where each member is assigned a unique
identification number.
Simple Random Sampling
Then, random numbers are generated to
determine which elements to include in the
sample.
Random nos. may be generated by using any
techniques of randomness, e.g., computer
routine, table of random numbers, fishbowl or
draw of lots, every ith no. beginning with a
random number etc.
In the fishbowl/draw of lots method sampling
may be with replacement or without
replacement.
Simple Random Sampling
With replacement ensures equal probability of
each element and also equal probability of each
possible sample. e.g., N=10, n=2; p of each
element is 2/10=.20.
No. of possible samples of n size (n=2) is
Nn=102 =100
Without replacement probability of elements
decline successively. e.g., 1/10, 1/9 etc. and so
is the No. of possible samples, i.e.,
N!/(N-n)! n! 10!/(10-2)! 2!
(10x9x8x7x6x5x4x3x2)/8x7x6x5x4x3x2x2 = 45
Simple Random Sampling: Assumptions
The random sampling is most representative but
if certain assumptions are met.
• Members/elements of population are
enumerated/listed or catalogued in some way.
• Form of dist of the trait in the population is
known to be normal.
• The population is known only in general terms.
Each characteristic of the population may not
be known specifically.
Simple Random Sampling: Evaluation
Advantages
1. Most representative: drawn all through the
population.
Each element has equal and independent
probability.
2. Exact composition of population need not to
be known. Theoretically it represents all
important characteristics of the population.
3. Easiest and simplest technique to draw
random sample.
Simple Random Sampling: Evaluation
Advantages
4. Basic and fundamental all other random
sampling techniques.
5. Sampling error of any randomly drawn sample
can easily be assessed.
6. No risk of classification errors. The population
is not classified in to narrower strata.
Disadvantages
1. Difficulty in getting/having appropriate sample
frame. No lists are available with complete
coverage of the population.
Simple Random Sampling: Evaluation
Disadvantages
2. Risk of exclusion of elements existing in small
numbers.
3. Risk of disproportionate sampling. Even the
knowledge of the researcher about segments
of popu goes unutilised.
4. Greater sampling error. As compared to other
random sampling techniques of n size sample.
It is so because it arrives at heterogeneous
sample.
2. Stratified Sampling
It is a type of random sampling, which ensures
representativeness and avoid bias.
Useful when the population is composed of
subgroups or strata of certain demographic
factors, e.g., sex, residential background (rural
vs. urban, or education.
To ensure representativeness, elements are
drawn randomly from each strata in accordance
with the size of the subgroups.
Stratified Sampling
Main feature of the stratified sampling is that
each of the stratum is homogeneous internally,
and there heterogeneity between strata.
Therefore, it leads to greater precision or lower
sampling error.
Initially, after a survey of the population, it is
divided into certain relevant strata along one or
more factors.
Then, requisite sample is drawn from each of the
stratum using any of the techniques of random
sampling as in simple random sampling-fishbowl,
computer routine, table of random numbers.
Stratified Sampling
Two basic reasons for stratification are:
a. Provides convenience in sampling elements
from each possible subgroup.
b. Increases the precision in estimating the
attribute of the whole population. The chances
of variation within stratum are minimum and by
combining them, we can have more precision.

Types:
1. Proportionate stratified sampling.
2. Disproportionate stratified sampling.
Stratified Sampling
1. Proportionate stratified sampling.
Equal proportion of elements are drawn from
each stratum. It ensures representation of
each of the sections according to size of the
population.
Advantages
2. Greater representation of the population. Even
the elements in few numbers are represented.
3. Lesser sampling error. Due to within stratum
homogeneity.
Advantages
3. Equal weightage to each population
characteristics.
Disadvantages
1. Difficulty in stratification. It requires in-depth
understanding of the population.
2. Time consuming.
3. Risk of classification error.
Stratified Sampling
2. Disproportionate stratified sampling.
As name suggests, it does not sample units in
proportion of their population size.
Smaller population may be over represented
to have some minimum number of elements in
the sample.
Selection procedure is same as in
proportionate sampling.
Advantages
Sufficient/appropriate weightage may be given to
stratum having thin population.
Disadvantages
1. Risk of bias in deciding n from the various
strata.
2. Over or underrepresentation of certain strata.
3. Thorough knowledge of the composition of the
population is required.
4. Risk of misclassification.
3. Cluster Sampling
Also known as area or multistage sampling.
It has origin in agriculture and farming research
to study the effect of various kinds of fertilizers,
soil treatment, planting methods and irrigation
etc. on the yield of the crop.
In cluster sampling it is assumed that the
population is indigenously divided into certain
geographical divisions, like countries, states,
districts/province, tehsiles, blocks etc.
Investigator selects randomly the requisite no. of
geographical units and sample may be drawn….
Cluster Sampling
For example, one is interested in studying the
attitude of people of Haryana towards dowry.
Then he/she may select some districts at
random from all dist. of Haryana. Each dist. has
a no. of Tehsiles. Some of the Tehsiles may be
drawn randomly. The same way there are
further divisions of tehsiles and so on and so
forth.
Finally, one draws sample from selected smaller
geographical units of population using the tech.
of simple random sampling.
Cluster Sampling
Contrary to stratified sampling, clusters are more
or less similar in characteristics, hence
homogeneity between clusters. But
there is heterogeneity within each cluster.
Utilities:
a. When enumeration of population is not
possible/inaccessible, cluster sampling is the
best way.
b. Public opinion/polls are easily conducted.
c. Convenience of large scale surveys of social
behaviour.
Cluster Sampling
Advantages
1. Efficient in covering larger geographical area.
When popu covers larger area, list of the units
may be inaccessible, then cluster sampling
easier.
2. Substitutes of missing cases are readily
available. Because clusters rather than
individuals are sampled in the cluster
sampling.
3. Economical in time, effort, and cost.
Cluster Sampling
Advantages
4. Flexible. Since it is a multistage sampling,
technique sampling may be changed as per
requirement. For example, at 3rd stage one
may stratify the population and draw sample
accordingly.
5. Convenience and ease of procedure.
Cluster Sampling
Disadvantages
1. Greater degree of sampling error. Within
cluster heterogeneity is as high as in simple
random sampling.
2. Cluster sizes are usually unequal and therefore
may lead to bias in sampling. Researcher does
not have control over the size of cluster.
3. Threat to independence. Individuals drawn
from one cluster may not be independent of
individuals from other cluster. There is
possibility of mobility between clusters.
Non-Probability/Non-Random Sampling
Non-Probability/Non-Random Sampling refers to
sampling of readily or easily available
individuals. Primarily it is a matter of
convenience rather than any theoretical
consideration.
It simply based on the principle of first-
come first-serve basis.
Researcher broadly defines the population and
decides the sample size.
Then individual units are drawn from who so ever
is available.
Non-Probability Sampling
1. Incidental sampling
Also called accidental sampling, is a popular
method of non-random sampling.
The term incidental refers to easily or
readily available units.
There is no systematic way to accomplish
sampling, who so ever happen to be available
or meet is selected, hence, incidental.
Its always a matter of chance who is readily
available. Sample is drawn on first-come first-
serve basis.
Incidental Sampling
Used when preliminary survey is being conducted
or one is interested only in approximation.
Generalizations based on such sampling are
often misleading.
Advantages
1. Alternate to random sampling when
cataloguing of the population is not feasible.
2. Most convenient and easy method of
sampling.
3. Economical. In respect of time, effort and
cost.
Incidental Sampling
Disadvantages
1. Risk of bias. It is the subjective decision of the
Researcher to select the units as a matter of
convenience.
2. Lack of generalizablity. From incidental
samples nothing can be generalized with
confidence.
Because every next sample would have
different properties.
3. High degree of sampling error. There is no
coherence in the sampling units.
Incidental Sampling
Disadvantages
4. Weak in reliability and validity.
Non-Probability Sampling
2. Quota sampling
Quota sampling is one of the important
methods of non-random sampling.
It may be considered as non-random
version of stratified sampling.
Like stratified sampling, in
quota sampling also the popu is categorized
into certain strata, like, sex, race, religion etc.
Then the no. or quota of individuals is decided
for each of the stratum.
Quota Sampling
Individual units sampled from each of the
stratum using incidental sampling procedure
until specified quota is not met.
Generally, quota from each stratum is decided in
proportionate to the size of population.
In this respect, it is similar to stratified sampling.
Evaluation
Advantages
1. Most satisfactory procedure when quick and
approximate results are desired.
Quota Sampling
Advantages
2. Good alternative to Probability sampling. When
it is impossible to have random sampling, it is
a good alternative.
3. Greater precision. As compared to other
methods of non-random sampling.
Disadvantages
1. Lack of representativeness. Incidental
selection does not ensure sampling of all
possible population parameters.
Quota Sampling
Disadvantages
2. Lack of generalizablity. From incidental
samples nothing can be generalized with
confidence.
Because every next sample would have
different properties.
3. Risk of bias. It is the subjective decision of the
Researcher to select the particular units as a
matter of convenience.
4. Risk of misclassification.
Non-Probability Sampling
3. Purposive sampling
Also known as judgmental sampling. As name
indicates, the units which seem to serve the
purpose as per the judgment of the researcher
are selected.
Sample typical of the population to be surveyed/
studied. For example, if the researcher wants
to study certain aspects of criminals’
behaviour, he may take purposive sample of
the police personnel or jail warders.
Purposive Sampling
Researcher believes that they might know the
aspect of criminals’ behaviour he is interested in.
Similarly, a researcher interested in the study of
stealing habits in young children may take a
sample of teachers in elementary/primary
schools.
Here, representativeness and suitability of the
sample depends upon the accuracy of
researcher’s judgment in meeting the
requirement.
Otherwise, there is potential risk of bias.
Purposive Sampling
Evaluation
Advantages
1. Very convenient and simple.
It is most convenient among all
non-random sampling procedures.
It is in the hands
of the researcher to decide the sample/units
to be sampled.
2. Relevance. Purposive sampling guaranties, to
some extent, that the selected sample meet
the requirement of the design of the study.
Purposive Sampling
Disadvantages
1. Lack of representativeness.
Since it is not drawn randomly, many of the
units have no chance to be selected.
2. Risk of bias.
Too much reliance is placed on
the wisdom/ability of the researcher to judge
the suitability of the sample. This leaves much
scope for subjectivity in the sampling.
3. Lack of external validity. Results are valid only
to the judgmental sample, not to others.
Thank You!