SAMPLING TECHNIQUES
Various methods available for selecting samples can be grouped
under two major heads: (i)Random sampling (Probability sampling)
and (ii)Non random sampling (Non - Probability sampling).
Probability sampling (Random sampling)
Probability sampling design is based on probability for the selec
tion of cach item. Probability sampling is also known as random sam
pling or chance sampling. In this method each item has its own chance
for being selected,
Inprobability sampling we can measure the errors of estimation.
Also it ensures the law of statistical regularity, This is why probability
sampling isconsidered as the best technique of selecting representative
sample.
Probability sampling may be simple random sampling and
comt
C.9
(mixed random sampling) random
plex om sampling. Stratified
sampling,
systematic sampling, cluster sampling,
sequential sampling etc. are
complex random sampling designs.
Non -Probability sampling
Non - probability sampling is that sampling procedure
not which does
afford any basis for estimating
the probability for each item
included in thesample. to be
Deliberate sampling
(judgement sampling),
Quota sampling, Convenience
Sampling etc., are non - probability sam
pling, In non probability sampling
generally the person selecting the
sample purposively selects a sample which is suitable
for his study.
(1) Simple random sampling
A simple random sample is a sample selectedfrom a
population in
such a way that every member ofthe population has an
equal chance of
being selected and the selection of any individual does not influence
the selection of any other. The selectionpurely depends on chance. So
the personal bias of the investigator be present in the sample
willnot
selected. Simple random samples may be selected(a)by Lottery method
(b) from table of random numbers.
(a) Lottery Method: Under this method all items of the universe
are numbered or named on separate slips ofpaper of identical size
and shape.
The slips are then folded and mixed up in a container. A blind
fold selection is then made of the number of slips required to con
stitute the desired sizeof sample. While preparing slips it must be
seen that slipsare of identical size, shade, colour etc. Otherwise
there is possíbility of selecting a particular slip. The selectionof
itemscan be made by some mechanical process also.
(b) Table of Randomnumbers: Using random number table we
can select random samples in an easy manner. Several standard
tables of random numbers are available (1) Tippet's random num
ber tables (2) Fisher and Yates tables (3) Kendall and BB smith
tables (4) C. R. Rao's tables ete. Tippets table of random numbers
contains series of random digits arranged in rows and columns.
Random numbers avoid any bias in selecting the sample. The
random numbers are so prepared that they fulfil the criterion of
complete randomness. The use of the tables of random numbers
involves the following steps (1)each member of the population is
assigned a number. (2) From some random point of the table of
random numbers, the random numbers are read out. The item
C. 10
whose number is identical with the random number, read
out,is
selected. (3)The selection of the items is continued tillthe desired
number of items are selected.
Since the selection of items in a simple random sample depends
entirely on chance, there is no possibility of personalbias. Ifsize of the
sample is increased it willrepresent the population in a better manner
Assumptions in asimple random samnple
Following assumptions is made in the selection of a simple ran
dom sample.
All the items of
sample are selected independently of one another
and all the items in population have the same chance of being
included
in the sample. The selection of a particular item in a
draw has no
influence on the probabilities of selectionof any other draw.
Merits ofSimple random sampling
1. There is no possibility of personal bias.
2. The simple random sample usually
represent population particu
larly when the sample size is
large.
3. The accuracy of the estimates based on the
simple random sam
pling can easily be assessed since it is
possible to measure the er
rors of estimation
random sampling. in
4. Random sampling ensures the law of statistical
regularity which
states that the sample willhave
the same characteristic ofthe popu
lation if the sample is selected at
random.
Demerits ofSimple random
sampling
1. often difficult to have an
It is
to - date list of all the items of
up
the population to
be sampled.
2.
Ifthe size ofthe sample is
3. small, the resultsmay not be
If the item in the reliable.
universe are not
homogeneous selection of a
Sample by simple random sample
method cannot beadopted.
(2) Stratified
sampling
If the
field of
enquiry is not
items,this type homogeneous and contains variety of
of
sampling is adopted. In this
the population into method we first divide
different sub
items in each populations known as strata, such that
stratum are
such a way that homogeneous. Various strata are formed in
items within every stratum are
homogeneous in charac-
ter and strata
between themselves
tum, items are selected are
hetrogeneous. From each stra-
by simple random
ample: we want to study the
if sample method. For ex-
Kerala, Kerala consumption pattern of the people in
may be divided into
a number of
zones. From each zone
C.11
be taken at
sample may random. This method
hasthe advantage of a
random sample method and also
simple the different groupsin the
get representation. popu
lation Stratified sampling
method reduces time and
to a great extent.
expense There is little possibility of
any essential
of the population being
group completely excluded. As the items in
ch stratum are homogeneous, stratified
sampling ensures greater ac
curacy.
Merits ofstratified sampling
1 It is more representative than simple random sample. There is
very little possibility of any essential group of the populationbeing
completely excluded.
2. Itensures accuracy since each stratum consists of homogeneous
items.
3. Stratified samples can be more concentrated geographically. Units
from the different strata may be selected in such a way that all of them
are located in one geographical area. This will reduce the time and
COst.
Demerits ofstratified sampling
1. If proper stratification is not done, the sample will have the effect
of bias. If the units selected in a stratum are not homogeneous the
results may not be reliable.
2. In the absence of random selection of items from each stratum
which cannot be ensured in many cases, the results are not reliable.
(3) Systematic sampling
This method is popularly used in those cases where complete list
of the population from which sample is to be drwn is available. Under
this method the items in the population are
included in intervals of
by random sample
magnitude K. From every interval select an item
method. Therefore between twoitems selected,there is an interval K.
For example:when there are 1000 items in the population and
we want
to draw a sample of 200 items, then the interval
5. From the first =
interval select an item by random sample
method. Thereafter the items
selected will be with a gap of 5, If first selected is 3rd item then the
items in the sample will be 3rd, 8th etc.,
and may be
Systematic sampling is relatively simple technique
in
more efficient than simple random sampling. The time and works
smaller. This method
volved in sampling by this method are relatively
if some hidden periodicities are
will not give a representative sample
existing in the universe.
C. 12
is an easier and less costlier method of
Systematicsampling case sam-
used even
of
in
large
pling and can be conveniently populations.
(4) Cluster Sampling
procedure in y which the
sampling is a sampling
Cluster elements
from the population in groups or
chosen clusters
for the sample are
consists in forming suitable
Cluster sampling clus-
rather than singly.
in the sample of clusters selected are
All the units
ters of units.
natural or administratie
veved. The clusters used are pre existing
factories etc.
groups of the population such as schools, colleges,
Cluster sampling can be adopted when there would be no reliable
Jist of the units of the population or when it would be very expensive to
collect such a list.
Cluster sampling is used in many surveys as it is very convenient
and less costly. Tribal survey can be done by cluster sampling tech
nique. Tribes are generally settled in particularareas. Simple random
sampling will fail here. In such surveys few settlementsare selected as
clustersand all the units of the selected clusters are surveyed.
(5) Multistage Sampling
This is a sampling procedure carried out in several stages. The
population is regarded as made up of a number of first stage sampling
units each of which made of second stage units etc. At first, the first
is
stage units are sampled by some suitable method like simple random
sampling. Then a sample of second stage units is selected from each of
the selected first stage units by some suitable method.
This procedure
is extended. The method has the advantage of
being flexible.
(6) Area sampling
Inarea
sampling, groups or clusters are
formed on geographic
basis such as
sectors, randomly and then all
the units within the block
are included in the
sample. Area sampling is
frame does not include every suitable when sampling
member of the population but tests
geographical areas are of
available. The draw back of
this method of
sampling is that the units
in the block
may be usually similar.
The differencebetween
in the former
multistage and
next stage units are multiphase sampling is
that
selected in each draw
latter at each
phase phase a whereas in the
the previous phase, subsample ofthe sampling
is drawn. units selected in
(7) Convenience
Sampling
Aconvenience
sample is obtained by
lation. For example, for the selecting a convenient
study ofthe popu
spending habits of students, if
C. 13
select a sample of studentsfrom a nearby college the sample is con
yenience sample.
(8) Quota Sampling
This method of sampling is
adopted in making surveys of
public
oninion. Only definite quotas of persons in different
socialclasses,
different age groups, different regions etc. are
selected for interviews.
Convenience sampling methods are sometimes not
representative since
they are biased and unsatisfactory.
Under quota sampling the interviewersaresimply
given quotas to
he filled from the different strata with some restriction they
are to be
fled. This type of sampling is convenient and less expensive.
(9) Sequential Sampling
This type of sampling is adopted in acceptance sampling plans in
statistical Quality Control. When a particular lot is to be accepted or
rejected on the basis of a single sample, it known as single sampling.
is
When the decision is to be taken on the basis oftwo samples, it is
known as double sampling. When it is based on more than two samples
it ismultiple sampling. Thus in this method samples are taken one
after another as desired. The ultimate size of the sample under this
technique is not fixed in advance. This type of sampling design is
called sequential sampling.
(10) Purposive Sampling (Judgement sampling) (deliberate sam
pling)
Under purposive sampling the investigator exercises his discre
tion ín the matter of selecting the items that are to be included in the
sample. Therefore, method items are selected on the judgement
in this
of the researcher. For instance, to study the economic conditions of
people ina country, few towns or villages can be deliberately selected
which the researcher feels that will represent the country. Under pur
of items to be included in the
posive sampling, therefore the selection
sample entirely depends upon the direction of the investigatorand no
formula or principles ís followed. Ifthe investigator is wise and unbi
ased,the sample selected in this manner will be representative. But the
chances of personal prejudice and bias are always there. This sampling
design is rarely used in large enquiries. This method is not at all scien
tific.
Difference between purposive sampling and random sampling
The basic difference between purposive and random sampling,
therefore, is that in the former the selection of items to be included in
