Sampling

Dr. Raji
Sampling may be defined as the selection of some part of an aggregate or totality on the basis of which a
judgement or inference about the aggregate or totality is made. In other words, it is the process of obtaining
information about an entire population by examining only a part of it. In most of the research work and surveys,
the usual approach happens to be to make generalisations or to draw inferences based on samples about the
parameters of population from which the samples are taken.
 SOME FUNDAMENTAL DEFINITIONS

1. Universe/Population: From a statistical point of view, the term ‘Universe’ refers to the total of the items or
units in any field of inquiry, whereas the term ‘population’ refers to the total of items about which
information is desired. The attributes that are the object of study are referred to as characteristics and the
units possessing them are called as elementary units. The aggregate of such units is generally described as
population. Thus, all units in any field of inquiry constitute universe and all elementary units (on the basis of
one characteristic or more) constitute population.
The population or universe can be finite or infinite. The population is said to be finite if it consists of a fixed
number of elements so that it is possible to enumerate it in its totality. For instance, the population of a city, the
number of workers in a factory are examples of finite populations. The symbol ‘N’ is generally used to indicate
how many elements (or items) are there in case of a finite population. An infinite population is that population in
which it is theoretically impossible to observe all the elements. Thus, in an infinite population the number of
items is infinite i.e., we cannot have any idea about the total number of items. The number of stars in a sky,
possible rolls of a pair of dice are examples of infinite population.
2. Sampling frame: The elementary units or the group or cluster of such units may form the basis of sampling
process in which case they are called as sampling units. A list containing all such sampling units is known as
sampling frame. Thus, sampling frame consists of a list of items from which the sample is to be drawn. For
instance, one can use telephone directory as a frame for conducting opinion survey in a city.
3. Sampling design: A sample design is a definite plan for obtaining a sample from the sampling frame. It refers
to the technique or the procedure the researcher would adopt in selecting some sampling units from which
inferences about the population is drawn. Sampling design is determined before any data are collected.

4. Statisitc(s) and parameter(s): A statistic is a characteristic of a sample, whereas a parameter is a characteristic

of a population. Thus, when we work out certain measures such as mean, median, mode or the like ones from
samples, then they are called statistic(s) for they describe the characteristics of a sample. But when such
measures describe the characteristics of a population, they are known as parameter(s). For instance, the
population mean(µ) is a parameter, whereas the sample mean (X) is a statistic. To obtain the estimate of a
parameter from a statistic constitutes the prime objective of sampling analysis.

5. Sampling error: Sample surveys do imply the study of a small portion of the population and as such there
would naturally be a certain amount of inaccuracy in the information collected. This inaccuracy may be termed
as sampling error or error variance. In other words, sampling errors are those errors which arise on account of
sampling and they generally happen to be random variations (in case of random sampling) in the sample
estimates around the true population values. The meaning of sampling error can be easily understood from the
following diagram:
The magnitude of the sampling error depends upon the nature of the universe; the more homogeneous the
universe, the smaller the sampling error. Sampling error is inversely related to the size of the sample i.e.,
sampling error decreases as the sample size increases and vice-versa. A measure of the random sampling error
can be calculated for a given sample design and size and this measure is often called the precision of the
sampling plan. As opposed to sampling errors, we may have non-sampling errors which may creep in during the
process of collecting actual information and such errors occur in all surveys whether census or sample. We have
no way to measure non-sampling errors.
6. Precision: Precision is the range within which the population average (or other parameter) will lie in
accordance with the reliability specified in the confidence level as a percentage of the estimate ± or as a
numerical quantity. For instance, if the estimate is Rs 4000 and the precision desired is ± 4%, then the true value
will be no less than Rs 3840 and no more than Rs 4160. This is the range (Rs 3840 to Rs 4160) within which the
true answer should lie. But if we desire that they should not deviate from the actual value by more than Rs 200
in either direction, in that case the range would be Rs 3800 to Rs 4200.

7. Confidence level and significance level: The confidence level or reliability is the expected percentage of times
that the actual value will fall within the stated precision limits. Thus, if we take a confidence level of 95%, then
we mean that there are 95 chances in 100 (or .95 in 1) that the sample results represent the true condition of the
population within a specified precision range against 5 chances in 100 (or .05 in 1) that it does not. Precision is
the range within which the answer may vary and still be acceptable; confidence level indicates the likelihood
that the answer will fall within that range, and the significance level indicates the likelihood that the answer will
fall outside that range. We can always remember that if the confidence level is 95%, then the significance level
will be (100 – 95) i.e., 5%; if the confidence level is 99%, the significance level is (100 – 99) i.e., 1%, and so on.
 Types of Sample Design
Sampling is divided into two types:
• Probability sampling: In a probability sample, every unit in the population has equal chances for being
selected as a sample unit.
• Non-probability sampling: In the non-probability sampling, the units in the population have unequal or
negligible, almost no chances for being selected as a sample unit.
Probability Sampling Techniques
1. Random sampling
2. Systematic random sampling
3. Stratified random sampling
4. Cluster sampling
5. Multistage sampling
Non-probability Sampling Techniques
1. Deliberate sampling
2. Shopping mall intercept sampling
3. Sequential sampling
4. Quota sampling
5. Snowball sampling
6. Panel samples
• Random Sampling
Simple random sample is a process in which every item of the population has an equal probability of
being chosen.
There are two methods used in the random sampling:
1. Lottery method: Take a population containing four departmental stores: A, B, C and D. Suppose we
need to pick a sample of two stores from the population using a simple random procedure. We write
down all possible samples of two. Six different combinations, each containing two stores from the
population, are AB, AD, AC, BC, BD, CD. We can now write down six sample combination on six
identical pieces of paper, fold the piece of paper so that they cannot be distinguished. Put them in a
box. Mix them and pull one at random. This procedure is the lottery method of making a random
selection.
2. Using random number table: A random number table consists of a group of digits that are arranged
in random order, i.e., any row, column, or diagonal in such a table contains digits that are not in any
systematic order.
There are three tables for random numbers
(a) Tippet's table
(b) Fisher and Yate's table
(c) Kendall and Raington table.
The table for random number is as follows:

40743 39672
80833 18496
10743 39431
88103 23016
53946 43761
31230 41212
24323 18054

Example: Taking the earlier example of stores. We first number the stores. 1 A 2 B 3 C 4 D
The stores A, B, C and D have been numbered as 1, 2, 3 and 4. We proceed as follows, in order to
select two shops out of four randomly:
Suppose, we start with the second row in the first column of the table and decide to read diagonally.
The starting digit is 8. There are no departmental stores with the number 8 in the population. There
are only four stores. Move to the next digit on the diagonal, which is 0. Ignore it, since it does not
correspond to any of the stores in the population. The next digit on the diagonal is 1 which
corresponds to store A. Pick A and proceed until we get two samples. In this case, the two
departmental stores are 1 and 4. The sample derived from this consists of departmental stores A and
D.
In random sampling, there are two possibilities (a) Equal probability (b) Varying probability.

(a) Equal Probability: This is also called as the random sampling with replacement.

Example: Put 100 chits in a box numbered 1 to 100. Pick one number at random. Now the population
has 99 chits. Now, when a second number is being picked, there are 99 chits. In order to provide equal
probability, the sample selected is being replaced in the population.

(b) Varying Probability: This is also called random sampling without replacement. Once a number is
picked, it is not included again. Therefore, the probability of selecting a unit varies from the other. In
our example, it is 1/100, 1/99, 1/98, 1/97 if we select four samples out of 100.
• Systematic Random Sampling
There are three steps:
1. Sampling interval K is determined by the following formula:
K = No. of units in the population
No. of units desired in the sample
2. One unit between the first and Kth unit in the population list is randomly chosen.
3. Add Kth unit to the randomly chosen number.
Example: Consider 1,000 households from which we want to select 50 units.
K = 1000
50
Calculate
To select the first unit, we randomly pick one number between 1 to 20, say 17. So our sample begins
with 17, 37, 57………….. Please note that only the first item was randomly selected. The rest are
systematically selected. This is a very popular method because we need only one random number.
• Stratified Random Sampling
A probability sampling procedure in which simple random sub-samples are drawn from within
different strata that are, more or less equal on some characteristics. Stratified sampling is of two types:
1. Proportionate stratified sampling: The number of sampling units drawn from each stratum is in
proportion to the population size of that stratum.
2. Disproportionate stratified sampling: The number of sampling units drawn from each stratum is
based on the analytical consideration, but not in proportion to the size of the population of that
stratum.
Sample Proportionate
• If N is the size of the population.
• n is the size of the sample.
• i represents 1, 2, 3,…………..k [number of strata in the population]
P = n1/N1=n2/N2=n3/N3………=nk/Nk=n/N
n1 is the sample size to be drawn from stratum 1
n1 + n2 +………… nk = n [Total sample size of the all strata]
 Example: A survey is planned to analyse the perception of people towards their own religious practices.
The population consists of various religions, viz., Hindu, Muslim, Christian, Sikh, Jain, assuming a total
of 10,000. Hindu, Muslim, Christian, Sikh and Jains consists of 6,000, 2,000, 1,000, 500 and 500
respectively. Determine the sample size of each stratum by applying proportionate stratified sampling,
if the sample size required is 200.

Solution: Let us determine the sample size of strata n1

Total population, N = 10,000
n1=n/N*N1 =200/10,000 *6000= 120
Population in the strata of Hindus
N1 = 6,000 Sample size for Strata n2
Population in the strata of Muslims N2 = 2,000 n2=n/N*N2 =200/10,000* 2000 = 40
Population in the strata of Christians Sample size for Strata n3
N3 = 1,000
n3=n/N*N3 =200/10,000* 1000 = 20
Population in the strata of Sikhs N4 = 500
Population in the strata of Jains N5 = 500 Sample size for Strata n4
Proportionate Stratified Sampling n4=n/N*N4 =200/10,000* 500 = 10
P= n1/N1=n2/N2=n3/N3=n4/N4=n5/N5=n/N Sample size for Strata n5
n5=n/N*N5 =200/10,000* 500 = 10
n=n1+n2+n3+n4+n5 = 120+40+20+10+10
= 200
Sample Disproportion
Let is the variance of the stratum i, where i = 1, 2, 3……….k.
ri σ𝑖
The formula to compute the sample size of the stratum i is ∑ri σ𝑖
σ𝑖 is the variance of the stratum i,
where size of stratum i,
ri = Sample size of stratum i
ri = Ni/N
r = Ratio of the size of the stratum i with that of the population.
Ni = Population of stratum i
N = Total population
Example: The Government of India wants to study the performance of women self help groups
(WSHGs) in three regions viz. North, South and West. The total number of WSHGs is 1,500. The
number of groups in North, South and West are 600, 500 and 400 respectively. The Government
found more variation between WSHGs in the North, South and West regions. The variance of
performance of WSHGs in these regions are 64, 25 and 16 respectively. If the disappropriate stratified
sampling is to be used with the sample size of 100, determine the number of sampling units for each
region.
Solution:
Total Population N = 1,500
Stratum Size of ri = σ𝑖 ri σ𝑖 ri σ𝑖
Size of the stratum 1, N1 = 600
Number the Ni/N ∑ri σ𝑖
Size of the stratum 2, N2 = 500 stratum
Size of the stratum 3, N3 = 400 Ni
Variance of stratum 1, σ1^2 = 8^2 = 64 1 600 0.4 8 3.2 54
Variance of stratum 2, σ2^2 = 5^2 = 25 2 500 0.33 5 1.65 28
Variance of stratum 3, σ3^2 = 4^2 = 16 3 400 0.26 4 1.04 18
Sample size n = 100 Total 100
• Cluster Sampling
The following steps are followed:
1. The population is divided into clusters.
2. A simple random sample of few clusters is selected.
3. All the units in the selected cluster are studied.
The above mentioned cluster sampling is similar to the first step of stratified random sampling. But
the two sampling methods are different. The key to cluster sampling is decided by how homogeneous
or heterogeneous the clusters are.
A major advantage of simple cluster sampling is the case of sample selection. Suppose, we have a
population of 20,000 units from which we wish to select 500 units. Choosing a sample of that size is a
very time-consuming process, if we use Random Numbers table. Suppose, the entire population is
divided into 80 clusters of 250 units each, we can choose two sample clusters (2 × 250 = 500) easily by
using cluster sampling. The most difficult job is to form clusters. In marketing, the researcher forms
clusters so that he can deal with each cluster differently.
Example: Assume there are 16 households in a locality.

Cross Houses
1 X1 X2 X3 X4
2 X5 X6 X7 X8
3 X9 X10 X11 X12
4 X13 X14 X15 X16

We need to select eight houses. We can choose eight houses at random. Alternatively, two clusters,
each containing four houses can be chosen. In this method, every possible sample of eight houses
would have a known probability of being chosen – i.e. chance of one in two. We must remember that
in the cluster, each house has the same characteristics. With cluster sampling, it is impossible for
certain random sample to be selected. For example, in the cluster sampling process described above,
the following combination of houses could not occur: X1 X2 X5 X6 X9 X10 X13 X14. This is because the
original universe of 16 houses have been redefined as a universe of four clusters. So only clusters can
be chosen as a sample.
• Multistage Sampling
The name implies that sampling is done in several stages. This is used with stratified/cluster designs.
An illustration of double sampling is as follows.
The management of a newly-opened club is solicits new membership. During the first rounds, all
corporates were sent details so that those who are interested may enroll. Having enrolled, the second
round concentrates on how many are interested to enroll for various entertainment activities that club
offers such as billiards, indoor sports, swimming, gym etc. After obtaining this information, you
might stratify the interested respondents. This will also tell you the reaction of new members to
various activities. This technique is considered to be scientific, since there is no possibility of ignoring
the characteristics of the universe.
• Area Sampling
This is a probability sampling, a special form of cluster sampling.
Example: If someone wants to measure the sales of toffee in retail stores, one might choose a city
locality and then audit toffee sales in retail outlets in those localities.
The main problem in area sampling is the non-availability of lists of shops selling toffee in a particular
area. Therefore, it would be impossible to choose a probability sample from these outlets directly.
Thus, the first job is to choose a geographical area and then list out outlets selling toffee. Then follows
the probability sample for shops among the list prepared.
Example: You may like to choose shops which sell the brand-Cadbury dairy milk. The disadvantage of
the area sampling is that it is expensive and time-consuming.
Non-probability Sampling Techniques

• Deliberate or Purposive Sampling

This is also known as the judgment sampling. The investigator uses his discretion in selecting sample
observations from the universe. As a result, there is an element of bias in the selection. From the point
of view of the investigator, the sample thus chosen may be a true representative of the universe.
However, the units in the universe do not enjoy an equal chance of getting included in the sample.
Therefore, it cannot be considered a probability sampling.
Example: Test market cities are being selected, based on the judgment sampling, because these cities
are viewed as typical cities matching with certain demographical characteristics. Judgment sample is
also frequently used to select stores for the purpose of introducing a new display.
• Shopping Mall Intercept Sampling
This is a non-probability sampling method. In this method the respondents are recruited for
individual interviews at fixed locations in shopping malls.
Example: Shopper's Shoppe, Food World, Sunday to Monday.
This type of study would include several malls, each serving different socio-economic population.
Example: The researcher may wish to compare the responses of two or more TV commercials for two
or more products. Mall samples can be informative for this kind of studies. Mall samples should not
be used under following circumstances i.e., if the difference in effectiveness of two commercials varies
with the frequency of mall shopping, change in the demographic characteristic of mall shoppers, or
any other characteristic. The success of this method depends on "How well the sample is chosen".
• Sequential Sampling
This is a method in which the sample is formed on the basis of a series of successive decisions.
They aim at answering the research question on the basis of accumulated evidence. Sometimes, a
researcher may want to take a modest sample and look at the results. Thereafter, s(he) will decide if
more information is required for which larger samples are considered. If the evidence is not
conclusive after a small sample, more samples are required. If the position is still inconclusive, still
larger samples are taken. At each stage, a decision is made about whether more information should
be collected or the evidence is now sufficient to permit a conclusion.
Example: Assume that a product needs to be evaluated.
A small probability sample is taken from among the current user. Suppose it is found that average
annual usage is between 200 to 300 units. It is known that the product is economically viable only if
the average consumption is 400 units. This information is sufficient to take a decision to drop the
product. On the other hand, if the initial sample shows a consumption level of 450 to 600 units,
additional samples are needed for further study.
• Quota Sampling
Quota sampling is quite frequently used in marketing research. It involves the fixation of certain quotas,
which are to be fulfilled by the interviewers.
Suppose, 2,00,000 students are appearing for a competitive examination. We need to select 1% of them
based on quota sampling. The classification of quota may be as follows:
Example: Classification of Samples

Category Quota
General merit 1,000
Sport 600
NRI 100
SC/ST 300
Total 2,000

Quota sampling involves the following steps:

1. The population is divided into segments on the basis of certain characteristics. Here, the segments are
termed as cells.
2. A quota of unit is selected from each cell.
• Snowball Sampling
This is a non-probability sampling. In this method, the initial group of respondents are selected
randomly. Subsequent respondents are being selected based on the opinion or referrals provided by
the initial respondents. Further referrals will lead to more referrals, thus leading to a snowball
sampling. The referrals will have demographic and psychographic characteristics that are relatively
similar to the person referring them.
Example: College students bring in more students on the consumption of Pepsi. The major advantage
of snowball sampling is that it monitors the desired characteristics in the population.
• Panel Samples
Panel samples are frequently used in marketing research. To give an example, suppose that one is
interested in knowing the change in the consumption pattern of households. A sample of households
is drawn. These households are contacted to gather information on the pattern of consumption.
Subsequently, say after a period of six months, the same households are approached once again and
the necessary information on their consumption is collected.