Sampling design:
In order to obtain adequate and accurate data the researcher conduct investigation into a given
universe.
Statistical data can be collected either by census enquiry or by sample enquiry. In a census
enquiry all the units of universe are studied where as in a sample enquiry only selected number
of units are observed and conclusions are drawn about the universe from their study.
Advantages:
✓ A sample survey is more adaptable than a census enquiry, because small number of
trained investigators can collect the whole data.
✓ Sample enquiry needs less time and money.
✓ It is much more scientific than census enquiry. Because, the extent of reliability of the
results can be known.
A proper procedure is adopted for evaluating sample plan in order to select representative units
so that the influence of change and probability can be estimated.
The selected respondents constitute a sample and the selection process is called sampling
technique or sampling design.
Objectives of sampling:
➢ To get desired information about the universe with a prescribed reliability at the
minimum cost or with a maximum reliability at a given cost.
➢ The data collected should be precise and reliable which can attained through the sampling
techniques.
➢ To obtain the best possible values of the parameters.
Steps in sampling design:
A researcher must pay attention to the following points while selecting or developing a sample
design.
1. Define type of universe:
Clearly define the objects of the sampling known as the universe to be studied. Universe
can be of two types:
I. Finite universe: finite universe contain a definite number of units.
II. Infinite universe: infinite universe in which the number of units is infinite.
Universe can also be classified as,
I. Hypothetical: does not consist of concrete objects. E.g. throwing dice
II. Existent: refers to a population of concrete objects. E.g. income
2. Sampling unit:
While selecting a sample, we have to decide about a sapling unit may be of various types.
E.g. it may be geographical area, one such as state, district, village etc. or a constitute unit
such as house, flat etc. or it may be social units-family, club, school etc.
The researcher has to take the decision regarding the units that he has to select to his
research study.
3. Source list or sampling frame:
It contain the names of all items of a universe especially in case of a finite universe only.
4. Size of sample:
It deals with the number of items to be selected from the universe which constitute a
sample.
The size of sample should neither be excessively large no too small.
The size should be optimum; such size of sample should meet the requirements of
efficiency, representativeness, flexibility and reliability. The large the size of population
usually the bigger sample is needed. The parameter of the research study must be kept
in view, while deciding the size of the sample. Time, money and energy are other factors
which affect the size of the sample.
5. Parameters of interest:
In developing a sample design, the population parameters which are of interest must be
considered. Parameters indicate various statistical measure like mean, SD, correlations
etc. in the universe.
6. Organization of data:
The data collection, editing, classification, tabulation and studying the tabulated material
are involved in the organization of data.
7. Budgeting provision:
Cost consideration from practical point of view have an important impact eye on the size
of the sample and its type.
8. Sampling procedure:
Large number of sample design are available out of which the researcher has to select
one which is suitable for his research study. The cost, time and energy aspects should also
be considered while selecting a particular sample design.
Salient feature of a good sample design:
1. Sample design must be a truly representative sample.
2. It should be free from random sampling error.
3. It should be free from error due to bias.
4. It should not suffer from incomplete coverage of units.
5. Sample design must be viable in the context of funds available for the research study.
6. As far as possible, independent units are included in the sample.
Types of sample design:
There are different types of sample design based on two factors viz,
✓ The representation bias and
✓ The element of selection methods
On the basis of representation, the sample may be either probability sampling or non-probability
sampling.
Probability sampling is based on the principle of random selection where as non-probability
sampling is non-random sampling.
The concept of probability is very much useful in selecting a sample from the population and
helps to know the choice of degree of likeness between the sample and the population. In other
words, it helps to know the probability of the sample being taken from a given universe.
In case of non-probability or deliberate sampling or judgement representing most effectively.
Probability sampling methods:
1. Simple random sampling
2. Stratified random sampling
3. Cluster sampling
Non-probability sampling methods:
1. Systematic sampling
2. Quota sampling
3. Incidental or accidental sampling
4. Judgement sampling
5. Purposive sampling
6. Convenient or chunk sampling
7. Self-selected sampling
Probability sampling methods:
1. Simple random sampling
This technique is the most simple procedure of drawing a sample from a given universe.
In this sampling each unit has an equal chance of being selected. It is also called as equal
probability sampling. The process of selection is a random one but in practice we note
that random sample usually is a representative sample and gives reasonably accurate
data.
Random sampling ensures the law off statistical regularity which state that if on an
average the sample chosen is a random one, the sample will have the same composition
and characteristics as the universe. On account of this characteristics random sampling is
considered as the best technique of selecting a representative sample.
The salient features of SRS are:
✓ It provides 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 an equal probability of being chosen.
In SRS, the population units are denoted as P₁, P₂, P₃ ….. Pᴺ and sample units as S₁, S₂,
S₃…..Sn
Where N and n denote the size of population and sample respectively.
A random sample of 6 units may be x⁴, x⁵, x⁷, x⁸, x⁹, x″ or any such sample.
It is obvious that S= 2⁴, S=x’⁰…
The probability of selecting of each unit is 1/N at the first draw, 1/N-1 at the second draw,
1/N-2 at the third draw and so on.
The SRS is conducted in two different ways viz,
Simple random sampling with replacement
Simple random sampling without replacement
In the former case repetitions are not possible while in the latter are
Advantages:
✓ SRS requires only a minimum knowledge of the universe in advance
✓ It is free of error in classification but the selection of sample does not become
representative.
Disadvantages:
✓ The technique is not applicable when the units are heterogeneous in nature.
2. Stratified random sampling:
Under this technique the selection of sample requires prior knowledge of the universe.
The universe may be divided into strata and then from each stratum some units may be
selected by random sampling.
The stratified random sampling consists of the following steps:
I. The universe is stratified according to the chosen control factors.
II. A separate random sample is drawn from all the units of each stratum.
III. Mean is computed separately for each sample and is approximately weighed to
estimate universe parameter.
IV. Variance is calculated for each sample separately in order to estimate population
variance.
Thus, this sampling method means a number of random samples, one from each stratum
of the universe. Instead, it is a composite of several random samples.
E.g. 500 – 4 group ---100 samples draw 10% from each group
Under stratified random sampling no unit/ group remains unrepresented. A few units
would serve the purpose in case of a perfectly homogeneous stratum.
When the size of different strata are an equal attainment of correct population becomes
difficult and improper stratification may lead to bias in the sample.
3. Cluster sampling:
Involves clustering/ grouping the population and then selecting the groups or the clusters
rather than individual elements for inclusion in the sample.
If KVG bank wishes to sample its farmer borrowers. It has issued credit cards to 15,000
farmers. The sample size has to be kept say, 450
For cluster sampling, this list of 15,000 card holders could be formed with 100 cluster of
150 borrowed each.
Three clusters might then be selected for the sample randomly.
The sample size must often be larger than the simple random sampling procedure.
Advantages:
➢ It is economical. Because it is much easier and less expensive in selecting the units
from the cluster or areas.
➢ It consist of the units of both random and stratified sampling.
Limitations:
➢ It is precise than random sampling and suffer from the element of bias.
Non-probability sampling methods:
1. systematic sampling:
Is one in which every Nth item is selected in a list representing the universe. The number
of ‘N’ is called sampling interval. The first unit is selected at random and the remaining
units are selected automatically in a pre-determined manner.
E.g. A systematic sample of five household from a block of 75 household on random.
First the number of households are listed out and then the 75 are divided into five.
It means every 15th household would be selected. First household is selected from the
number one to fifteen. Suppose think 5th is selected, then of select of every 15th
household beginning with fifth household comes to 5, 20, 35, 50, 65 to complete the
sample.
2. Quota sampling:
Under this type of sampling technique the interviewer is instructed to collect information
from an assigned number of quota of individuals in each of the several groups. The groups
may be specified on the basis of some characteristics such as sex, age, income or others.
Advantages:
✓ This method is very much popular in market surveys and public opinion polls.
✓ Cheaper than random sampling.
Disadvantages:
✓ It lacks representativeness as the interviewer may select people according to his
own experience.
3. Incidental sampling:
Also called as accidental sampling. Here researcher pick up the causes that fall to hand
continuing the process till the size of the sampling.
If 100 samples are needed, he meets on any one of the path of a strict who are ready to
be interviewed. People of different walks of life like teacher, worker, and wholesaler etc.
selected.
4. Judgement sampling:
In order to have good judgement on the basis of sampling the judgement sampling is
taken by investigator with an appropriate strategy.
E.g. if a sample of 10 students is to be selected from a class of 100 for analyzing the
smoking habits of the students, the investigator would select 10 students who is his
opinion represent the class.
5. Purposive sampling:
Here items of the sample are selected in accordance with purposive principle.
Advantages:
✓ In this method some criterion of selection is first laid down and items are selected
in accordance with it.
Disadvantages:
✓ Here the probability of inclusion of some units of the universe in the sample is very
high while probability of other very poor.
6. Convenience sampling:
Also called as chunk sampling. Here samples are selected as per the convenience of the
researcher.
Here any part of the population is selected by the researcher either by probability or by
judgement or by convenience.
7. Self-selected sampling:
Here sample is not selected by the researcher but the members of the universe or
population offers themselves for a selection.
E.g. to get opinion about a particular phenomenon the people who are interested they
come forward on their own to act as subject of the study.
Conclusion:
Normally a researcher should research of simple random sampling. Because under it bias is
generally eliminated and the sampling error can be estimated. But in case of small universe and
its intensive study purposive sampling is considered more appropriate. While selecting any
sampling design the purpose, reliability, accuracy etc. along with its costs should be taken to
account by the researcher.