Unit 4
Teena Y. Sharma
Introduction
Theory of Sampling: It is the study of relationships existing between a population and various populations drawn from this population. The sample so selected has to be truly representative of the population. Sampling is one of the most fundamental concept
underlying any research work. A sample enables a researcher to intelligently estimate the population parameters. Most of the researchers utilize the concept of Sampling. The other method used by researchers is Census Method, but Sampling is more predominant.
Teena Y. Sharma 2
Theory of Sampling
Mainly studies relationship between a population and
the samples drawn form that population. It helps in moving from particular to general concept, by moving from a sample to population Generally applicable to random samples only.
Teena Y. Sharma
Sample v/s Census
Census Method
Has almost 100% accuracy, as
Sampling Method
Has good accuracy (+) Only some units, which
all elements are studied (+) All units are studied (+) Consumes a lot of resources (time &money) (-) Expensive (-) Unmanageable when population is large (-) No cautions required for choosing elements, as all are considered for study (+)
Teena Y. Sharma
represent all units are studied (-) Saves resources Reasonable (+) Sample is of a reasonable size, so manageable (+) Caution is required for selecting elements which will fall in the sample, so that they truly represent the population (-)
4
Sample v/s Census
Parameter
Budget Time Available Population Size Variance in Characteristics of Elements Cost of Sampling Errors Cost of Non-Sampling Errors Attention to Individual Cases Required
Teena Y. Sharma
Conditions favoring use of
SAMPLE Small Short Large Small Low High No CENSUS Large Long Small Large High Low Yes
5
Concepts in Sampling
Population/ Universe- Finite / Infinite
Census Sample
Sample Survey
Parameters Statistics Sampling Unit Sampling Frame Sampling Error
Teena Y. Sharma 6
Concepts..
Non-sampling Errors
Sample Size Random error
Bias
Precision Non-response Errors
Teena Y. Sharma
Objectives of Sampling
Statistical estimation: The prime objective of
Sampling is to intelligently estimate population parameters Testing a hypothesis: Sampling is also used to test a statistical hypothesis. A sample is drawn and the data collected from the sample is analyzed so as to accept or reject the hypothesis based on the difference between the two values. (hypothesis value and sample statistic) Statistic Interference: Generalizations regarding populations and accuracy of these generalizations can be done using the theory of Sampling.
Teena Y. Sharma 8
Need of Sampling
Sampling saves time and money. A Sampling study is
usually less expensive than census study and produces results at relatively faster speed. Sampling usually enables more accurate measurements for a study, as it is generally conducted by trained and experienced. Sampling remains the only way when population contains infinitely many members. It also remains the only way when a test involves the destruction of the items under study. Sampling usually enables to estimate sampling errors and thus assist in obtaining information concerning some characteristics of this population.
Teena Y. Sharma 9
Limitations of sampling
Need of specialized knowledge
Need of discipline Chances of bias
Need of large samples as there are Errors due to small
samples Complicated sampling plans Sampling errors Difficulties in sticking to a sample Impossibility of sampling
Teena Y. Sharma 10
Sample Design
Definition : Sample design is a definite plan for
obtaining a sample from a given population It refers to the technique or procedure the researcher would adopt in selecting items for the sample Sample design also lays down the number of items to be included in a sample. It is determined prior to data collection.
Teena Y. Sharma
11
Steps in Sample design
Type of Universe: Whether finite or infinite.
Sampling unit: Whether geographic, social, or
construction or any other. Source list: Also known as Sampling Frame , and contains all names of the universe. Hence, it should be comprehensive, correct, reliable and appropriate. Size of sample: should neither be large nor too small. It should be optimum. Before deciding about the sample, researcher should determine the desired precision needed and the parameters to be estimated.
Teena Y. Sharma 12
Steps..
Parameters of interest
Budgetary constraints Sampling procedure
Teena Y. Sharma
13
Characteristics of a good sample design
Sample design must result in truly representative
sample. It should result in small sampling errors. Sample design must be viable in context of funds available for research study. It should be such that systematic biases can be controlled. Sample should be such that the results of sample study can be applied for the universe with a reasonable level of confidence.
Teena Y. Sharma 14
Types of Sampling Techniques
Teena Y. Sharma
15
Different types of Sampling Techniques
Element Selection Technique Probability Sampling Non-probability Sampling
Unrestricted Sampling
Simple Random Sampling Haphazard Sampling or Convenience Sampling
Restricted Sampling
Complex Random Sampling ( Cluster Sampling, Systematic Sampling, Stratified Sampling)
Purposive Sampling ( Quota Sampling, Judgment Sampling)
Teena Y. Sharma
16
Probability Sampling
Concept of Probability: Probability is termed as the
chance of a certain event happening or not happening. Probability Sampling:
It gives each element in the population an equal probability of
getting into the sample. All choices are independent of each other. Probability of any bias taking place is generally done away. If we design a random sample of size n from a finite population N then NCn possible samples has same probability. Each element has 1/ NCn of being chosen.
Teena Y. Sharma 17
Uses of Probability Sampling method
Generally applied in the case of critical decisions
where accuracy is at premium, regardless of time and cost involved.
Teena Y. Sharma
18
Merits of Probability Sampling
It is the only approach where in it is possible to
formulate determinable representative sampling plans. The population parameters are more accurate and reliable. Most widely applied method in descriptive research studies aiming at quantitative estimates. It results in more apt representative samples
Teena Y. Sharma
19
Demerits of Probability Sampling method
For reliable results , and generalizations, the size of
sample required is considerably large. Cannot be used in situations with budgetary constraints.
Teena Y. Sharma
20
Types of Probability Sampling method.
Probability Sampling
Simple Random Sampling
Systematic Sampling
Cluster Sampling
Area Sampling
Multi-phase Sampling
Stratified RS
Teena Y. Sharma
21
Random Sampling
Simple Random Sampling There is total absence of human judgment. Each element has equal chances of getting selected It does not mean picking up in haphazard manner There are various methods present to do so:
Lottery Method Using Random number tables Use of Computers
Teena Y. Sharma
22
Advantages of Random Sampling
Produces least biased and most representative
samples. Simple, as researcher does not need to make a decision criteria Does not require prior knowledge of composition of population
Teena Y. Sharma
23
Systematic Sampling
Also known as Quasi-random or Pseudo-random sampling. Is a special form of Simple Random Sampling. Is not a true random sampling in the sense in selecting a
sample of n units from a population of N units, only first element is selected randomly and thereafter every (N/n)th is selected for inclusion into the sample. The number (N/n) is designated as i and is termed as sampling fraction. First, element is k, a random number between 1 to i, later all elements are chosen corresponding to k. Hence, we have the elements as k, k+(i), k+2(i)..k+z(i)
Teena Y. Sharma
24
Cluster Sampling
Generally used when , the population element are
spread over a wide area. In this case, sampling unit is a cluster. Certain clusters are selected. For each cluster a simple or stratified sampling method is applied to arrive at a sample. This method is widely applicable in test-marketing of products, socio-economic surveys, demographic studies, public opinion polls etc.
Teena Y. Sharma 25
Steps in Cluster Sampling
Divide the population in sub-groups, call a group of
sub-groups as a Cluster. Identify a cluster based on objective of study and distribution of population. Examine cluster for intra-cluster homogeneity. Determine the stages single or multi-stage sampling.
Teena Y. Sharma
26
Area Sampling
It is a peculiar type of cluster sampling in which
samples are clustered together on the basis of geographical area basis
Teena Y. Sharma
27
Multiphase or Sequential Sampling
A multi-phase sample is such a sample which is
designed in such a way that some information is collected from the entire sample and the other information is collected only from a part of the sample.
Teena Y. Sharma
28
Stratified Random Sampling
If a population from which a sample is to be drawn
does not constitute a homogenous group , Stratified technique is generally applied to obtain a representative sample. Here a population is divided into several subpopulations that are individually more homogenous than total population, and then select samples from each stratum.
Teena Y. Sharma
29
Major issues in Stratified Sampling
How to form a strata? How many stratum should be formed? The sample size within each stratum. How to form a strata? It is determined according to the objective of the study or variable being studied. Number of stratum: If more than always beneficial But budget is also to be considered Hence, most cost-effective number should be thought of.
Teena Y. Sharma 30
Issues
Sample Size within each Stratum: Can be done in two ways:
Proportionate stratified random sampling: Here cases are drawn from each sample in same proportion as they appear in the original population. Disproportionate stratified random sampling: Depends on the variability, size, characteristics of the stratum.
Teena Y. Sharma
31
Non-probability Sampling
These are methods which are not based on the concept
of probability. These methods are also called non-random sampling techniques. Here there is no rule or formulae or method by which one can determine the chance or probability that a specific element is selected in a sample.
Teena Y. Sharma
32
Non-probability
Can be of following types
Non-probability sampling
Purposive/Judgmental Sampling
Convenience/ Accidental sampling
Quota Sampling
Snowballing
Teena Y. Sharma
33
Purposive / Judgmental Sampling
Elements are selected on a pre-determined criteria.
This criteria of decision is given by some experts. It is a kind of sampling where, special elements are put
into the sample.
These elements are special because, they possess some
knowledge, which is not possessed by any other element of the population.
Teena Y. Sharma
34
Convenience/ Accidental Sampling
Convenience of the researcher, convenient location for
field work is the prime criteria for selection.
Teena Y. Sharma
35
Quota Sampling
Most widely used non-probability sampling method.
It is a non-probabilistic version of stratified method. Population is divided into Stratum, and each strata has
a quota, reserved to give a sample in the total sample. Here clear understanding of the population is required. The parameters should also be known. Then the quotas can be decided.
Teena Y. Sharma
36
Snowballing
Here initial members of the sample lead the researcher
to newer sample constituents. The first element, gives you information about the next element.
Example: Suppose a survey is being conducted on the
problems faced by dentists who treat small children, the first dentist you identify can tell you about the other dentist who specializes for treatment of dental diseases in children.
Teena Y. Sharma 37
Teena Y. Sharma
38
