SAMPLING & SAMPLING DISTRIBUTION

WAYS OF OBTAINING INFORMATION

 Complete enumeration survey (census)

 Sample survey

ADVANTAGES OF SAMPLE SURVEY

• Reduced cost and enlarged scope

• Organization of work
• Greater accuracy
• Urgent information required
• Feasibility

STAGES OF SAMPLE SURVEY

• Objective of the survey
• Defined the target population
• Data to be collected / select a sampling frame
• Determine types of sampling chosen
• Plan procedure for sampling units
• Determine sample size
• Selection of sample
• The Pre-test
• Organization of the field work

ADVANTAGES OF SAMPLING
• Reduced cost and enlarged scope
• Organization of work
• Greater accuracy
• Urgent information required
• Feasibility

SAMPLING BREAK DOWN

Who do you want to generalize it? The Theoretical Population

What population can you access to? The Study Population
How can you get access to them? The Sampling Frame
Who is in your study? The Sample

1|P a g e
BASIC CONCEPTS:

• Population: Any complete group of entities that share some common set of characteristics.
Collection of all the elements of interest.

• Sample: Subset of the population.

• Sampling: The technique of selecting a sample from population.

• Sample Frame: The list of all the units from which a sample may be drawn is called the
sampling frame or working population. All the sampling units in the sampling frame have
identification particulars.

• Sampling unit: An element or a group of elements on which the observations can be taken is
called a sampling unit.

• Sampling Error: The error occurs at sampling

• Non Sampling Error: Procedural error which is common in both sampling & census

• Parameter: Population Characteristics

• Statistic: Sample Characteristics

Sampling process may encounter the problem of systematic errors and sampling biases.
Systematic errors can be defined as incorrect or false representation of the sample

SOURCES OF THE SAMPLING ERRORS ARE:

•
Sample size
•
Faulty selection of sampling technique
•
Substitution
•
Wrong choice of statistic

SOURCES OF THE NON-SAMPLING ERRORS ARE:

• The data specification may be inadequate and inconsistent with the objectives of the
survey or census.

2|P a g e
 • Due to imprecise definition of the boundaries of area units, incomplete or wrong
identification of units,
• Faulty methods of enumeration etc., the data may be duplicated or may be omitted.
• The methods of interview and observation collection may be inaccurate or inappropriate.
• The questionnaire, definitions and instructions may be ambiguous.
• The investigators may be inexperienced or not trained properly.
• The recall errors may pose difficulty in reporting the true data.
• The scrutiny of data is not adequate.
• The coding, tabulation etc. of the data may be erroneous.
• There can be errors in presenting and printing the tabulated results, graphs etc.

METHODS OF SAMPLING

• PROBABILITY
• NON PROBABILITY

PROBABILITY SAMPLING:

The selection of units in the sample from a population is governed by the laws of chance or
probability. The probability of selection of a unit can be equal as well as unequal.

Advantages:
This sampling technique reduces the chance of systematic errors & the chance of sampling
biases.
A better representative sample is produced using probability sampling techniques.
Inferences drawn from sample are generalizable to the population.
Disadvantages:
The techniques need a lot of efforts, lot of time & too expensive.

NON PROBABILITY SAMPLING:

The selection of units in the sample from population is not governed by the probability laws.

Advantages:
The techniques need less effort, need less time to finish up and are not much costly.

Disadvantages:
The sampling techniques are prone to encounter with systematic errors and sampling biases.
The sample cannot be claimed to be a good representative of the population.
Inferences drawn from sample are not generalizable to the population.

3|P a g e
PROBABILITY SAMPLING TYPES:
SIMPLE RANDOM, STRATIFIED RANDOM, SYSTEMATIC, CLUSTER & MULTISTAGE SAMPLING

PROBABILITY SAMPLING TYPES:

CONVENIENCE, PURPOSIVE, JUDGMENTAL, QUOTA & SNOWBALL SAMPLING

SIMPLE RANDOM SAMPLING

In this method each and every element of the population has an equal chance of being selected in
the sample.
Benefits:
• No sampling biases
Draw Backs:
• In many circumstances it is not possible to get or prepare an exhaustive list of elements.
• Sample sizes may need to be large to be representative
• It may be very costly and time consuming
Example: The Manager of a Company wants to know the satisfaction level of employees in
regards to performance Management Systems.

STRATIFIED RANDOM SAMPLING

Used when population is heterogeneous. So the sub groups/Strata are formed that are
homogenous. Samples are selected on the basis of a representative strata or segment
Benefits:
• Still random but more focussed which may give more relevant information
• May be more cost effective
Draw Backs:
• Needs a lot of efforts, costly and time consuming
• If the criterion characteristic/ variable used for classification is not selected correctly, the
whole research may go in vain.
Example:
The retail chain of India wants to a study employee’s satisfaction

SYSTEMATIC SAMPLING
This type of sampling is also used for homogenous population like SRS but not giving equal
probability to every element been included. In this sampling the first element is chosen at
random and the subsequent elements are selected at a regular interval
Benefits:
• It provides the way to get a random and representative sample in the situation where prior
listing up of elements is not possible.
Draw Backs:
• It may be very costly and time consuming

4|P a g e
 • It needs a lot of efforts especially for a large population.
• If the order of the list is biased in some way, systematic error may occur.
Example:
When list of elements cannot be possible: Promotion campaign of any shopping mall
When list of elements is possible: Satisfaction of employees

CLUSTER SAMPLING
The group of elements residing in one geographical region is called as cluster & sampling of
clusters is called as cluster sampling. This sampling technique is used when the elements of
population are spread over a wide geographical area.
Benefits:
• Less cost, less time and efforts as compare to other random sampling
Draw Backs:
• It may sometimes lead to sampling biases and systematic errors.
• If clusters are not homogeneous among them, the final sample may not be representative
of the population.
Example:
Ministry of Higher Education wants to inspect quality of education in Universities

MULTISTAGE SAMPLING
It is a sampling technique where two or more probability techniques are combined. It is used
when the elements of population are spread over a wide geographical region and it is not possible
to obtain a representative sample with only one aforementioned technique.
Benefits:
• It increases cost and time efficacy. Also useful in overcoming the heterogeneity problem
within the clusters.
Draw Backs:
• If the selected clusters do not capture the characteristic diversity of population, the
sample would not be representative of the population.
• If the characteristic variable used for making strata (in case of heterogeneity) at any stage
is not appropriately selected depending on the nature of investigation, the whole research
may go in
Example:
The purpose of a research is to find out the best seller brands

CONVENIENCE SAMPLING:
Samples are selected at the convenience of the investigator
E.g., Friends, family, neighbours, acquaintances, Students in a class or co-workers in a
workplace

5|P a g e
A researcher wants to investigate how men and women are different in their preference (Own
Organizations)

PURPOSIVE SAMPLING:
Samples are selected from specific target segments who will be able to provide information.
E.g., the purpose of a research is to investigate which type of clothing middle age women prefer

JUDGMENTAL SAMPLING:
Samples are selected from the respondent who is in the best position to provide the desired
information.

QUOTA SAMPLING:
This type of sampling method is used when population is heterogeneous i.e. every element of
population does not matches all the characteristics of the predefined criteria. Instead the elements
differ from one another on a characteristic.
It is a form of proportionate stratified sampling in which a predetermined proportion of elements
are sampled from different groups in the population, but on convenient basis

SNOWBALL SAMPLING (REFERRAL/ CHAIN SAMPLING):

Samples developed from contacts of existing customers –‘word of mouth’ type approach!

NOTE:

1. Stratified random sampling is used when population is heterogeneous while in cluster

sampling the population is homogenous. In stratified random sampling the sub groups are
made on the basis of variable that makes the population heterogeneous; while in cluster
sampling the grouping is done on the basis of geographical allocation of the population. Both
stratified sampling and cluster sampling involve two major steps: first groups are made and
then selection of the sample is randomly done. However the sample is selected only for once
in both the techniques while in multistage sampling, the sample is selected at least twice
using different types of sampling techniques at each stage

2. Both the Stratified Sampling and Quota Sampling techniques are used when the target
population is heterogeneous. In stratified random sampling, participants are selected
randomly while in quota sampling, participants are selected non-randomly.

3. If target population is homogenous and an exhaustive list of elements id available or possible

to prepare, simple random sampling is appropriate.

6|P a g e
4. If target population is homogenous and list of elements cannot be produced for any reason,
systematic sampling is appropriate.

5. If target population is heterogeneous and is not very much widely spread geographically,
stratified random sampling is used

6. If your target population is homogenous but is spread in a wide geographical area, cluster
sampling is used.

7. If you target population is spread over a wide geographical area and is also heterogeneous,
multistage sampling is preferable.

8. If shortage of time and/or efforts cannot be utilized in search of participants, volunteer

sampling is most feasible.

9. If target population is commonly found everywhere and not precisely defined, convenient
sampling is used.

10. If target population is heterogeneous and it is important to take representation from every sub
group, quota sampling is used

11. If target population is the type of people who are not easily available, lesser in number and
rare to find, snowball sampling technique is used.

DETERMINATION OF SAMPLE SIZE SELECTION

FACTORS IN CHOOSING A SAMPLE

1. Variance (or Heterogeneity)

A heterogeneous population has more variance (a larger standard deviation) which will require a
larger sample.
A homogeneous population has less variance (a smaller standard deviation) which permits a
smaller sample.
2. Magnitude of Error (Confidence Interval): 10%, 5%, 1%.....
How precise must the estimate be?
3. Confidence Level – 95%, 99%....
How much error will be tolerated?

FOR SAMPLE MEAN

1. Specify the amount of error that can be allowed

7|P a g e
 2. Specify the desired level of confidence
3. Determine Z at desired confidence level
4. Estimate the standard deviation of the population
5. Calculate the sample size considering finite or infinite population

FOR SAMPLE PROPORTION

1. Specify the amount of error that can be allowed
2. Specify the desired level of confidence
3. Determine Z at desired confidence level
4. Estimate the population proportion
5. Calculate the sample size considering finite or infinite population

FORMULAE FOR SAMPLE SIZE ESTIMATION

Z 2 2
n For infinite population
e2

Z 2 2 N
n
( N  1)e 2  Z 2 2 For finite population

Z 2 p(1  p)
n For infinite population
e2

Z 2 p(1  p) N
n For finite population
( N  1)e 2  Z 2 p(1  p)

CRITICAL VALUES OF Z

8|P a g e
ESTIMATING SAMPLE SIZE

Factor analysis
10 items a sample, 1:2, size of 200 is required

Multiple regression analyses

15 to 20 observations for each predictor
If there are N age categories in your analysis this translates to (N-1) parameters. For
example to find the predictors of dependent variable with predictors on independent
variable of 4 categories other of 2 categories
n = ((4 + 2)-1) x 20 = 100 participants required in the study.

Green (1991): n > 50 + 8p where p is the number of predictors

Cluster Analysis
Large sample

Estimating Sample Size at different confidence level

Prepared by Dr. Manas Kumar Pal

9|P a g e