2. Objectives
TO UNDERSTAND
• The purpose and the benefits of sampling.
• The different stages and techniques of sampling.
• The importance of sample size and methods for
determining the sample size.
3. Introduction
¨
• The basic idea behind sampling is by studying a
subgroup of the population, the characteristics of the
population can be ascertained.
• The question of ‘how many’ is therefore concerned with
determining the size of the sample that would by and
large adequately identify the characteristic under
study.
4. Why Sampling?
• Major benefits derived from sampling:
• Reduced costs
• Reduced time
• Greater accuracy
• Greater flexibility of scope
5. Sampling Stages
1. Defining the population: involves specifying
• the purpose of the study
• all alternative populations that may be considered
• the sampling unit
• the area or location
• time
• convenience of selection
6. Sampling Stages
2. Identifying the Sampling Frame: This is a detailed
record or list of members of the population under
study.
3. Defining the sampling unit
4. Identifying the sampling element,i.e., the actual
respondent from the sampling unit
…cont.
7. Sampling Stages
5. The sampling method: choice between probability
and non-probability based procedures depends on
• the accuracy desired
• the time available for data collection
• the costs involved, and
• the extent of control sought over selection bias
• the purpose of research
6. The sample-size
7. Sampling plan
…cont.
8. Sampling Techniques
Non-probability sampling : here the sample selection is based
on convenience, time, cost, and the researcher’s perception
of the respondent’s knowledge of the subject under study.
The choice between probability and non-probability sampling
procedures depends on :
o Kind of information needed
O Error tolerance accepted
o Likely size of non-sampling errors
o Homogeneity of the population
o Cost of errors
…cont.
9. Sampling Techniques
Probability sampling : method of sample selection
where each unit in the population has a definite, non-
zero chance of selection in the sample, following some
objective statistical rule.
Advantages: possible to
1. determine the size of the sampling error
2. estimate the population parameter,
and 3. deliver a relatively small selection-bias
10. Sampling Techniques
Methods of Sample -Size Determination
• Ad-Hoc Methods:
(1) Rule of Thumb: 1% of the total population
(2) Budget-constraints: A sample-size that can be
accommodated within the given budget
(3) Time –constraints: A sample-size that can be
achieved within the given time
(4) Comparable Studies: Sample –size similar to that
chosen for comparable studies
…cont.
11. Sampling Techniques
Specific Sampling Techniques
SAMPLING METHODS
PROBABILITY NON- PROBABILITY
RANDOM CONVENIENCE
SYSTEMATIC SNOWBALL
STRATIFIED QUOTA
CLUSTER JUDGMENTAL
…cont.
12. Sampling Techniques
Determining the Sample Size
For sampling to be meaningful, it is essential to choose
a sample that is large enough to approximate all the
variation likely in the population, and yet small enough
to provide significant savings in cost and time.
Factors influencing sample –size determination:
• The number of groups and subgroups within the sample
that will need to be studied separately.
• Value of the information required
• Costs involved
• Extent of variation in the population data
…cont.
13. Sampling Techniques
1.2. The hypothesis-testing approach suggests
1. Specify the values of µ under hypotheses H0 and H1
2. Specify the acceptable probabilities of the two types of
error α and β.
3. Determine the standard errors corresponding to these
probabilities
4. Provide some estimate of the population standard
deviation
5. Calculate the sample size ‘n’ that meets the α and β
error requirements by defining the critical values as
(µ0+Zα/√n) =C.V= ( µ1 -Z β / √n ) , solving for
( Z α + Z β)2 2
n = ------------
(µ1 -µ0)2
…cont.
14. Sampling Techniques
• Traditional Methods:
(1) Neyman –Pearson Approach:
1.1. The confidence-interval method suggests
Sample-size ‘n’ = (x z/e)2
Where x = estimate of variation in the population,
z = value of the normal variable for a given
confidence –interval
e = level of sampling –error acceptable
…cont.
15. Sampling Techniques
(2) Bayesian Approach: Based on the principle of ‘expected
net monetary gain’ from research, by
• determining the expected value of information from
samples of various sizes
• estimating the cost of obtaining each of these samples
• calculating the net gain from information obtained
from each sample choosing the sample which provides the
largest gain
It is thus based on ‘cost-benefit analysis’.
…cont.