Sampling Theory
Introduction
All items in any field of Enquiry constitutes Universe or
population.
A complete enumeration of all items in the ‘population’
is known as a census inquiry.
If the universe(population) is large and observations are
more census is Impractical.
When field studies are undertaken in practical life,
considerations of time and cost almost invariably lead
to a selection of respondents i.e., selection of only a few
items.
The respondents selected should be as representative of
the total population as possible in order to produce a
miniature cross-section.
The selected respondents constitute what is
technically called a ‘sample’ and the selection process
is called ‘sampling technique
The survey so conducted is known as ‘sample survey’.
Algebraically, let the population size be N and if a part
of size n (which is < N) of this population is selected
according to some rule for studying some
characteristic of the population.
The group consisting of these n units is known as
‘sample’.
Researcher must prepare a sample design for his
study i.e., he must plan how a sample should be
selected and of what size such a sample would be.
Steps in Sample Design
Type of universe
Sampling unit
Source list
Size of sample
Parameters of interest
Budgetary constraint
Sampling procedure
CHARACTERISTICS OF A GOOD SAMPLE DESIGN
Sample design must result in a truly
representative sample.
Sample design must be such which results in
a small sampling error.
Sample design must be viable in the context
of funds available for the research study.
Sample design must be such so that
systematic bias can be controlled in a better
way.
Sample should be such that the results of the
sample study can be applied, in general, for
the universe with a reasonable level of
confidence.
DIFFERENT TYPES OF SAMPLE DESIGNS
Let us consider Mysore city council wants to
know the responses of the people regarding
the withdrawal of subsidy of LPG
The council decides to send questionnaire to
each Individual in Mysore and 5% response
rate is expected,
Non-Probability Samples
Convenience
Sample selected simply for ease
Example – Mysore city Council go to KR Circle and
hand questionnaires to everyone they encounter
until they have 1000 complete.
Quick and cheap
Non-Probability Samples
Example 2:Want to know views on bringing Metro to Mysore.
Go to Railway St at 6:30 PM on Friday afternoon and ask 1000
people for their views
Go to suburban Bus stand at 6:30 on Friday afternoon and ask
1000 people for their views
Bias is a major problem – sample unlikely to be
representative
Non-Probability Samples
Quota Sampling
Subjective choice of sample based on
what researcher thinks is representative.
Example – Mysore council sends 3
researchers knocking on doors in
Nanjangud, Mysore city, Hunsur until they
have responses
Age broken
Men down
Women as:
10 to 20 120 100
20 to 30 100 100
30 to 40 55 50
40 to 50 70 65
50 to 60 80 90
60+ 60 110
Quota Sampling
Quicker and cheaper than probability
sampling
Large non response
Still may not be representative
What do base your quota on – age, gender,
ethnicity, education, tenure, religion?
Probability Sampling
Example – 1000 households randomly selected
across Mysuru
More expensive and slower
Non response a problem – but resources can be
targeted and extent of non response bias can be
estimated.
Enables precision of final statistics to be
assessed.
Sample selection method is objective, specified
and replicable.
Sampling Frames
List of all units/people that could be included in
sample
Sample is only as good as the sampling frame
o Eligible units/people not on frame cannot be
selected – leads to coverage error
o Units/people on frame more than once changes
probability of being selected
o Ineligible units/people on frame can lead to
final sample being smaller than intended.
Sampling Frames
Frames can be created:
From pre-existing lists (UK postcode address
file)
Geographically with multi-stages
Through time
Simple random sampling
Sampling method completely random based on
random numbers.
Is easy to understand, can be expensive
Example – every household in Mysuru assigned a
number. 1000 random numbers chosen between
1 and 204,683. These numbers identify which
households are in sample.
Tables of random numbers
[Link]
Excel function ‘=rand()’
Systematic Sampling
Uses a ‘random’ start on the sampling frame and
then selects every i’th unit/person.
Easy to understand, quick and easy to implement.
Can lead to some stratification depending on how
the list is ordered.
Can be expensive
Need to be careful on how list is ordered to avoid
bias.
Systematic Sampling
Sampling Frame Example
1 16 31 46
2 17 32 47 Total units on sampling frame =
3
4
18
19
33
34
48
49
60
5 20 35 50
6 21 36 51
7 22 37 52 Want sample of 10
8 23 38 53
9 24 39 54
10 25 40 55
11 26 41 56
Interval size is 60/10 = 6
12 27 42 57
13 28 43 58
14 29 44 59 Select random start between 1
15 30 45 60
and 6
Select every sixth unit
Stratified Sampling
Units/people are aggregated into subgroups
called strata. A certain number of units are
sampled from each stratum.
United Kingdom
(60m)
England (51m)Northern Ireland Scotland (5m) Wales (3m)
(2m)
Guards against unusual samples
Stratification information has to be available
Stratified Sampling
Proportionate Stratification
Chance of inclusion in sample is same for
all units/people regardless of strata
Population Sample Size Sampling Fraction
United Kingdom 61 000 000 6 100 0.01%
England 51 000 000 5 100 0.01%
Northern Ireland 2 000 000 200 0.01%
Scotland 5 000 000 500 0.01%
Wales 3 000 000 300 0.01%
Stratified Sampling
Disproportionate Stratification
Chance of a unit/person being included in the
sample depends on the strata they are in.
Often used to target small sub groups to help
analysis
Population Sample Size Sampling Fraction
United Kingdom 61 000 000 6 100 0.01%
England 51 000 000 3 100 0.01%
Northern Ireland 2 000 000 1 000 0.05%
Scotland 5 000 000 1 000 0.02%
Wales 3 000 000 1 000 0.03%
Cluster Sampling
What if there are a small number of units
across a large area?
Divide population into clusters along
geographic boundaries (e.g. wards)
Randomly sample clusters
Measure all units within selected clusters
Reduces costs for face to face interviews
But bias can be a problem if what you
want to measure depends of geographic
location.
Cluster Sampling
Example – Select 100 random postcodes
within Edinburgh and interview all
households within the 100 postcodes
Reduces costs for face to face interviews
But bias can be a problem if what you
want to measure depends of geographic
location.
Multistage Sampling
Larger units are randomly selected
Smaller units within the larger unit are then
randomly selected
Examples – Mysuru council randomly selects
40 Areas and then selects 10 households
within each Area
Multistage Sampling
Can be useful when no sampling frame is
available
Reduces costs for face to face interviews
Similar bias problems to clustering