SADC Course in Statistics Estimation in Stratified Random Sampling (Session 07)
07 Sampling
-
Upload
stacy-dsouza -
Category
Documents
-
view
7 -
download
0
description
Transcript of 07 Sampling
02/10/2013
1
Zaki Rashidi
Sampling Techniques
Slide 7.2
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Selecting samples
Population, sample and individual cases
Source: Saunders et al. (2009)
Figure 7.1 Population, sample and individual cases
Slide 7.3
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
The need to sample
Sampling- a valid alternative to a census when
A survey of the entire population is impracticable
Budget constraints restrict data collection
Time constraints restrict data collection
Results from data collection are needed quickly
Slide 7.4
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Basics of sampling I
A sample is a
“part of a whole
to show what the
rest is like”.
Sampling helps to
determine the
corresponding
value of the
population and
plays a vital role
in esearch.
Samples offer many benefits:
Save costs: Less expensive to study the sample than the population.
Save time: Less time needed to study the sample than the population .
Accuracy: Since sampling is done with care and studies are conducted by skilled and qualified interviewers, the results are expected to be accurate.
Destructive nature of elements: For some elements, sampling is the way to test, since tests destroy the element itself.
Slide 7.5
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Basics of sampling II
Limitations of Sampling
Demands more rigid control in undertaking sample operation.
Minority and smallness in number of sub-groups often render study to be suspected.
Accuracy level may be affected when data is subjected to weighing.
Sample results are good approximations at best.
Sampling Process
Defining the
population
Developing
a sampling
Frame
Determining
Sample
Size
Specifying
Sample
Method
SELECTING THE SAMPLE
Slide 7.6
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Sampling Cycle
Population
Draw a Sample
Compute Statistics
Apply inference
Estimate Parameter
02/10/2013
2
Slide 7.7
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
The sampling frame
The sampling frame for any probability sample is
a complete list of all the cases / units in the
population from which your sample will be
drown.
What is the difference between population and
sampling frame?
Slide 7.8
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Sampling: Step 1
Defining the Universe
Universe or population is the
whole mass under study.
How to define a universe:
What constitutes the units of
analysis (HDB apartments)?
What are the sampling units
(HDB apartments occupied in
the last three months)?
What is the specific designation
of the units to be covered (HDB
in town area)?
What time period does the data
refer to (December 31, 1995)
Sampling: Step 2 Establishing the Sampling
Frame
A sample frame is the list of all elements in the population (such as telephone directories, electoral registers, club membership etc.) from which the samples are drawn.
A sample frame which does not fully represent an intended population will result in frame error and affect the degree of
reliability of sample result.
Slide 7.9
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Step - 3
Determination of Sample Size
Sample size may be determined by using:
Subjective methods (less sophisticated methods)
The rule of thumb approach: eg. 5% of population
Conventional approach: eg. Average of sample sizes of similar
other studies;
Cost basis approach: The number that can be studied with the
available funds;
Statistical formulae (more sophisticated methods)
Confidence interval approach.
Slide 7.10
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Choice of Sample Size - Large
Populations Sample Sizes
% Margin of Error 95% Confidence 99% Confidence
± 1 9,604 16,590
± 2 2,401 4,148
± 3 1,068 1,844
± 4 601 1,037
± 5 385 664
± 6 267 461
± 7 196 339
± 8 151 260
± 9 119 250
± 10 97 166
Source :Parker & Rea, Designing and Conducting Research
Table 1
Slide 7.11
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Choice of Sample Size - Small Populations
Sample Sizes
95% Level of Confidence 99% Level of Confidence
N ± 3% ± 5% ± 10% ± 3% ± 5% ± 10%
500 250 218 81 250 250 124
1000 500 278 88 500 399 143
1500 624 306 91 750 460 150
2,000 696 323 92 959 498 154
3,000 788 341 94 1,142 544 158
5,000 880 357 95 1,347 586 161
10,000 965 370 96 1,556 622 164
20,000 1,014 377 96 1,687 642 165
50,000 1,045 382 96 1,777 655 166
100,000 1,058 383 96 1,809 659 166
Source : Parker & Rea, Designing and Conducting Research
Table 2
Slide 7.12
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Sample size determination using statistical formulae:
The confidence interval approach
To determine sample sizes using statistical formulae, researchers
use the confidence interval approach based on the following
factors:
Desired level of data precision or accuracy;
Amount of variability in the population (homogeneity);
Level of confidence required in the estimates of population values.
Availability of resources such as money, manpower and time
may prompt the researcher to modify the computed sample size.
Students are encouraged to consult any standard marketing
research textbook to have an understanding of these formulae.
02/10/2013
3
Slide 7.13
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Probability sampling Probability of each case / unit being selected from
the population is known (and usually equal to all
cases).
This means that it is possible to answer research
questions and to achieve objectives that require you
to estimate statistically the characteristics of the
population from the sample.
Consequently, probability sampling is often
associated with survey and experimental research
strategies.
Slide 7.14
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Non probability samples The probability of each case being selected from the
total population is not known and it is impossible to
answer research questions or to address research
objectives that require you to make statistical
inferences about the characteristics of the
population.
You may still be able to generalize from non
probability samples about the population, but not on
statistical grounds
Slide 7.15
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Overview of sampling techniques
Sampling techniques
Source: Saunders et al. (2009) Figure 7.2 Sampling techniques
Probability Sampling
Slide 7.17
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Probability sampling
The four stage process
1. Identify sampling frame from research objectives
2. Decide on a suitable sample size
3. Select the appropriate technique and the sample
4. Check that the sample is representative
Slide 7.18
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Identifying a suitable sampling frame
Key points to consider
Problems of using existing databases
Extent of possible generalisation from the sample
Validity and reliability
Avoidance of bias
02/10/2013
4
Slide 7.19
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Sample size
Choice of sample size is influenced by
Confidence needed in the data
Margin of error that can be tolerated
Types of analyses to be undertaken
Size of the sample population and distribution
Slide 7.20
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
The importance of response rate
Key considerations
Non- respondents and analysis of refusals
Obtaining a representative sample
Calculating the active response rate
Estimating response rate and sample size
Slide 7.21
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Selecting a sampling technique
Five main techniques used for a probability sample
1. Simple random
2. Stratified random
3. Systematic
4. Cluster
5. Multi-stage
Slide 7.22
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Simple random(Random sampling) Involves you selecting at random frame using either random
number tables, a computer or an online random number
generator such as Research Randomizer.
Slide 7.23
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Stratified random sampling Stratified random sampling is a modification of random sampling
in which you divide the population into two or more relevant and
significant strata (groups) based on a one or a number of
attributes.
Sampling frame is divided into a number of subsets.
A random sample (simple or systematic) is then drawn from each
of the strata.
Consequently stratified sampling shares many of the advantages
and disadvantages of simple random or systematic sampling
Slide 7.24
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Three approaches
a) Proportional Allocation
b) Disproportional Allocation
c) Neyman’s Allocation
02/10/2013
5
Slide 7.25
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Systematic sampling
Systematic sampling involves you selecting the sample at regular
intervals from the sampling frame.
1. Number each of the cases in your sampling frame with a unique
number . The first is numbered 0, the second 1 and so on.
2. Select the first case using a random number.
3. Calculate the sample fraction.
4. Select subsequent cases systematically using the sample fraction to
determine the frequency of selection
Slide 7.26
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Cluster Sampling Similar to stratified as you need to divide the population into
discrete groups prior to sampling.
The groups are termed clusters in this form of sampling and
can be based in any naturally occurring grouping.
For example, you could group your data by type of
manufacturing firm or geographical area
Slide 7.27
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Cluster Sampling For cluster sampling your sampling frame is the complete list
of clusters rather than complete list of individual cases within
population, you then select a few cluster normally using
simple random sampling.
Data are then collected from every case within the selected
clusters.
What is the difference in the groups of
stratified sampling and cluster sampling?
Slide 7.28
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Multi-stage sampling
(multi-stage cluster sampling It is a development of cluster sampling
It is normally used to overcome problems associated with a
geographically dispersed population when face to face contact is
needed or where it is expensive and time consuming to construct
a sampling frame for a large geographical area.
However, like cluster sampling you can use it for any discrete
groups, including those not are geographically based.
The technique involves taking a series of cluster samples, each
involving some form of random sampling method.
Non Probability Sampling
Slide 7.30
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Non-Probability Sampling
1. Quota Sampling
2. Purposive Sampling
1. Extreme case Sampling
2. Heterogeneous /Maximum Variation
3. Homogeneous Sampling
4. Critical case Sampling
5. Typical case Sampling
3. Snowball Sampling
4. Self-selection Sampling
5. Convenience Sampling
02/10/2013
6
Slide 7.31
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Non- probability sampling
Key considerations
Deciding on a suitable sample size
Selecting the appropriate technique
Slide 7.32
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Quota sampling It is entirely non random and it is normally used for
interview surveys.
It is based on the premise that your sample will represent
the population as the variability in your sample for
various quota variables is the same as that in population.
Quota sampling is therefore a type of stratified sample in
which selection of cases within strata is entirely non-
random
Slide 7.33
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Quota sampling Divide the population into specific groups.
Calculate a quota for each group based on relevant and available
data.
Give each interviewer an ‘assignment', which states the number of
cases in each quota from which they must collect data.
Combine the data collected by interviewers to provide the full
sample.
Slide 7.34
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Purposive sampling
Purposive or judgmental sampling enables you to use your
judgment to select cases that will best enable you to answer your
research question(s) and to meet your objectives.
This form of sample is often used when working with very small
samples such as in case research and when you wish to select cases
that are particularly informative.
Purposive sampling can also be used by researchers adopting the
grounded theory strategy. For such research, findings from data
collected from your initial sample inform the way you extend your
sample into subsequent cases.
Such samples, however can not be considered to be statistically
representative of the total population.
Slide 7.35
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Continued The logic on which you base your strategy for selecting cases
for a purposive sample should be dependent on your research
question(s)and objectives.
Select information-rich cases in purposive sampling in
contrast to need to be statistically representative in
probability sampling.
Slide 7.36
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Extreme case or deviant sampling
Extreme case or deviant sampling focuses on unusual or
special cases
You will learn the most to answer your research
question(s) and to meet your objects more effectively.
02/10/2013
7
Slide 7.37
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Heterogeneous or maximum variation
sampling
Heterogeneous or maximum variation sampling enables
you to collect data to explain and describe the key
themes that can be observed.
To ensure maximum variation within a sample it is
suggested to identify diverse characteristics (sample
selection criteria) prior to selecting your sample.
Slide 7.38
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Homogenous Sampling In direct contrast to heterogeneous sampling , homogenous
sampling focuses on one particular sub-group in which all the
sample members are similar.
This enables you to study the group in great depth.
Slide 7.39
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Critical Case Sampling Critical case sampling selects critical cases on the bases that
they can make a point dramatically or because they are
important.
The focus of data collections to understand what is happening
in each critical case so that logical generalizations can be
made.
Slide 7.40
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Continued A number of clues that suggest critical cases can be
summarized by the questions such as:
If it happens there, will it happen everywhere?
If they are having problems, can you be sure that
everyone will have problems?
If they cannot understand the process, is it likely that no
one will be able to understand the process?
Slide 7.41
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Typical case sampling In contrast of critical case sampling, typical case sampling is
usually used as a part of a research project to provide an
illustrative profile using a representative case.
Such a sample enables you to provide an illustration of what is
‘typical’ to those who will be reading your research report and
may be unfamiliar with the subject matter.
Slide 7.42
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Snowball sampling Is commonly used when it is difficult to identify members of
desired population. For example people who are working while
claiming unemployment benefit you therefore, need to:
1. Make contact with one or two cases in the population.
2. Ask these cases to identify further cases.
3. Ask theses new cases to identify further new cases (and so on)
4. Stop when either no new cases are given or the sample is as
large as manageable
02/10/2013
8
Slide 7.43
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Self selecting sampling It occurs when you allow each case usually individuals, to
identify their desire to take part in the research you therefore
1. Publicize your need for cases, either by advertising through
appropriate media or by asking them to take part.
2. Collect data from those who respond
Slide 7.44
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Convenience sampling Convenience sampling (or haphazard sampling) involves selecting
haphazardly those cases that are easiest to obtain for your sample,
such as the person interviewed at random in a shopping centre for a
television programme or the book about entrepreneurship you find
at the airport.
The sample selection process is continued until your required
sample size has been reached.
Although this technique of sampling is used widely, it is prone to
bias and influences that are beyond your control, as the cases appear
in the sample only because of the ease of obtaining them.
Slide 7.45
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Choosing probability vs. non-probability sampling
Probability Evaluation Criteria Non-probability sampling sampling
Conclusive Nature of research Exploratory
Larger sampling Relative magnitude Larger non-sampling
errors sampling vs. error
non-sampling error
High Population variability Low
[Heterogeneous] [Homogeneous]
Favorable Statistical Considerations Unfavorable
High Sophistication Needed Low
Relatively Longer Time Relatively shorter
High Budget Needed Low
Slide 7.46
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Sampling vs non-sampling errors
Sampling Error [SE] Non-sampling Error [NSE]
Very small sample Size
Larger sample size
Still larger sample
Complete census
Slide 7.47
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009