Slide 7.

CHAPTER 7
SELECTING SAMPLES

1
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.2

7.1 INTRODUCTION
Population, sample and individual cases

Source: Saunders et al. (2009)

Figure 7.1 Population, sample and individual cases

2
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.3

7.1 INTRODUCTION
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

3
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.4

7.1 INTRODUCTION
OVERVIEW OF SAMPLING TECHNIQUES
Sampling techniques

Source: Saunders et al. (2009)

Figure 7.2 Sampling techniques
4
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.5

7.1 INTRODUCTION
Non-Probability Sampling
Probability Sampling
(Non-Random Sampling/
(Representative Sampling)
Judgemental Sampling)

Random, each unit has an equal Non-random, based on judgment or

probability of being selected convenience

Generalizability is higher because the Generalizability is lower because the

sample represents the entire population sample does not necessarily represent
the entire population
Results can be generalized, high Easier to conduct, less costly, suitable
accuracy and reliability when the entire population cannot be
determined
Often more costly and complex to Results may be biased and not
implement representative of the population

Simple Random, Systematic, Stratified Quota, Snowball, Convenience,

Random, Cluster, Multi-stage Purposive, Self-selection
Figure 7.2 Sampling techniques
5
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.6

7.1 INTRODUCTION

Exploratory Studies Descriptive Studies Explanatory

Studies

Probability Not commonly Frequently used to Mainly used to

Sampling
used as ensure representative ensure reliability
representative samples and and
samples are not generalizable results generalizability of
needed results
Non-Proba Frequently used, Can be used but less Rarely used as it
bility
Sampling such as common does not ensure
Convenience representative
Sampling samples

Figure 7.2 Sampling techniques

6
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.7

7.2 PROBABILITY SAMPLING

The four stage process
Stage Description

1. Identify sampling - Identify the target group for the research.

frame from research
- Ensure the sampling frame includes all units
objectives
necessary to answer the research questions.
2. Decide on a - Identify the number of samples needed to ensure
suitable sample size
the accuracy and reliability of the results.
- Consider factors such as resources, time, and cost.
3. Select the - Select a suitable sampling method (random,
appropriate technique
non-random).
and the sample
- Conduct sampling according to the chosen
technique to ensure representativeness.
4. Check that the - Assess whether the selected sample adequately
sample is
reflects the characteristics of the population.
representative
- Adjust if necessary to ensure the sample's
representativeness. 7
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.8

7.2 PROBABILITY SAMPLING

S1: IDENTIFYING A SUITABLE SAMPLING FRAME - Key points to consider
Factor Explanation Example

Problems of Databases can be A researcher uses the company's customer

using existing incomplete, inaccurate, or database, but it is not updated with the
databases outdated. latest customer information, leading to
inaccurate results on customer satisfaction.

Extent of Ensuring that the sample A study on online shopping habits only
possible accurately represents the uses a sample from major cities like Hanoi,
generalization population to generalize the but the results are applied to the entire
from the findings. population of Vietnam, including rural
sample areas.

Validity and Data must be accurate and A survey on employee satisfaction uses the
reliability consistently measured to be same set of questions and methods in
trusted. multiple surveys to ensure results
accurately reflect reality and are
comparable.

Avoidance of Ensuring data collection and When conducting a customer satisfaction

bias analysis are free from survey, the researcher uses random
personal or procedural sampling to select participants from the
biases. customer list, instead of choosing
8
acquaintances or easily accessible people.
th
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5 Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.9

7.2 PROBABILITY SAMPLING

S1: IDENTIFYING A SUITABLE SAMPLING FRAME

9
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.10

7.2 PROBABILITY SAMPLING

S2: DECIDING ON A SUITABLE SAMPLE SIZE

(1) General Principles: Generalizations from probability samples are based on

statistical probability. Larger samples reduce error when generalizing to the population.
(2) Considerations for Sample Size:
● Confidence Level: The certainty needed that the data characteristics represent the
population.
● Margin of Error: The accuracy required for estimates made from the sample.
● Types of Analysis: The number of data required for analysis, as many statistical
techniques have minimum data thresholds (e.g. chi square, Section 12.5).
● Population Size: The size of the total population from which the sample is drawn.
(3) Judgement in Sample Size: Final sample size is a balance of these
considerations, influenced by statistical analysis requirements and the need for normally
distributed data (normal distribution, Chapter 12).
(4) Statistical Thresholds: Stutely’s (2003) advice of a minimum number of 30 for
statistical analyses provides a useful rule of thumb for the smallest number in each
category within your overall sample. 10
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.11

7.2 PROBABILITY SAMPLING

S2: DECIDING ON A SUITABLE SAMPLE SIZE

11
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.12

7.2 PROBABILITY SAMPLING

S2: DECIDING ON A SUITABLE SAMPLE SIZE

Population = 3000, Margin of error = 5%, Sample size = ?

Applying Linear Interpolation Formula

12
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.13

7.2 PROBABILITY SAMPLING

Estimating response rates and actual sample size required

Janice was a part-time student employed by a large manufacturing company.

She had decided to send a questionnaire to the company’s customers and
calculated that an adjusted minimum sample size of 439 was required. Janice
estimated the response rate would be 30 per cent. From this, she could
calculate her actual sample size; Actual sample size?

13
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.14

7.2 PROBABILITY SAMPLING

The importance of a high response rate
Factor Explanation Example

Non-respon Non-respondents are those who do not A customer satisfaction survey might face
dents and participate in the survey, leading to issues when some customers refuse to
analysis of potential bias in the data. Analyzing participate. Analyzing these refusals might
refusals refusals helps identify the reason for show that they are dissatisfied with the data
non-response and address any bias. collection method, prompting a change in
approach.

Obtaining a Ensuring that the sample accurately When studying customer shopping habits, it is
representati represents the population to make necessary to select a sample that represents all
ve sample valid generalizations. ages, genders, and income levels so that the
results can be generalized to the entire
population.

Calculating The active response rate is calculated In a telephone survey, out of 500 people called,
the active by excluding ineligible and only 300 had valid and reachable phone
response unreachable respondents from the total numbers. Among those, 150 responded,
rate sample size. creating an active response rate of 150/300 =
50%.

Estimating Estimating the likely response rate A researcher anticipates a response rate of
response helps determine the necessary sample 30% from 1000 people invited to participate in a
rate and size to achieve reliable results. survey. Therefore, they need to send out 1000
sample size surveys to achieve at least 300 responses,
ensuring reliable results. 14
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.15

7.2 PROBABILITY SAMPLING

S3: SELECTING A SAMPLING TECHNIQUE
Technique Description

Probability Simple Each unit in the population has an equal chance of being
Sampling: Random
each
selected.
element in
Systematic Samples are selected using a fixed system, e.g., every 10th
the
population person is chosen.
has an equal
chance of Stratified The population is divided into groups (or 'strata') and
being Random
samples are randomly taken from each group.
selected.
Cluster The population is divided into groups (clusters), and a
random selection of these clusters is chosen for data
collection.
Multi-stage Combines several probability sampling methods.

15
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.16

7.2 PROBABILITY SAMPLING

S3: SELECTING A SAMPLING TECHNIQUE

Aspect Stratified Sampling Cluster Sampling

Key Points Population divided into subgroups (strata) Population divided into clusters,
based on characteristics, random samples random selection of entire clusters.
drawn from each subgroup.

Grouping Basis Based on specific characteristics (e.g., Based on natural groups (e.g.,
age, gender). geographic location, schools).

Sample Ensures representation of all subgroups. Ensures representation of selected

Representation clusters only.

Data Requires detailed population information. Requires knowledge of cluster

Requirements boundaries.

Bias Potential Low within each stratum, higher if strata Higher if clusters are not
not well-defined. representative of the population.

Example Sampling students from different grades. Sampling schools and surveying all
students in selected schools.

Stratified Sampling vs. Cluster Sampling 16

Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.17

7.2 PROBABILITY SAMPLING

S3: SELECTING A SAMPLING TECHNIQUE

Aspect Simple Random Sampling Systematic Sampling

Key Each element in the population Randomly select a starting point,

Points has an equal chance of being then select every k-th element from
selected. the population list.

Implemen Purely random selection. Selects elements at regular intervals

tation (every k-th element).

Bias Low if truly random. Can be biased if there's an

Potential underlying pattern in the population.

Ease of Can be time-consuming for large Simpler to implement, especially for

Use populations. ordered lists.

Example Drawing names from a hat. Selecting every 10th person from a
list.

Simple Random Sampling vs. Systematic Sampling 17

Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.18

18
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.19

19
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.20

7.2 PROBABILITY SAMPLING

S4: CHECKING REPRESENTATIVENESS
▪ Compare with the population: Compare the key characteristics of the sample with
those of the target population to ensure representativeness.
▪ Distribution check: Ensure that the distribution of important variables in the sample
reflects the distribution of those variables in the population.
▪ Use statistical measures: Use statistical measures such as chi-square tests to
compare the distribution of the sample and the population.
▪ Consider biases: Identify and adjust any biases in the sample to improve
representativeness.
▪ Evaluate response rate: Check the response rate to ensure that the responding
sample is large enough and representative of the entire population.
▪ Non-response analysis: Analyze non-response cases to determine if there are any
differences between respondents and non-respondents.
▪ Use mixed methods: Use mixed sampling methods, such as stratified or cluster
sampling, to enhance the representativeness of the sample.

20
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.21

7.3 NON - PROBABILITY SAMPLING

Technique Description Example

Quota
A sample that reflects certain Surveying a specific number of male
characteristics of the and female customers in a shopping
population but is not random. mall to ensure gender balance.
Snowball
The sample grows via Starting with a few known participants
referrals from initial in a niche market and asking them to
participants. refer others for a marketing study.
Convenience
Sample is chosen based on Interviewing customers at a local cafe
convenience, not randomly. because it is easily accessible.
Purposive
Sample is selected based on Selecting expert employees in a tech
the research purpose and company to study innovation practices.
the researcher’s
understanding of the
population.
Self-selection
Individuals volunteer to Posting an online survey link and
participate in the study. allowing anyone interested to
21
participate.
Saunders, Lewis and Thornhill, Research Methods for th
Business Students, 5 Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.22

7.3 NON - PROBABILITY SAMPLING

Selecting the most appropriate sampling technique and the sample

Control Over
Likelihood of Sample Being
Sample Type Types of Research in Which Useful Relative Costs Sample
Representative
Contents

Quota Reasonable to high, although Where costs constrained or data Moderately Relatively
dependent on selection of quota needed very quickly, so an high to high
variables alternative to probability sampling reasonable
needed

Purposive Low, although dependent on Where working with very small Reasonable Reasona
researcher’s choices: samples ble
extreme case focus: unusual or special
heterogeneous focus: key themes
homogeneous focus: in-depth
critical case focus: importance of case
typical case focus: illustrative

Snowball Low, but cases will have Where difficulties in identifying Reasonable Quite low
characteristics desired cases

Self-selection Low, but cases self-selected Where exploratory research Low Low
needed

Convenience Very low Where very little variation in Low Low

population
22
Sources:
Saunders, Lewis developed
and Thornhill, from
Research Kervin
Methods (1999);
for Business Patton
Students, 5th (2002).
Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.23

7.3 NON - PROBABILITY SAMPLING

Technique Description Example

Quota + Use quota sampling to ensure representation Use quota sampling to select samples by
Convenience
of certain characteristics, then use gender, then choose the most accessible
convenience sampling within each quota. individuals within each gender group.

Snowball + Start with purposive sampling to select Select a few experts in a field using
Purposive
participants meeting specific criteria, then use purposive sampling, then ask them to refer
snowball sampling to have initial participants other experts.
refer others.

Self-Selection + Allow participants to self-select by responding Post an advertisement on social media for
Convenience
to an advertisement or call for participation, participants to self-select, then choose those
then use convenience sampling among those who respond quickly or are most accessible.
self-selected.

Purposive + Use purposive sampling to identify participants Select individuals with experience in a
Quota
with specific characteristics or experiences, specific field using purposive sampling, then
then apply quota sampling to ensure balance ensure representation of each age group
among subgroups. using quota sampling.

Convenience + Start with convenience sampling to gather Select participants from a specific event or
Snowball
easily accessible participants, then use location using convenience sampling, then
snowball sampling to expand the sample by ask them to refer others who also attended
23
asking initial participants to refer others. the event or location.
th
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5 Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.24

7.3 NON - PROBABILITY SAMPLING

Deciding on a suitable sample size
▪ Research questions and statistical analyses determine
sample size thresholds.
▪ Ensuring the sample distribution is normal is crucial for
robust results.
▪ A sample size of at least 30 is a useful rule of thumb for
many statistical analyses.
▪ For populations with fewer than 30 individuals, data
should be collected from all cases in the category.

24
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.25

Stratified Sampling Quota Sampling

Definition Divides the population into different strata Divides the population into groups and selects samples
(groups) and randomly selects samples based on fixed quotas for each group, without
from each stratum. randomness.

Process 1. Identify strata in the population. 1. Identify groups in the population.

2. Randomly select samples from each 2. Select samples from each group until the quota is
stratum. met, without randomness.

Purpose Ensures that each stratum is represented Ensures that important groups are represented
in the sample, increasing the accuracy of proportionally, but does not require randomness.
the results.

Randomness Involves randomness in selecting samples Does not involve randomness, based on criteria and
from strata. convenience.

Representativene High, as each stratum has a chance to be Medium, depending on the selection of quota variables
ss
sampled. and the accuracy in tracking quotas.

Examples Randomly selecting samples from different Selecting samples from different age groups until a fixed
age groups within an organization. number of samples for each group is reached.

Advantages Ensures high representativeness, reliable Easy to implement, saves time and costs.
and accurate results.

Disadvantages Requires significant time and effort to Lacks randomness, prone to bias due to subjectivity in
identify strata and randomly select the sampling process.
samples. 25
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.26

7.4 SUMMARY

▪ Choice of sampling techniques depends upon the

research question(s) and their objectives

▪ Factors affecting sample size include:

- confidence needed in the findings
- accuracy required
- likely categories for analysis

26
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.27

7.4 SUMMARY

▪ Probability sampling requires a sampling frame and

can be more time consuming

▪ When a sampling frame is not possible, non- probability

sampling is used

▪ Many research projects use a combination of sampling

techniques

27
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.28

7.4 SUMMARY
Criteria Non-probability Sampling Probability Sampling

Definition Samples do not have an equal chance of being Samples have an equal chance of being
selected. selected.

Representativen Does not ensure that the sample represents the Ensures that the sample represents the
ess
population. population.

Generalizability Difficult to generalize results from the sample to Easy to generalize results from the sample
the population. to the population.

Sampling Based on the subjective judgment of the Based on probability and randomness.
Method
researcher.

Examples Convenience sampling, quota sampling, Simple random sampling, stratified random
purposive sampling, snowball sampling. sampling, cluster sampling, systematic
sampling.

Application Often used in exploratory research and when Often used in descriptive and inferential
resources are not available for probability research that requires high
sampling. representativeness.

Advantages Easy to implement, saves time and costs. Reliable results that can be inferred to the
entire population.

Disadvantages Lack of representativeness, difficult to check Requires more time and resources to
sampling error. implement.

Control over Difficult to control and assess sampling error. Easy to control and assess sampling error.
Sampling Error 28
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.29

7.4 SUMMARY

All choices depend on the ability to gain

access to organisations

29
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009
Slide 7.30

THANK YOU

30
Saunders, Lewis and Thornhill, Research Methods for Business Students, 5th Edition, © Mark Saunders, Philip Lewis and Adrian Thornhill 2009