CHAPTER 2

Basic Concept
of
Sampling
1
2.0 Basic Concept of Sampling

 2.1 Technical Term

 Element/sampling unit:
an object on which a measurement is taken
E.g.: in a certain community, an opinion poll was
conducted to determine public sentiment toward a
child abuse issue.
Element is the public in the community

2
Example 2:

In a large farm, a study on the

age of trees was conducted.

tree
What is an element?

1 year, 2 years,
What is data value? 3 years, etc.
(age)

3
 Response variable:
The thing/element that you are measured in the
study
E.g.: in a certain community, an opinion poll was
conducted to determine public sentiment toward a
child abuse issue.
Response variable is the opinion of the community
Eg2: a study was conducted to know the
achievement of SQS students in Inferential Statistics
course.
Response variable is the score of Inferential
Statistics course
4
 Population:
A collection of element about which we wish to
make an inference
E.g.: collection of all public in the community
 Frame:
Elements have
A list of sampling units
an equal likely
E.g.: a list of all public in the community to be chose
Elements drawn
 Sample: using certain
amount of
A collection of sampling units drawn from a frame judgement
Can be random or non-random
E.g.: a part of all public in the community

5
 How large sample should
be?

Population size Margin of error Confidence level Standard deviation

Depends on the data that

- How confidence are you you are want to collected
- More bigger the - No sample will perfect.
to decide the sample be Get from the previous
population, more bigger - Determine how much accurate to the actual study.
the sample needed. error to allow value If there is no previous
study, how?

Common confidence
interval:
- 90%
- 95%
6
- 99%
 Sample design:
A clear specification of all sample of a given
type with their corresponding probabilities.
E.g.: suppose we select a sample of n units
with equal probability WITH REPLACEMENT.
The sample design consist of Nn possible
Number of possible sample = 100(30) = 3000
samples with as the probability of selection
Probability = 1/3000
for each of them.

Population =100 Sample = 30

7
In sampling n units
without
replacement, what
is the number of
possible sample be
selected? And what
is the probability?

8
 Excersice 1:

Consider the following situation:

An industry is composed of many small plants located throughout
Kulim. An executive wants to survey the opinions of employees on
the vacation policy of the industry. She sampled employees randomly
as a sampling unit.

a. What is population?
b. What is sampling frame?
c. Suppose the industry has 12 plants and he decides to select one-
third of the population as a sample. What is the probability of each
sampling unit being selected if he used the sampling with
replacement? 9
 Solution:

Population: All small plants in Kulim

Sampling frame: List all of small plants in Kulim

Number of possible
sample = 4(12)=48

Probability = 1/48 =
0.0208 10
 Exercise 2:
 An experimenter wants to estimate the average water
consumption per family in Jitra
What is an element?
Family

What is a population?
All family in Jitra

What is the response variable?

Water consumtion (in liter)

What
List ofis a family
the frame?in Jitra

What
Part ofisthe
a family
sample?in Jitra
11
 Exercise 3:
A forester wants to estimate the total number of trees
on a tree farm that posses diameter exceeding 12
inches. A map of the farm is available.
What is an element?
Tree
What is a population?
All trees on the farm
What is a frame?
A map of Jitra
What is a sample?
Part of map of Jitra
12
Sample size determination

13
Sample size determination
In SQQS2063 (Inferential Statistics) we learnt how to make inference for
population mean and proportion.

We estimate the point and interval estimation.

Interval estimation for mean:

Interval estimation for proportion:

14
Sample size determination

15
Sample size determination

Example:
You are investigating the use of mobile phones for online
banking and want to estimate what proportion of the
population uses their phones in this way at approximate 95%
confidence level. As this is a preliminary study, you are
prepared to accept a margin of error of ±5%. What is the
minimum sample size should be collect?

16
 Solution:

𝛼=0.05 ; 𝜎 =1.5 ; 𝐸=0.25

𝑍𝛼 𝜎
2
𝐸=
√𝑛
1.96(1.5)
¿
√ 138
¿ 0.2503
138 ?
2 2
𝑍𝛼 𝜎
1.96 ( 1.5 )
2 2
2
𝑛= = =138.3
𝐸2 0.25 2

139 ?
𝑍𝛼 𝜎
9 𝐸=
2

√𝑛
1.96(1.5)
¿
√ 139
¿ 0.2494
 Type of
Sampling

Non-
Probability
probability

Simple
Stratified Systematic Cluster Multi-stage Convenience Judgmental Quota Snowball
random

18
 Type of sampling

 Probability sampling:
 The classical estimation
 Requires that randomness
 Itemsin the sample are chosen on the basis of known
probabilities
 Simple random sampling (chapter 3)
 Stratified random sampling (chapter 4)
 Systematic sampling (chapter 6)
 Cluster sampling (chapter 7)
 Two-stage cluster sampling (chapter 8)

19
Nonprobability sampling
Items included are chosen without regard to
their probability of occurrence
Not select the sample at random at all
The findings from the study of the sample
CANNOT be confidently GENERALIZED to the
population.

20
Type of non-probability sampling

Non-
probability
sampling

Convenience Judgmental Quota Snowball

21
Convenience sampling
Involve collecting information from
members of population who are
conveniently available to prove
information

Obtain information from those who

are most conveniently available

Might be necessary to obtain

information from specific target.
22
Example:

Suppose a researcher want to study the

previous semester result of SQS students.
Let say there are 1500 students in SQS, then
a researcher need 10% of the population as
the sample. How to select the sample using
convenience sampling?

23
Solution:

10% out of 1500 is 150 students should be

selected as the sample. By using
convenience sampling he just invite any
student who like to do the survey until it
achieve 150 students.

24
 Advantage and disadvantages of convenience sampling

advantages disadvantages
• Very easy to carry • May be bias
out – few rules for • Undermines your
collecting sample ability to make
• Cost and time are generalizations to
small as the population.
compared to
probability
sample 25
Judgment sampling

Also known as purposive sampling

Used judgement to select from

the population

The choice of subjects who are

well equipped with information
that will be relevant to the focus
26
Judgment sampling (cont..)
Employed when the desired
population for the study is
uncommon or very difficult to
locate and employ.

Focus on particular group of people

whose background expertise relates
to the objective of study

Used when a limited number of

individuals poses the trait of
interest.
27
Example1:

Suppose a researcher want to study the

previous semester result of SQS students.
Let say there are 1500 students in SQS, then
a researcher need 10% of the population as
the sample. How to select the sample using
judgement sampling?

28
Solution:

10% out of 1500 is 150 students should be

selected as the sample. Supposed that there
is very difficult to get the information/data
from various programs in SQS so judgement
sampling is suitable. By using judgement
sampling he can select only student from
industrial statistics program to do the survey
until it achieve 150 students.

29
 Advantage and disadvantages of judgmental sampling

advantages disadvantages
• Wide range of • Can be highly prone
sampling techniques to researcher bias.
that can be used. • Difficult to convince
• Provide researchers the reader – in terms
with justification to of the selecting
make generalization sampling units.
from the sample
• Can be used in a
critical case study 30
Quota sampling
A form of proportionate stratified
sampling but not on convenience basis.

Aims to obtain a representative sample

from a selection of individuals not
necessarily random.

Need to select the relevant variables to

the object of our study.

Need to know the distribution of those

relevant variables in the population.
31
Example:

Suppose a researcher want to study and

compare the previous semester result of SQS
students. Let say there are 1600 students in
SQS which are 800 of them are from
Decision Sciences course, 500 from Statistics
Industry course and the rest are from
Management Mathematics. The researcher
need 10% of the population as the sample.
How to select the sample using quota
sampling?

32
Solution:

Because decision science students is more

then statistics industrial and management
mathematics, researcher may be can chose
the students from decisions science in the
ratio of 3:2:1.
3 2
𝐷𝑒𝑐 𝑆𝑐= ×1 6 0=80𝑆 𝑡𝑎𝑡 = ×1 6 0=53.3 ≈ 53
What is 6the number of 6students for each
1
course?
𝐷𝑒𝑐 𝑆𝑐= ×1 6 0=26. 7 ≈ 27
6
So the researcher has to select 80 student from
Dec. Sc, 53 from Stat Ind. and 27 from Mgt Math.

33
 Advantage and disadvantages of quota sampling

advantages disadvantages
• Helps create an accurate sample • Impossible to detect a potential
of the population when a sampling error, since the sample is
probability sample cannot be not chosen via random selection.
obtained. • Even if a proportion of the
• Since quota sampling does not population is estimated correctly,
need a sampling frame, it is easier the sample selection may be
and quicker to perform biased.
• Gives a better representation of • Since statistical inferences cannot
certain groups within the be made from the sample to the
population, without over- population, it leads to
representing them. generalization problems.
• Difficult to have an accurate quota
frame due to the unavailability of
current information on certain
topics
34
Snowball sampling

Also called chain referral sampling

The technique where existing study

subjects recruit future subjects
from among their acquaintances

The sample group is said to grow

like a rolling snowball
35
Example:

Suppose a researcher want to study the

previous semester result of SQS students.
Let say there are 1500 students in SQS, then
a researcher need 10% of the population as
the sample. How to select the sample using
snowball sampling?

36
Solution:

10% out of 1500 is 150 students should be

selected as the sample. By using snowball
sampling he can select one student to do
the survey. Then the student select another
student and so on until it achieve 150
students.

37
 Advantage and disadvantages of snowball sampling

advantages disadvantages
• It can be difficult to • Snowball sampling is inexact,
identifying units to include in and can produce varied and
your sample, perhaps because inaccurate results - Bias.
there is no obvious list of the • Contradicts many of the
population you are interested assumptions supporting
in. conventional notions of
• The sensitivity of coming random selection and
forward to take part in representativeness – not
research is more acute in such random.
research contexts. • As the subjects locate the
• There may be no other way of hidden population, the
accessing your sample, making research has very little control
snowball sampling the only over the sampling Method
viable choice of sampling 38
strategy.
 Closure: what we studied today….

Probability Non-probability
Terms
sampling sampling
• Element • Simple • convenience
• population random • Judgmental
• Frame • Stratified • Quota
• Sample • Systematics • snowball
• Sample design • Cluster
• Multi-stage

39
 Any question?

Next class lesson:

Chapter 3: Simple Random
sampling

Get ready for 1st quiz (may be

next week)

40