Professional Documents
Culture Documents
Sampling
Sampling
m
@
Ud
em
y
Sa
m
@
Ud
em
2
y
Sa
m
3
@
Ud
em
y
What is Sampling?
y
○ 16% of total population (approx. 13m) – Time/Cost consuming!!
em
○ Select a suitable sample from 13m population
by Location – Munich, Berlin, Frankfurt
Ud
■
■ by Department – Interior, Finance, Defence
@
m
Sa
4
Sampling Terminology
y
Sample/Sample Size – Selected sample (n)
em
employees in Germany
Ud
Sampling Design – How sample is selected
@
Sampling Unit – Each employee is an Unit
m
Sa
Sample Statistics – Findings from sample
5
Sampling Consideration
y
Variation in Sample population
em
Larger the sample
characteristics
Ud
@
Accurate the findings Greater uncertainty
m
Sa
6
Sa
m
7
@
Ud
em
y
Sampling Types in Quantitative Research
y
▷
em
Ud
@
m
Sa
8
Random/Probability Sampling (1/2)
Randomly selected sample – each has equal and independent chance of
selection
Equal Independent
Probability of each sample being No sample is dependent on other
y
selected is Equal sample
em
Total employees: 100; Total employees: 100;
Agreed: 80, Department 5; Each department: 20 pax.
Ud
Refused: 20 Department 2 says all 30 needs to be
(Not equal) included else none of them will
participate.
@
(Not Independent; Not acceptable)
9
Random/Probability Sampling (2/2)
Sample selection technique
y
Narrower focus for larger sample size (e.g. German Public Sector)
em
e.g. Clustered by department and select 1000 from each
Cluster
Narrow focus with similar characteristics e.g. gender, age, earning,
Ud
Proportionate: near equal distribution. of 100 - 50 men 50 women
Stratified Disproportionate: varied distribution: of 100 - 20 men, 80 women
@
Choose from random numbers (manual/software)
Simple/Lottery
m
Sa
10
Non-Random/Non-Probability
Non-Random – Sample features are wide/unknown
Select sample as you get Identify who you think Network sampling. Find
can provide one. He/she will find few
y
until you reach total
em
number correct/unbiased more
information
Ud
Quota Sampling Judgemental Snowball
@
m
Sa
Pre determined sample size Quan
Application
11
Mixed Sampling
y
em
○ Calculate interval distribution (30/6 -› 5)
○ Spread the sample in intervals ( 5 in each interval)
Ud
○ Take 1st sample who best suits (3rd from 5 in first internal)
@
○ Take 2nd sample (must also be from 3rd position in next interval)
m
Sa
12
Sample Size Calculation (1/2)
y
em
○ acceptable amount of error (generally 5%)
■ 60% of population would go for bus over car with 5% margin of error
Ud
■ i.e. given entire population surveyed, you are confident between 55%-65% would go
for bus
@
▷ Step 3 – Set Confidence Level
○
m
Claim your confidence that you will meet your margin of error (typically 90%,
95% or 99%)
Sa
■ i.e. given entire population surveyed, you are 99% confident between 55%-65%
would go for bus
y
em
▷ Step 6 – Use the formula
for known population for unknown population (Step 1 invalid)
Ud
○ ○
𝑧2 × 𝑝 1−𝑃 𝑧 2𝑥 𝑃 1−𝑝
■
@
ⅇ2 ⅇ2
■ 𝑧2 𝑥 𝑃 1−𝑝
1+
ⅇ2 𝑁 2⋅58 2 × 0.5 1−0.5
m ■
2.582 × 0.5 1−0.5 2⋅58 2
Sa
0.052
■ 2⋅582 × 0.5 1−0.5 665 sample
1+
.052 × 1000
(N ) population size
(z ) z-score
400 sample
(e ) margin of error
(p ) standard deviation
14
Sampling Criteria (Quan vs. Qual)
▷ Quantitative
○ Unbiased
○ Represent specific group of population being researched on
y
em
▷ Qualitative
○ Researcher powered
Ud
■ Easily available
@
■ Extend of knowledge
■ Sample related to specific phenomenon/issue
m
Sa
15
Sa
m
@
Ud
em
16
y