Week 2

12-Sep-20
COMPANY
LOGO
BUSINESS DECISION MAKING
Nguyen Minh Thu
Faculty of Statistics – National Economics University
Email: [email protected]
BUSINESS DECISION MAKING
Week 2:
THE COLLECTION OF DATA
Business Decision Making – Nguyen Minh Thu – [email protected]
Outline
▪ Data
▪ Types of Data
▪ Data Measurement
▪ Sources of Data
▪ Sampling Methods
1
12-Sep-20
Data
• Data are the facts and figures that are collected, summarized,
analyzed, and interpreted.
• Raw data is a term used to describe data in its most basic
digital format.
Types of data
Cross-sectional Data Time Series Data Panel Data
Time is fixed! Object is fixed! Multi object over time
Eg: 63 provinces’ Eg: Hanoi’s annual Eg: 63 provinces’

economic economic annual economic
information in 2019 information from information from 2000
2000 to 2019 to 2019
Types of data
2
12-Sep-20
Structure of the Dataset

Variables
English Maths
No. Name Sex Age …
Mark Score
1 Anderson M 19 A 8 …
Observations
2 Berky F 20 C 9 …
3 Charles M 20 B 7
… … … … … …. ….
• Observation: the person or thing on which measurements

are taken
• Variable: A measured characteristic of an observational
unit
• Datum: each value in the dataset
Types of Variables
Qualitative (categorical)
attribute
Quantitative (Numerical)
number
Types of Variables
Quantitative:
Qualitative:
3
12-Sep-20
Quantitative Variables
Discrete or Continuous
Result from: Result from:

- counting or labelling observations - measuring (length, weight etc) to
any degree of precision.
• The number of cars sold in the UK in 2017 • The salary of employees

• The age of a person in years • The distance between places
• The size of clothes (S, M, L,...) • The productivity of workers
10
Data Measurement
11
Nominal Scale
• The labels are numerically coded

• Have no logical order among labels and numbers
Sex Code
Male 1
Female 2
12
4
12-Sep-20
Ordinal Scale
The data classified can be ranked or ordered.
LEARNING STATISTICS IS INTERESTING!
Strongly Strongly
Disagree Neutral Agree
Disagree Agree
1 2 3 4 5
Strongly Strongly
Disagree Neutral Agree
Disagree Agree
5 4 3 2 1
13
Interval Scale
There is no natural zero point
→ cannot calculate the ratio
14
Ratio Scale
The interval level with a natural zero

starting point
Can use every function
15
5
12-Sep-20
Levels of Measurement
16
Levels of Measurement
Qualitative (Categorical) Quantitative (Scale)
Nominal Ordinal Discrete Continuous

Listing, Listing, Interval Ratio
Grouping Grouping,
Sorting, Listing, Grouping, Sorting
Maybe ± Math operation: ±, , ÷, …
Coded by numbers Used to rank
17
Sources of Data
Primary Secondary
Source Source
Both must be:
• Relevant
Already exists,
Collected for
• Accurate collected for
the particular purpose
• Current some other purpose
• Impartial
18
6
12-Sep-20
Sources of Data
19
Primary vs Secondary Data?
• Focus group
• Statistical Yearbook of Vietnam
• Survey
• Interview
• Trade Association Report
20
Primary vs Secondary Data?
Where can you find these data?

• Vietnamese population by province
• Key financial indicators for enterprises
• Student’s satisfaction with the service quality of IBD programs
• GDP of ASEAN countries
• Customer satisfaction with university cafeteria food services
21
7
12-Sep-20
Primary Data Source
• Census vs Sampling
• Sampling process
22
Population vs Sample
Population:
A set of all interested elements
N represents the population size, maybe infinite
Sample:
A part of the population that is selected to
represent the entire group
n represents the sample size, finite
23
Census vs Sampling
A census is a study of every unit, Sampling is a method of

everyone or everything, in a studying from a few selected
population items, instead of the entire big
number of units
24
8
12-Sep-20
Reason to take sample
• Collecting information from the entire population is sometimes
impossible
• Enable research/surveys to be done more quickly/timely
• Less expensive and often more accurate than large census
• Allows for minimal damage or lost
• Be used to validate census data
25
An important requirement
A sample must be representative of the population.
26
Sampling process
Define Population
Specify Sampling Frame
Determine Sampling Method
Probability Sampling Non-Probability Sampling
Determine Appropriate
Sample Size
Execute Sampling Design
27
9
12-Sep-20
Moving from population to sample
Population
Sampling frame
(a list of all items of
the population)
Sample
28
Types of Sampling
29
Probability vs Non-probability Sampling
PROBABILITY NON-PROBABILITY
FEATURE
SAMPLING SAMPLING
The researcher selects
Subjects of the population get
samples based on the
an equal opportunity to be
Meaning subjective judgment of the
selected as a representative
researcher rather than random
sample
selection
Alternately known as Random sampling Non-random sampling
Basis of selection Randomly Arbitrarily
Opportunity of
selection Fixed and known Not specified and unknown
Research Conclusive Exploratory

Result Unbiased Biased
Method Objective Subjective
Inferences Statistical Analytical
Hypothesis Tested Generated
30
10
12-Sep-20
Probability Sampling
31
Simple random sampling

• Informal method: randomly picking. Easiest way and can be applied to a
small population (picking a name out of a hat, choosing the short straw,
lottery draw,…
• Formal method: use the table of random numbers, software programs
32
Simple random sampling
• Five steps in applying this method

i. Obtain a complete sampling frame
ii. Give each case a unique number starting at one
iii. Decide on the required sample size
iv. Select numbers for the sample size from a table of random numbers
v. Select the cases that correspond to the randomly chosen numbers
• Example: Randomly call a few students to take attendance
33
11
12-Sep-20
Table of random numbers
54033 93539 74902 57237 83940 03830 70718 70015 45487 45727 98085 14512 38614
92744 53223 90608 36942 71382 71368 65638 24113 92374 39008 76553 79286 14332
17716 95690 21584 44015 67622 95328 21217 20944 30226 73254 40506 38808 50946
99153 06630 48287 63905 43610 97537 15845 17295 39327 21392 84739 74071 80258
32607 84109 56169 87115 94217 93041 81437 84223 38925 77017 80482 70788 93096
25123 11307 88876 15580 35470 15266 92263 49508 59603 82354 93782 24775 57586
62173 29061 68582 76463 26261 68616 77488 61533 16777 98307 56249 29970 96282
60706 30534 75614 81804 10239 76535 51098 78806 24059 43888 30521 30119 18724
34
When do we apply this?
▪ Have a good sampling frame
▪ Population is geographically concentrated
▪ Data collection technique does not involve travelling
35
Systematic random sampling
Choose every “kth” individual to be a part of the sample
36
12
12-Sep-20
Systematic random sampling
Steps to obtain a systematic sample:

• Obtain a sampling frame
• Determine the population size: N
• Determine the sample size required: n
• Divide population of N individuals into groups of k individuals:
N
k=
n
• Randomly select one individual from the 1st group
• Select every kth individual thereafter
37
Stratified random sampling
• Population is divided into two or more groups called

strata
• Subsamples are randomly selected from each strata
38
Stratified random sampling
• The sampling procedure is more complicated

• Steps to take a stratified sample
• Select the stratifying variable
• Divide the sampling frame into strata or categories
• Draw a systematic or random sample of each stratum
39
13
12-Sep-20
Cluster random sampling
▪ Divide the population into separate groups, called clusters.

▪ Two types of cluster sampling:
• One stage cluster
• Two stage cluster
40
Cluster random sampling
▪ One stage cluster

• Randomly select subsets
• Sample entire participations in
the selected subset
▪ Two stage cluster
• Randomly select subsets
• Conduct simple random
sampling for participations in
the selected subset
41
Multi-stage random sampling
• To be a complex form of cluster and stratified sampling

• Carried out in stages
• Using smaller and smaller sampling units at each stage
42
14
12-Sep-20
Probability sampling
Technique Advantages Disadvantages
- Easy to conduct
- Identification of all members of the
- Not require any additional information
population can be difficult
Random except the contact info
- Can be expensive and unfeasible for
- Meets assumption of many statistical
large population
procedures
- Easy to construct, execute, compare, and
Systematic understand - High sampling bias if periodicity exists
- Spread over population
- More accurate sample - Problem if strata not clearly defined
Stratified
- Effective representation of all subgroups - Complex to apply in practical levels
- Time efficient
- Cost efficient: reduce field cost - May not be representative of whole
Cluster
- Applicable where no complete list of population
units is available
43
Non-Probability sampling
The process of
selecting sample
without using statistical
probability theory
44
Quota sampling
▪ Similar to stratified sampling: population is divided into

subsets
▪ Select the participations from each subset based on
specified proportion
45
15
12-Sep-20
Purposive sampling
- Also known as: Judgmental Sampling, Selective Sampling,

Subjective Sampling
- Rely on the judgement of the researcher
Subjective Selective
Sampling Sampling
46
Volunteer sampling
▪ Participants self-select to become part of a study because
they volunteer when asked, or respond to an advert
▪ Two types of volunteer sampling:
- Snowball
- Self selection
47
Snowball sampling
▪ Known as network or chain-referral sampling
▪ Existing participations recruit future participations among
their acquaintances
48
16
12-Sep-20
Self selection sampling
▪ Individuals identify their wish to take part in the study
▪ Individuals volunteer to be part of the sample
49
Convinience sampling
▪ Known as Haphazard or Accidential sampling
▪ Sample units are only selected if they can be accessed
easily and conviniently
50
Non Probability sampling
Technique Advantages Disadvantages
- Low cost, time and administrations - Dependent on subjective decisions

Quota
- No need for list of population elements - Not possible to generalise
- Select only individuals whom are - No guarantee that chosen sample are
Purposive relevant to research purpose true representative of the population
- Less costly, more convenient - Limited generalisability
- May have an interest in the subject so - Over-representation of a particular

they are less likely to give biased network
Volunteer
information - Take a long time to get enough
- Doesn’t require a lot of screening people to do experiment
- High levels of simplicity and ease - Highest level of sampling error

Convinience - Less time and cost required - Sample is not representative of
- Usefulness in pilot studies population
51
17
12-Sep-20
Types of Sampling
52
Group work
Let’s play a role to choose a few students in our

class according to one of the sampling methods!
53
Determine Sample Size
Key issues to consider:

• Financial
• Informational
• Statistical
54
18
12-Sep-20
Financial issue
• Larger samples cost more, however, the sampling

error will decrease
• Budget available often poses a financial constraint
on sampling and sample size
→ It is important to find the optimal sample
that best balances the sample sizes, costs
and budget available
55
Informational issue
Number of subgroups to be analyzed often poses

challenges in determining optimal sample size
56
Statistical issue
Statistical approaches for determining sample size

for probability samples:
• An estimate of the population standard deviation
• The acceptable level of sampling error
• The desired level of confidence that the sample will fall
within a certain range of the true population values
57
19
12-Sep-20
Statistical issue
Determining sample size based on:

• Margin of error (MOE)
• Confidence level/Z-score
• Cost per interview (CPI)
• Budget
• The population standard deviation
58
Formula to determine sample size
From MOE: use in nearly every case
Z2 σ2 Z2 p(1−p)
n= n=
ε2 ε2
where:
• Z: level of confidence expressed in standard errors (95% or 1.96)
• σ2 : population standard deviation for means
• p(1-p) = population standard deviation for proportions (.25 or .5)
• ε = acceptable amount of sampling error (min. +/-5%)
59
Formula to determine sample size
From budget and CPI:
Budget
n=
CPI
60
20
12-Sep-20
But…will this have an acceptable MOE?
Z 2 σ2 Z 2 p(1 − p)
ε= ε=
n n
where:
• ε = acceptable amount of sampling error (min. +/-5%)
• Z = level of confidence expressed in standard errors (95% or 1.96)
• σ2 : population standard deviation for means

• p(1-p) = population standard deviation for proportions
61
Example 1
You are in charge of designing and running a research study with a

MOE of +/-5% and a confidence level of 95%. P=.50. What sample
size is needed?
Z 2 p(1 − p)
n=
ε2
Values:
• Z = Confidence Interval: 1.96 or 2 standard errors (95% confidence)
• p(1-p) = Standard Error of Proportions: .25
• ε = Random Sampling Margin of Error: 5%
62
Example 2
You are working with a confidence level of 95% and p=.80. Your
CPI is $25.00 and your boss has given you a budget of $10,000.
What sample size is needed to meet the budget constraints? Does
this have an acceptable MOE?
Budget Z 2 p(1 − p)
n= ε=
CPI n
Values:
• Z = Confidence Interval: 1.96 or 2 standard errors (95% confidence)
• p(1-p) = Standard Error of Proportions: .16
63
21
12-Sep-20
Exercise
You are embarking on a marketing research project for your company
and will subcontract to an outsourced firm for $10 per response. Your
boss wants to be 95% confident of the results. Assume a sample statistic
of .25 and that the level of tolerable sample error is 2.5%. They’ve given
you a budget of $5,000 to complete this survey.
a. What is the sample size given these parameters?
b. Given the budget, should the researcher take the project?
c. Why or why not?
d. How much more or less sample would the research need to meet the
budget?
e. Assume sample size of 500. What would the margin of error be?
64
22

Week 2

Uploaded by

Copyright:

Available Formats

Week 2

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Week 2

Uploaded by

Copyright:

Available Formats

12-Sep-20

BUSINESS DECISION MAKING

Nguyen Minh Thu

Faculty of Statistics – National Economics University

BUSINESS DECISION MAKING

THE COLLECTION OF DATA

Business Decision Making – Nguyen Minh Thu – [email protected]

Business Decision Making – Nguyen Minh Thu – [email protected]

analyzed, and interpreted.

• Raw data is a term used to describe data in its most basic

Business Decision Making – Nguyen Minh Thu – [email protected]

Cross-sectional Data Time Series Data Panel Data

Time is fixed! Object is fixed! Multi object over time

Eg: 63 provinces’ Eg: Hanoi’s annual Eg: 63 provinces’

Business Decision Making – Nguyen Minh Thu – [email protected]

Business Decision Making – Nguyen Minh Thu – [email protected]

Structure of the Dataset

• Observation: the person or thing on which measurements

Business Decision Making – Nguyen Minh Thu – [email protected]

Business Decision Making – Nguyen Minh Thu – [email protected]

Business Decision Making – Nguyen Minh Thu – [email protected]

Result from: Result from:

• The number of cars sold in the UK in 2017 • The salary of employees

Business Decision Making – Nguyen Minh Thu – [email protected]

Business Decision Making – Nguyen Minh Thu – [email protected]

• The labels are numerically coded

Business Decision Making – Nguyen Minh Thu – [email protected]

The data classified can be ranked or ordered.

LEARNING STATISTICS IS INTERESTING!

Business Decision Making – Nguyen Minh Thu – [email protected]

There is no natural zero point

→ cannot calculate the ratio

Business Decision Making – Nguyen Minh Thu – [email protected]

The interval level with a natural zero

Can use every function

Business Decision Making – Nguyen Minh Thu – [email protected]

Business Decision Making – Nguyen Minh Thu – [email protected]

Qualitative (Categorical) Quantitative (Scale)

Nominal Ordinal Discrete Continuous

Business Decision Making – Nguyen Minh Thu – [email protected]

Business Decision Making – Nguyen Minh Thu – [email protected]

Business Decision Making – Nguyen Minh Thu – [email protected]

Primary vs Secondary Data?

Business Decision Making – Nguyen Minh Thu – [email protected]

Primary vs Secondary Data?

Where can you find these data?

Business Decision Making – Nguyen Minh Thu – [email protected]

Primary Data Source

Business Decision Making – Nguyen Minh Thu – [email protected]

Business Decision Making – Nguyen Minh Thu – [email protected]

A census is a study of every unit, Sampling is a method of

Business Decision Making – Nguyen Minh Thu – [email protected]

Reason to take sample

• Collecting information from the entire population is sometimes

• Enable research/surveys to be done more quickly/timely

• Less expensive and often more accurate than large census

• Allows for minimal damage or lost

• Be used to validate census data

Business Decision Making – Nguyen Minh Thu – [email protected]

A sample must be representative of the population.