MSO174X

Quantitative Methods for

Business

Week 3
Collecting Data 2
 Recap
Last week we looked at data:
• Sources of data
– Primary
– Secondary
• Types of Investigation
– Experiment (and the Aspirin Experiment)
– Observational study
– Survey
• Experiment Example - Aspirin
 Objectives
Having worked through this unit, you should:

• be able to explain the term: ‘population’;

• be able to explain what a statistical sample is;

• understand why we may need to take a sample;

• be able to explain steps involved in sampling

methods.
 Populations
• A population of interest is the collective group of
people or things under investigation. Could be:
– People
– Companies
– Animals

• What was the population for Hennekens’ aspirin

experiment?
 Who was included?
• Did Hennekens use every qualifying male heart
attack survivor?
– Around 22,000 men were represented in the final
result
– Does this represent everyone who could have taken
part?
 Who was included?
• Did Hennekens use every qualifying male heart
attack survivor?
– Around 22,000 men were represented in the final
result
– Does this represent everyone who could have taken
part?
 Sampling
• sample - selection of items from population
which will be questioned/measured.

• A sample should ‘look like’ population:

– relevant factors reflected in sample;
– proportions of factors found in the population.
Why Should we Sample?
 Why sample?
• census rarely genuinely complete;

• census expensive to administer in terms of

money and time;

• census out of date by the time the data properly

collected;
Eg, 2011 UK census not published until September 2013
 Why sample?
• Testing something to destruction means nothing
left!

• Population difficult/impossible to identify.

E.g. all those with a particular genetic disorder.

• Population difficult/impossible to access.

E.g. people residing in the country illegally.
 Representative Sample
A sample must be representative
• Representative sample
– the relevant characteristics of the sample members
are generally the same as the characteristics of the
population.
• Bias
– A study suffers from bias if its design or conduct tends
to favour certain characteristics or results.
• Selection bias
– Occurs whenever researchers select their sample in a
biased way.
 Example
Population
• All Middlesex University students.
Census
• All Middlesex University students.
Sample
• Representative subset which depending on
requirements should consider:
– Gender;
– Level of study;
– Subject area;
– Campus.
 Examples
What is the average amount of time each week
that Middlesex University students spend in the
library?
Population
–All Middlesex University Students
Possible samples
–Students registered for a 1st Year statistics seminar;
–Students attending a postgraduate lecture.

• Would samples generate answer representative of

population?
• Problem?
 Examples
What proportion of the components coming off a
production line is defective?
Population:
–All products produced
Possible sample:
–Every 50th item produced.

• Would sample generate answer representative of

population?
 Examples
Every ten years, the UK government sends out a
census to every household in the country.
Population
Everyone
Possible samples
In theory, there shouldn’t be one.

• Why is this unlikely to be realistic?

• How useful or accurate do you think the results are?
 Examples
A food manufacturer is conducting research into the
shopping habits of its customers, and wants to collect data
from them while they are shopping. To get the widest range
of customers, it conducts a survey near the checkout
section of several supermarkets.
Population
-Manufacturer’s customers
Possible samples
-To think about…..

• What are the potential problems with this method of

data collection? How easily can they be overcome?
 Sampling Frame
• Some sampling methods require all members of the
population to be identifiable.

• Structure supporting identification called a sampling

frame. It should have following characteristics:

– Completeness. Subgroups identifiable?

– Accuracy. Is information correct?

• List up to date?
• Sampling frame available?
• Does each member appear only once in the list?
 Examples of Sampling Frame
• Two readily available sampling frames for the UK
population:
– council tax register (list of dwellings) and
– the electoral register (list of individuals).

Can you think of any problems with these?

 Sampling methods
•Simple random sample;
•Stratified sampling;
•Quota sampling;
•Systematic sampling;
•Popular Non-Random Sampling Methods:
–Snowball Sampling;
–Self-Selection Sampling.
 Sampling methods
•Simple random sample;
•Stratified sampling;
•Quota sampling;
•Systematic sampling;
•Popular Non-Random Sampling Methods:
–Snowball Sampling;
–Self-Selection Sampling.
 Simple Random Sample

• Every item of the population has equal chance of

selection.

• Putting names in a hat.

 Process (SRS)
1. Obtain a list of the population - Sampling Frame
 Process (SRS)
1. Obtain a list of the population - Sampling Frame.

2. Number each individual or item in the list.

 Process (SRS)
1. Obtain a list of the population - Sampling Frame.

2. Number each individual or item in the list.

3. Generate a random number and select the

person/item from the list that corresponds with the
generated number.
 Process (SRS)
1. Obtain a list of the population - Sampling Frame.

2. Number each individual or item in the list.

3. Generate a random number and select the

person/item from the list that corresponds with the
generated number.
4. Repeat to the required sample size.
 Process (SRS)
1. Obtain a list of the population - Sampling Frame.

2. Number each individual or item in the list.

3. Generate a random number and select the

person/item from the list that corresponds with the
generated number.
4. Repeat to the required sample size.
 Stratified Sampling

• Identifies important characteristics.

• Preserves relevant proportions of population in

sample.

• Selections made within strata using simple

random sampling.
Population Sample
Population Sample
 Example - Stratified Sampling
Popular method in market research

Type of employee Number of employees

Manual 200
Administrative 70
Manager 30
300
 Example
If a sample of size 50 was required the sample
would be made up as follows:
Type of employee No. of employees No. in sample
Manual 200
Administrative 70
Manager 30
300
 Example
If a sample of size 50 was required the sample
would be made up as follows:
Type of employee No. of employees No. in sample
Manual 200
Administrative 70
Manager 30
300
 Example
If a sample of size 50 was required the sample
would be made up as follows:
Type of employee No. of employees No. in sample
Manual 200
Administrative 70
Manager 30
300
 Quota Sampling
• Quota: interviewer/team of interviewers given set
number to interview.

• Interviewers - well trained to avoid selection bias.

• Stratified sampling can be used in quota sampling.

• Method of sampling - not really random.

 Example
You are taking a quota survey of men and women
subdividing both into “upto 40” and “40 or older”.
The survey is being taken at a motorway service
station on a Saturday morning.

• Around 10 a.m. 25 men between the ages of 18 and 35

arrive more or less simultaneously.
• Would this make a good quota sample for males under 40?
• What do you think has happened?
Systematic Sampling
Systematic Sampling
Systematic Sampling
 Convenience Sampling
The convenience sample makes use of most readily
available sources of data: usually the people nearby.
 Example: Convenience Sample
A sample is needed to represent the following
populations:
1.All 1st year university students;
2.All Middlesex University students;
3.All 1st year Middlesex University students;
4.All students studying this module.
Convenient sample: choose students attending the
lecture.

• How good a sample would this be for any of the

above populations?
 Non-Random Sampling…
Snowball Sampling

• Snowball samples get bigger because respondents

are asked to suggest other participants.
– chain sampling or
– referral sampling.

• Possibly biased sampling.

 Example: Snowball Sampling
You are asked by an interviewer for your views on
the closure of a local hospital, and are then asked if
you know someone else who might be interested in
being interviewed.

• Who would you ask?

• Are there advantages to this method?

 Non-Random Sampling…
Self-selection
• Results of a poll conducted by a TV or radio
programme: self-selection.

• People invited to take part - biased subgroup.

• Advantages and disadvantages?

 Sample Size
• Larger sample - more accurate result.

• 10-20% - sample size:

– Sample size of 30 (approx);
– Stratified/Quota samples may need much greater
sample proportion if population small;
– If you approach 100 people, or broadcast to 2 million,
much less than that are likely to respond.

• Large population - 10% sample not needed.

 Constraints on Sample Size
•Money and time available;

•Aims of the study and required precision;

•Number of subgroups required.

 Recap
• Understand terms ‘population’ and ‘sample’.
• In certain situations it may be necessary to take a sample.
• Sampling methods:
– Simple random sample;
– Stratified sampling;
– Quota sampling;
– Systematic sampling;
– Snowball Sampling;
– Self-Selection Sampling.
• Sample results will only generalise to be a valid population
result if the sample is unbiased.
 Before Next Week’s Workshop
• Read through Week 3 Lecture Notes; and

• Attempt all the Exercises at the end of the

Week 3 Lecture;

• It is important to prepare these exercises in

advance. Your results will be discussed in
next week’s Workshop.