GCSE: Data Collection &

Sampling
Dr J Frost ([email protected])
www.drfrostmaths.com

Last modified: 31st October 2016

Questionnaires
In GCSE papers, you tend to have to do two things to do with questionnaires.
1. What’s wrong with a given question.
2. How you would rewrite it.

Q Ben wants to find out what food Q Naomi wants to find out how often
people like to eat in restaurants, so people go to the cinema. She uses
asks his family: this question on a questionnaire.
“Do you agree that pizza is better
than pasta?” “How many times do you go to the
What’s wrong with his survey? (2) cinema?”
□ Not very often □ Sometimes
□ A lot
2 of the following:
a. Biased question. a) Write down two things wrong
b. Restricted sample of with this question. (2)
people. ?
c. Doesn’t specify a range of 2 of the following:
foods. a. No time-scale.
d. Nothing to do with eating
habits. boxes.
?
b. Non-exhaustive response

c. Labels too vague.

Criticising Questions

Q “What is your age in years? Q “How many texts have you sent on
your mobile phone?
□ Under 20
□ 20-30 □ 0-10
□ 30-40 □ 10-20
□ 40-50 □ 20-30
□ over 50 □ 30 or more”

What is wrong with this question? (1) List two things wrong with this
question. (2)

?
Overlapping regions.
•
• ?
Overlapping regions.
No time frame.
Improving/Designing Questions
Q Naomi wants to find out how often people go to the cinema. She uses this
question on a questionnaire.

“How many times do you go to the cinema?”

□ Not very often □ Sometimes □ A lot

a) Write down two things wrong with this question. (2)

b) Design a better question for her questionnaire to find out how often adults
go to the cinema. You should include some response boxes. (2)

How many times did you go to the cinema last month?

□0 □ 1-2 □ 3-5 □ >5
______________________________________________________

What do you think the mark scheme ?is looking for?

• Time period must be included. (1 mark)
? boxes. (1 mark)
• At least 3 non-overlapping response
Improving/Designing Questions
Q Valerie is the manager of a supermarket.
She wants to find out how often people shop at her supermarket.
She will use a questionnaire.
Design a suitable question for Valerie to use on her questionnaire.
You must include some response boxes. (2 marks)

e.g. How many times each week do you shop at this supermarket?
0, 1, 2, 3, 4 or more

•
?
B1 for an appropriate question with a reference to a time period
OR a question with time period implied by responses.
• B1 for at least 3 non-overlapping boxes (ignore if not exhaustive)
Do not accept frequency tables or data collection sheets.
Question Bias
How much does the wording of a question change the results?

English/Welsh National Census British Social Attitudes Survey

(2001): (2002):
“What is your religion?” “Do you regard yourself as
belonging to any particular
religion?”
?
15.7% - No religion

?
50.5% - No religion
Tally Charts
Phillip is going to carry out a survey of the football teams supported by each of his
friends. In the space below, draw a suitable data collection sheet that Phillip could use.
(3 marks)

Team Tally Frequency

1 mark for each column heading.

Sampling Motivation
I wish to find out the proportion of animals in my zoo
that have contracted the deadly disease ‘Chilcotius’.
Sampling Motivation
I could test the entire population of animals at the zoo.
?
(Looking at the entire population is known as a census)
But this would be time consuming.

We therefore
instead want to
just test some of
the animals, and
hope that the
percentage of
these animals with
the disease
accurately reflects
the whole
population of
animals, e.g. if
20% have the
disease within our
selected group, we
assume roughly
20% of all animals
have the disease.
Sampling Motivation
?
I could select just some of the animals to test. This selection is known as a sample.
We hope that the proportion of animals with the disease in the sample
accurately reflects the whole population.

We therefore
instead want to just
test some of the
animals, and hope
that the percentage
of these animals
with the disease
within the sample
accurately reflects
the whole
population of
animals, e.g. if 20%
have the disease
within our selected
group, we assume
roughly 20% of all
animals have the
disease.
Random Sample
But how do we get our sample?
We could just animals randomly.

Disadvantage of
random sampling:
It’s possible not all
the different types
of animals are
?
represented. In the
sample on the left,
no dolphins were
included.
Exam Questions on Random Sampling
“Describe a random sample”
?
a sample where each thing in the population is equally likely to be chosen.
Bro Tip: If you are asked in an exam what a
random sample is, the key phrase they’re
looking for is “equal chance”.

“You want to take a random sample a student’s favourite TV programmes at

school. Describe how you could achieve a random sample.”

Since it is a random sample, you need to ensure each thing is equally

likely to be chosen. Mark schemes would expect an approach such as:
1. Put all student names into a hat and pick them out to decide who to
sample. ?
2. Use a random number generator where each number represents a
student.

“Dave wants to determine the spending habits of the UK population.

He stands outside HSBC bank’s headquarters and asks the first 10 people he sees. Suggest two
reasons why his sample will likely be poor.”

• Sample not random: People working for a bank likely to earn more money.
• Sample size too small. ?
Sample Bias Question on your Sheet
Melanie wants to find out how often people go to the cinema.
She gives a questionnaire to all the women leaving a cinema.
Her sample is biased.
Give two possible reasons why.

1. Only women were asked/you need to ask men.

ask people in different places.

?
2. Only people leaving the cinema were asked/you need to
Stratified Sampling
Instead of sampling animals completely randomly, we might want to ensure that we sample the same
proportion/percentage from each group, so that each type of animal is fairly represented. (With
random sampling, it’s possible to avoid having any lions!) This is known as a stratified sample.

If we want to do a
stratified sample
where we sample
25% of all animals,
how many do we
sample from each
group?

Pandas
4?
Lions
1?
Dolphins
2?
Elephants
1?
Stratified Sampling Example Questions
! “Describe a stratified sample”
In stratified sampling, the population is divided into groups, and random samples
are taken from each stratum. ?
(Stratum means group, and literally means ‘layer’)

Typical Question:
In a school, I wish to sample 40 people from Year 8 to Year 10
for their favourite chemical compound.
I wish to do a stratified sample by year group. Bro Tip 1: You can only
sample a whole number of
Year Group Total Students things, so you MUST round
8 184 your final answer.

9 154 Bro Tip 2: Check the your

10 238 answer looks sensible.
Here it does, as slightly
under half the students are
How many students should I sample in Year 10? in Year 10.

Proportion Sampled: ? Num to sample: people

?
Test your understanding

50/258 of students sampled.

26 x (50/258) =?5.039
So 5 students.

25 + 48 + 62 = 135 females.
135 x (50/258)? = 26.16
So 26 students.
Exercises (on your sheet)
Question 1

students ?
Question 2

students ?
Exercises (on your sheet)
Question 3

a)
b) 9
c) 28
?

Question 4

𝟏𝟎𝟎
× 𝟏𝟖 → 𝟓
𝟑𝟒𝟎 ?
Exercises (on your sheet)
Question 5

? ?
? ?
? ?
? ?
TV Ratings

What method of sampling does it sound like the UK uses?

“These panel homes are drawn from a

household sample that is designed by
RSMB to remain representative of all
television households across the UK.
This means it always encompasses the
full range of demographic and TV
reception variations, amongst other
variables, that are found across the
country and in different ITV and BBC
regions.”
Quickfire Questions
Do these in your head...

Out of a herd of 200 big cats, I want to sample 50 of them. There

Q
are 60 lions. How many lions should I sample?

?
15 lions.

Class A has 20 people, Class B 10 and Class C 30. I want to sample

Q
10 people. How many people do I sample from Class C?

?
5 people

In the land of Frostonia there are 1 million people. I want to

Q
sample 50,000 to determine their TV watching habits. If there are
6,000 Indians in Frostonia, how many of them should I sample?

?
300 Indians.