Brief Description
This unit explains the basic techniques involved in opinion polls. It is centred around
a practical poll to find out pupils' favourite records, and the wider implications to all
polls are stressed throughout. Both random and stratified sampling are introduced.
Design time: About 6 hours
Aims and
Objectives
Pupils will learn how to conduct an opinion poll and identify possible pitfalls. On
completion of the unit they will know that a sample needs to be representative and will be
able to choose a random sample and a stratified sample to help ensure this.
They meet examples of samples which are unrepresentative through being too small,
self-selected or clearly biased in nature. They carry out one com- plete survey and plan
another. They will be more aware of the practical problems that arise in data collection
and the relationship between sample and population. They practise filling in tables, using
random number tables, drawing bar charts and writing a report on a statistical experiment.
Prerequisites
Pupils should be able to use random number tables, draw bar charts, make tally charts,
multiply whole numbers by a fraction and round to the nearest whole number.
Equipment
and Planning
Each pupil will need a sheet of random number tables. For the school poll, class lists
for the first five years and the total numbers of pupils in each year are required. A list
of the current 'Top 30' long-playing albums is also needed. These are published in New
Musical Express, Melody Maker and Sounds. Graph paper is required
in Sections C and D. A calculator will be useful for C6.
Sections B2, B3 and C7 are optional. It is possible to end the
unit with Section C and return to Section D at a later date.
Detailed
Notes
Section A
This section introduces the opinion poll and the idea of a sample. It sets the scene
for the rest of the unit. As such, it would be valuable to use this section for discussion
on the purpose of polls, what sort of information may be required and how it may be used.
Surveys are conducted on many diverse topics. A list of those reported to the Survey
Control Unit is published in Statistical News. Most of those in A1 are
selected from these lists. Two major companies involved in opinion polls are National
Opinion Polls and Gallup Polls, whose most publicized polls are those concerning
prediction of election results.
A1
Pupils are invited to match up the organizations who commissioned the polls with
the topics and the relevant population. The poll on the reliability of cars was run by the
Consumers' Association, who questioned drivers (2). The Department of Industry carried out
the poll on the Recycling of Waste Materials by interviewing adults (3). The Ministry of
Agriculture, Fisheries and Food asked farmers about growing lettuce under glass (4).
Doctors asked mothers about the decline in breast feeding (5).
A2
The poll on the reliability of cars is used to introduce the purpose of and the
background work to a poll, including the need to take a sample. The purpose of the poll
which the pupils will carry out in C8 is also introduced. It is important to get
the purpose of the poll clearly defined, since this will determine who should be asked and
the nature of the questions. The questions raised can act as a basis for discussion to
help pupils appreciate how each stage of the unit will lead up to the poll. No definite
answers are expected: these will emerge as work through the unit progresses. Some points
relating to these questions are made in D2.
Section B
This section deals with whom and what to ask to obtain the information you need. After
discussion, it may be suitable for homework.
B1
All questions here relate to the proposed school record library. Asking pupils
which albums they would like to borrow has the advantage that it links directly with the
purpose of the library. The disadvantage is that pupils may not know what is available.
Asking a local shop about sales has the advantage of showing which is popular. A
disadvantage is that sales are made to people of all ages, not just school pupils. Asking
pupils which albums they like best has the advantage of finding out pupils' preferences. A
disadvantage is that some will write down more titles than others and their views will be
over-represented. Asking pupils which they would buy has the advantage of finding the
pupils' favourite records. A disadvantage is that the records they might buy are not
necessarily those that they might want to borrow.
A list of the 'Top 30' restricts choice and simplifies the analysis of results. Any
that do not appear in the 'Top 30' will almost certainly not be the main favourites at the
moment.
*B2
This section is optional and allows for further work on collecting the required
information by referring to the weekly 'Top 30' singles charts.
Asking a sample of record shops about their sales is the method least open to abuse. It
is the method that is used, with the check that if a record goes up more than 10 places,
the returns for that record are scrutinized to see if one particular shop or area is
involved.
The number of requests is a self-selected sample, and ignores people who do not write
in. It can also be easily 'fixed' by phoney requests. The panel of disc jockeys may not be
representative and could be affected by self- interests. Sampling record buyers would be
too time-consuming to repeat each week. The number of times a record is played on the
radio may well reflect the views of an individual producer or disc jockey.
*B3
B3 is optional and raises the problem of how to choose which shops to
put in the sample. In choosing a local sample, one could choose at random from the larger
shops' these should be more representative than the smaller ones. A wid e geographical
spread o f shops is needed to ensure full representation of national views.
The 'Top 30' singles gain much publicity from the TV and radio programmes featuring
them. Other 'Top 30s'are not featured to that extent on programmes.
B4
Here we identify the populations for a survey. It is important to determine whom
to ask and start with the correct total population (sampling frame), although this may be
difficult or even impossible. It may be different for different purposes. The population
could be as follows:
School dinners: (i) all pupils in the school, (ii) all pupils who take dinner,
(iii) all pupils and staff, etc.
Airports: (i) all travellers using the airport on a given day, (ii) all people
in the airport on a given day, etc.
Parks: (i) all people in the town, (ii) all people who use a park in the town,
etc. (To ask people in the park at a particular time would yield a biased sample.)
Section C
This section contains the actual record album poll in C8. The early parts of
the section deal with the distinctions between good and bad samples. A random sample and a
stratified sample are taken and compared to see how well they represent the population. A
report of the poll is written.
Some of the work in C3 and C4 may be omitted or replaced by
discussion. For the pupil poll, the simplest way may be to take samples from pupils in the
first five years, depending on your school. The total population should be under 1000,
otherwise four-figure random numbers are needed. To avoid this, you could use only the
first three or four year groups. The class lists of these years and the total number in
each year will be required.
C1
- It is extremely time-consuming and difficult to question all the pupils in the school
and to analyse their replies. A sample is easier to do.
- A sample should be chosen fairly and be representative of the population. It is possible
that a fair method may give an unrepresentative sample (e.g. a random sample of pupils
could contain only girls from one year, although this is very unlikely in a large sample.)
- The fairest methods are to choose every tenth pupil and to ask one in ten pupils from
each year group. Slight variations on these are expanded later. The former is a systematic
sample, and a fuller condition of
fairness requires that a random process be used to select the person from the first
ten. Asking 50 pupils in the playground may give too many younger pupils, boys playing
football or girls skipping. Asking 50 pupils from one year group ignores all other year
groups. The school magazine gives a self-selected sample and ignores all those who do not
respond. Asking four pupils from each class is only fair if all classes are the same size.
Using volunteers from assembly gives another self- selected sample and will get those with
strong opinions. Asking only five pupils as they enter school gives too small a sample.
Inviting pupils to attend a meeting gives. another self-selected sample.
C2
There are many examples of bad sampling. It is important to bear them in mind.
Polling by telephone only includes people with a telephone, and this principle led to a
wrong prediction in the 1936 US presidential poll. The Scottish newspaper example is a
self-selected sample, while the next sample is too small. The post office sample is likely
to be heavily weighted towards mothers. The first 10 people off a bus gives too small a
sample, whilst the last example is likely to omit those with knowledge about fashionable
teenage clothes. The parents may well be at work while the teenagers are at school or
work.
C3
This section provides pictorial clues to badly chosen samples. Each of the groups
at the bus stops can be related to particular activities at certain times, e.g. pupils
travelling to school, football or rugby supporters going to a match and housewives
shopping.
C4
Pupils may need more help in the use of random number tables and where to find
them. They are introduced in the Level One units Being Fair to Ernie and If
at first... It is usual to ignore repeats, although one could use the same pupil's
view twice. In this case both methods are fair: the distinction is between sampling
without and with replacement. Either can be valid, but it is important to decide which you
mean to do (and why) before you begin.
- The second method would lend itself more easily to a large population.
C5
A simple random sample is drawn. This may be done as a class activity with each
pupil (or pair of pupils) taking one class list with numbers allocated accordingly across
all the lists. The sample size of 50 may be varied, depending on what is a convenient
sample size for the stratified sample (in C6) in your particular school. As the
two samples are to be compared, it is simpler i f the sample sizes are equal. It may be
helpful to obtain an extra three or four 'reserves' for this sample, as reserves may be
needed for the stratified sample in C6. Reserves will be used in carrying out the
poll with the stratified sample if any of the sample are absent.
- If the stratified sample of C6 is to be used in the actual poll, it is only
necessary here to write down the names of pupils from one form or year.
The class will need to know the numbers of pupils in each year to complete column 5 of
Table 3. They may be able to decide whether each year is fairly represented (i.e. in
proportion to the numbers in each year of the population) simply by inspection. It is
unlikely that each year will be fairly represented, hence the suggestion of the stratified
sample in C6.
C6
A stratified sample is drawn up according to the numbers of pupils in each year.
It is possible to stratify according to sex as well as age if there are grounds for belief
that choice of record album will vary from boys to girls. This is more complicated but
could be considered by the most able classes.
- Selecting l0 from each year would be unfair if there were different numbers of pupils in
each year.
The method of finding the number of pupils to be asked in each stratum (year group) is
shown by a worked example using different numbers. A simpler example may help less able
pupils see the principle. For example:
Year |
No. of pupils |
Stratified sample no. |
1 |
70 |
7 |
2 |
120 |
12 |
3 |
100 |
10 |
4 |
90 |
9 |
5 |
120 |
12 |
Total |
500 |
50 |
A sample of 50 is 50/500 i.e. 1/10 of the
school, so we need 1/10 of each year. It may be necessary to adjust
the figures slightly if rounding leads to a sample size other than 50. Reserves for each
year should be chosen before carrying out the poll. It may be interesting to see if some
years needed significantly more reserves than others.
*C7
Bar charts are drawn to enable the comparison of the numbers in each year in each
of the two samples (from C5 and C6) to be made with the numbers of
pupils in each year of the population. An alternative way would be to use percentages and
draw up a table. The comparison should show that the stratified sample is the more
representative, and so this gives the list of pupils to be used in the poll in C8.
C8
It is important that the actual poll is carried out and the results analysed. The
actual questioning of the sample can be carried out by the pupils at registration or other
convenient time, and their results brought to the next lesson to be collated. An
alternative is to enlist the help of form teachers. The collation can be carried out as a
class activity with each pupil recording the number of votes for each album as they are
read out. A discussion of the poll is beneficial as a prelude to writing the report. This
should be one piece ofcontinuous prose, not a list of answers to the listed points which
are given purely as a guide to the structure and content of the report.
Section D
D1
Here some artificial results are presented of l000 pupils to enable a comparison
between random and stratified sampling. This section could be done in groups and the
results pooled, with some doing the random sample, some the stratified sample and others
obtaining the overall results. The stratified sample may be obtained by taking every
twelfth (or fourteenth), having chosen the first out of the first twelve (or fourteen) at
random, rather than use random numbers.
Bar charts are used to help make the comparison between the random and stratified
samples, although percentages and/or pie charts could be employed. Stratification does
make a difference as the first year prefer B, the second year prefer A, while the
third year prefer C. The overall result is 348/311/341.
D2
This section applies general principles in specific polls. This should help
pupils remember the wider implications of polling and would make suitable homework
material. You may like to add alternative suggestions to the list of possible polls. These
may be either topical or of particular interest in your school, e.g. uniform, youth clubs
and year councils.
Page R1
Opinion poll on: |
Information wanted by: |
Whom to ask: |
1 Holiday intentions |
English Tourist Board |
Adults |
2 Reliability of cars |
|
|
3 Recycling waste materials |
|
|
4 Growth of lettuce under glass |
|
|
5 Decline in breast feeding |
|
|
Table 3 - Opinion polls
1 |
2 |
3 |
4 |
5 |
Year |
Number of pupils |
Calculation |
Stratified sample size |
Random sample size |
1st |
|
|
|
|
2nd |
|
|
|
|
3rd |
|
|
|
|
4th |
|
|
|
|
5th |
|
|
|
|
Total no. of pupils = _____
Table 4 - Numbers of pupils in each year
Table 5 - Number of votes for each group (from a random sample)
|
Number of votes |
Year |
A |
B |
C |
1 |
|
|
|
2 |
|
|
|
3 |
|
|
|
Totals |
|
|
|
Table 6 - Number of votes for each group (from a stratified sample)
Test
Questions
Here are three opinion polls that might be carried out:
- Find the views of pupils on school dinners.
- Find out whether Scotland wants a separate government.
- Find out whether a new type of washing powder will sell.
- For each poll decide who might be interested. Choose from this list:
Manufacturers
Pilots
School caretakers
Housewives
Parliament
School cooks
Poll a
Poll b
Poll c
- For the same polls decide whom you might ask. Choose from this list
Scottish people
Housewives
Farmers
Pupils
Teachers
Tourists in Scotland
Poll a
Poll b
Poll c
- Expiain why the following samples chosen for the three polls are not representative.
Poll a: Ask the first 25 pupils in the dinner queue.
Poll b: Ask people who wrote to their MP about the subject.
Poll c: Ask only three housewives at a supermarket.
- A youth club has a free ticket for a concert each week. Which of these three methods is
the fairest way of deciding which members should have it?. Explain why the other two are
not fair.
- Give it to the first member to arrive.
- Put the names of the members in a hat and draw one.
- Let the youth leader decide.
- 600 people attend a concert. Their seats are numbered from 1 to 600. How would you
choose a random sample of 25?
- A school has 50 teachers. It also has 600 pupils of whom:
300 are aged 11 to 13; 200 are aged l4 to 16; 100 are aged l7 to 19.
You want to find out the views of the pupils and teachers (together) on classical music.
- Give one reason why you might use a stratified sample.
- In a stratified sample of size 65, how many teachers would you interview?
- You are to do a survey on which magazines should be in the school library. You decide to
ask a sample of pupils.
- What do you want to find out?
- Who might use the information you collect?
- Whom will you ask?
- How will you choose your sample?
- What problems may occur in interviewing? How would you overcome them?
- How will you present your results?
Answers
1 |
a |
School cooks |
|
b |
Parliament |
|
c |
Manufacturers |
|
|
|
2 |
a |
Pupils |
|
b |
Scottish people |
|
c |
Housewives |
|
|
|
3 |
a |
It depends on who rushes fastest (or whether there are 'early' dinners
for teams or a rotating system of forms is used). |
|
b |
This is a self-selected sample. |
|
c |
This is too small a sample. |
|
|
|
4 |
|
b is the fairest. In the others, each member does not have equal chances. |
|
|
|
5 |
|
Use a 3-digit random number table. Use nos 1-600, ignore 601-999. |
|
|
|
6 |
a |
To make sure each group is represented proportionately. |
|
b |
5 |
|
|
|
7 |
|
Possible answers are as follows: |
|
a |
Which magazines would be read by pupils if the library stocked them. |
|
b |
The school librarian. |
|
c |
A sample of pupils from the whole school. |
|
d |
A sample stratified according to age (and/or sex). |
|
e |
Pupils absent; try later. Pupils refusing to answer; report this or
perhaps use reserves. Pupils not being aware of possible magazines; give a list of
magazines for a choice to be made. |
|
f |
In a ranked list, or a bar chart or (if appropriate) a pie chart and as a
written report. |
Connections
with Other Published Units from the Project
Other Units
at the Same Level (Level 3)
Car Careers
Multiplying People
Net Catch
Phoney Figures
Cutting it Fine
Units at
Other Levels in the Same or Allied Areas of the Curriculum
Level 1
Being Fair to Ernie
Leisure for Pleasure
Wheels and Meals
Level 2
Authors Anonymous
Opinion Matters
Level 4
Sampling the Census
Smoking and Health
Retail Price Index
This unit is particularly relevant to: Humanities, General Knowledge, Mathematics,
Commerce.
Interconnections
between Concepts and Techniques Used In these Units
These are detailed in the following table. The code number in the left-hand column
refers to the items spelled out in more detail in Chapter 5 of Teaching Statistics
11-16.
An item mentioned under Statistical Prerequisites needs to be covered before
this unit is taught. Units which introduce this idea or technique are listed alongside.
An item mentioned under Idea or Technique Used is not specifically introduced
or necessarily pointed out as such in the unit. There may be one or more specific examples
of a more general concept. No previous experience is necessary with these items before
teaching the unit, but more practice can be obtained before or afterwards by using the
other units listed in the two columns alongside.
An item mentioned under Idea or Technique Introduced occurs specifically in
the unit and, if a technique, there will be specific detailed instruction for carrying it
out. Further practice and reinforcement can be carried out by using the other units listed
alongside.
Code No. |
Statistical Prerequisites |
Introduced in |
1.3g |
Random number tables |
Being Fair to Ernie |
2.1a |
Constructing single variable frequency tables |
Wheels and Meals
Authors Anonymous
Opinion Matters |
2.2a |
Bar charts |
Leisure for Pleasure
Authors Anonymous |
|
Idea or Technique Used |
Introduced in |
Also Used in |
1.2c |
Problems of classification of data |
Wheels and Meals
Opinion Matters
Car Careers
Phoney Figures |
Leisure for Pleasure
Retail Price Index
Authors Anonymous |
5k |
Variability of estimates |
Car Careers
Net Catch |
|
|
Idea or Technique Introduced |
Also Used in |
1.3a |
Sampling from a small well-defined population |
Authors Anonymous
Net Catch |
1.3b |
Sampling from a large population |
Car Careers
Seeing is Believing
Retail Price Index |
1.3d |
More sophisticated sampling techniques |
Smoking and Health |
1.3e |
Finding appropriate data |
Being Fair to Ernie
Cutting it Fine
Car Careers
Smoking and Health
Net Catch |
1.3h |
Biased samples |
Car Careers
Net Catch |
1.4c |
Using own questionnaire |
Opinion Matters
Sampling the Census |
5i |
Estimating population figures from samples |
Car Careers
Retail Price Index
Smoking and Health |
5w |
Large samples are better for inference |
Net Catch |
|