2. Agenda
1. Some questions answered about polling
2. The RED C Approach to getting polls right
3. Why do different polls give different results?
4. Current party support
5. What the future holds
5. How many people here
today have been surveyed
for an opinion poll?
6. How come I am never called for a poll?
Over
4 Million
people!
Only about 15
national polls are
published each year
Approx. 60,000
people contacted and
15,000 people
interviewed
Less than 1% of the
population.
7. How can 1000 people represent the whole
population?
“If you don't
believe in random
sampling….
…next time you
go for a blood test
ask them to take it
all!”
8. What is the possible error of opinion polls?
AIMRO guidelines =
minimum sample of 1000
interviews for a national poll.
Sample error of just + or
-3%
Local or regional polls = 500
interviews.
This does increase the
possible error to + or – 4.5%
9. Does it matter when a poll was done?
Even bad publicity can give parties a boost!
Events can have a dramatic impact on
poll results.
10. Do people not just lie to pollsters?
1. Why bother? Easier to tell
the truth
2. If people did often lie,
opinion polls would never
get their predictions correct.
3. Code of conduct by proper
industry players means
results are collected on an
unattributable basis, so
there should be no reason
to lie.
11. How were the interviews conducted?
The key here is that the poll is
conducted outbound by the
company – not a phone-in poll.
These types of polls are
normally conducted face to face
or by phone.
Online polls should still be
treated with caution in Ireland –
over 55’s not represented.
13. RED C Method for Getting Polls Right
RED C conducts polls by phone,
because we believe they are
more accurate.
Telephone interviewer has no
control over who they interview.
This makes respondent selection
truly random as the interviewer
doesn’t know if the person is in
an apartment or country house!
Provides a base sample that is
more representative.
14. How do we make sure the sample is accurate?
1. Random Digit Dialling
Initial random sample of directories nationwide
– then +1 or +3 added to number – ensures
ex-directory households included
2. Mobile sample
25% of the Irish Population now mobile
only
Half sample with landlines and half with
mobiles – reach is 98% of the population
3. Quotas
Quotas set based on census data for gender,
age, region and social class to ensure the
sample is representative.
15. How important are the questions that are asked?
Question wording and order of
questions are vital
RED C always ask voting
intention first, as other questions
such as party leader ratings may
bias results.
On the phone it is also important
to prompt for ALL parties.
16. RED C techniques employed to maximise accuracy
Likelihood to Vote
Not everyone votes –
turnout 67% at the last
election….
…this needs to be taken
account of in the analysis
Only look at results by
those likely to vote (on a ten
point scale exclude those
who say they are 1-3)
Turnout GE 2007
67%
17. RED C techniques employed to maximise accuracy
Weighting by past vote
Demographics do not necessarily
give us a 100% accurate sample
Sometimes recall of how people
voted last time is different to the
result.
Part faulty recall ….BUT part
suggesting a biased sample
RED C compares the declared past
votes to the actual result of the last
general election and weights to the
mid-point of the two.
18. The benefits of Regular Tracking
Statistical theories prove that, in a random poll of 1,000, 19
times out of 20 a poll will be accurate to within 3 per cent.
But this does leave the one in 20 chance of the result being
outside this margin of error – a Rogue Poll
It is sensible to weigh up the results of several different polls
when calculating the most likely outcome.
19. 33 33
67 67
How accurate are opinion polls then?
RED C had four of the five most
accurate pre-election polls in
2007.
Final GE 2007 poll had an
average sample error of just
1.4%.
2009 local elections saw an
average error of just 2.1% - as
accurate as the exit poll.
The Lisbon Treaty result was
predicted with 100% accuracy.
Yes – in favour
No - against
Last
RED C
Poll
%
Final
Result
%
Lisbon Treaty Sep 2009
20. Why are do polls from different companies show
different results?
Volatile nature of electorate
Events that happen during or
before polls.
Different method – phone vs.
face to face
Different question wording &
order.
Analysis techniques – likely
voters & past vote weighting
A Rogue Poll!
Normally they
don’t!
1.Last general election all
were within average 1-2%
error of the final result
2.Biggest gap for any party
across all the pollsters was
just 2%
BUT POSSIBLY…..
21. Fianna Fáil
Fine Gael
Labour
Sinn Féin
Green Party
PDs
Independent
Current trends in 1st
preference vote intention
(Base: All adults likely to vote 18+)
31%
24%
23%
10%
3%
9%
Banking
Crisis
Starts
Cowen
FF Leader
TD’s Pay
Increase
+ Tribunal
Economy
Worsens
Lisbon
Treaty 2
Budget
2009
FG
Heave
Local
Elections
22. What can we expect in the run up to the election?
1. The electorate are volatile – 50% of those still undecided voted
Fianna Fail at the last election
2. The ability to handle the economy is the most important aspect for
voters in the run up to next election, BUT…
… no major party currently has that support – only 30% of the
electorate believe Fine Gael/Labour coalition can handle the
economy despite joint 1st
preference of 56%!
3. Current support for Fine Gael and Labour may not be built on very
strong foundations – however the desire to punish Fianna Fail
remains.
4. 1 in 5 voters only made up their mind in the last week of the election
campaign in 2007.
…plenty of room for more changes
before an election is held