Government factors that influence election outcomes and voter participation
Wisdom of the Crowd Vs. The Pollster
1. Wisdom of the Crowd versus Pollster
Pollsters are responsible for reporting findings based on a calculated process of gathering
results among a representative sample on a given topic. For the most part, where the
question is not contentious, we can be confident of the result within a particular margin of
error by ensuring the sample is representative and robust. So for example if we ask 1,000
people who have been selected to represent the total population accurately what their
favourite colour is, we can be confident the result will be accurate to within +/- 3.2%.
However what if the question involved reveals what a person may deem to be a socially
‘undesirable’ response? While some people are proud of their stance, and have no issues
about an attitude or belief that perhaps goes against the masses, this is not the case for all
– those who are less inclined to reveal their attitudes are referred to in polling terms as ‘shy
voters’.
It is not to say that ‘shy voters’ lie in their response to interviewers, however they do have
some difficulty in revealing to a stranger their disposition on a particular topic. As a result
they generally suggest they have not yet decided how they will vote, but some may claim
they will vote along the same lines as the masses.
The learning for researchers is that if you find yourselves with a high level of ‘Don’t Know or
Undecided’ respondents to your survey, it should be an immediate red flag that some level
of “shy voting” is at play.
So, here in lies the difficulty; if the very source of truth we rely on cannot be deemed to be
100% accurate how does a pollster overcome this to predict an accurate outcome?
In advance of the recent Marriage referendum all pollsters, including RED C, were reporting
high levels of support for the yes side, however there were concerns that with such a
contentious topic that those who supported the No side would be underrepresented in the
polls, either because they would be far more likely to tell us that they didn’t know how they
would vote or because they simply refused to take part on the poll at all. This was clearly
seen to be the case when we look at the final “standard” polls vs. the result, where the No
side were clearly underrepresented.
2. Including Undecided
Excluding
Undecided Sample Dates
Yes No Undecided Yes No
RED C 69% 25% 6% 69%* 31%* 1,009 11
th
– 13
th
May
Millward
Brown
53% 24% 23% 69% 31% 994 2
nd
– 15
th
May
Behaviour &
Attitudes
63% 26% 11% 71% 29% 927 1
st
– 11
th
May
Ipsos MRBI 58% 25% 17% 70% 30% 1,200 13
th
– 14
th
May
*(RED C did not re-allocate D/Ks, but suggested they would all vote NO)
Having witnessed the significant effect the ‘shy voter’ had recently on pollsters in the UK,
with actual outcome of UK elections versus poll outcomes quite different it was clear this
group couldn’t be ignored and needed some other technique to give a voice to the shy
voter.
RED C decided to use a technique known as ‘Wisdom of the Crowds’ to uncover the shy
voter. This method asks the respondent to consider all the conversations they may have
had, or heard friends and family having and on this basis to try and predict what they
believe the percentage outcome will be for and against the marriage referendum.
This removes any discomfort from the situation in which a shy voter may find themselves
and instead put the focus on the general population as opposed to one individual. The
result was a resounding success, as it predicted the outcome with 100% accuracy, 62% in
support and 38% against.
3. This “Wisdom of Crowds” method allowed the shy voter, whether in support of or against
the marriage referendum to lend some level of their own viewpoint, while disguised within
how the ‘masses’ will vote. As such it bridged the gap between standard polling and a
potential issue of the shy voter influencing the results.
So how does this transfer to general insights we gather for commercial clients or regular
political polling? The first thing we have to do is to always be aware of the topics we are
researching and the way that people may respond to them, and always ask are we doing
enough to get behind the topline stock answers we receive. For instance when we ask
people how likely they are to buy a product or service this can often be oversubscribed,
particularly among the genial Irish, who are likely to say they “might” buy something, even
if in reality they wont go on to do so. So we should also always be asking how likely they
think people in the population generally will be to buy that service or product as well, and
where large gaps exist take this into account in our analysis.
Within party support this also applies, and RED C use several techniques over and above
the standard questions to get the most accurate result possible – all parties are read out,
unlikely voters are removed, ‘undecided’ voters re-allocated and finally we look at past vote
behaviour and layer it with the current claimed intention to ensure an accurate sample.
This serves not to alter results, but to allow us report the most accurate results possible.
4. But maybe even this wont be enough? With the Irish elections looming we may also need
to look further at techniques such as the Wisdom of Crowds on party support to try to give
the ever present shy voter a voice.
It is not to suggest that standard polling is defunct, quite the contrary, it simply serves to
inform us as pollsters that although research can be quite systematic and indeed scientific,
we cannot forget the human in all of us and how best to overcome the nature of the
reluctant respondent.
Ciara Regan is an Associate Director with RED C Research.