There was some interesting news out of Clontarf this morning: polling company RED C is using a new method to predict the results of General Election 2016, and it’s suggesting something quite different to other polls. Notably, it suggests that the Labour Party could receive 13% of first-preference votes – 5% more than in Sunday’s RED C poll for the Sunday Business Post.
This new predictor uses what is called a ‘Wisdom of Crowds’ model, which was piloted during the 2015 Marriage Equality Referendum. In a post on the RED C website, CEO Richard Colwell describes the process used in the pilot:
‘The process involved asking people, who by then had spent time with family and friends discussing how they would vote, what they thought the outcome would be, rather than how they would vote. This allowed us to identify the fact that people felt the No vote would be higher than voters were telling us in the standard polls.’
This would certainly be one potential explanation for why RED C’s prediction – which estimated a significantly lower margin of victory for the Yes side than every other poll – was so accurate.
The ‘Shy Voter’ Phenomemon
For those who are interested in psephology, this points up to a possible solution to a long-standing polling conundrum: opinion polls might be able to broadly predict the outcome of a vote (Yes versus No, biggest party, viable coalitions, etc.) but they are rarely successful in predicting results (margin of victory, exact seat numbers, etc.).
One possible source of this inaccuracy is what is called the ‘Shy Voter’ phenomenon. Particularly prevalent in discussions about British elections and the so-called ‘Shy Tory’, this simply refers to a voter who will not disclose their true voting intentions for fear of stigma or shame. Commentators have attributed the surprisingly strong Tory performances in 1992, 2010, and especially 2015 to this phenomenon.
RED C tells us that the same experience can be found in Irish cases, particularly ‘Shy Fianna Fáilers’ in the 2011 General Election and ‘Shy No voters’ in the 2015 Marriage Referendum. Due to the perceived social stigma generated by both the unpopularity of the Fianna Fáil government in 2011 and the large coalition built around the Yes side in 2015, it is believed that people with those voting intentions underreported their genuine views, thus explaining their ‘overperformance’ in the actual voting relative to opinion polls.
The Wisdom of Crowds?
Although RED C’s predictor is certainly an innovation in Irish polling, neither the ‘Shy Voter’ nor the ‘Wisdom of Crowds’ are new ideas. The latter was referenced in work as early as Aristotle’s Politics and has been influential in cognitive science, legal science (particularly analysis of jury decision-making), and marketing. Crowd-curated information websites such as Wikipedia also operate from the principle that the opinions of the many are more accurate than those of the few.
However, not everyone is convinced by the Wisdom of Crowds argument. Alternative explanations for all of these cases can be found and put forward, such as the ‘Lazy Labour’ theory for Britain’s 2015 election. It also does not necessarily account for the evidence that suggests polls do not just reflect voting intention, they influence it – meaning that the succession of polls indicating low support for Labour might well reduce its vote on the day.
Despite the criticism, the Wisdom of Crowds is still a very intuitive model; it ‘feels right’ to say that there are more people supporting the incumbent parties than the polls indicate. Ultimately, we will have to wait until the weekend’s count to determine if the Wisdom of Crowds has, once again, seemingly beaten the traditional polls. In particular, Labour will be watching closely to see if its haul is close to the 13% prediction – 5% higher than in the most recent Sunday Business Post/RED C poll, more than twice the 5% predicted in the Irish Times/Ipsos MRBI poll, and three times the 4% predicted by the Sunday Times/Behaviour & Attitudes poll.