During the run up to the 2015 UK General Election the opinion polls were consistently wrong and led many people to believe there would be a hung parliament. The polls in 2015 underestimated support for the Conservative Party. Research companies failed to allow for the greater difficulty in sampling such voters who are generally busier than Labour voters.
Some commentators even suggest that David Cameron only agreed to include a promise to hold a EU referendum on the basis that it would be blocked by his coalition partners as he did not expect the Conservative Party to win an overall majority. This raises the questions of whether opinion polls could also have unwittingly influenced the final result. The media often argue that they only reflect society. In this case they may have been communicating a warped and inaccurate view of public opinion.
The role of mass media?
There is widespread agreement that the mass media in general can influence peoples’ knowledge, attitudes and behaviour. What is debatable is the degree to which this occurs and whether opinion polls in particular can drive changes in voting patterns.
The evidence from research into the impact of opinion polls is mixed. Probably because of the nature of the subject and a reliance on self-reporting. However, a study into the 1998 Canadian election concluded that there was evidence that polls affected voters’ perceptions of the chances of different parties winning and they did influence voter behaviour. The latter was primarily the result of tactical voting where people switched their support away from a party that had little chance of winning and towards a party with a better outlook.
Facebook comments double after exit poll result:
The mass media also feeds into our social networks. There is indisputable evidence to show that social networks influence our behaviour as we are very social creatures. Our herd behaviour is at the heart of the dispersion of ideas and behavioural change.
Herd theory suggests we are more heavily influenced by the actions of others in our network, and we like to be associated with the same values and norms as our social networks. As a result word of mouth can be very influential. We are driven by what we think other people in our social networks are doing or saying.
After the shock exit poll results, comments about the election on Facebook doubled confirming that polls are certainly a talking point. Opinion polls may form another piece of the jigsaw to help us determine what our wider network are intending to do.
Does the accuracy of information matter?
People often prefer to follow their gut instinct rather than listen to data. Maybe it does not matter? But people also hate uncertainty and suppress ambiguity because it slows our thought processes. The concern is that we are heavily influenced by the What You See Is All There Is (WYSIATI) rule. During the campaign the polls consistently told the same story.
There was also a great deal of uncertainty as the Conservative’s played on fear that Labour would have to form a coalition with the SNP to form a Government. If people had known the Conservatives were heading for a slim majority then this would have become a non-story.
In his book The Wisdom of the Crowd, James Surowiecki examined how stock market bubbles arise. He noted that:
He outlined experiments with students conducted by Jack Treynor where they were asked to estimate a quantity of items. In one experiment all participants were given additional information that pointed in one direction. Although the information was true, adding the additional guidance reduced their collective wisdom. It meant their average guess was off by 15% with just 3% in experiments where no extra information was given to participants.
The experiment by Jack Treynor could partly explain why the ‘wisdom poll’ conducted by ICM put the Conservative’s ahead of Labour, but still underestimated the magnitude of their lead. Overall, it seems clear that widely publicised and inaccurate information can skew our personal forecasts of an outcome. This in turn may inevitably change the behaviour of some voters. Whether it be the decision to vote or the party they ultimately voted for.
Given the evidence that opinion polls do influence voter behaviour. There is a strong case to ban polls during the run up to the election. However, some might argue that polls assist people in tactical voting. This may help prevent a party gaining a landslide victory that they might use to abuse once in power.
Featured image: 2015 UK General Election Opinion Poll
Neal is Chief Optimisation Consultant at Conversion Uplift Ltd and a CXL Instructor. He has over 10 years experience of A/B testing and improving conversion rates for major international brands across multiple sectors, including gaming and financial services.