Let me say straight away that I’m not talking about wriggly things in the garden that are doing good things for soil quality. I’m referring to the measure of voter approval that is sometimes displayed during election debates.

In the last UK general election we had a television debate amongst the three main party leaders for the first time. As has happened elsewhere, the broadcasters added three lines on the screen, one for each candidate, and plotted a ‘sample’ of voters satisfaction with the points being made by the politicians in real time.

You might think that in something as significant as the leadership of one’s country people would be keen to make their own informed opinion. But as has been shown in numerous other studies, we can’t help but be influenced by the knowledge of what others are doing or what we believe they’re thinking.

This was demonstrated in the case of the TV election debate worm by a team of researchers from two British Universities. They invited a group of students to watch the TV debate and placed their own ‘worm’ on the screen. Theirs was not being produced from a sample of voters, but was created by the researchers to favour one candidate over the others.

Two versions were used, one reflected favourably on Gordon Brown, the other on Nick Clegg.

The results showed an amazing capacity to influence people’s perceptions. In the debate where the results were biased towards Nick Clegg 79% of participants thought that he had won the debate (ahead of Brown on 9% and David Cameron just 4%).

When the results were biased towards Gordon Brown, he emerged as the winner of the debate (47%), with Clegg receiving 35% and Cameron 13%.

This influence raises a number of very important issues about the way in which the media influences the electorate. Firstly, the sample used to provide the responses for worms are often very small (sometimes as low as 12 according to the research). Often they are of people who declare themselves to be undecided about how they will vote. In neither case can it be argued that the views collected are statistically representative of the wider population.

Even if a representative sample were to be drawn, I would argue that the election itself is the moment when people should find out what everyone else thinks, rather than letting notoriously unreliable opinion polls (particularly those conducted more than a day or two prior to an election) or other media-feeding research tools (like worms) shape perceptions.

And before anyone claims election polls are now very accurate, they’re not. Just last week the Scottish electorate embarrassed the pollsters again: YouGov, who had trumpeted themselves as being the most accuate pollsters in the 2003 Scottish Elections, conducted a poll that ended on May 4 (the elections were on May 5) and found that the SNP would get 42% of the constituency vote and 31% of the regional vote. They concluded that the result would leave the SNP “well short of an overall majority”. The SNP ended up with a majority of nine in the 129 seat parliament.

In fact, the SNP obtained 45% of the constituency vote and 44% of the regional vote

Whilst people are rightly suspicious of opinion polls, it’s hard not to be swayed when a number of different polls appear to be showing more or less the same picture, and when journalists implicitly treat the data from them as factual.

I would argue that the path of democracy is damaged further by the use of opinion polls to inform policy. Research is self-influencing: in Consumer.ology I reference a number of examples where the same issue is researched independently and, because of the influence of the questions, diametrically opposed results are obtained. Frequently there is no answer that can be extracted reliably by asking questions. Yes, you will get an answer, but only one that is a by-product of having asked the question and the way you have gone about doing so.

Going back to the worm, which is essentially asking only one question, you would almost certainly obtain a different outcome if you asked people to track how dissatisfied they were with the candidates, or how reasonable, or how suitable for the post they are competing for. Even if the worm was drawn from a representative sample, how are we to know that momentary ‘satisfaction’ with a politician is the measure that will correlate with a voting decision in the forthcoming election?

Source: Davis CJ, Bowers JS, Memon A (2011) Social Influence in Televised Election Debates: A Potential Distortion of Democracy. PLoS ONE 6(3): e18154. doi:10.1371/journal.pone.0018154

Image courtesy: Theresa Thompson