Polls can shape reality, not just reflect it

The pollsters and the media have to make decisions based on public opinion – but those decisions can then shape us.

Sign Up

Get the New Statesman's Morning Call email.

For daily news, analysis and predictions on next year's election, follow May2015.com.

An experiment this weekend by ComRes has supported what many pollsters had already found: including Ukip alongside the three traditional major parties increases their support.

The pollster ran a second poll alongside their typical monthly survey for the Independent. This included Ukip when the pollster asked respondents how they would vote in a general election tomorrow. Usually ComRes, like six of the other seven major British pollsters, doesn’t do this.

Only Survation do, and their most recent poll, published last weekend in the Mail on Sunday, put Ukip on 25% – the highest figure any poll has registered for them so far.

This encouraged ComRes to test the impact of “prompting” Ukip. Their results uncovered a 5 point difference: when unprompted, and left as part of the “other” parties which voters had to specifically select, Ukip polled 19 per cent. When prompted, they managed 24 per cent.

As YouGov’s Anthony Wells discussed yesterday, this doesn’t mean prompting for Ukip inevitably leads to a 5 point spike in their support. 1,000-person polls have a 3 per cent margin of error, which could account for some of this effect. We would need ComRes, or other pollsters, to run this experiment repeatedly before we could more precisely quantify the effect of prompting for Ukip.

But the finding reinforces what Survation’s surveys have shown: doing so does create some spike in support for the party.

There is a reason other pollsters don’t do this. They have found prompting for smaller parties seems to overstate their support: smaller parties do less well at elections than polls that prompted for them would have suggested.

When voters are presented with a party they may not have considered, like Ukip or the Greens, they are more likely to choose them.

When voters are presented with a party they may not have considered, like Ukip or the Greens, they are more likely to choose them. But when they are in a ballot booth, and these smaller parties are muddled with many other candidates and parties, their support for these parties has faded.

That does not mean Ukip will always be left unprompted by most pollsters. The Lib Dems were once also excluded, but not prompting for them was eventually found to underestimate their support. All pollsters prompted for Ukip ahead of the European elections because the party’s support seemed likely to be too great to exclude.

That judgement can't be made according to anything as simple as a threshold of polling support: it is a decision made across a number of factors. One of them is the classifications given to each party by the BBC and Ofcom, which we were given an indication of by the recent leaders’ debates proposals.

But these decisions are circular. The broadcasters will have included Farage partly because of the polls – but the pollsters may be more likely to prompt for Ukip if he is in the debates.

To what extent do polls reflect or determine reality? Getting decisions like whether to prompt for Ukip right are key if pollsters are simply to report on public opinion rather than shape it.

These decisions are circular. The broadcasters will have included Farage partly because of the polls – but the pollsters may be more likely to prompt for Ukip if he is in the debates.

It is the same challenge faced by the broadcasters. Nick Clegg is thought to have partly won so much support during the debates in 2010 because he was on a stage with Brown and Cameron. By being given a podium he was presented as an equal and viable alternative.

Like the pollsters, the broadcasters not only have to consider whether putting Farage a stage will inflate Ukip support, but now, after some public outcry and much dithering from Cameron and Miliband, whether including Natalie Bennett would hand the Greens undue popularity.

The media – through the debates they run or the polls they report on – cannot always remain impartial.

Lord Ashcroft is now in a similar position thanks to his marginal polls. While some of the parties have internal polling, he is, as we detailed on Friday, the only pollster with the funds to publicly show how the parties are faring in Britain's battlegrounds.

If a poll of his shows an unexpectedly close or distant race, it could shape how much attention a party pays to the seat. Indeed, if his and Survation's polls in Heywood & Middleton had shown a closer race, Ukip may have plunged far more resources into the seat; or they could have been persuaded by our forecast, which used past by-elections to suggest the race would be within 6 or 7, rather than 20, points (Labour ended up winning by just over 2).

In the end Ukip lost by 617 voters and won Clacton by more than 12,000. If Nigel Farage had spent his final campaigning days north of Manchester, rather than east of London, Ukip might now have two MPs.

Explore May2015.com.

Harry Lambert was the editor of May2015, the New Statesman's election website.