OPINION16 April 2010
This year’s election is a tough one to call, and it looks like the role of research in trying to predict and understand the outcome will be more prominent than ever.
The introduction of TV debates, the first of which took place last night, provides a hook for yet more polling and analysis, courtesy of YouGov, ComRes, FizzBack, Populus and ICM among others. Ipsos Mori’s Ben Page provided analysis on the BBC, and Conquest is due to reveal results on ITN tonight using a new technique to analyse which candidate has the potential to generate the most buzz among the electorate.
The first of the three planned debates has shown how quickly opinion polls can respond. YouGov claim to have done “the fastest representative survey ever conducted in Britain”, with results out within a few minutes of the end of the debate, while Ipsos tracked the responses of a panel watching the event live. In a world where you can exchange opinions on the internet in real-time, viewers and broadcasters are coming to expect this sort of turnaround from pollsters. We are often reminded that polls are only a snapshot in time, so the opportunity for them to be a snapshot from ten minutes ago rather than two days ago, can only help.
Social media data provides countless more opportunities for measurement and analysis. Voters are interacting with candidates directly online, and the blogs and social networks are responding fast to events, pulling apart the claims and counter-claims, and defacing the posters that campaign teams have spent so much care putting together. One of the companies seeking to make sense of all this data is Tweetminster – not a research agency but a website that tracks UK political happenings on Twitter. It’s running an experiment to see if it can predict the outcome based on what people are saying on Twitter. Having never done polling before, Tweetminster can afford to have a go without risk of losing face.
Meanwhile numerous websites are providing data to voters and mashing it up in useful ways, resulting in an electorate that is more engaged with the numbers and how they are crunched. No need to check the polls when you can go straight to the poll of polls. For the basics on your local representative, check out Theyworkforyou.com. Unsure who to vote for? Voteforpolicies.org.uk and whoshouldivotefor.com will tell you, based on the policies you support. And to find out just how much (or in most cases, how little) that vote is worth, head to Voterpower.org.uk.
How all this affects research depends on how close the pollsters get to the final result, but also on how results are used and discussed by the media and the public. On the first point, we won’t know until May 6th. On the second, we can already see signs of how things are going.
For example, figures from last night’s “Who won the debate?” polls have been misinterpreted by some as being comparable to voting intention polls – leading some commentators to overstate the rise in support for the Lib Dems. The Telegraph says senior Tories have complained that YouGov’s online polls are biased in favour of Labour, which is interesting given that just two years ago it was Labour complaining that the pollster was biased in favour of the Tories (a complaint that was abandoned when the Tories won). And yesterday The Times used a single focus group as the basis for an article on voters’ feelings, extrapolating the changing opinions of the nine people in the group to show how the same swing would affect the final result. Such a sample, the writer noted, “is too small to be representative”.
Lyndsay Peck from Engage Research is among those to have warned of the election’s potential to undermine confidence in consumer research. The research industry needs to make clear the difference between the raw results of opinion polls as published in the press, and the deeper understanding that comes from a relationship between a research agency and its client, she argues.
On a positive note, there is evidence of researchers being appreciated for their nouse as well as their numbers. Peter Kellner of YouGov and Ben Page of Ipsos are among the most prominent commentators on the election.
Last night, Page was commenting with the help of three ‘worms’ – red, blue and yellow lines at the bottom of the screen that wiggled up and down depending on the responses of a panel of viewers with handheld devices. The interesting thing was that Page’s analysis was far more informative than the squiggly lines, which really only there to give the sense of a factual basis for Page’s views. Quantitative data often finds itself exalted at election time, but the lesson of Page’s appearance is that the value of research is in analysis and insight provided by people, not just in the numbers.