90 per cent of the press coverage pollsters get comes in these few weeks – so they’re anxious to make the most of it
You may already be sick of the general election campaign but for Britain’s pollsters, Christmas has come early.
Now is the time when their work – questioning the nation on their voting intentions, preferences and prejudices via phone calls, online polls and focus groups – is at the heart of almost every TV bulletin and political interview.
“This is about 90 per cent of the press coverage our industry gets,” says Ben Page, chief executive of Ipsos Mori, whose firm also provides the most dramatic moment of election night itself.
When the clocks strike 10pm on December 12 and polling stations around the country close their doors, Huw Edwards, Tom Bradby and Dermot Murnaghan will begin their night of coverage by delivering the result of the exit poll conducted for the BBC, ITV and Sky News.
It is the most expensive and largest survey that happens during the election – taking the views of up to 80,000 voters outside 130 polling stations – and costs hundreds of thousands of pounds.
The political triangle
In the run-up to the big day, market research companies will be undertaking polls and for media organisations, campaign groups and the parties themselves.
According to Page, the questions focus on three areas. “There’s a sort of political triangle – the policies, the parties and the leaders.” For the past 40 years, polling on the party that is seen as best for the economy and the leader regarded as the best future prime minister has been as telling as surveys on actual voter intention.
General election polls 2019: here’s what the latest UK opinion polling says – and why they’re not always right
A decent sample size is usually regarded as 1,000 people – which would give a margin of error of plus or minus 3 per cent – but pollsters must also secure a representative spread of voters across age, gender, class, race and region.
Volunteers – who are usually rewarded with cash or vouchers – can be recruited by buying the data of individuals who have opted to be contacted, knocking on doors, asking people in the street or phoning landlines and mobiles at random.
Once questions have been answered, data is “weighted” to correct for any imbalances in the sample – for example if 6 per cent of the UK is aged 20 to 24, but only 2 per cent of those surveyed happen to be.
Polling firms are also commissioned, often by parties, to undertake focus groups of eight to ten people in a room.
“They are not designed to be statistically accurate,” says Page. “They tell us not how many people think things; they tell us why people think things. We recruit focus groups literally in the street and door-to-door. So if you’re doing one in a constituency and you want wealthier people, you would go to areas of large, semi-detached houses. If you want to interview poorer people, you’ll probably be recruiting in council estates and in markets.”
He adds: “Things like [the Tories’ slogan] ‘Get Brexit Done’ come straight out of a focus group. It resonates and the people moderating see that it’s something that the rest of the group pick up. A lot of these lines will be tested.”
Focus groups also provide a rare chance for both sides of the Brexit divide to engage in a calm and measured conversation.
An unwarranted kicking
Chris Hopkins, head of political research at ComRes, tells i: “If you want an active discussion between leavers and remainers, then that is possible, but it has to be facilitated in a sensitive way and there is the potential for that to get quite heated. The flipside is that just having leavers in a room and just having remainers in a room means that you can get a bit of groupthink – and they’re less likely to be able or willing to listen to the opposing view, so you don’t always get a high-quality discussion or any form of debate.”
The industry has come in for much criticism in recent years. Most pollsters failed to predict a Tory majority in the 2015 election and then forecast one in 2017, when we ended up with a hung parliament. Most polls also put remain ahead in advance of the 2016 EU referendum, though Hopkins says “the final result was well within the margin of error for a lot of the final polls”.
According to polling by Ipsos Mori, only 54 per cent of us trust the polling professionals to tell the truth – less than for “the ordinary man or woman in the street”.
“Pollsters have come in for a slightly unwarranted kicking,” says Hopkins. “One thing that we need to be careful about is spending too much time over-correcting for what happened last time and not enough effort concentrated on what is going to happen this time.”
The main problem came down to the companies’ anticipation of turnout – and Page says they are constantly updating their calculations.
When trying to get your head around a poll’s results, it is wise to first check whether it was conducted by a legitimate market research company, who will have carefully sought a representative sample and – if a member of the British Polling Council – will have had to post its data tables and methodology online.
And when a single outlying poll offers up a spectacular outcome, it is also worth remembering Twyman’s Law, named after media-research analyst Tony Twyman: “If a statistic looks interesting or unusual, it is probably wrong.”