There are things polls can tell us, and things they can’t – so let’s not confuse the two, argues Ivor Gaber
In the run-up to the general election, all eyes are on the opinion polls. They play an important part in any campaign. Don’t believe politicians who say they never pay any attention to them. They all do – obsessively. Good polls lift activists and poor polls depress them. But polls can be confusing in what they tell us –and what they don’t.
Newspapers, particularly local ones, public relations companies with something to sell, campaign groups and even political parties are notorious for putting out phoney polls. Questioning the first 20, 50 or 100 people encountered outside the office or – even worse – selected from an email address book, provides no useful measure of public opinion. The same goes for focus groups, which are good for getting a flavour of what the public is thinking, but no more.
Polls can be conducted face-to-face (increasingly rare), by phone (the most usual method) or online (a growing trend, as it’s quick and cheap). At one time, academic experts and polling companies only regarded face-to-face polls as reliable, but there is now substantial evidence that telephone and online polling can give equally reliable results, if they are properly conducted.
The sample size has to be large enough (1,000 is regarded as the minimum to measure opinion nationally), respondents must be selected to reflect the whole population and the questions must be worded neutrally.
Pollsters then “re-weight” the results – this is what makes polling more an art than a science. Re-weighting involves ensuring that the sample more or less reflects the broad political and social make-up of the country. The number of those who said they voted Labour, Conservative and so on at the previous election should accurately reflect the number who actually did. If, for example, the sample contains too many people who said they voted Labour in 2005, their responses are adjusted downwards to reflect the real result. Thus pollsters can ensure their figures reflect how opinion has changed since 2005.
Polls don’t predict, pollsters do. Polls don’t tell you who is actually going to win an election. They ask: “If there was an election tomorrow, how would you vote?” But there isn’t an election tomorrow and how people might vote in three months might be different from their current intentions
Although people tell pollsters that they definitely intend to vote, many don’t. And people don’t always tell pollsters the truth. It’s not that they deliberately lie, but they sometimes give what they think is the most socially acceptable answer.
Finally, the percentage share of the vote doesn’t decide who gets into Downing Street – only seats in the House of Commons do and there’s a less than perfect fit between the two. The current conventional wisdom is that, because of the distortions of our electoral system, the Tories need a lead of around 10 per cent to ensure a majority of one. But that assumes an even swing to the Conservatives across the country, while the Tories are banking on doing better in their targeted marginal seats. Like so much else, whether they are right will become clear on May 7 – probably.
Ivor Gaber is professor of political campaigning and reporting at City University London. He will be writing regularly about the election campaign over the coming months
About The Author
