Q: What does polling actually entail?

A: Polling is simply that branch of market research concerned with testing people’s opinions and attitudes towards a part or whole of the political process. It used to be done using face to face interviewing, but to save cost and time it has been replaced by telephone and, latterly, online interviewing. The process of conducting a political poll varies from company to company but we pollsters all set out to model the prospective voting behaviour of people by measuring propensity to vote, how people will vote who are likely to do so, and then try to account for people who are reluctant to say whether or how they would vote. There is also a knack to weighting the sample to ensure it is sufficiently representative of the voting public. When you see a poll published in a newspaper it has (or should have) there- fore gone through several stages of weighting to be demographically and politically representative, and the ‘don’t knows’ will have been removed.


Q: How has polling changed over the years?

A: The art of polling has changed as a result of technology and changing voter behaviour. The media wants its polling done as quickly and cheaply as possible, which has accelerated the adoption of new technology. So, as telephone use increased, so did its use as a cheaper alternative to face to face interviews. We have more recently had to adapt to the wider use of mobile telephones and the internet has again helped with both speed and cost. Aside from political polling, these methods lend themselves more to some uses than others. Voters have changed too and at any one time there is likely to be differential willingness among various voter groups to admit their voting intentions. This used to be the ‘shy Tory’ phenomenon but Tony Blair post-Iraq and Gordon Brown changed all that.


Q: How accurate can polling be?

A: When done properly, it is extremely accurate. Assuming the absence of other methodological factors, on a sample size of 1,000 people, 19 out of 20 polls will be accurate to within about 3% points of the ‘true’ figure. In practice, every polling firm wants to be the most accurate and so we all review our methodologies regularly to give ourselves the best chances of success. This is underlined by the fact that we tend to be remembered more for our failures than our wins. For instance, most polling companies over-estimated the Liberal Democrat vote share in the 2010 General Election but the accuracy for the two main parties’ vote shares was exceptionally good. That accuracy doesn’t stick in people’s minds for long though.


Q: How important is polling to the democratic process?

A: Politicians take it extremely seriously. When the polls are against them, they revert to a well-rehearsed script: “On the doorstep people are telling me that we’re doing much better than the polls indicate”. But we know full well they are relying on their own polling data to help steer their campaigns! Political parties themselves are heavy users of private opinion polls, notwithstanding that the media is stuffed full of the things. Parties use them to test their popularity and help hone and target their campaign themes and messages. Whether or not polls are helpful to the democratic process is another matter. Every few years someone raises the old chestnut of whether their publication should be banned in the run-up to elections. Aside from the fact that the internet renders such a ban obsolete, it would likely lead to bookmakers and perhaps investment banks or wealthy individuals commissioning their own polls and trading off the back of the results. A ban would put a premium on knowledge, but it would be bad for the democratic process to allow some groups access to such information and not others.


Andrew Hawkins is the Chief Executive of ComRes