What does it mean to 'win' a debate?

May 16, 2011, 3:27 AM GMT+0

What does it mean to ‘win’ a TV debate? A plain numerical winner, which the polls attempt to get, will be highly affected by who is the most popular candidate beforehand. The 'score' reflects not only the performance but the bias of the committed supporters. Another way of looking at it is how each leader performs relative to expectations of their performance. Even better, we might judge the winner as the leader who convinces the most people to change their opinion.

Another question pollsters face is to what concept they should weight their sample; are we trying to represent the TV audience, or the whole nation? The nation may seem more relevant, but the whole nation did not watch the debate, so could not have been affected in the same way as those who did. YouGov and ICM both weighted their respondents to reflect the debate audience (the makeup of which we derived from other polls beforehand), whereas ComRes weighted their responses to reflect the general public. There is no ‘correct’ approach for this. One might even have recruited a panel of non-partisan ‘citizen debate judges’, for example.

Overall, the second leaders’ debate provided a fascinating evening. YouGov were first out with our results, and incurred the usual flood of complaints and conspiracy theories. Along with our unparalleled record of accuracy, we also have the impressive accolade of being accused of bias, deception, and duplicity by Labour, Conservative and Liberal Democrat supporters within a short space of time. LibDem supporters accusing us now of fixing our results are the same LibDem supporters that were quoting our polls when they showed their surge in support in the previous week. And this morning, even as the attacks on one poll mounted, the Liberal Democrat high command was on the the BBC Today programme extolling the virtues of our other polls, which show that 49% of voters would vote Lib Dem if they thought that the party had a “significant chance of winning the election”.

It is probably worth clearing up a number of issues that have emerged from the more enthusiastic conspiracy-theory seekers. In the interests of full disclosure (and for those who are interested in such things), our political polling is run by long-standing and publicly committed Labour supporter Peter Kellner, while his team includes a member of the Liberal Democrats, a Conservative Councillor, and a Young Fabian. Nadhim Zahawi, currently the Conservative PPC for Stratford-upon-Avon, was YouGov CEO until February 22nd of this year. He remains a non-executive director until May 6th. I myself ran as a Conservative PPC in 1997. Of course we all love politics. To say that pollsters should have no political interests – as one commentator did – is strange. To suggest that those interests would make us distort our results and hurt our company is plain barmy.

We are aware that the coverage polling receives during an election campaign can excite the passions of those who do not like it when the results are not going their way. We have tried to be as transparent as possible about our polling activities: last week we explained how political message-testing works, and we always publish the full results of our media partner’s questions on our voting intention surveys, regardless of whether our media partners choose to publish them themselves.