When opinion polls go bad

by Stephen Tall on June 25, 2007

For the last few months, no opinion poll of voting intentions worthy of the name has been complete without the inevitable tagged-on question, ‘And how would you vote if Gordon Brown were Labour leader?’ To which the answer of Labour-inclined voters has invariably been: less likely than if Tony Blair were still Labour leader.

Anthony Wells, at the indispensable Polling Report website, has recorded the figures of the pollsters’ beloved anticipatory question. None of the 26 polls conducted since David Cameron became Tory leader have projected a Labour lead on the presumption that Mr Brown succeeds Mr Blair. The average Tory lead in the nine polls conducted in 2007 would suggest Mr Cameron should be some 10.5% ahead the moment Mr Brown formally moves into No. 10.

This has led to all sorts of silly speculation in the media that Mr Brown is a vote loser for Labour. And it has sparked an amusingly unmerited degree of triumphalism among Tories, who somehow imagine they will cruise to a general election victory merely by virtue of having elected a leader who manages, in contrast to most of his colleagues, to look and sound relatively normal.

Yesterday’s opinion poll in The Observer, showing a 3% Labour lead just as Mr Brown accedes to the premiership, has poured a bucket of cold water over the heads of those who, rather foolishly, had taken too much to heart the pollsters’ misleading findings.

As opinion pollsters will be the first to point out, their findings present only a snapshot of public opinion. Moreover, the question they ask to solicit the public’s views is a hypothetical one: how would you vote if there were a general election tomorrow.

Those who are asked the question (1) know there isn’t actually a general election tomorrow; and (2) have not been subjected to a month-long election campaign which will, in many cases, alter their voting intentions before they finally cast the only ballot that actually counts. This inevitably distorts the findings in quite unquantifiable ways.

To add, as the newspapers who commission the polls insist the pollsters do, an extra layer of hypothesis – how do you believe you will vote in an imaginary general election when a man who is not yet Prime Minister becomes Prime Minister – is a step too far.

The public which answers this question knows full well that what they are really being asked is: what do you think of Gordon Brown? And those who are sceptical about him will take the pain-free opportunity to voice their objections in their answer. This, too, distorts the findings in quite unquantifiable ways.

Finally, you must factor in the statistical fact that opinions polls can only predict with 95% probability to within a margin of error of +/-3% the headline figures which the newspapers will brandish on their front pages with cast-iron certainty. So the Observer poll which I’ve used as evidence in this article may just as well point to a Tory lead of 3% as it does a Labour lead of 3%. Or it could be one of the 5% of opinion polls which is complete bollocks outside the margin of error. You pays your money, you takes your choice…

Like many observers of the British political scene, I expect Mr Brown to enjoy a poll bounce over the summer, and that Labour will continue to be tied with, or build a small lead over, the Tories. How long this lasts will, rightly, depend on the success or failure of Mr Brown’s government.

In part, this will be because Labour has, contrary to all the pundits’ predictions (including my own), accomplished a bloodless transfer of power. It will also be because the media, believing their own rogue polls, and the Tories, believing their own anti-Brown propaganda, have so far lowered expectations of our PM-to-be that he cannot help but clear the low hurdles he has been set.

Indeed, so successful has this ‘dampening down’ strategy been, it’s almost tempting to believe that Mr Brown, a man of native political cunning, may secretly have commissioned some of those polls himself.