There was an interesting presentation at the offices of ComRes yesterday by Rob Hayward, the polling expert and former Conservative MP. It was looking at prospects for the General Election this year. Mr Hayward’s conclusion that the result was uncertain was unremarkable but within that overall judgement were a number of fascinating insights.
Most startling was an analysis of opinion polls undertaken for elections which took place last year. They tended to understate the Conservative vote and to overstate the Labour vote.
There were ten by-elections last year, for which there were 10 opinion polls undertaken. On average the Conservative vote was understated by 1.8 per cent. The Labour vote overstated by 3.7 per cent.
For the Euro Elections there were six polls. On average the Conservative vote was understated by 2.2 per cent. The Labour vote overstated by 2.0 per cent.
The council elections saw the Conservative vote overstated in the average of polls which had them at 32 per cent. The Rallings and Thrasher National Equivalent Vote measured the actual result at 30 per cent. However the polls overstated Labour by more five points – 36 per cent against the 31 per cent vote share that Labour actually obtained.
The pollsters can point out that those results are within the margin of error. But in a tight race those four or five points could make the difference between victory and defeat.
The pollsters could also reply that they offer snapshots not predictions. Again a fair point. But then why would there always be a late falling away in support for Labour in the final few days?
This is familiar territory for Mr Hayward. The Sunday Times of April 12th 1992 reflected on the Conservative victory which had just taken place in the General Election, in contradiction to the polls:
“They (John Major and others) were influenced by a paper sent a week ago by Rob Hayward, the Commons in-house Tory polling buff. The paper compared opinion polls with actual results for the 1987 General Election and subsequent by-elections. He found a clear trend. Both the Tories and Lib Dems did better in actual elections than the polls. Labour performed worse.”
Pollsters concluded that there were “shy Tories” and sought to improve their methods to allow for this. Mr Hayward’s analysis might suggest that the “shy Tories” are still with us – or have returned.
There could be other explanations. For instance that Labour supporters are less likely to vote. Apparently men especially tend to claim to pollsters to be emphatic than they really are. Huge numbers who insist they are “certain to vote” end up not bothering.
Then we have the Green Party having their vote understated by the polls last year. Are there shy Greens?
The conventional wisdom is that the Conservatives need to be around four points ahead of Labour in the popular vote to get more seats in the House of Commons. That causes some to reflect that if the polls remain neck and neck, or even if they show a very narrow Conservative lead, then Labour will win. This analysis by Mr Hayward suggests that if the Conservatives are just a couple of points ahead in the polls that could be enough.