The conventional opinion polls got the London Mayoral Election so badly wrong in 2000 over-estimating Livingstone’s support that it’s no wonder that they’ve mostly avoided this June’s battle.
There were six main polls in 2000 which showed leads over Steve Norris ranging from 34%-57% when on the day the margin was just 11.9%. The smallest error was a staggering 22%. Apart from one ICM survey reported in February the polling this time has been left to the internet pollster, YouGov.
Will YouGov’s unique methodology, so criticised by the other polling firms, make a better job? For if it does it could have far-reaching implications.
A close prediction of the London Mayoral Election will be a good pointer to the accuracy of YouGov’s General Election polls which have consistently shown higher shares for the Conservatives than the other opinion polls.
Following its accurate 2001 General Election forecast YouGov’s best performance was at last May’s elections for the Scottish Parliament. Then they got Labour, the Conservatives and the SNP to within one per cent. The only failure was with the Lib Dem vote which YouGov predicted at 19% against an actual 16%.
The big YouGov success was predicting the Conservative vote in Scotland. All the other polls had the party at either 11 or 12%. YouGov put them at 16% against an actual 17%. With Labour the other polls over-estimated the share by 4-8%. YouGov’s overestimation was 1%.
Getting an accurate Tory share has been a nightmare for many pollsters for many years. This has been put down to Tories being less reluctant to admit their allegiance to a compter as with YouGov, rather than to an opinion poll interviewer as with the conventional polls.
Will YouGov be able to repeat this in London on June 10? A challenging issue for all pollsters is the turnout – in Scotland last year it was 49.5%. At the London Mayoral Election in 2000 it was just 34% and the signs are that it could be lower this time. The conventional wisdom is that the lower the turnout the better it is for the Tories.
Political gamblers seem to be ignoring YouGov’s latest London poll that had a Livingstone lead of just 5% amongst those “certain to vote”. With such a gap in a normal election market you would expect prices of 1.5 for the leader and 2.5 for the challenger especially as the margin has been narrowing with every poll. Here even the betting exchanges still have Ken at 1.22 and Norris at 6.
UPDATE – WHITE HOUSE 2004.Prices are edging towards Kerry. On the Iowa Electronic Exchange, which we have explained before, Kerry is within a hair-breadth of Bush. The prices are rated out of 1 and Keery is on 0.49. On the UK markets Bush is still a clear favourite but the gap is tightening.