When a 48% polling lead became an actual 11.9%?

My most profitable political wager of all time up to eight years ago was a spread bet that Ken would get less than 50% in first preference votes in the first London Mayoral contest. For all the signs were that Ken, then standing as an independent, was going to overwhelm his opponents by a staggering margin and many took it as read that he would get 50% on the first round.

For the evidence from the last two elections is that Ken’s supporters appear much more willing to say they will vote for him than in practice they actually do.

The pie-chart on the right from March was from six weeks before the 2000 election when Ken’s share topped the 60% mark with an amazing 48% lead over the person who was to finish second, the Tory, Steve Norris. As it turned out when the votes were counted Ken’s eventual lead on first preferences was just 11.9% – some difference.

It got a little bit better by polling day but not much. The eve of poll survey in the Evening standard had these shares Livingstone 51%: Norris 17%: Dobson 16%: Kramer 12%.

Compare those numbers with the actual votes Livingstone 39%: Norris 27.1%: Dobson 13.1%%: Kramer 11,9%. So Ken’s eve of election polling lead was 34% which was almost triple what actually happened.

In that election every single poll from every single poster that carried out surveys came up with huge over-estimates of Ken’s margin over the Tory.

Move on four years to 2004 and most of the polling was done by the internet pollster, YouGov. The only survey by a phone pollster produced an inflated margin for Ken though not on the scale of 2000. Thus the equivalent to yesterday’s YouGov survey had Livingstone 41%: Norris 27%: Hughes (LD) 17%

The final YouGov poll four years ago had it at Ken 37% to Norris’s 26% on first preferences. In the election the shares were Ken 35.7% to Norris’s 28.2%. So the predicted YouGov margin was 11% compared with the 7.5% that actually happened.

It is perhaps worth reminding ourselves when we see current polling predictions in the 40s that last time Ken’s share of the first preferences was not much more than a third.

My default position on the 2008 race is that Ken is still enjoying a polling bonus and if the final surveys have them level-pegging I would still bet against him.

Latest betting prices are here.

Mike Smithson