Would he have had a bigger lead with a different turnout approach?
With the publication of two polls showing very different pictures of the UK’s biggest political betting event for years there’s going to be even more scrutiny of how the figures are produced. There are matters in relation to the YouGov approach that I have written to Peter Kellner about that I hope to discuss in the next few days
In the meantime we now have the detailed data from ICM which on Wednesday night reported top-line figures of a Johnson margin of just one per cent.
As I discussed at the time the survey caused me to close down a large part of my betting position on the Tory. Now, after looking at the detail from ICM, I am back backing Boris again.
The challenge for pollsters in this election is dealing with lowish turnouts. For how can pollsters be sure that those who say they will vote in a particular way actually do so? There’s not a very good history here for in the 2000 and 2004 elections turnouts were in the mid-30s which led to some skewed figures from the phone pollsters with big over-statements of Ken’s position.
Like in all ICM voting intention polls those Londoners interviewed for the Mayoral race were asked to rate on a scale of 1 to 10 how certain it is that they would vote. Reproduced above is the relevant section of data and as can be seen Boris supporters 59% rated themselves as “certain” compared with 49% for Ken. So if the poll was just restricted to just them then Boris has 43% to Ken’s 36%. Taking the 9/10s into account as well the split becomes Boris 41% to Ken’s 35%.
Some pollsters weight responses so an 8/10 answer is only worth four-fifths of the value of a 10/10. ICM don’t do this – everybody rating themselves as 7/10 or more is included and given the same value.
Now this approach might be right for general election voting intentions but will it work in a much lower turnout election like the one on May 1st? If I was a Ken backer I would be worried by the lower certainty ratings. We’ll know when the results are declared but the ICM data has given me the confidence to put four figures bets on Boris again.
Latest mayoral betting is here.
New format betting section. The betting odds search engine, Bestbetting.com, has revised our betting odds pages and added new facilities such as an historical odds feature. The firm pays a commission share which helps keep the site going. Please use these links for all your betting needs – not just politics.