Has ICM adopted PB’s turnout weighting plan?
Is this putting the squeeze on the “others”?
Well done to Anthony Wells as UKPollingReport for spotting this but ICM seem to have introduced a significant methodology change in this week’s Guardian poll which looks as though it will reduce the shares it reports for UKIP/BNP/GRN.
For the firm has introduced a change that I’ve been pressing for since 2007 – that greater weight be given when assessing likelihood to vote to those respondents who have a record of voting.
This is how the firm explains it: “If someone is 10/10 certain to vote, they are given a weight of 1.0. If someone is 9/10 certain to vote they are given a weight of 0.9 etc. People who tell us they are likely to vote in the next General Election but did not vote in the last, are further down-weighted. If someone says they are 10/10 certain to vote but did not vote in 2005, they are given a weight of 0.5. If they say 9/10 certainty, the weight becomes 0.45 etc.”
This is not as radical as my original proposal in August 2007 but it is going a long way down the route of recognising that the people most likely to turnout next time are those who did so at the last election.
My initial reaction is that the parties most likely to suffer will be UKIP/BNP/GRN where a much higher proportion of their support in previous surveys has come from 2005 non-voters. In this latest poll with the new calculation these three parties between them only get 4%.
Perhaps we ought to look again at the vote betting markets from 