Looking back to those final GE15 phone polls one thing stands out about LAB backers…
There were more non-GE10 voters in the LAB totals than in CON
The British Polling Council inquiry into what went wrong with the GE15 is well under way and no doubt many will be putting forward theories about what caused them to be so wrong.
One of the factors that I believe was partly responsible for the overstatement of LAB shares is featured in the chart above – a larger part of its support according to the surveys was coming from those who did not vote for the main parties at the 2010 election.
From the data that’s made available by the pollsters it’s not possible to identify whether these were non voters or not but given that 90%+ of the overall 2010 vote went to LAB-CON-LD it is a reasonable assumption that the above is a good pointer.
Part of this is accounted for by those in the 18-23 age bracket who, clearly, could not have voted at the previous general election and they were more likely to be LAB backer but that is only a partial explanation.
The same happened in much of the by-election polling over the past year. LAB was attracting more non-voters at the previous election than other parties.
- All the evidence is that the best guide to whether you will vote in the next election is whether you voted in the last one.
The BPC needs to address the way pollsters deal with likelihood to vote. It should be more than just telling the interviewer that you are 10/10 certain.