As the polling industry gather for its election post-mortem….
Pollster | Firm | Prediction |
---|---|---|
Andrew Hawkins | ComRes | Con majority 32 |
Ben Page | Ipsos-MORI | Con 25 seats short |
Andrew Cooper | Populus | Con majority of 10+ |
Peter Kellner | YouGov | Con majority 20-30 |
Robert Salvoni | Harris | Con majority 2-10 |
Andy Morris | Angus Reid | Con majority 40-50 |
Martin Boon | ICM | Con majority 20 |
John Heald | Opinion Research | Con majority 40+ |
…remember these April 11th predictions?
With almost all the leading figures in the polling industry getting together today for their general election post-mortem I thought I’d dig out the above predictions that many of them made to the Independent on Sunday on April 11th 2010 just three and a half weeks before the big day.
As can be seen only Ben Page of MORI was against the flow but in a comment on UKPR later that day he said that his personal view as well was that it’d be a Tory majority.
What’s remarkable is that at the time the polls were edging in Labour’s direction and almost none of the surveys were pointing towards an overall Tory majority.
Thus the ComRes poll published in that edition of the paper had the Tories on 37%, Labour on 32% and the Lib Dems down on just 16%. On election day the votes cast split C37: L29.7: LD 23.6.
This snap-shot of expert opinion highlights two major perceptions at the time – that David Cameron would get a huge boost from the first TV debate that was to take place four days later and that the Tory work in the key marginals would deliver more seats than a simple uniform swing calculation suggested.
As we all know it didn’t quite happen that way and I am looking forward to today’s discussions for, hopefully, more insight into what happened and how the polling firms performed. The conference itself has been jointly organised by the British Polling Council and the National Centre for Research Methods.
Today on PB: There’s a guest slot lined up for this afternoon and this evening I’m hoping that we’ll see the first ICM poll for a month.