Thursday Round Up

D minus 28 Tonight's polls: YouGov/Sun (7th-8th Apr) CON 40%(+3), LAB 31%(-1), LDEM 18%(-1)

Just the one new GB poll today, with YouGov returning to similar figures to their polls earlier in the week. There was also a new Ipsos MORI poll of marginal seats for Reuters which found a 5.5% swing in the marginal seats the Conservatives would need for a majority, short of the 6.9% the Tories would need. Sadly we don't have a recent national MORI poll to compare it to in order to tell whether it is still suggesting the Conservatives doing better in marginals than nationally.

The political news today was largely dominated by the issue of National Insurance (where there is sadly still a shortage of good polling), and the Conservative announcement of their new "National Citizen Service" policy, for which YouGov found solid support - in tonight's poll 77% thought it was a good idea, 14% a bad idea.

There are two models and seat projections out today. On PoliticHome they have the Poll Centre projection model, which is a UNS projection with the additions of a seperate swing in Scotland, a swing of 2 points more in marginal seats and factoring in the random variation in swing between one seat and the next (as developed by David Firth and John Curtice for the 2005 exit poll). All of this I thoroughly endorse, and it's a similar sort of thing I would do if I had the time and the mathematical ability. Details of the model are here, and their current projection has the Conservatives just short of a majority on 315 seats.

The second is from Danny Finkelstein here, or more specifically from Henry Stott, Ian Graham and Keith Simpson, who have been modelling football stats for the Times. They seem to have done something quite complex using transistional matrixes in particular types of seats, full details of which are not yet on the site. Their current projection though is not much different from PoliticsHome's, with the Conservatives on 318 seats.