An exclusive Politics Home study of 35,000 voters in 238 marginal constituencies, based on fieldwork carried out by YouGov predicts that if an election were held now the Conservatives would win 398 seats, Labour 160, and the Lib/Dems 44.
This is one of the largest surveys of marginal seats ever attempted.
You can read a report summarising the results here,)
Marginal seats have been grouped into 17 categegories, and an overall swing based on the total votes in that category applied to Anthony Wells' estimate of the notional 2005 results.
In terms of the overall national position the evidence provided by this survey is quite powerful, but in terms of individual seats, where the sample sizes were between 100 and 400, not too much weight should be given to it.
For example, one of the categories consists of the four Labour-held seats in Cumbria, where 536 adults took part, of whom on a standard voting intention 39% intended to vote Conservative, 36% Labour, and 16% Lib/Dem. (They asked several different versions of this question, which got slightly different results. On the basis of this, the survey predicted that there would be a 6.5% swing from Labour to Conservative in Cumbria and that two of the four Labour held seats in the county would go Conservative.
In many respects this appears to be an interesting and informative survey, but my scepticism of the reliability of the individual constituency predictions is somewhat skewed because the notional base for my own constituency, Copeland, is inaccurate.
If there really is a 6.5% swing from Labour to Conservative in Copeland, we will be into recount territory with the seat way too close to call. Boundary changes shift from Workington to Copeland three rock-solid Conservative wards, and one three-way marginal which has councillors from all three parties. Allowing for the impact of this the Labour majority falls to about 13% and therefore a 6.5% swing would leave Copeland very close indeed.
However, the study doesn't start from this point.
Anyone who knows the area concerned realises that the four Allerdale wards which transfer into the Copeland seat are basically the most Conservative part of the former Workington constituency. They are represented on Allerdale council by the Conservative leader of the council and three of his most senior colleagues. But unfortunately for anyone trying to estimate exactly what is likely to happen in an election, these wards are not always contested - basically Labour and the Lib/Dems often don't bother to put anyone up except in Keswick, which is the three-way marginal, and didn't last year.
So in the absence of appropriate recent local election data, the notional 2005 result for Copeland on Anthony Wells' site was created by transferring a block of electors with Conservative and Labour vote shares equivalent to the Workington average into Copeland. This probably seems reasonable to anyone who lives three hundred miles away from Cumbria, but it means that a block of net Labour supporterts have been put into the Copeland estimates increasing the notional Labour majority when it is almost certain that what the boundary changes have actually happened was to transfer in a block of net Conservative supporters, making Copeland more marginal (and Workington safer for Labour.)
On that basis the study calls Copeland as a Labour hold, but if the suggested Cumbria swing is approximately right the situation is far too close for anyone to be able to predict accurately which way the seat would go if an election were held this autumn. And of course, there could be anything up to 21 months until the election, so there is probably still plenty of time for the position to change.
Well, we'll all find out what the real situation is when the General Election finally arrives some time between now and June 2010, but I believe there is everything to play for in Copeland.