Lord Ashcroft’s latest round of constituency polling was most notable for its Scottish surveys, but away from the land of the 20% swing, there were some interesting (if less dramatic) seats in England and Wales.
The three English seats all showed similar swings of around 5%, larger than the swing nationally or in the seats polled in December. On the face of it, this would suggest Labour outperforming where it counts – in the marginals, in contrast to what previous marginals polls have shown. However the picture isn’t quite so straightforward, as two of them are very atypical.
As Anthony Wells points out, Norwich North was taken by the Conservatives in a by-election in July 2009 and held in 2010. This is significant for a couple of reasons. Firstly, as Anthony writes, the “incumbency bounce” had already happened before the election, whereas in most marginals any benefit would be felt now. Tory MPs tend to get a smaller incumbency benefit than those of other parties, but it’s premature to suggest that there is none at all purely because the first and second questions give similar results. As I’ve written before, there could well be an effect that the first question is picking up and/or that the second question isn’t picking up entirely.
Secondly, and perhaps more significantly, is the effect that the by-election would have had on the 2010 election, over and above the question of incumbency. If we compare the 2010 result to the notional 2005 result (i.e. accounting for the 2010 boundary changes), we see an enormous swing of 12.9%, compared to just 5.6% across England as a whole. It appears that the by-election (as often happens) had a sustained effect on seat, including at the general election less than a year later, which has since worn off, hence a bigger swing the other way.
Colne Valley can reasonably be called a Tory-Labour contest, as the 2015 race is between those two parties, despite the fact that in 2010 it was actually the Lib Dems that came second with 28.2% of the vote. As I mentioned before, most of the red-blue battleground had disproportionately few Lib Dem voters in 2010, meaning that Labour, who have taken many of those votes since the last election, are acheiving smaller swings in the marginals than across the rest of England. This seat actually has significantly more Lib Dems than the average English seat – in Ashcroft’s previous Labour-Tory seat polling, seats with a 25% or higher Lib Dem vote or higher have seen swings, on average, 1.5% smaller than the battleground as a whole, so a bigger swing is what you’d expect.
High Peak has a few more 2010 Lib Dems than most marginals, but on the whole it’s a pretty straightforward contest, so this seat is a bad poll for the Tories. But the December batch generally showed them outperforming in the marginals, by more than usual, so it’s too soon to say that the pattern has been reversed.
Vale of Glamorgan shows a much smaller swing than GB-wide polls, but is also untypical, because of where it is. Swings in Wales have typically been smaller than in England, due to Labour’s poor performance there. (If you’re interested in the reasons for this, I recommend this interview with Professor Roger Scully of Cardiff University). This swing of 1.5% is a point or two smaller than in recent Welsh polling.
Where does this leave the skew (which, for new readers, is what I call the GB-wide Conservative lead over Labour at which the two would have equal seats)? It has many moving parts and is particularly sensitive to the assumptions you make about all sorts of things, but I’d expect it to be very small, and probably less than a point. Others seem to concur, John Curtice says one point, the Electionforecast numbers seem to suggest a bit more, the PSA survey a bit less, but they suggest something very similar. May2015’s model and Peter Kellner both imply that it’s gone slightly negative.
I’ve also been asked by readers what sort of lead the Conservatives would need for an overall majority, after a couple of polls this week showed them with four-point leads. The answer is about double that, again depending on the assumptions you make. My guess would be that a four-point lead would give the Tories about 300 seats, down only slightly from last time, due to a more efficient vote distribution than in 2010.
Lord Ashcroft’s next batch of seats will be out on Wednesday.