Lord Ashcroft has released a second tranche of polls in Con-v-Lab ultramarginals, full details are here. It’s a repeat of his first polling of Con-Lab seats in April and now as then, Ashcroft polled the 12 most marginal Conservative held seats with Labour in second place, plus two other seats (South Thanet and Great Yarmouth) that he thought might have interesting UKIP results. Note that this wasn’t what we’ve often called a “marginal poll” in the past (a single poll of a group of marginal seats), it was 14 separate polls, one in each seat, individually sampled and weighted.

In April, the average swing in the 12 Con-Lab ultra marginals was 5.5% from Con to Lab, the same as in the national polls at the time. In other words, Con-Lab ultra-marginals were behaving right in line with national polling.

The average picture in the national polls is unchanged since April – the national swing from Con to Lab in GB polls conducted during the time Ashcroft did his latest fieldwork was 5.4% from Con to Lab. Lord Ashcroft’s findings in the marginal polls however have shown a slight weakening in the Labour position relative to the Conservatives, looking at just the 12 ultra-marginal seats the Conservatives are down by about 1.5% on average, Labour down by about 2.7%, UKIP up by about 4.5%. This looks like a European boost – the previous poll was done before the European election campaign, this one started in mid-June. The average swing in the ultra-marginals is now 4.9 from Con to Lab, meaning an increase in UKIP support has slightly hurt Labour. While the difference is too small to really make a fuss about, it also means that Ashcroft’s polls now show a slightly smaller swing in Con held marginals than nationally – in line with what we’d expect given the normal incumbency effect.

In two of the seats polled, Thanet South and Thurrock, UKIP were ahead in voting intentions, which will be enormous help for them in convincing voters they are a viable general election vote in the local area.

When you sit down to do a marginals polls one consideration is where you draw the line: what is a marginal? The thing you want to avoid is under or overshooting the real battleground – the risk is that you poll lots of seats that need a swing of up to 5% and find a swing of 10%, enough to win lots of seats you didn’t bother polling.

That’s the sort of thing that’s happened in the third of Lord Ashcroft’s three sets of marginal polls – full details here. He polled the four most marginal LD -v- Lab seats, Norwich South, Bradford East, Brent Central and Manchester Withington. These need a swing of up to 2.1% to go from LD to Lab, which unsurprisingly Labour get easily. The average LD => Lab swing in these seats was 15%, confirming that the Lib Dems are doing much worse where they are up against Labour and easily enough to unseat almost all Lib Dem MPs with Labour in second place. In practice of course we can’t actually be that confident that voters in a tight LD-Lab marginal will behave the same way as in a seat where the Lib Dems have a 20% majority, so it’s a bit of a shame Ashcroft didn’t include some more challenging LD-Lab fights like Cambridge, Hornsey & Wood Green or Bermondsey.

An interesting thing to note is that the Greens are doing notably well in a couple of these seats. In Norwich South they are in second place on 20%, but that was one of their target seats anyway, they are also doing well in Manchester Withington, up 8 points on 10%.

While it’s hardly a LD-Lab battle, Lord Ashcroft also polled Brighton Pavilion, Caroline Lucas’s seat. Voting intentions there with changes from the general election are CON 18%(-6), LAB 33%(+4), LDEM 5%(-9), GREEN 32%(+1), suggesting an extremely tight race between Labour and the Green party.


Three new polls so far today – the two regular GB polls from Ashcroft and Populus and ComRes’s monthly marginals poll. Lord Ashcroft’s weekly poll today has topline figures of CON 33%, LAB 31%, LDEM 9%, UKIP 15%. This is the first Conservative lead in a poll since that brief narrowing in mid-May. As ever, don’t get too excited about a single poll, it could be repeated in other polling and be the first sign of a genuine movement… or it could just be a blip. Meanwhile Populus’s twice weekly poll this morning had rather more typical figures of CON 33%, LAB 37%, LDEM 10%, UKIP 12% (tabs here.)

Finally ComRes have released the second of their regular polls of marginal seats. I wrote about the first here. ComRes’s regular marginal polls cover the 40 most marginal Con v Lab seats, 25 held by the Tories, 15 held by Labour. At the last election the vote in these seats was CON 37%, LAB 37%. Today’s poll has voting intentions of CON 31%, LAB 36%, LD 7%, UKIP 17%. This is a swing of 2.5% from Conservative to Labour, the equivalent of a 2 point Conservative lead in a national poll. Hence it suggests the Conservatives are doing better in the Con-Lab marginals than in national polls (the poll shows some movement towards Labour in the marginals since May, but the previous poll was taken during that brief narrowing of the polls in mid-May, so the change reflects the national picture). The ComRes marginal polls remain in contrast to Lord Ashcroft’s most recent marginal polls which suggested the marginal picture is much the same as the national picture.

Having polled Lab -v- Con seats last month, Lord Ashcroft has now done a similar exercise in ultra-marginal seats between the Liberal Democrats and Conservatives. He polled 17 seats. 6 are Conservative held seats where the Lib Dems came a close second last time and need not unduly delay us, all show a shift from Lib Dem to Conservative and Conservative holds, the most interesting ones being Watford (which was a three way marginal in 2010 and remains so in this poll) and the two Cornish seats in the sample which both put UKIP in second place, more on that later.

Turning to the Lib Dem seats polled, Lord Ashcroft did fieldwork in the ten most marginal Lib Dem seats with the Conservatives in second place. He also polled the special case of Eastleigh because of the by-election and UKIP, but we’ll set that aside for now. The ten Lib Dem ultra-marginals represent seats with majorities of up to 6.4%, so they’d need a swing of 3.2% from LD to Con for the Tories to take them all at the next election. Based on national polling the Conservatives should do that easily (the national LD=>Con swing is about 5.5%), but as we know, Liberal Democrat MPs rely far more upon their personal support and tactical voting than MPs from other parties, and tend to be better able to confound a national swing.

Across the ten LD ultra-marginals the average swing from LD to Con was 3.4, so the Lib Dems continue to do far better in their own seats than in the country as a whole. However, “doing better” doesn’t mean completely immune from loss. If that was repeated across all LD-Con marginals they’d still lose all those ten to the Tories plus perhaps Eastborne. In practice there is variation between seats, so some seats with smaller majorities the Lib Dems would cling on to, some more distant targets they’d probably lose. Looking at these particular seats Ashcroft found the Lib Dems doing far better than average in Sutton & Cheam, which this suggests they’d hold with ease, and better than average in Cheadle which this also suggests they’d hold. Those two are the most urban of the two seats polled – but with limited data points its difficult to tell if that’s significant. North Cornwall’s swing isn’t far from the average, but would be too close to call.

If the Liberal Democrats lost only 10 or 11 seats at the next election they’d probably be quite pleased… but remember, the Lib Dems also have around 10 English seats where Labour is the challenger and 11 seats in Scotland that could be vulnerable to either Labour or the SNP, so this is not the only battleground for them. It’s the largest Lib Dem battleground, but not the one where they are most vulnerable.

Two other things to note. Seats that have had a by-election are changed by it, so Eastleigh is probably representative of nothing but itself, but for the record Ashcroft found voting intentions there of CON 27%, LAB 10%, LDEM 39%, UKIP 22%. Also worth noting is how well UKIP are doing in the Cornish seats included here. Because it’s a close LD v Con battleground Ashcroft’s sample happened to include five of the six Cornish constituencies, and UKIP were running in second place in three of them. Take the UKIP scores with a slight pinch of salt because the timing of the fieldwork (it was mostly done during or in the fortnight following the European elections, so when UKIP were on a bit of a publicity high), but it’s another potential pointer as to where they could do well.

Lord Ashcorft has published some new polling of marginal seats, full details here. As with the ComRes marginal poll in the week the seats polled were mostly ultra-marginal seats – in this case, the 12 most marginal Con-Lab seats, the 12 most marginal Lab-Con seats, but whereas the ComRes poll was a single sample representing the most marginal 40, these were 24 individual samples, one from each seat. Ashcroft also added two seats that are less marginal, but thought to be good for UKIP: Thanet South and Great Yarmouth.

The fieldwork for Ashcroft polls was done between the 31st March and the 18th May. During that period the average Labour lead in the national polls was about 3.5 points: that’s the equivalent of a uniform swing of 5.25%. The average swing in the twelve CONSERVATIVE ultra-marginals seats was 5.5%, The average swing in the twelve LABOUR ultra-marginals was 6.5%.

That means that in contrast to the the ComRes poll in the week, the swing from Con-to-Lab in Conservative ultra-marginals is pretty much in line with the national swing, a fraction of a percentage point better for Labour compared to the national figure. In Labour-held ultra-marginals the swing is a little larger, which is what we’d expect to find (parties do a little better in seats they hold due to the incumbency effect of the local MP).

It’s not a very exciting finding – swing in Conservative marginals not vastly different to other seats – but it’s one that gives me some confidence in the poll. The reality is that come general elections marginals as a group are not usually vastly different to other seats. The swing is sometimes a little bigger or smaller, new incumbents normally do a little better, but the contrast isn’t normally vast.

While I excluded them from the sums above (as they were selected because they were unusual, so would have skewed the sample) I should comment on those two extra seats polled – Thanet South and Great Yarmouth. Both were chosen because there was an expectation that UKIP would be doing well, and in both cases it proved to be true – both had them in a very strong third place, with 28% in Great Yarmouth and 27% in Thanet South. Their strongest performance though came in a seat that was part of the normal sample of ultra marginals – Thurrock, where Ashcroft found them at 29% and in second place behind Labour. Thurrock was also one of the seats where UKIP did extremely well in Thursday’s local elections.

UPDATE: Actually I’ve just spotted that the fieldwork in the Tory held seats was done earlier than the fieldwork in the Labour held seats. So comparing the swing in Con-Lab seats to the swing in national polls at the time the polls were done shows no difference at all (both show swing of 5.5%). Comparing the swing in Lab-Con seats to the swing in national polls at the time those polls were done shows Lab doing about 1.5 points better in seats they already hold.