May2015 currently predicts the Lib Dems will win 26 seats in three weeks. Four other forecasters—a pair of academic models, the Guardian and the bookies—agree: all five of us say 26-29 seats. Only Oxford’s Steve Fisher, who runs Elections Etc, disagrees; he says 22 seats.
But a new poll of the Lib Dems’ 14 South West seats, conducted by ComRes for ITV, suggests that the Lib Dems could win far fewer than we are all predicting, and end up with just 19 seats across the UK.
ComRes’ poll shows a swing from the Lib Dems to Tories of 13 percentage points. That compares with the 4-5 point swing shown by Lord Ashcroft’s polls in the region (he has polled almost all of these seats). If we apply this 13-point swing since 2010 uniformly across all 14 seats the Lib Dems would lose every one.
At the moment they are forecast to lose just seven by both us and Election Forecast, the academic forecasters used by Newsnight and FiveThirtyEight.
ComRes’ poll shows a 13 point swing to the Tories, compared to 4-5 points in Ashcroft.
Initially it seemed that this might all be accounted for by ‘incumbency’ – the clearly demonstrated effect in Ashcroft’s polls where Lib Dem MPs perform better when voters are asked to consider their parliamentary constituency and the “candidates likely to stand there”.
But ComRes then made clear they had asked a very similar question to Ashcroft, which referred to the candidates likely to stand in a voter’s seat (indeed their research showed that Lib Dem voters are voting for their candidate more than for the national party).
So what accounts for this sharp difference, and who is to be believed? On hearing the results one pollster quipped, “I think I marginally prefer a 1,000 person sample per seat, but maybe that’s just my irrational prejudice.” ComRes can only have quizzed around 70 people in each seat, whereas Ashcroft has polled a representative group of at least 1,000 – and more than once in some seats. (ComRes don’t seem to have released detailed tables of their results, so we can’t compare them in more depth.)
It would be preemptive to suggest the Lib Dems are really going to lose seven more seats than we expect, and therefore fall from 26 to 19 seats overall.
One pollster quipped, “I think I marginally prefer a 1,000 person sample per seat.”
Take Ashcroft’s polls. We can divide the Lib Dems’ 14 seats into five categories. First, there are the two likely Lib Dem ‘holds’, in Thornbury & Yate and Cheltenham. Ashcroft put the Lib Dems ahead by 8 in the latter in November, and internal party polling shows the party even further ahead, while Ashcroft put the Lib Dems ahead by more than 20 in the former that month.
Then there are Bath and Yeovil, neither of which we have polls for (but here’s a May2015 longread on the battle for Bath). Both of these seats are expected to stay Lib Dem, but we can’t rebut the implications of ComRes’ poll so easily in these seats.
In three others—St Ives, Cornwall North and Torbay—the Lib Dems seem to be locked in a tight race, as Ashcroft showed as recently as a fortnight ago. He put the Lib Dems ahead by 3, 2 and 1 point in these seats, having put them behind by 1 in St Ives last June. It seems unlikely the Lib Dems really trail by around 20 points in these seats, as ComRes’ poll implies.
Three others are also close, but Ashcroft has shown that the Lib Dems are trailing in St Austell & Newquay, Devon North and Taunton Deane. But again, he hasn’t showed them trailing by nearly as much as ComRes’ numbers imply.
He suggested the Lib Dems are within 6 points in St Austell earlier this month, and had them within 8 last June. (The Lib Dems privately think they are ahead in the seat.) Similarly, he put them behind by 7 in Devon North, and had them within 1 in November, while he put them within 4 of holding Taunton Deane in late 2014.
The Lib Dems could win 4 or 10 of their 14 South West seats. We have limited information.
There are four other Lib Dem seats in the South West – Somerton & Frome, Wells, Dorset North and Chippenham. A pair of Ashcroft polls have put the Lib Dems fairly far behind in all four of these.
In other words, the Lib Dems are forecast to keep seven seats, and are competitive in another three, but could in turn win none of those three and even win closer to four out of 14. These races are very close and we have limited information.
The Lib Dems will greet today’s poll with calm. They are convinced they poll better when candidates are actually named; they don’t agree with Ashcroft that alluding to candidates helps them as much as actually naming them. And their private polling, which is being carried out by Survation, shows them faring better in these seats than Ashcroft’s polls suggest, let alone this ComRes poll.
More regional polls by other national pollsters, such as YouGov, could help confirm today’s findings. But they will remain difficult to reconcile with Ashcroft, whose 1,000-person strong and fully weighted polls will continue to power May2015’s election forecast.