Since launching in September, May2015 has tried to become a home for all the election data you might need. We began with a ‘Poll of Polls’ to keep on top of the 10-11 polls coming out each week. Then we used a version of uniform swing to turn that into a seat prediction.
In December we added ‘The Drilldown’: our unique insight into the polls, which allows you to break down voters’ attitudes to the economy, government and different issues by age, class, gender and political ID.
But our method for predicting seats was still too crude. Like all traditional models, it didn’t use all the polling Lord Ashcroft was doing of individual seats. So in January we launched a real election-forecasting machine: now we combine all the latest national and constituency polls to make our prediction. But we still have a problem.
That prediction is reliant on public polls. There is no way to plug in what we might know about individual seats. The parties are doing their own constituency polling, but those polls aren’t publicly released.
The Lib Dems are polling more strongly than public polls suggest.
But sometimes we can get a sense of how parties are faring in those polls. And we can add this to other information we have about specific seats – on how strong the parties’ ground campaigns are, how much money is being dedicated to each seat, and how favourable demographics are for different parties in certain places. By doing all this, we can expand on May2015’s polling data and offer a more precise forecast.
This is what we will now be doing until election day. May2015’s objective forecast, based purely on the polls, will still be our main model, but we’ll have a second prediction that we’re calling the “New Statesman Political Index”.
It will pool all the information gathered by the New Statesman’s political team, from May2015 editor Harry Lambert to NS deputy editor Helen Lewis, political editor George Eaton, ‘Staggers’ editor Stephen Bush, and NS writers Anoosh Chakelian and Tim Wigmore.
We will soon explain and add the Report’s ratings to May2015’s seat lists. But ahead of that, we can reveal that the Lib Dems are polling more strongly than public polls suggest.
The party has spent around £350,000 on private polling of marginal seats, conducted by the pollster Survation.
Current election forecasts, from academic models to the betting markets, predict the party will win just 23 to 28 seats in May. In other words, they will lose at least half of their 57 MPs. But the New Statesman Political Index now predicts they will win at least 30 seats.
In half a dozen seats – St Austell & Newquay, Cardiff Central, Solihull, Bermondsey, Leeds North West, St Ives – the party has reasons to be confident.
It is competitive in races where forecasters have written them off. The odds of a Lib Dem win in St Austell & Newquay are just 37 per cent. In Cardiff Central and Solihull, they’re even lower – 27 per cent and 24 per cent. (Via Firstpastthepost.net.) But the New Statesman Political Index now considers all three seats “toss ups”.
Current election forecasts predict the party will win just 23 to 28 seats in May.
The party is being helped by an ‘incumbency effect’ that May2015 first highlighted in early September. That effect may be ever stronger than Lord Ashcroft’s public seat polls suggest.
Ashcroft asks two questions: a generic and abstract national voting question (“Who would you vote for in an election held tomorrow?”), and a specific local question (“Thinking about your own seat…and the candidates likely to stand there…”). By comparing answers to these questions we can test whether Lib Dems MPs are out-polling their national party.
We have showed how they are, and how that is hurting Tory hopes of winning many Lib Dem seats. But the Lib Dems’ believe there is an even greater effect if an MP’s name is included in the question (as it is on election day), and this is giving them confidence in many marginal seats.
The party is also encouraged by its success in reaching out to young voters and women. They are seeing the significant impact that direct campaigning can have – a reminder to all forecasters that this election still needs to be fought.
The Lib Dems are not recovering in the national polls, and aren’t holding up well in seats they didn’t win in 2010 (that’s nearly 600 seats). But they think they can compete in almost every seat where they have an incumbent MP.
There is reason to believe the party could win as many as 35 seats.
47 of the Lib Dems’ 57 MPs are standing in May, and the party still believes it can hold onto 40 seats, although the rise of the SNP in Scotland has made that harder. The Lib Dems aren’t resigned to losing any Scottish seats, but nor are they certain they will win many.
The SNP are hurting them both directly and indirectly. Lord Ashcroft has shown how they trailing the SNP in Gordon and Inverness, Danny Alexander’s seat. But the nationalists’ rise is also hurting the Lib Dem in seats like East Dunbartonshire, a Lib Dem-Labour marginal until recently.
However the party’s most important battle will be with the Tories. They are competing with their coalition partners in 31 seats. That compares to 14 Labour contests, 11 against the SNP and one against Plaid Cymru.
The New Statesman Political Index has rated every single one of these contests. Doing so gives us a new overall prediction for the party: 30 seats. (There is reason to believe the party could win as many as 35 seats.) That contrasts with May2015’s polling-based prediction of 26, and all other forecasts for the party.
The difference may seem trivial, but four seats could prove pivotal in May.