One of the few certainties when it comes to general elections is the near-unanimity of pollsters in declaring the eventual result unclear. Self-preservation plays a large role here: no-one wants to be the one who calls it early, only to have their credibility later blown to shreds.  However, in the case of the upcoming UK general election, this hedging of bets is also recognition of the fact that – with Labour and the Tories in deadlock, and the unpredictable wildcards of UKIP and SNP support – predicting a definitive result from current headline polling figures is near impossible. 

However, this year’s rune-gazing is enhanced by the intriguing and extensive constituency-level polling conducted by the Tory peer Lord Ashcroft. A Conservative strategist during the 2005 and 2010 campaigns, Ashcroft has continued to bankroll polls in marginal seats, now available for public consumption. 

The Ashcroft Method

So how does this differ from the typical headline voting figures we see in the media? The Ashcroft approach has the benefit of granularity, and negation of the difficulties inherent in interpreting nationally representative polls. This arises from the fact that, however accurate a nationally representative poll may be, the First Past the Post system means we then have to translate share of vote into actual seats. The standard method here is ‘Uniform Swing Projection’ - which is to assume that changes in voting patterns seen at national level will be the same at local level in every seat. A useful shorthand, but one that does not account for variations in swing by constituency.  In fact, the last general election result showed wildly different swings in individual seats, where a local issue, candidate or campaign often meant the result bucked the national trend.

At the same time, the Ashcroft methodology involves a trade-off: narrowing in on marginal seats means greater depth at the expense of less breadth. Combining the two approaches, however, allows us to be more confident in talking about trends in opinion.  Let’s take two examples – the threat posed by UKIP to the Tories, and that posed by the SNP to Labour.

The UKIP threat to the Conservatives

If we take the current ‘poll of polls’ national headline figures as reported on Antony Wells’ excellent UKPollingReport site (Cons – 32%; Lab – 33%; Lib –  8%; UKIP –15%) and plug them into the Electoral Calculus Website which applies a Universal Swing ,we end up with a prediction that UKIP would not win a single seat  - despite increasing their vote share by twelve points since 2010. This comes down to the question of ‘efficiency’ of vote distribution: while the increase in national share of vote looks dramatic, it would not be sufficient to win a single seat, given the extent to which UKIP trailed in even its ‘better’ seats in 2010. 

Ashcroft, however, shows us that UKIP are adopting a tactic of targeting key marginals and achieving much wider gains at local level: for example, as early as July last year, an Ashcroft poll of Tory-held Thurrock was showing UKIP’s vote up from 7% to 36% - enough to comfortably take the seat.

Arguably, then, the question of if and how many seats  UKIP will win will be determined by their ability to maintain efficient local campaigns in the manner that has allowed the Lib-Dems to maximise their seats for years – rather than their national performance.

SNP and Labour

Perhaps even more intriguingly, Ashcroft is now turning his attention north of the border, polling key constituencies to examine the extent of the losses Labour faces to the SNP. The possibility of a Labour wipeout has previously been based on stark headline polling that shows the SNP has consolidated support built up during last year’s independence referendum – for example, a Daily Record/Survation poll in December showed 48% of Scottish voters planning to vote SNP in May, twice as many as intended to vote Labour – certainly enough to lead to a Labour wipeout on Uniform Swing.

Ashcroft’s first round of polling in Scottish constituencies, the results of which were released earlier today, has focused on seats with a high ‘yes’ vote in the referendum: to test the theory that this is a large predictor of the SNP’s surge. Indeed, 13 of the 14 seats polled show results that imply an SNP gain – sometimes with an extremely large swing (for example, Dundee West shows an SNP gain of 30 points from the last election, with Labour losing 24 points). The true extent of the problem Labour faces in Scotland should be revealed in Ashcroft’s next round of Scottish polling which will focus on seats with a lower ‘yes’ vote in the referendum. If these also show a swing to the SNP, Labour could be in real trouble.

The local-level analysis supplied by Ashcroft may not get us any closer to being able to concretely predict a result come May – but, in adding to our understanding of how national trends play out, it is certainly a welcome addition.

Chris Levy
Chris Levy

Chris Levy, director of APCO Insight in Europe, is based in APCO’s London office. APCO Insight is the opinion research consultancy at APCO Worldwide. Read More