Traditional methodology might not be applicable
Tomorrow afternoon Lord Ashcroft is publishing his poll for Thursday’s Newark by-election which will be the only the second survey that’s been carried in what’s turning out to be a humdinger of a fight between UKIP and the Tories. Both have got historical baggage that a win could help them shed.
For you have to go back to William Hague’s victory in the N Yorks seat of Richmond in 1989 to find the last time that the blues held on to a by-election seat whilst in office. UKIP, of course, have never managed to win a Westminster constituency either in a general or a by-election. Their best performance was the 27.8% at Eastleigh in February last year.
It is important to recall that all the polls in that contest, as the chart shows, understated the purples by quite some margin. None of them had UKIP any higher than third place.
It was a similar pattern in Corby in November 2012 when the Tories were trying to hang on to the seat following Louise Mensch’s decision to quit politics. The final Ashcroft poll had UKIP on just 6% – they ended up on 14.3%.
By-elections, of course, are susceptible to late swings which is what can make them so exciting. But I believe there are other methodological reasons why the polls have struggled with UKIP in by-elections in particular the reallocation of don’t knows or refusers to what they said they did at the previous general election.
I’ve been in correspondence with Lord Ashcroft about tomorrow’s Newark poll and have suggested that he highlights the pre-reallocated numbers. At Eastleigh that would have got his final poll a bit nearer to what UKIP achieved.