Ukip defectors and the opinion polls that don’t add up

A Ukip supporter attaches a poster to a window during May’s general election campaign. Matt Singh suggests that pollsters may have miscalculated the number of defections to the party from previous Labour and Tory voters. Photograph: Ben Stansall/AFP/Getty Images

Alan Tiplady (Letters, 17 November) suggested that no one had considered the possibility that the pollsters got the general election result wrong due to fear of the Scottish National party prompting Ukip defectors from the Conservatives to return, while those from Labour remained.

I can assure him that pollsters and analysts have considered this – recontact data strongly suggests that there was no net effect from people changing their minds at the last minute, as Tom Clark noted in his article (Study shows why polls got election so wrong, 13 November).

However, Mr Tiplady is on the right track. Analysis by Number Cruncher Politics published last week concludes that the relative proportions of “Labour” and “Conservative” Ukip defectors had been mismeasured – more Ukip voters had come from Labour than the pollsters thought, and fewer from the Tories. This was very likely caused by an imbalance in the pollsters’ samples, combined with problems weighting the data.
Matt Singh
Founder, Number Cruncher Politics

• Join the debate – email guardian.letters@theguardian.com