BBC voting pundit claims a ‘rogue poll’ gave SNP lead


Polling expert and regular BBC Scotland pundit Professor John Curtice has claimed that a recent IPSOS Mori poll that gave the SNP a slender lead in the race for Holyrood was probably flawed.

Professor Curtice, who is also Professor of Politics at University of Strathclyde, was speaking on BBC Radio Scotland’s Newsweek Scotland this weekend where he was giving his views on two polls that gave markedly different results.

One poll carried out by IPSOS Mori on 10-13th February showed the SNP with a 1% lead in Holyrood voting intentions.  However a  Yougov poll carried out one week later gave Labour an incredible 9% lead.

The Yougov poll resulted in controversy after it emerged that the raw poll had in fact showed the SNP with a 13 point lead but that Yougov had altered this result using their controversial Westminster weighting system.

Speaking on the programme professor Curtice acknowledged that the Yougov methodology had been flawed.  However the academic claimed that too many SNP supporters had been allowed to participate and this is why the raw data showed an SNP lead.

Professor Curtice explained that the two polls were done differently, whilst the Mori poll giving the SNP a lead was carried out by random telephone, the Yougov poll that favoured Labour was carried out over the internet by a panel of people selected by the polling organisation.

Professor Curtice confirmed that the Yougov poll was “very, very heavily weighted to an unusual degree”.  The academic claimed that the weighting was due to Yougov making ‘a mistake’ when they selected the panel and that the organisation had included “too many SNP supporters” and “too few Labour supporters”.

Professor Curtice said: “The problem we discovered that happened with this survey is that it was very, very heavily weighted to an unusual degree and what we’ve discovered is that actually when Yougov originally issued its invitations to participate in that survey they made a mistake, they invited too many SNP supporters to participate in the survey and too few Labour people.”

“Therefore this survey has ended up being weighted to an unusual degree.  Now that doesn’t necessarily mean it’s wrong, it certainly doesn’t necessarily mean to say that it’s biased but what it does mean is that although on the face of it, it is a survey and interview of twelve hundred people.  In statistical terms actually it is not as powerful as that because there were almost half as many Labour people in the sample as there should be, then the basis on which the polls estimate of what Labour identifiers are going to do in May is therefore more subject to sampling error.”

However, despite listing the problems with the Yougov poll Professor Curtice claimed that in his opinion it was the IPSOS Mori poll that was probably inaccurate.  His reasoning being that it was at variance with previous polls that gave Labour a lead.

Professor Curtice added: “I think the truth is, is what we’re having to accept at the moment is that the Mori survey, the earlier survey, the one that showed the SNP narrowly ahead is the one that looks at variance with everything else.

“Given the mistakes that Yougov made and given that perhaps its weighted in such a way that the SNP feel it’s actually disadvantageous, even if one accepts all that you have to realise that the Yougov survey for the Greens was weighted in exactly the same way as another survey that Yougov did for The Scotsman at the backend of October.

“And so even if we could argue about the weighting itself, what should be true is that if indeed there’s been a swing away from Labour towards the SNP you would have at least expected the Labour lead to be smaller in the more recent Yougov poll than in the earlier one and in fact that isn’t true, in fact the opposite is true.”

“Now it may still be, and this is the caveat, it may be that Yougov failed to identify this swing and because of that problem about how it issued the sample and because therefore they’ve had to weight it so much, for the moment at least as it were, the weight of evidence seems to suggest that probably it’s the Mori survey that may well be the one that’s out rather than the more recent Yougov one.

“Of course it may well be that the truth is somewhere between the two.”

Professor Curtice’s remarks are sure to cause controversy amongst other polling experts given that he appears to have ignored criticisms of Yougov’s weighting system whereby Westminster voting intentions are used.

Other experts have claimed that Scotland is a particular problem for Yougov in that the organisation has thus far failed to take into account the very real differences in voting habits amongst Labour supporters when it comes to Holyrood, with many opting for the nationalists.

His remarks come as results of yet another poll, this time for Westminster, show the SNP having made significant advances amongst voters at the expense of Labour and the Lib Dems.  The survey, from an Angus Reid / Vision Critical, saw the SNP increase support from the last Westminster election by 13 points to 33% as Labour fell to 41% and the LibDems down to 6%.

To hear Professor Curtice go to the following link and fast forward 19 mins and 30 secs.