Survation, the regular (internet) pollsters for the London newsdesk of the Mail on Sunday (though not of the desk in Scotland), have entered the referendum polling fray for the first time this weekend.
They put the Yes vote at 32%, No at 52%, with 16% saying Don’t Know (after taking into account people’s reported likelihood of voting). Once the Don’t Knows are excluded, that means Yes are on 38%, No on 62% – almost exactly in line with the average reading of the six polls conducted since the publication of the Scottish Government’s White Paper in November. Of course, as Survation have not previously conducted a poll on referendum vote intentions, we do not know whether or not the poll is further confirmation that there has been a swing to Yes.
However, there is one feature of the way in which this poll has been reported that should be noted. The poll has been weighted by how people said that that they voted in the 2010 UK general election. On the face of it that seems a sensible correction, as the sample contains slightly more people who said that they voted SNP in 2010 (33%), than Labour (32%), when of course in reality Labour were ahead by 42% to 20%. However, this correction only makes sense if we can be sure that some people were not muddling up what they did in 2010 with what they did in 2011, when it was indeed the SNP who were (well) ahead.
Two previous polls that asked the same people how they voted in both the 2010 Westminster and the 2011 Holyrood elections discovered that people’s memories of what they did on the latter occasion seemed to be more accurate than their recollection of what they did in 2010. Both ICM and Panelbase discovered that what people said they did in 2011 proved to be much closer to the actual outcome than what they said they did in 2010. That evidence strongly suggests that people have a better memory of what they did in 2011 – and leads some to tell pollsters that they had voted SNP in 2010 when in fact they had not done so. If that is the case then weighting by how people said they voted in 2010 can introduce a bias rather than correct one. Certainly all other pollsters who opt to weight their samples by past vote have for some months now all chosen to do so by asking people what they did in the Holyrood election of 2011 rather than how they voted a year earlier
The effect of Survation’s weighting, predictably, has been to increase the estimated No lead considerably. Before the data were weighted at all (including to make sure that the sample matches the demographic profile of Scotland), the Yes tally stood as high as 43% (after the Don’t Knows are excluded). The weighting has knocked as much as five points off that figure. It looks highly likely that if Survation had followed the same practice as most other pollsters, the reported Yes vote in this poll would have been over 40% – just as it was in last weekend’s ICM poll and is in this weekend’s TNS BMRB poll.
Survation’s poll also asked people how they would vote if the Conservatives were expected to win an overall majority at the 2015 UK election. That had only a modest effect on the proportion who said they would vote Yes and No, increasing the former tally by three points and reducing the latter by three. A more substantial effect was found when people were asked what they might do if they thought the Conservatives would be in power for the next 15 years. In that event 38% said they would vote Yes, 47% No – implying a 45% Yes vote once the Don’t Knows are left aside. But quite how anyone will know that we are about to enter a period of Conservative hegemony is not clear . A revival in Tory fortunes by September might perhaps be of some value to Mr Salmond, but he would probably be wise not to assume that the reaction north of the border would be sufficient to carry him over the winning line.
This blog is an expanded and corrected version of that published yesterday, in which I had inadvertently misread one of the columns in Survation’s tables.