TO be honest, I was extremely surprised that the new Scot Goes Pop poll, conducted by Find Out Now, has produced an outright majority in favour of independence.
This is the sort of result that we were seeing across all polling firms in the late autumn, just after the Supreme Court ruled that the UK Government essentially had a veto on the holding of a conventional independence referendum.
Since then, however, most firms have shown a swing back to No, and although there's been no consensus on how large that swing is, it's certainly been big enough to push No into the lead in almost all polls conducted so far in this calendar year.
So does this sudden bucking of the trend mean there has been a sudden resurgence for Yes, possibly caused by the buzz around the SNP leadership contest? Can we now expect a flurry of Yes-majority polls? My guess would be probably not, but that's not necessarily such bad news.
The only other poll in 2023 to show Yes in the lead was also conducted by Find Out Now, and in fact the new poll shows a slight dip in the Yes vote since that previous survey, so the trend is not totally out of line with what other firms are showing. The only thing that is unusual is the ongoing Yes lead, and that probably has a methodological explanation.
Like most polling firms, Find Out Now absolutely do adjust their result by past vote. I had the opportunity to watch the raw numbers come through in real time as people responded to the poll, and one thing that was painfully obvious throughout the process was that people who had voted SNP in the 2019 General Election were vastly over-represented in the unweighted sample.
Find Out Now entirely corrected for that, drastically reducing the number of SNP voters in the sample to bring them into line with the actual 2019 election result, which of course automatically reduced the size of the Yes majority in the published results. But what didn't happen, as far as I can see, was any adjustment by how people recalled voting in the 2014 independence referendum, and in my view that's absolutely as it should be. That kind of adjustment may have made sense in 2016 or even 2018, but nine years after the referendum it's getting a bit silly. It's hard to think of any other example of a polling firm using a past vote weighting based on a democratic event that took place around a decade ago, and yet most firms still do that.
One other firm that doesn't adjust by how people recall voting in the 2014 referendum is also one of the most prestigious, Ipsos UK. And it's notable that Ipsos also have a track record of reporting a higher Yes vote than other firms. So I'll just leave you with a tantalising thought that has entered my head a few times before.
Could it be that Ipsos and Find Out Now are getting it right by leaving out a 2014 adjustment, and other firms are getting it wrong? If so, there could be a No-friendly distortion in the vast majority of polls that we see, and it may be leading us to miss a fairly steady underlying Yes majority in the real world that has been around for years.
That's an exciting possibility in one sense, but it's also a terrifying one, because increasingly there's a political narrative that opinion polls can be considered reliable enough to be used as a barometer for whether there is sufficient Yes support to justify a second independence referendum. Why we have allowed that narrative to take root is something of a mystery, given the long and dismal history of polling failures in the UK.
