If you don’t know what I’m talking about, there was recently a Scottish subsample from a GB-wide YouGov poll that had the SNP in third place. The exact figures were: Labour 35%, Conservatives 20%, SNP 19%, Greens 11%, Liberal Democrats 11%, Reform UK 1%. Unionists have tried to put their gloating into the ‘mature’ context that subsamples carry a health warning and that this is merely an indication of which way the wind is blowing. But actually it’s not even that.
In other circumstances I might take a YouGov subsample a little more seriously, because unlike most other firms, YouGov seem to structure and weight their Scottish subsamples correctly, leaving the small sample size as the only remaining problem. In the absence of full-scale Scottish polling, then, YouGov subsamples can often be the most useful straws in the wind about what is going on. But that doesn’t apply in this case, because it just so happens that YouGov conducted a full-scale Scottish poll at exactly the same time as the GB-wide poll that the subsample is taken from. The fieldwork dates for the Scottish poll were 17th-20th April, while the GB poll was conducted on 18th-19th April. The Scottish poll showed an SNP lead of nine percentage points on the Westminster ballot, with Labour in second place and the Tories in a distant third. So the subsample can be safely disregarded.
In fact, even without the presence of a simultaneous full-scale Scottish poll, there would still be plenty of evidence that the subsample was a freak result caused by sampling variation. The poll it was taken from had the SNP on just 2% of the GB-wide vote, whereas every other poll from every other firm in this calendar year (including several polls conducted after the YouGov poll) has had the SNP somewhere between 3% and 6%.
Incidentally, a certain “newspaper” has claimed that if the YouGov subsample is right, the SNP would be totally wiped out in terms of Westminster seats. Even accepting for a moment the wildly implausible hypothetical of the SNP slumping to 19% of the vote, I don’t think that would wipe them out in the real world. It’s true that they suffer from first-past-the-post due to their vote being less geographically concentrated than Labour’s, and we know from past experience that it’s possible for a significant share of the vote to translate into zero seats (the Scottish Tories were on 17% when they were wiped out in 1997), but in practice I think the SNP would hold onto a seat or two in Dundee, and possibly Angus MacNeil might cling on with the help of his personal vote in Na h-Eileanan an Iar, a constituency that is often wholly decoupled from mainland trends.
Of course the crisis that the SNP are engulfed in has further intensified in recent days, but my guess is that a poll right now would still show a small SNP lead. But for the reasons I explained in my blogpost a couple of days ago, a small lead is no cause for celebration. Given the special difficulties the SNP face as a Scotland-only party in Westminster campaigns, it’s likely that a small lead a few weeks before polling day would translate into an election defeat of some description. The SNP need to face up to that problem by addressing the current crisis. That will require a bit of long-overdue glasnost, a bit of perestroika, and above all else a new leadership election, this time conducted fairly and without any efforts by HQ to load the dice in the continuity candidate’s favour.
