As requested, here's more statistical detail on why Stew's nutty claim that the SNP will win an all-time record of 65 constituency seats, and no list seats at all, is so obviously wrong

Several of you have asked me to go into more statistical detail about my blogpost on Sunday, which pointed out why Stuart "Stew" Campbell was so obviously wrong in the angry claims he made about the supposed certainty that the SNP will fail to win any list seats at all in next year's Holyrood election.  

So I'll do that, but first of all it's worth pointing out that three days have now passed since that Sunday blogpost, and in that time Stew has continued to post repeated tweets about me (so good luck to him the next time he tries his "I don't stalk James, I barely even mention the guy" schtick!), and one of those tweets even references the Sunday blogpost, so it's highly likely that he read it.  But what we haven't seen from him is any substantive reply to the points I made in that post, most importantly the specific constituency seats I identified that the SNP are actually unlikely to win even though Stew insisted they were nailed-on certainties.  I think it's fair to say that if Stew was actually capable of providing a credible rebuttal of those points, he'd have done so by now.  The fact that all we've seen from him instead is a continuation of his usual "James is a raving lunatic" repertoire ought to tell his cult followers something rather important.

There was a huge amount of padding in what Stew called his "stats post", but his basic claim can be condensed to the following:

* The SNP are certain to reach the target of 65 seats for an overall Holyrood majority, and they will do it on constituency seats alone.

* They will have so many constituency seats that for the first time in their history they will not be allocated any list seats at all, and therefore all of their list votes will be "wasted".

That's an extraordinary and wildly implausible claim in umpteen ways.  It directly contradicts Stew's own insistence from five months ago that there is "zero" chance of a pro-indy majority at Holyrood after the election, let alone a single-party SNP majority.  It means he is predicting that the SNP will win a majority for only the second time in their history, and will do it on constituency seats alone for the first time ever (when Alex Salmond led the SNP to their only majority to date in 2011, he was nowhere near the target of 65 on constituency seats and required substantial numbers of list seats to get over the line).  And Stew is saying that all of this will happen in spite of the fact that the SNP are only polling at 33-34% of the constituency vote at present, which is between thirteen and fifteen percentage points lower than they received in the 2016 and 2021 elections, when they failed to win a majority.  It's all, to put it mildly, a bit bonkers.

The best way of visualising Stew's nutty claim that the SNP are guaranteed to win at least 65 constituency seats is to look at the eight other constituency seats that he is conceding they won't win or might not win. (By the way, all of this is massively complicated by the fact that there's a boundary revision going on, but it hasn't been completed yet so we just have to work with what we've got.)  As I understand it, the Stew Eight are: 

Dumbarton
Edinburgh Southern 
Caithness, Sutherland & Ross

North East Fife

Orkney

Shetland

Edinburgh Western

Ettrick, Roxburgh & Berwickshire

This means by Stew's own admission, if you can find seats the SNP are unlikely to win that are not on the above list, it debunks his claim that the SNP are guaranteed to win 65 constituencies and by extension invalidates his claim that the SNP will win zero list seats and that SNP list votes will be "wasted".

Finding examples of other seats is not at all hard to do, because literally every single seats projection from every single opinion poll is showing the SNP failing to win a number of constituencies from outwith the Stew Eight - hardly surprising given the big drop in the SNP constituency vote.  Here are some hard examples - 

East Lothian constituency, 2021 result:

SNP 39.2%

Labour 36.7%

Conservatives 20.7%

Liberal Democrats 3.4%

Now, come on, you don't need me to hold your hand here - you can see why the SNP's position is so vulnerable.  In a situation where both the SNP and Labour vote may fall, all that would have to happen to allow Labour to gain the seat is for the SNP vote to fall just that little bit more than Labour's.  In reality, opinion polls are suggesting the SNP's vote will fall far more than Labour's - the most recent Survation poll had the SNP down fifteen percentage points from 2021 and Labour down only three points.  On a uniform swing, Labour would gain East Lothian easily, thus driving a coach and horses through the Stew Theorum.  In truth, there's a good reason for thinking Labour might even outperform a uniform swing, because there's a substantial Tory vote in the constituency which they may be able to tactically squeeze.

Galloway & West Dumfries constituency, 2021 result:

Conservatives 47.0%

SNP 39.9%

Labour 7.9%

Greens 2.6%

Liberal Democrats 2.5%

It would be an exaggeration to call this a safe seat for the Tories, but they do have a bit of a cushion over the SNP and there would need to be a significant Tory to SNP swing for the SNP to be able to win here.  In reality, the polls are suggesting a swing from the SNP to the Tories.  Although the most recent Survation poll shows a big eleven point drop in Tory support since 2021, that's dwarfed by the fifteen point drop in SNP support, which means that there's a modest net swing from SNP to Tory of 2%.  The Tories ought therefore to hold Galloway & West Dumfries with a bit to spare, and the same is also true of three other Tory constituencies which Stew is claiming the SNP are certain to gain, namely Eastwood, Aberdeenshire West and Dumfriesshire.

Ayr constituency, 2021 result:

SNP 43.5%

Conservatives 43.1%

Labour 11.0%

Liberal Democrats 1.9%

Although this is an SNP-held seat, it's an ultra-marginal.  The 2% swing to the Tories implied by the recent Survation poll would overwhelm the SNP lead in the constituency and put it back in the Tory column.  For the same reason, the Tories would be likely to gain Banffshire & Buchan Coast from the SNP (even without any help from Christina "Of The Blood" Hendry).

All of these seats are outwith the Stew Eight, and therefore demonstrate why the SNP are likely to fall well short of 65 constituency seats and will thus have much more scope to pick up compensatory list seats than Stew's fraudulent "analysis" suggests - as long, of course, as SNP supporters actually vote SNP on the list and don't waste their votes on no-hoper fringe parties in the way that siren voices such as Stew's are trying to persuade them to do.

I've spoken to three or four stats-minded people since the weekend and all of them are utterly baffled as to how Stew arrived at some of what might laughably be called his "constituency projections", most particularly for the several seats that ought to be Tory but that he's inexplicably awarded to the SNP.  There's a lot of smoke and mirrors in his post - he goes into extensive detail when it suits him to give the impression that he's engaged in some kind of rigorous Curtice-quality statistical analysis, but for the substantial parts of his claim that make no sense whatsoever, he tries to paper over the cracks with throwaway, bombastic statements such as "there is no sane way to imagine [the Tories] holding the four seats we’ve turned SNP yellow".  Er, yes there is, Stew.  Yes, there is, and I've just explained it.

I confidently predict that Stew will provide no substantive response to any of the points raised above.  I also confidently predict that he'll instead treat us to another sixty-seven "James Kelly is a shambolic wreck of a gibbering imbecile" tweets.  It really is great fun, I'm not going to deny that I always thoroughly enjoy his epic meltdowns and temper-tantrums.  But all I'd say to his dwindling band of cult followers is this: if you enjoy that kind of thing as much as I do, fine, but don't ever come to me again and try to pretend with a straight face that your guy is some kind of serious, credible, rigorous "analyst" (let alone a "journalist").