General article(s) on Polls

[The Bishop]

Amidst all this hand wringing, let us recall there was an important part of the UK where Labour support wasn't overestimated by much if at all.

How come Scotland was able to get it basically *right*? What applied there that didn't elsewhere, or indeed vice versa.

And tbh I wouldn't overdo the idea of a universal international trend either.

If there really is a generic "overstatement" at present, it is arguably of populist right parties like the FN, SD and AfD.

[carlton43]

Aug 29, 2024 10:23:26 GMT The Bishop said:

Amidst all this hand wringing, let us recall there was an important part of the UK where Labour support wasn't overestimated by much if at all.

How come Scotland was able to get it basically *right*? What applied there that didn't elsewhere, or indeed vice versa.

And tbh I wouldn't overdo the idea of a universal international trend either.

If there really is a generic "overstatement" at present, it is arguably of populist right parties like the FN, SD and AfD.

And you make that final assertion of what evidence?

[The Bishop]

The evidence of actual recent elections in France, Germany and Sweden?

[iain]

I also find the idea of a general international trend pretty unconvincing, with cherry picked examples. The recent general elections in France and Spain both had a polling error which overestimated the right for example.

[London Lad]

Aug 29, 2024 10:23:26 GMT The Bishop said:

Amidst all this hand wringing, let us recall there was an important part of the UK where Labour support wasn't overestimated by much if at all.

How come Scotland was able to get it basically *right*? What applied there that didn't elsewhere, or indeed vice versa.

Perhaps Scotsmen (and women) are just more honest when it comes to answering polls? - the 'shy Tory' doesn't exist

[The Bishop]

Aug 29, 2024 10:30:07 GMT iain said:

I also find the idea of a general international trend pretty unconvincing, with cherry picked examples. The recent general elections in France and Spain both had a polling error which overestimated the right for example.

They seem to lean quite heavily on the 2020 US election, when polling was very obviously complicated by Covid.

[jamie]

Aug 29, 2024 10:30:07 GMT iain said:

I also find the idea of a general international trend pretty unconvincing, with cherry picked examples. The recent general elections in France and Spain both had a polling error which overestimated the right for example.

I’m also not sure where this idea the AFD is systematically overestimated comes from. A quick check shows there’s a couple of nationwide elections where they were slightly underestimated, at the last federal they were slightly overestimated, and they were noticeably overestimated at basically every COVID era state election. 

[The Bishop]

Aug 29, 2024 21:14:07 GMT London Lad said:

Aug 29, 2024 10:23:26 GMT The Bishop said:

Amidst all this hand wringing, let us recall there was an important part of the UK where Labour support wasn't overestimated by much if at all.

How come Scotland was able to get it basically *right*? What applied there that didn't elsewhere, or indeed vice versa.

Perhaps Scotsmen (and women) are just more honest when it comes to answering polls? - the 'shy Tory' doesn't exist

"Shy Tories" have not meaningfully existed as a polling factor since the 1990s, and their continued invoking is a classic "zombie" myth that stubbornly refuses to die. But if such a thing *was* significant, you would surely expect it to be *more* of a thing in Scotland rather than less (for fairly obvious reasons)

[john07]

The difference between polls in England and those in Scotland appears blindingly obvious to me. The longer the campaign went on the clearer it became that the Tories were going to get a hiding in England. That would have been factored in by many potential Labour tactical voters who will have decided that their vote could safely stay with the Lib Dems, Greens, etc.

In Scotland there was not the same where of certainly about the collapse in SNP support. I only became convinced from the excellent updates provided by ntyuk1707, and many of those came late in the campaign. Hence the Labour tactical voters were less included to switch back late on.

[Sibboleth]

Aug 29, 2024 8:21:58 GMT hullenedge said:

A bit more from yesterday:-

There's a global bias to the Left.

If we start from 2010 or thereabouts the obvious observation is that the British polling industry is, frankly, shit, and that the question is generally not whether it will be wrong or not but exactly how it will be wrong. That's probably a little unfair: what they produce usually tells you which way the wind is blowing. But almost never to what degree.

[hullenedge]

Last chance saloon?

Let's hope that it is fixed. The continentals are far superior with their polling accuracy.

[The Bishop]

Again, the "across nations" bit does not convince at all. And even here, neither 2019 or (especially) 2017 had a "left wing bias" in their polling.

Looks like some pollsters are over-reacting tbh - especially since this wasn't a GE where they got the actual *outcome* wrong.

[Sibboleth]

Aug 31, 2024 11:13:25 GMT The Bishop said:

Again, the "across nations" bit does not convince at all. And even here, neither 2019 or (especially) 2017 had a "left wing bias" in their polling.

Looks like some pollsters are over-reacting tbh - especially since this wasn't a GE where they got the actual *outcome* wrong.

I think it's in part a reflection of the commercial pressures that the industry is under: there are clearly far too many polling companies presently in existence conducting far too many polls and this at a time when the industry's main sponsors are themselves increasingly less than commercially viable. But it is always eyebrow raising to see exhortations about the need to be frank and honest be accompanied by claims that are neither of these things: since the first election in which a significant proportion of the electorate was not reachable via traditional means (2010) polling has been pretty bad - except in the loose 'which why is the wind blowing sense', which, as you say, is actually quite valuable - with the solitary exception of the very odd, very sui generis 2019 election. And the errors have not been made in a consistent direction. Of course that doesn't lead itself to any quick and easy 'fixes', but then if the situation is as bad for the industry as it quite probably is, then I'm not sure how peddling false solutions is especially helpful either.

[aargauer]

Aug 31, 2024 11:13:25 GMT The Bishop said:

Again, the "across nations" bit does not convince at all. And even here, neither 2019 or (especially) 2017 had a "left wing bias" in their polling.

Looks like some pollsters are over-reacting tbh - especially since this wasn't a GE where they got the actual *outcome* wrong.

To me 2024 had the feeling of a late swing of sorts. It was obviously heading towards a heavy labour win, so people like myself dragged ourselves out and voted basically against a hypermajority rather than for the Tories, whilst a lot of weak labour pledges stayed at home.

The electorate were supporting Labour by greater than the votes cast. Differential turnout on the day reduced that margin.

It's going to make a swing back hard.

[hullenedge]

[robert1]

Who will come up with the fastest poll and most crass question to provide an 'analysis' of what the public think of the Budget? 

Suggestions in advance greatly appreciated. 

[cathyc]

Oct 30, 2024 9:45:18 GMT robert1 said:

Who will come up with the fastest poll and most crass question to provide an 'analysis' of what the public think of the Budget? 

Suggestions in advance greatly appreciated. 

Almost certainly it will be Matt Goodwin.  I suspect it's already written.