You know this Yes 48%, No 52% Panelbase poll that unionists are crowing about? Does anyone know if it's the Alba-commissioned poll, or is there another one? Because I think there may be a discrepancy in the numbers.
— James Kelly (@JamesKelly) October 16, 2022
According to Wikipedia it's the Alba-commissioned poll (yes 46%, no 49% before DKs removed). Another one commissioned by Business for Scotland had both tied on 45% each.
— Mike Auld (@mikeinsdevon) October 16, 2022
If this is true, I've seen the data tables of the Alba-commissioned poll and the figures are Yes 49%, No 51%. So unless Panelbase have had to revise the numbers for some reason, it looks like people are lying about it.https://t.co/yI03cWlbPG
— James Kelly (@JamesKelly) October 16, 2022
— Mike Auld (@mikeinsdevon) October 16, 2022
What the...? Had some random guy on Twitter unilaterally decided to change the published numbers? Has the Express based their entire article on that? Have four unionist politicians gone orgasmic just because of that? This is nuts.
— James Kelly (@JamesKelly) October 16, 2022
The only other explanation I can think of, and I hope it's the correct one because if it's not we're going to have to fire off more complaints to IPSO, is that there's been a third Panelbase independence poll this week that took place more or less simultaneously with the Alba-commissioned one. But I've got the data tables of the Alba poll right in front of me, and I can tell you they undoubtedly show Yes 49%, No 51%. I've checked the raw numbers and they are correctly rounded to those percentages. (After weighting, 407 respondents were Yes voters and 431 were No, which works out as Yes 48.6%, No 51.4%.)In relation to my three previous tweets, I don't think I can adequately convey how bizarre and extraordinary this is. It really does look like some guy has changed the results of a poll on a whim, the Express have reported it, and unionist politicians are celebrating it. Insane.
— James Kelly (@JamesKelly) October 16, 2022
Looks to me like British Electoral Politics has taken the rounded numbers with don't knows included and calculated percentages for don't knows excluded. 49/(46+49)=0.5158 which he/she has rounded to 58%.
ReplyDeleteExactly - the snag being that the correct figures for Don't Knows excluded had already been published (The National were the first to publish them, as far as I know).
DeleteThe wikipedia pages show the sample size as "1000~" i.e. it's an estimate. Meanwhile Panelbase themselves have not (yet) listed the Alba poll on their own website.
ReplyDeleteA "Believe in Scotland" poll is shown in wikipedia as being split 45/45 (including DKs) with a sample size of 1017. However, a "Sunday Times" poll is on the Panelbase website with the same sample size (1017) as wikipedia has for "Believe in Scotland". The data tables on the former, however, do not have any Independence question, just the party voting intentions at the next UK General Election (see https://drg.global/our-work/political-polls/). Both have field work dates as 4-7 October.
Cockup or conspiracy?
Given the 'Matchetgate' fiasco a couple of years ago and more recent Sun and Express reportage of the Deltapoll survey of 14-16 September you can't rule out the conspiracy possibility.
Further to this I have been in touch directly with Panelbase.
DeleteThey have informed me that The Sunday Times and Believe in Scotland surveys published on 7th October were in fact a single survey but with different questions asked by sponsors - the Independence question was posed by BiS. Fieldwork was 5-7 October. The data tables for the BiS questions are now up on their website.
The Alba Party poll details have now also been published on Panelbase' website. These are also part of "omnibus" survey jointly commissioned with The Sunday Times. These confirm the 49/51 Yes/No split once don't knows are removed. There were 1018 people surveyed with fieldwork from 7-11 October.
So I think that should clear up the confusion (although wikipedia, as this point in time, has still not been updated to reflect all the Alba polling sample size correctly).
I make it that the Panelbase polls shows Yes at
49% (Apr), 51% (Jul), 49% (Aug), 50% (7th Oct) & 49% (11th Oct).
That is, effectively and as you say, no change.
Really depressing there is no change given the current state of the British government.
Delete