Thursday, November 15, 2018

Palpably pleasing Panelbase poll puts support for independence at eighteen month high

It's not very often that I'm given advance sight of a full-scale Scottish poll, so I was very grateful to the Scottish Independence Foundation for giving me a sneak peek a few days ago at the new Panelbase poll they've funded.  I was able to contribute some analysis for the press release they've just sent out.  Of most interest is the fact that support for independence is at its highest level in any Panelbase poll for eighteen months.

Should Scotland be an independent country?

Yes 45% (+1)
No 55% (-1)

In normal circumstances, 45% would be a disappointing Yes showing, but it's high by recent Panelbase standards.  Paradoxically, what used to be the most Yes-friendly polling firm during the indyref is now very much on the No-friendly end of the spectrum.  Yes support has been hovering at 43% or 44% in Panelbase polls since the spring of 2017.  Obviously a small increase to 45% is not statistically significant and may be caused by random sampling variation, but the fact that this result is even possible gives considerable reassurance after a recent Survation poll that put Yes on an unusually low 45%.  (Survation's normal Yes range is a bit higher than Panelbase's.)  So it may well be that Panelbase are just randomly showing a slightly higher Yes vote than usual, and that Survation just randomly showed a slightly lower Yes vote than usual, and that in reality nothing much has changed at all.

Slightly embarrassingly, even though I've already seen the Westminster and Holyrood numbers, I can't actually post them just at the moment, because I'm on my mobile phone and I can't seem to open the Excel file properly!  However, from memory, the SNP are on 37% for Westminster, which is a statistically insignificant 1% down on the last Panelbase poll.  Although 37% is exactly what they received at last year's general election, their lead over both the Tories and Labour is slightly higher than it was in June 2017.  On a uniform swing, the 9-point lead over the Tories would be enough to win back Stirling, and the 12-point lead over Labour would be enough to win back four Labour seats.  North-East Fife would remain on a knife-edge between the SNP and Lib Dems, meaning that the SNP would end up with either 39 or 40 seats, up from the current 35.

On Holyrood voting intentions, the SNP are two points down on the constituency ballot since the last Panelbase poll, but two points up on the all-important list ballot.  I ran the numbers through a couple of seat projection models, and they both put the SNP on 57 seats (significantly better than the 52 seats projected by the Record from their recent Survation poll) and the Greens on 4 seats.  So the pro-independence parties would have 61 seats in combination - just 4 short of maintaining their overall majority.

*  *  *


  1. Should Scotland be an independent country?

    Yes 45% (+1)
    No 55% (-1)

    Somewhat disappointing from my perspective. Perhaps James could shed some light on the following:

    1. One would think that nearly all EU foreign nationals who voted NO in indyref1 (the great majority) would vote yes today. That should translate to a mighty boost in the YES vote showing in the polling? True/false? Unless of course the data are not weighted in a manner that takes account of the change in the voting preference of that specific cohort (around 163k IMS) - I don't know - perhaps they are not being sampled or specifically identified. 163,000 is a lot of votes. Perhaps I've got the completely wrong, somewhere.

    2. The cohort of voters 65 and above voted 75% NO if I recall correctly, in 2014. There will be a natural rate of attrition and replacement, where those replacing the departed are coming from a pool of once younger voters with a historical voting bias more favourable to YES in 2014. Now, if as this younger cohort transitioned to 65+, the on the whole, maintained a more YES-friendly ratio than that of the 2014 cohort, one would expect to see a drift northward of the YES vote, which over the four years might be cumulatively significant?

    3. Even when not referencing brexit in the question would it nevertheless be reasonable to expect to see *some* uptick in the YES vote given the political and economic chaos and uncertainty abroad "in the zeitgeist" created by brexit?

    These factors taken together lead me to expect ... well, maybe better YES outcomes in the polls than we are seeing.


  2. How can a polling firm be palpably no-friendly , I thought this was meant to be science? You imply they are doctoring the survey to suit one side , can you explain how they can do this? I'm not looking for a 'select the respondents' well reply please. Genuinely interested.

    1. No, I did not imply that they are "doctoring the survey to suit one side".