Wednesday, June 7, 2017

Election Eve : Polling Roundup

We're midway through the barrel-load of GB-wide polls we're going to see tonight, so I'll use this post for rolling updates.  So far we've only seen three of the most Tory-friendly pollsters, and the picture they've been painting is one of stability -

* ICM have the Tory lead edging up 1% to 12%, but that could well just be margin of error noise.  It's also only 1% better than the lowest Tory lead ICM have reported at any point during the campaign.

ICM Scottish subsample : SNP 47%, Conservatives 30%, Labour 21%, Liberal Democrats 2%

* ComRes have the Tory lead falling from 12% to 10%, which could also be margin of error noise - but on the other hand it may be more significant than that, because 10% is the lowest lead ComRes have reported so far.

ComRes Scottish subsample : SNP 45%, Conservatives 27%, Labour 16%, Liberal Democrats 9%

* Panelbase show the gap steady at 8%, which means the Tories have failed to reverse the big drop in the last poll.  It's hard to make a meaningful comparison with the Panelbase polls in the early part of the campaign, because the firm's methodology was overhauled a couple of weeks ago to make it more similar to ComRes and ICM (which automatically made it more Tory-friendly).  However, it seems likely that if the new methodology had been used throughout, the Tory lead would never have been lower than it is now.

Panelbase Scottish subsample : SNP 40%, Conservatives 30%, Labour 24%, Liberal Democrats 4%

UPDATE : Here comes more...

* Kantar/TNS have a potentially significant poll showing the Tory lead cut in half from 10% to 5%.  That may exaggerate the trend because the previous poll from the firm showed a slight increase in the Tory lead, but nevertheless 5% is comfortably the lowest gap they've reported so far, and arguably moves the race into hung parliament territory.  Some of the fieldwork is a little out of date, though - although the poll concluded today, it's been running for a full week.

Kantar/TNS Scottish subsample : SNP 38%, Labour 31%, Conservatives 21%, Liberal Democrats 7%  (Note : This is a very rough estimate based on the raw numbers in the datasets, and doesn't take turnout weighting into account.)

* YouGov, which in recent weeks has become regarded as one of the most Labour-friendly pollsters, has unexpectedly shown the Tory lead jumping back up from 4% to 7%.  That's a bolt from the blue, because this afternoon's update of the YouGov projection model continued to show a 4% gap in the popular vote.  The fieldwork for the new poll is probably very up-to-date, so we may be looking at the first evidence of a late and decisive swing away from Labour...or it may be meaningless noise caused by random sampling variation. (Update : It turns out YouGov have made methodological changes for this poll, so that may be the real explanation.)

YouGov Scottish subsample : SNP 41%, Conservatives 28%, Labour 22%, Liberal Democrats 7%

* For what it's worth, some firm I've never heard of called Qriously is actually putting Labour ahead by around 3%.  The poll is being billed as "the first poll of the campaign to show a Labour lead", but as this firm are presumably very inexperienced in the field of political polling, a large helping of scepticism is warranted.  It's highly unlikely we'll get Scottish subsample figures, but the SNP are on 3.8% of the GB-wide vote, which perhaps puts them somewhere in the 40s in Scotland.

* Leaving aside unheard-of newcomers, Survation have firmly established themselves as the most Labour-friendly pollster in recent days.  They haven't followed YouGov back into the pack tonight - their final poll continues to show a miniscule Tory lead of only around 1%.  In the unlikely event that this is a fully accurate call, Jeremy Corbyn would probably become Prime Minister with the SNP's help.

Survation Scottish subsample : SNP 41%, Labour 29%, Conservatives 25%, Liberal Democrats 5%

* Oddly, The Herald (yes, our Herald) have commissioned a GB-wide poll from BMG and it shows a 13% Tory lead.  I don't think that really tells us anything other than that they must have used similar methodology to ICM and ComRes.

*  *  *

The average Tory lead in tonight's polls is 8% if you don't include Qriously, and around 6.6% if you do.  The latter figure is on the fringes of potential hung parliament territory, but would be more likely to produce a modest overall majority.  However, there's no particular reason to attach magical significance to the average lead - it may well be that the most accurate pollster will be the one on either the Labour-friendly or Tory-friendly extreme edge of the spectrum.  (And of course most pundits will tell you it's going to be the latter.)

The average of the six Scottish subsamples that have appeared tonight is...

SNP 42.0%
Conservatives 26.8%
Labour 23.8%
Liberal Democrats 5.7%

Those numbers are eerily close to the findings of recent full-scale Scottish polls.  Due to the even split in the unionist vote, they'd probably produce a pretty decent result for the SNP in terms of seat numbers.


  1. what size majority 8- 10%


  2. The Independent reported today that a record breaking number of people registered to vote just before the deadline.
    It's not unthinkable that these are youngsters who plan to vote for J-Corb.
    Please God......

    1. Aye, but the youngsters do actually have to get to the polling station! I remember watching the independence referendum from here in NC and being stunned that there was a party in George Square on voting day.

  3. There was an interesting thing I read suggesting that voter registration among younger voters (18-34) was up quite a bit while older generations had fallen, will see if I can find it.

  4. Hi James - It just looks to me as if YouGov are totally hedging their bets. If the election is close, then they will point to their "fabulous" model which shows a hung Parliament. But if there is a big Tory win, then they can point to the final poll and Anthony Wells' projection that the Tories will increase their majority. I know pollsters are nervous about repeating their mistakes of 2015 (and 2016!) - but it all becomes meaningless if they can predict two completely contradictory results...

    1. Yes, that thought occurred to me. Anthony Wells' piece places suspiciously large emphasis on the seats projection rather than on the lead in the popular vote (which they probably privately suspect they're still underestimating). But even on the seats projection, they're now simultaneously forecasting a hung parliament and an increased Tory majority. It'll be interesting to see if there's a sudden change in the projection model tomorrow.

  5. I have found the polls particularly unilluminating this election.

    Is there likely to be a final Scottish poll from anyone?

  6. I'm a little disappointed with the UK Wide Polls, even though Kantar, Survation and Qriously report its very close and YouGov possibly covering their arse however the SNP Vote looks to have held up reasonably well in the face of a massive Unionist onslaught and I'd love to see the Tories end up in 3rd..... only one poll matters now folks!

  7. James

    I think that BMG have also published a Scottish poll - 42% SNP, 27% Cons and 21% LAB and 8% LIbDems. But fieldwork is quite old...

    1. Is there a link? I've had a look, but I can't find it mentioned anywhere.

    2. James - It's referred to in the Britain Elects Twitter feed. However I couldn't find any data tables for it.

  8. So essentially the final sub sample averages are:

    SNP: 42 Con: 27 Lab: 24 Lib: 6

    Talking to young people in my college class today, most of them are independence supporters who usually vote SNP but are voting Labour as they believe only they will stop the Tories. I've been getting jittery that there could be a last miniute swing to Labour that will put central belt seats at risk, however there's been no evidence in the recent polls or YouGov model for this, the last sub sample averages look pretty stable and in line with full scale polling also. The SNP lead over Labour appears large enough to save all or nearly all seats still. Obviously my college class doesn't represent everyone, but I think it's very likely Labour are indeed ahead among the youngest voters in Scotland.

  9. There's also a final Scottish telephone poll from Survarion:

    SNP: 39%
    LAB: 29%
    CON: 26%
    LDEM: 6%

    No idea of the fieldwork dates. But since Survation are at the Labour friendly end of the spectrum in GB it's very possible the same is also true of Scotland too. However if this turn should out to the case then some Central belt seats could be at risk.

  10. Fin that would result in a lower national vote share for the SNP, but zero loss of seats. We just can't afford to have young people voting Labour in marginals like Perth ,Edin West, Aberdeen etc.

  11. Thank you for nice post.we think new idea for a new electoral system.