More on Glasgow East

Mike Smithson has managed to get hold of the tables for the Progressive Scottish Opinion poll in Glasgow East.

Looking at the tables, Progressive Scottish Opinion do not appear to have used any political weighting when weighting their sample. Politically their raw sample appears to have been pretty much the same as ICM’s, whose weighting had the effective of weighting the SNP sharply downwards and the Liberal Democrats very strongly upwards. As a result, the Progressive Scottish Opinion sample contained a lot more former SNP voters, and a lot fewer Liberal Democrats than ICM’s.

It also looks as though Progressive Scottish Opinion did not make any attempt to weight or filter people by their likelihood to vote. In a contest likely to be won on a very low turnout, this is questionable. However, while turnout filtering normally works against Labour, in ICM’s Glasgow East poll it helped them – their supporters were more certain to vote than the SNP’s.

The irony is that Progressive Scottish Opinion used a methodology that should have produced significantly better figures for the SNP than ICM’s method, yet they ended up showing a larger Labour lead than ICM did.

We’ll know if either of them are close to the result on Thursday. Personally I would be dubious about reading too much into any polls of the contest: the severe social deprivation and likely atrocious turnout make this constituency a formidable challenge to pollsters.


16 Responses to “More on Glasgow East”

  1. Sorry, I don’t quite understand – you say “Politically their raw sample appears to have been pretty much the same as ICM’s” – Did they actually ask how people voted in 2005? or were the unweighted crude figures similar to ICM?

    The raw data of the ICM poll showed SNP and Labour very close but the adjusted figures were more favourable to Labour partly because of the propensity to turnout but also because of the demographic adjustment (I do not know if ICM used a Glasgow East or Whole of Scotland mix)

  2. Sorry Stephen – I meant that the *unweighted* recalled past vote in the ICM poll was pretty similar to the *unweighted* recalled past vote in the PSO poll.

    Both ICM and PSO weighted their panel demographically to match the profile of Glasgow East (not Scotland as a whole, which would have been an unforgivable error) – the difference was ICM also weighted their sample politically (which hurt the SNP), but PSO didn’t.

  3. I thought this was nicely phrased, from the Gaurdian–

    ‘Indeed if Labour holds on to Glasgow East in the byelection this Thursday, Brown and his team are likely to see it as part of a general pattern of modest recovery. By all means let them enjoy their rare moment of success. But it is a weird week when Labour may celebrate the retention of its 25th safest seat as something of a triumph and when it is already treating its second worst ICM poll of all time as evidence that the Brown government is back on the bounce. Better, in the absence of stronger evidence, to remember the old saying that even a dead cat bounces once.’

    linka

  4. Realistically is there (and has there ever been) a chance that Labour would lose this?

    Polls have shown solid Labour leads, generally to the point that even large errors wouldn’t change the actual result.

    Political Betting has only really talked of the SNP being “good value” given the odds – very different to suggesting that they were going to win.

    There was the ePolitix “poll” of readers (effectively a voodoo poll Anthony given the self selected and predictive nature?) but other than that nothing.

    Yes it’s interesting from a polling perspective for a number of reasons but has there ever really been a basis for the media stories about Labour losing it?

    Or am missing something?

  5. My best guess is that Labour is in currently a few hundred voters ahead of the Nats.

    However this does not take account of undecided voters (do we have any number how many currently undecided voters are actually likely to vote?). Neither does it show about the strength of the conviction, and I would think that at the moment the SNP has a much larger chance of turning soft Labour supporters than the other way round.

    Where does that leave us? Labour goes into the polling booth with a slight advantage, but one that is possible to be overturned by very, very late decisions. The Scottish elections provided plenty of examples of both failed challenges (Cumbernauld) and surprise victories that are likely to have been caused by a strong SNP campaign first sowing the doubts with voters plumbing for change in the last minute (Edinburgh East).

    Still, I’ll go to bed early tomorrow, and get the result on breakfast telly (though if the Nats win even the neighbour’s wide screen will be too small to show all of Salmond’s grin)

  6. Here is my final prediction for this seat:-
    SNP = 42.0%
    Lab = 39.0%
    Con = 7.5%
    LD = 5.5%
    SSP = 3.0%
    Green = 1.5%
    Sol = 0.5%
    Ind = 0.5%
    F4C = 0.5%

    Turnout = 41.0%
    SNP Maj = 3.0%
    Swing = 24.0% Lab to SNP.

  7. Here’s my final prediction:

    Lab = 43.2% (-17.5)
    SNP = 35.1% (+18.1)
    Con = 8.7% (+1.8)
    LD = 7.3% (-4.5)
    SSP = 2.8% (-0.7)
    Green = 1.5% (+1.5)
    Sol = 0.7% (+0.7)
    Ind = 0.4% (+0.7)
    F4C = 0.3% (+0.3)

    Please note that the Green, Sol, Ind and F4C didn’t stand at the last General election in 2005. I have used the BBC style of regarding a party’s performance who didn’t stand at the previous election, as progress.

  8. Christian:

    As ever, I agree with every word you say. If the Nats win the Labour MP’s who helped in the campaign should go home and prepare for independence.

  9. SNP Win by just 305 votes but a 26% swing on a higher turnout than for Holyrood.

    Great result for us but close, still great for all those who have worked themselves in to the ground down there.

    Peter.

  10. Brian,

    Well done 42% to 39% was close to the final 43% to 41%.

    I think the difference is that you predicted the Tories to rise and they were static and LibDems and SSP were well down.

    Peter.

  11. Warmest congratulations to Cllr Peter Cairns and his party. Well done! :)

  12. Er Peter, a 22.4% swing? Interesting to note where the SNP swing came from – does this mean the SNP is effectively SDP+independence?

  13. So, two by-election polls, both widely off the mark (surely outside margin of error?).

    Was it that they asked the wrong people – Glasgow East just too difficult to poll? Was it late swing? Or something else?

    If it was undecided and soft labour support breaking to the Nats, then are there methods to find this out tghrough polls (or at the the possibility). For example, when asking for party preference, should pollsters also how strong the preference is (say on a scale of 1-10, similar to the question about likelyhood to vote at all)? And what about a forced choice between the front runners for undecided voters. Would that work?

  14. Christian – I don’t think it’s possible to say why the polls were so wrong, the socio-economic conditions in the constituency pose such a challenge to getting a decent sample in the seat that it’s impossible to rule that out as the explanation.

    I have to say I was very surprised that anyone did a poll in the seat. It was brave of ICM and SPO to tackle it, but I’m not particularly surprised it didn’t work out. I think in their shoes I’d have told the client that it would probably be impossible to get a decent sample and passed!

  15. I must confess I thought Labour would win. And I didn’t expect such a high turnout.

    This really is disasterous for Labour.

    The only very small crumb of comfort for Labour is that they might have narrowly won a seat with a similar majority in England because splinter further left parties like the SSP would not have got as many votes.

    I blogged yesterday in relation to the MORI July monitor about what Labour would need to do in terms of organisational behaviour to recover. The likelihood must be, particularly with the National Policy Forum coming up in the immediate future, that Labour faces recriminations and infighting, which will do it no good at all.

  16. Frederick:

    The SNP are a refuge for the disaffected leftish former Labour voter if the SSP and Solidarity arn’t available. One of these parties leadership is on the wrong side of perjury charges, but we don’t know which.

    Elsewhere they appear to have taken votes from SNP rather than Labour where they stand for the first time.

    If you don’t want independence, you vote against it in a referendum, not an election. There is no great harm in electing an SNP MP for a parliament he wants to leave.

    You are right about the organisational behaviour. That will do Labour more damage, especially if some are in denial, others panic and a few settle old scores irrespective of the damage it does to the party.

    Morale will collapse as will funding and election help.

    You don’t need anything more than that to lose an election.

    Surely nobody here believes elections are won do they? Or that the electorate has been persuaded of the new-found wisdom of the Conservatives, or that there is a surge in demand for independence in Scotland not showing in the polls?

    What do you think this result will do for “organisational behaviour” so far as the SNP are concerned?