CrosbyTextor have sent be the tables for the marginal poll in the Sunday Telegraph. The actual voting intention figures, with don’t knows and won’t votes excluded, is CON 49%, LAB 20%, LDEM 21%, Others 10% – representing a swing of 10.9% from the last general election.

It was carried out between the 16th and 29th June, when the national polls were showing Tory leads between 18 and 23 points, equating to swings between 10.5% and 12.5%. Without knowing things like what weighting was used we don’t know exactly how comparable the figures are, but it certainly doesn’t look as though the Conservative swing is any larger in the easiest marginals than in the county as a whole (and – in the present political circumstances – it doesn’t need to be).

24 Responses to “More on Sunday’s CrosbyTextor poll”

  1. Anthony,

    i don’t suppose there is any chance of you posting the respective tables?

    Oh and as an aside, I think it might improve the site if you considered a By Election Link under the “Seat guide”.

    That way rather than have to look up a seat every time there is a by election you could get to current or indeed recent ones from the current parliament with a single click.

    Just an idea.


  2. Oh and I don’t suppose that after talk only 1 Labour MP in Scotland (yes I know it’s silly but they do it after every By Election) we could have a list of SNP/PC Target seats.


  3. Would also appreciate the tables , Anthony . Are there any clues as to whether LibDem support is down in the Lab/Con marginals and steady in the LibDem seats as would seem logical .

  4. Anthony,

    I thought swing calculation was the sum of the variance in vote share of the respective parties from the respective lection divided by 2. In which case (using national vote share figures?) isn’t the swing herein in comparison to 2005 national shares 16% (Con +16, Lab -16)?

    I appreciate you might be using the 2005 figures that apply solely for the 30 constituencies but then how can you compare with the national polls as it does not seem like for like?

    Perhaps you can provide some further clarification?

  5. Political Betting is reporting the result of a Populus poll for tomorrows Times.

    Con 43% (+1)
    Lab 27% (-1)
    LD 17% (-1)

    The fieldwork for this was carried out over the weekend so any post Glasgow East impact should have been apparent but in line with all other recent polls it looks like things have settled for now.

  6. I agree with Peter Cairns and Mark Senior[?!], the tables would be useful when trying to interpret the expected outcome of a/any poll. Without it we are left with Jack W’s ARSE-adjusted Wells/Baxter results…!

  7. The correct figures are 43/27/18 see Times article .
    Abthony , Mike Smithson has kindly emailed me the table . Strange form of question wording ” Who would you vote for or lean towards ” sampling done Mid June .

  8. jsfl –

    The swing for a poll of these 30 seats should be from the result in those same 30 seats in 2005 – so 10% or thereabouts.

    The swing for the national polls needs to be compared to the national shares of the in 2005.

    You can compare the two swings no problem – same way we can compare the swing in one seat to the swing in another.

    Peter – good idea about the by-election list. I’ll try and stick the tables up tomorrow.

    Mark – I was hoping for a break for the Con/Lab seats and Con/LD seats, but as you’ll know now – there isn’t one.

  9. jsfl,
    In order to calculate the swing from 2005 in these marginal seats, we need to have the party shares that occurred in these constituencies – not the national shares.Given that Anthony is quoting a 29 % Con lead over Labour and a swing of approximately 11% from Lab to Con since 2005, the implication is that in 2005 the Tories enjoyed a 7% lead over Labour in these paricular seats!

  10. Abthony Ditto in hoping for Lab/Con and LD/Con split . Still puzzled by the question asked , seems to be a voting intention plus double squeeze with no difference in weighting towards those giving a definite voting question .

  11. Graham – The share of the vote in these seats at the last election would have been CON 38.1%, LAB 30.8%, LDEM 24.5%, Others 6.6%.

  12. Clearly even if the Lib Dem vote is holding up better in the Lib Dem Tory marginals – that doesn’ mean the Lib Dems would win these seats were a general election held tomorrow.

    I would guess that the figures are something like

    Con/Lab marginals
    was Lab 40% Con 38% Lib Dem 16% others 7%
    now Lab 27% Con 50% Lib Dem 12% others 11%

    Con/Lib Dem marginals
    was Con 38% Lab 12% Lib Dem 42% others 1%
    now Con 47% Lab 6% Lib Dem 38% others 9%

  13. We can all have a guess and come up with different answers , what we need are some clues but unfortunately this poll does not give us anyway .

  14. Well of course the poll gives some clues.

    If the Lib Dems have 21% accross 30 seats there are certain parameters.

    The seats are divided 1/3 Lib Dem/Tory 2/3 Lab/Conservative marginals.

    The Lib Dems polled 24.5% overall in these seats last time and a look at the election results can give you the exact score for the Lib Dems in both type of seat.

    Comparing the result last time with the poll means that overall the Lib Dems have lost 3.5% support.

    If their support in the Lib Dem/Con marginals is unchanged then support for them in the Lab/Con marginals must be down 5.25% (so the overall total is down 3.5%)

    If their support in Lab/Con marginals is unchanged, then their support in Lib Dem/Con marginals will be down by 10.5% to compensate. I doubt if anyone thinks Lib Dem support is higher in either type of seat than it was at the general election

    Labour are down 11% overall, that could be a uniform 11% drop, but more likely, the drop it will be less where they had less support in the first place.

    The Labour vote share in Con/Lib Dem marginals was about 13% in 2005, making an 11% swing impossible in some places.

    Therefore, Labour support is down by more than 11% in Lab/Con marginals, and correspondingly, conservative support will be up more in Labour/Conservative marginals. As a consequence, Conservative support will be up by less in Lib Dem/Con marginals.

  15. Surely all the polls are showing Lib Dem support down since the 2005 election, so it’s reasonable to assume that Lib Dem/Con marginals are affected aswell, unless there is some clear evidence otherwise.
    Looking at the Lib Dem losses to the Tories in 2007 and 2008 it would appear to contradict that.

  16. This question of LibDem support is vexing, partly because the dynamic of support over the electoral cycle is different and partly because their dynamic of support varies according to which of the larger two partoes is in power and whether that party in on the way in or the way out.

    The 1979 election is the best and most recent comparison for the third party, but even so the then Liberal party was not the beast (fluffy bunny?) it is now, having since amalgamated with and integrated a more leftish influence from the SDP.

    So to a large extent this is untested waters and we are all hypothesising irrationally about the impact of any swings.

    I don’t think any swings regarding the LDs are coupled to the fate of the government of the day as they consistently play on their ‘local’ credentials, so it would be odd for opinion in Con/LD marginals to motivated by anti-government sentiment to as large an extent as in Con/Lab marginals.

    My feeling is that in all marginals the third-party locally will be squeezed, thereby neutralising the effect of any swing for or against the third party nationally. The consequence of this would be to redraw the battlelines for every party so that they concentrate their resources where they are least likely to waste efforts, thereby modulating opinion polls swings and minimising both expected losses and expected gains.

    The additional factor which plays into the hands of the LDs and is continually underplayed by poll-watchers is how the additional publicity of a general election campaign equalises the benefit to all opposition parties. LDs have shown themselves to be able to make consistent gains in the polls through the course of elections which can probably be accounted for the DKs making their minds up disproportionately.

    So it is interesting to watch the number of DKs this far out from a GE, as this would imply potential growth territory for the third party the closer we get to the ballot.

    Therefore a high level of confidence among Conservative supporters in their ability to gain a strong overall majority may play into the hands of a scattering of LD targets seats, while conversely the threat of a hung parliament forces the question of preference and will likely favour the party with momentum.

    I suggest that current Conservative opinion poll leads are on the cusp of making a vote for the LDs safer and more profitable for potential LD supporters.

  17. Mark Senior

    We can all have a guess and come up with different answers.

    Yes! But we can all be silly. The facts we are playing with are vague. Otherwise, why are we here?

  18. The Lib Dems reached 20-21% in some polls during the closely fought 1992 campaign, and ended with 18.3%.

    The Lib Dem emphasis on local issues and local personalities is well known but is vulnerable in an election where people really want to change the government – if it comes about.

  19. Joe James B , there were negligible losses to the Conservatives by the LibDems in 2008 local elections from memory around 70 Con gains and 55 Con losses from/to the LibDems and the 3 LibDem marginal seats in these 9 which had local elections this year ( Eastleigh , Westmoreland and Cheltenham ) all had LibDem advancing and Conservatives losing ground . It is true that Romsey had partial elections in the Southampton wards and Conservatives did advance strongly there .

  20. jjb, are you trying to agree or disagree? It isn’t clear.

    I too can read the historical statistics provided as a resource here and interpret them independently, but you seem incapable of accounting for normal statistical problems such as margin for error and selectivity.

    Surely you accept that over the course of the 1992 campaign third party popularity did in fact match the general rule that it increased as the campaign wore on – or are you trying to propagate confusion and roguery by claiming exceptionalism as a more accurate rule? If so, please could you provide some evidence to support your otherwise obviously biased and inaccurate theory?

  21. “…but you seem incapable of accounting for normal statistical problems such as margin for error and selectivity….”

    That statement is uncalled for – in fact it is part of my job. Thomas should stick to facts and avoid abuse.

    On that basis, probably the majority (but not all) of LD ratings in the 1992 campaign from I think 15 to 21% were within a margin of error.
    The LDs started the 1992 campaign on 15-17% and ended on 18.3, which is still an increase, within a margin of error,
    but there was a clear point in the campaign when they reached 20-21 for a while (not just the odd rogue), and were squeezed at the end.
    (Robert Waller, who I have huge respect for, verbally told me he thought the LDs would [I think probably] reach 23% about 2 weeks before the 1992 election day].

    As for the local elections, we saw Lib Dems losing heavily in a number of authorities where they had done considerable damage to the Tories in 2003. They were lucky that the large 2007 round wasn’t fought in 2008.

  22. Ok – it wasn’t abusive,
    but I do resent your comment about incapable of dealing with stats as I’ve done it since I was a teenager.

  23. I can’t find the figures archived,
    but suspect the Alliance didn’t increase their support during the 1987 campaign.

    Labour actually entered the 1987 campaign from a lower position than in 1983, after Greenwich and a fall in the local government elections from 34.5% to 31%.
    The Alliance was about 25%, maybe a bit more, at the start, drifted down to 21% and ended up on 23.1

  24. I agree, JJB, I ought to withdraw the comment, but because I think it wasn’t entirely accurate, not because it was abusive (it wasn’t). I merely suggest as someone who is and hopes to remain unaffiliated that your decent analysis was allowed to be coloured by your partisan bias when you reached the stage of drawing a conclusion. So, perhaps it shames me slightly to admit being slightly provocative in attempting forthrightness.

    The Alliance was a different kettle of fish in ’87 being an unresolved and incoherent rump of discontents and should be treated as two halves of the party it essentially developed into. Therefore the logic which should prevail is that the the general rule of the third party to which I allude above would be inverted as the incoherence was exposed to the public and proved a vote loser.

    My interest in minor parties stems from how they change the dynamic of electoral calculations and turn a duopolistic zero-sum game into a valid competition, which the traditional big-two should ignore at their peril, though I suspect you and they each secretly habour conflicting viewpoints about how to deal with the reality of their existence and the additional complexity they provide.

    Thus subjects such as sleaze, corruption and institutional bias/failure should be treated as doubly risky by potential parties of government in that they undermine the basis of established norms and wider civic society – which can be seen in the growth of ‘others’. Any political capital which is gained at a competitors expense is a short-lived gain more akin to debt to be repaid with interest than any credit due to their own efforts.

    Anything less than a wholly accurate disinterment of fact is simple distraction from truth.