A poll with holes?

26th April, 2012 10:24 am

There has been predictable crowing from the Tories this morning over a poll that puts Boris Johnson 8 points ahead of Ken Livingstone with just a week to go in the race. That’s not a completely crazy result on its own – as I said when YouGov reported just a 2 point lead on Monday “I fully expected the polls to have worsened for Livingstone” – and that’s what ComRes is reporting.

However, when you look at the breakdown by borough you get some unusual results that require a few leaps of the imagination.

According to the ComRes numbers:

Boris leads in Hackney and Tower Hamlets

and

Ken leads in Wandsworth and Kensington and Chelsea (where ComRes reports he has 100% of the vote!)

I’m not saying that these numbers are definitely wrong – but London is a huge and complex electorate, and each borough has its own particular political micro-climate.

I’m not convinced the polling is necessarily picking up what is happening on the ground.

To report anything from the comment section, please e-mail [email protected]
  • Tom

    Eh, you can’t really do that sort of thing with sub-samples of polls.
    For a start, sub-samples involve tiny, tiny numbers of people.  Of the ones you mention, the largest sample is Wandsworth, with 34 people. Even if this was a representative sample, the margin of error for 95% confidence would be ±16%. But as it’s a sub-sample, even this would be underestimating it – they’re not demographically representative or weighted as the overall sample is.

    • GuyM

      Exactly, the results of the research only applies to the entire sample within the confidence limits and not sub-sections of it.

      If you wanted to investigate the views of residents of Wandsworth you’d need to set up a new representative sample within the agreed confidence limits and test again.

      • Carl

        Agree with both the above. Some wishful thinking going on in this blogpost. Sadly the underlying picture seems to be a Boris lead of 5-6 points.

      • GuyM

        In addition the weighting used across the pan London sample would need to be ammended for a different population base i.e. Wandsworth.

        Thus you can make no valid assumptions from sub-sections of the poll at all.

        • derek

          LoL! your answering yourself? but yip Bradford was a lesson. Don’t count your chickens before they hatch. 

  • My gut feel is that Boris is ahead – but not by that much.  I’ve a feeling that there are a number of people who like the Boris image and say they’re going to vote for him, but who won’t bother. A big worry for Labour has to be the very low polling for the Lib Dems and especially the Greens – most of whom might be assumed to have Ken as second preference.  UKIP’s vote will almost all go to Boris; if they do as well as some are predicting, that could prove crucial.

  • Pingback: UK Polling Report()

  • Carl

    See here: 
    http://ukpollingreport.co.uk/blog/archives/5255

    It would be appropriate to update the original post to link to the above.

    • AnotherOldBoy

      Thanks for the link.  That seems to be a conclusive answer to Mr Ferguson, who still gets a star for effort.

  • The writer needs to learn about polling before writing such a stupid article.

  • Amber Star

    Okay, maybe sub-samples are not relevant but Mark Ferguson may not be entirely wrong. This poll could be an outlier, or even a ‘rogue’ poll.

    There’s a YouGov poll due fairly soon, so that may confirm or contradict with the ComRes.

    Personally, I think Ken will win. Boris is depending on Labour voters choosing him. On the day, I think ‘Labour for Boris’ voters may decide not to vote or even demonstrate some Party loyalty by switching to Ken at the 11th hour.

    • postageincluded

       Boris’s biggest headache will be getting his luke-warm Labour supporters to make the walk to the polling station. I’m praying for rain.

  • Amber Star

    Okay, maybe sub-samples are not relevant but Mark Ferguson may not be entirely wrong. This poll could be an outlier, or even a ‘rogue’ poll.

    There’s a YouGov poll due fairly soon, so that may confirm or contradict with the ComRes.

    Personally, I think Ken will win. Boris is depending on Labour voters choosing him. On the day, I think ‘Labour for Boris’ voters may decide not to vote or even demonstrate some Party loyalty by switching to Ken at the 11th hour.

x

LabourList Daily Email

Everything Labour. Every weekday morning

Share with your friends










Submit