We’re back with the YouGov-ICM divide

We’re back with the YouGov-ICM divide


    YouGov has Labour 3% behind while ICM shows a 4% lead

In much of the period leading up to the 2005 General Election the UK’s two major polling organisations were showing very different pictures of the way the public were thinking. Alas by polling day findings from the two firms almost converged and it was hard to draw a significant conclusion about their respective methodologies.

  • Today a telephone survey by ICM in the SundayTelegraph has LAB 37: CON 33: LD 21. The only change on the firm’s Guardian survey on Tuesday is a one point decline in the Tory share.
  • Meanwhile a YouGov internet survey in the Sunday Times has CON 38 : LAB 35: LD 19 – the only changes on its last being being a one point decline in Labour support and a one point increase for the Lib Dems.
    • So how do you explain these totally different figures?

    We last looked at the differing methodologies two months ago when the two firms had very different Lib Dem shares.

    When the full datasets are available from the two pollsters we’ll be able to examine if there are any technical polling factors involved. ICM weight by recalled past vote and there might have been a change in the numbers they use.

    As far as I can see neither ICM nor Yougov are currently weighting by the relative likelihood of people voting – something they did in their General Election polls. This resulted in 1-2% drops in both their projected Labour leads.

    ICM also use what has become known as the “spiral of silence” adjustment that distributes 50% of don’t knows in the same way they said they voted did last time. In recent ICM polls this has boosted the Labour margin.

    The main issue with YouGov is that its polls are carried out on the internet amongst members of its “polling panel” which critics say might not be representative. Sometimes YouGov appears to magnify trends such as the UKIP support ahead of the 2004 Euro election and the 13% Lib Dem share in January.

    For whatever reason the people who respond to phone surveys like ICM’s are almost always more Labour-inclined than the electorate as a whole. The pollster has pioneered methods for adjusting for this but is this enough?

    YouGov also has a voting intention figure based on what people would do if Gordon Brown was Labour leader. This produced shares of CON 39% LAB 37%: LD 17%.

      So for the first time since December the pollster is showing that the party would do better under the Chancellor.

    Aside from the main voting intention both surveys have “sleaze” questions. ICM report that 73% think Lab is “”sleazy” – or more so – than John Major’s government. Almost two-thirds of the YouGov responses thought that Labour kept the loans secret because they were embarrassed by them.

    Mike Smithson

    Comments are closed.