• EU referendum tracker

    A period of relative stagnation in the EU tracker continues, with the headline findings the same as last week: Remain In 43%, Leave 39% with 17% unable to provide a view. However, when the DKs are stripped out the Leave campaigns inch forward to 48% while Remain In drop marginally to 52% on the basis of rounding considerations.

    This week has seen You Gov produce their findings on the polling miss of 2015, with methodological tweaks likely to manifest themselves in and around differential propensity to vote modelling and political interest weighting. On this, our own testing work has just a little to say, at least in context of the referendum tracker question.

    We have been (privately, to date) testing a new form of turnout modelling for our own Vote Intention work, but which does carry through (in currently unpublished data) to this EU tracker dataset should we desire it to do so – on the probably contentious supposition that turnout propensities in the European Referendum will be consistent with those (assumed, and always with limitations) to have occurred at the General Election.

    The double caveat just stated probably serves as a major health warning and I’ll save you from further theoretical or methodological headaches, but the outcome was interesting – the alternative turnout modelling made reduced the Remain In lead to 44% vs 42% (DK:15%) or to 51% vs 49% depending on your preference.

    It’s easy to analyse why: older people are more likely to vote, more likely to be Conservative voters and thus more likely to be to Leavers. Being upweighted in a new turnout scheme thus inevitably increases the Leave share in a poll.

    ICM will introduce a turnout question for the EU referendum tracker in the near future, and the new turnout weighting scheme may well follow, although it very much remains a work in progress.