Populus change their much criticised party ID weightings which should see much bigger UKIP shares
New Populus poll with new methodology that gives big boost to UKIP
Lab 36
Cons 33
LD 9
UKIP 15;
— Mike Smithson (@MSmithsonPB) February 7, 2014
@PopulusPolls say their new methodology will boost UKIP and see declines for LAB & LDs. Tories should stay the same.
— Mike Smithson (@MSmithsonPB) February 7, 2014
@PopulusPolls has been sharply criticised about its former party ID weighting model.
— Mike Smithson (@MSmithsonPB) February 7, 2014
The new weightings – see how they shift
Given the frequency of Populus online polls today’s change in methodlogy is an important event if only because the firm that traditionally had the lowest UKIP numbers will now be up there amongst the highest.
The firm is trying to ensure politically balanced samples by asking what party respondents usually associate with. They then apply weightings seen in the table above.
I think the broad move is the correct one though I’ve yet to study the detail of how they are doing it.
A big challenge is that the online pollsters seem to produce bigger UKIP shares than the long-established firms who poll by telephone. Quite why this should be is hard to say.
ICM, for instance, applies a turnout adjustment that discounts by half the “value” of those repondents who say they did not vote at the last general election. Given that the best indicator of whether someone will vote is whether they’ve done so in the past then, to me, that seems justified.
What I found worrying about the make-up of current UKIP support in some polls is that a siginficant proportion are non-voters from last time. I published this yesterday.
Whether UKIP indeed is going to atttract substantial support from non-voters we will have to wait and see. I’m sceptical.