More on the new PB poll
Will it be a good guide of the race to Westminster?
Further details of PB’s first poll from Angus Reid Strategies are now available for down-load here. The fieldwork took place online on October 15th and 16th.
It’s planned that with future surveys that there will be a much shorter period between the fieldwork ending and the results being published here. There are some key points to note;-
Clearly the firm has no track record in UK election polling although it has done well with Canadian elections. At the coming general election it will be one of three pollsters UK firm whose methodologies will be tested for the first time against actual results. The other two, ComRes and Ipsos-MORI have been around longer but have undergone major methodological changes since 2005.
Online polling remain controversial. There has been an ongoing debate about online polling and using a polling panel ever since YouGov started eight years ago. That firm’s response is usually to say “judge us by our results”. Angus Reid has recruited some senior YouGov staff for its UK operation and they work closely with head office in Vancouver. I’m of the view that the more ways of looking at public opinion the better and that it is good that an alternative in online polling to YouGov is being established.
Angus Reid’s does not use a “spiral of silence adjustment” to deal with “don’t knows”. ICM, Populus and ComRes have individual formulas for handling the preferences of what can be a key part of their samples – those who say they will be voting but won’t say how. They use the party ID or past vote responses to allocate a part of these to the total figures. Angus Reid, like MORI, takes a different approach. It asks a supplementary on “which party you are currently leaning towards” and then aggregates the responses with those who are definite.
The past voting weightings – a mechanism to ensure a politically balanced sample – are calculated on the actual 2005 result without taking into account any mis-remembering – something that is the normal practice here. The ARS approach is what is used in Canada and will be the subject of some debate. On top of that Angus Reid weights on newspaper readership.
We’ll know on the Saturday after the election which pollsters have got it right.
Mike Smithson