|Angus Reid for Politicalbetting||Mar 11||Feb 19|
|LAB to CON swing from 2005||8%||7.5%|
So why the difference between the two online pollsters?
Here’s the second poll of the night – the exclusive Angus Reid one for Politicalbetting – which is showing a radically different picture than YouGov a few minutes ago.
On the face of it this is odd – for both operate on-line confining their polling to members of their panel on whom they have a lot of data. Both use newspaper readership as a key weighting factor and Angus Reid’s UK operation is headed by a former YouGov man.
When we get the detailed data though for both polls I’m going to see if there any obvious differences in the way the weightings are operating.
One thought is that the AR panel was all recruited in the past ten months whereas the YouGov one has been built up over quite a long time. When the latter started having home internet access was very much the exception. Now it’s very much the norm.
The one big difference is the way the firm do their political weightings. YouGov has a complex system involving party ID – AR weights to precisely what respondents said they did in 2005.
It’s interesting as well that of the four firms now polling on-line three are reporting very high shares for “others”. YouGov is the exception.