It’s time for a cringing apology to Survation for implying that their recent poll of Camborne-Redruth was somehow less worthy than Ashcroft’s polls. True, the margin of polling error will be higher as the numbers polled are lower. However, I’m informed by Survation that they don’t weight the responses about previous vote to bring them in line with reality, as does Ashcroft.
Here are the relevant points from Survation –
‘Our tables show what respondents actually told us about their past vote and Ashcroft’s do not. Those figures you show look like the “post weighted to 2010” figures to me [they were], just slightly different to the actual targets due to something called “chop” – things move around a little bit when you do multi-variable weighting.
Our tables are more fun as you can see the false recall in its full glory (8% of Clacton by-election voters told us they voted UKIP in 2010 – party did not stand).
Here is a note from us about why we do not weight by past vote any more.’