Daily polling roundup 5th April

Today's big story turned out to the ComRes Thanet South poll that was leaked to the Mail on Sunday. As I wrote in last night's roundup:

In seat polling, tomorrow’s Mail on Sunday has a front page headline about a leaked private poll of Thanet South. This makes the front-page presumably because it shows a party leader trailing in his own seat, but as I’ve written before, private polling happens all the time, and all but the most favourable are usually buried.

The Mail reported the toplines as CON 31, UKIP 30, LAB 29. ComRes chair Andrew Hawkins confirmed that his house had conducted the poll. Under BPC rules, where private polling finds its way into the public domain (including via unauthorised leaks), the BPC member responsible for the polling is obliged to publish the details in full, which ComRes did this afternoon, accompanied with a statement.
Anthony Wells did a very thorough analysis to debunk some of the nonsense that was flying around on Twitter. There was nothing dubious about its methodology, which was the same as in ComRes’s national phone polling (which has given pretty similar results to other phone polls) except that as a constituency poll, Candidates were named and past-vote weighting wasn’t used.
Incidentally, if PV weighting had been used (as Lord Ashcroft does), the order of parties wouldn’t have changed, although the race wouldn’t have been quite so close – the Conservatives would be higher and Labour lower (the amount would depend on the degree of false recall adjustment used). Here’s a May2015 graph showing recent polls in the seat:

Another thing that should concern Nigel Farage is the proportion of UKIP support (31%, more than double the equivalent Labour and Tory figures) that didn’t vote in 2010. Pollsters have found that past non-voters are more likely to overstate their likelihood to vote, and some (ComRes is not one of them) weight them down further. If that’s the case here, UKIP will underperform their polls. It’s not certain to happen, but it is a risk.


In national polling, the tables for last night’s YouGov/Times poll were today released. The toplines showed a dramatic looking fall in Conservative and Labour combined vote share, but 67% is extremely similar to the 66.6% share of the mainland vote that the two largest parties got in 2010. As with Opinium, the percentage saying they watched the debate was disproportionately large, but this might have been due to a desirability bias (people saying they watched the debate when they hadn’t) rather than being purely down to response bias. This may have exaggerated some of the moves, but we’ll only know for sure when we get some more polls.
The Sunday Times also commissioned Panelbase polling, including Scottish Westminster VI:


With the Scottish debates debates approaching, and the increased focus on Nicola Sturgeon, expect more Scottish polls in the days ahead. Finally we also got the ICM/Telegraph wisdom index overnight:


The Press Association announced expected declaration times for election night. On the basis of this list, no seats likely to change hands before 1am. Time to stock up on coffee!
And finally:


Tomorrow we will not get Populus (out Tuesday) or Ashcroft (no national poll this week). YouGov/Sun polls tend not to come out on bank holidays either.





About The Author

Related Posts