It's beginning to look like the signs over the weekend and Monday that Remain were pulling clear in the EU referendum may have been illusory. Hot on the heels of YouGov and ICM polls suggesting that Remain have made no progress at all in online polling (and have possibly gone backwards), we have a new Survation telephone poll that puts the Remain lead at just 6%. As far as I can see from the records, that's the third worst showing for Remain in any phone poll from any firm - the only worse ones were a freakish ORB poll in March, and YouGov's rather artificial experiment earlier this month that was intended to illustrate that phone polls are inaccurate.
Should the United Kingdom remain a member of the European Union or leave the European Union?
Remain 44% (-1)
Leave 38% (n/c)
The change from the last Survation poll may look insignificant, but it has to be remembered that the previous poll was already the worst for Remain in the series by some distance - Survation were showing a gap of eleven points in March, and fifteen points in February. Perhaps more to the point, if Ipsos-Mori and ORB were correct that the Remain lead has dramatically surged recently, you wouldn't expect Leave to be narrowing the gap in any other phone poll - even by the smallest of amounts. So in spite of the claims from Downing Street that private polling corroborates the Remain-friendly findings from ORB, the evidence from public polls now looks very shaky - three out of five telephone firms (ICM, ComRes and Survation) are failing to support that narrative, as are three out of four online firms (YouGov, ICM and TNS).
It's also worth emphasising just how bad yesterday's ICM poll was for Remain, once you take account of the methodological tweaks. ICM have noticed that the earliest responders to their online polls are disproportionately likely to be Leave supporters, so they've introduced two changes to resolve that problem - they're downweighting early responders, and are also staggering survey invitations in the hope of attracting a more representative sample in the first place. Crucially, we're told that the weighting changes alone were responsible for the equivalent of a 2% swing to Remain - meaning that without them Leave would have had a 4% lead in the poll. But it's surely logical to assume that the staggering of the survey invitations also worked in Remain's favour, which may well mean that without any methodological changes at all, the Leave lead would have been even higher than 4% - and that would have been the best ICM poll for Leave to date. So in terms of the trend, ICM, YouGov, ComRes and Survation all seem to be singing from the same hymn-sheet - either no real change, or minor progress for Leave.
One thing that's got on my nerves during this campaign is the succession of smug articles from pollsters along the following lines : "Hmmm, yes, it does look like phone/online (delete as applicable) polls are turning out to be more accurate. But what's interesting is why that should be the case. Let's have a discussion about why online/phone (delete as applicable) polls are so useless at the moment." YouGov are at it today (arguing that online polls are best), and their ex-President Peter Kellner was at it a few days ago (ironically he took the side of phone polls). Seriously, chaps, what you're doing is putting forward speculative theories. It won't "turn out" or even "look like" one side or another is more accurate until we see the actual referendum results. Then the side that got it right can gloat to its heart's content, but it really is the height of arrogance to do it now when nobody actually has a sodding clue how things are going to play out on June 23rd.
* * *
SCOT GOES POP POLL OF POLLS
50/50 ONLINE/TELEPHONE AVERAGE :
Remain 45.9% (-0.6)
Leave 40.6% (-0.3)
ONLINE AVERAGE :
Remain 42.1% (-0.3)
Leave 42.2% (-0.5)
TELEPHONE AVERAGE :
Remain 49.6% (-0.9)
Leave 38.9% (-0.1)
(The Poll of Polls takes account of all polls that were conducted at least partly within the last three weeks. The online average is based on nine polls - four from YouGov, three from ICM, one from TNS and one from Opinium. The telephone average is based on seven polls - two from ORB, one from ICM, one from YouGov, one from Ipsos-Mori, one from ComRes and one from Survation.)