Leave back in the lead in stunning YouGov upset

This is a suitably surprising end to a bizarre day.  In the morning and early afternoon, the markets (of both the betting and financial varieties) were pointing heavily towards Remain, which led to some suspicions that there might be inside knowledge of forthcoming poll numbers.  Those suspicions seemed to have been borne out when Ben Page of Ipsos-Mori eccentrically revealed on Twitter that two of his firm's competitors were about to release polls putting Remain in the lead.  We knew that YouGov and ORB polls were pending, so those seemed like the most obvious candidates - which suggested a very real chance that a third successive YouGov poll might show a swing to Remain, and perhaps vindicate the over-excitable commentators who have been "calling" the referendum for Remain on the basis of seemingly no evidence whatever.  But as it turns out, one of the Remain leads was in fact a NatCen survey that took place over an extended four-week period, and that therefore tells us absolutely nothing about very recent trends.  Against all expectations, the new YouGov poll actually puts Leave back in the lead -

Should the United Kingdom remain a member of the European Union or leave the European Union?

Remain 42% (-2)
Leave 44% (+1)

The fieldwork was apparently conducted between Friday and Sunday, so was entirely after the murder of Jo Cox.  The YouGov poll immediately before the tragedy had an identical result to this one, so on the face of it there doesn't appear to have been any direct and lasting impact on public opinion.  (If there has been an impact, it's being masked by the quirks of sampling variation.)

The other half of what we were expecting to happen did happen - the ORB phone poll put Remain in the lead, but there's some pretty good news to be found for Leave even there.  As I've mentioned a number of times, the ORB phone series has been driving me mad because there don't seem to be any definitive 'headline figures' - the Telegraph (who commission the polls) headline the figures that exclude anyone who isn't absolutely certain to vote, while ORB themselves usually seem to prefer the unfiltered numbers.  On the latter count, the Remain lead has actually fallen from 5% to 2% over the last week, leaving us firmly in 'statistical tie' territory -

Remain 49% (n/c)
Leave 47% (+3)

Remarkably, that's the second-best showing for Leave in the unfiltered sample of any ORB phone poll since the series started in March.  (And in retrospect, the outright best showing for them looks like it was a fluke result, because no other phone pollster was reporting similar numbers at the time.)

Curiously, though, the turnout-filtered figures (which until now have generally been better for Leave) shoot off in completely the opposite direction -

Remain 53% (+5)
Leave 46% (-3)

As I always used to point out about TNS polls during the indyref, the trend shown by turnout-filtered figures tends to be much more volatile, because significantly fewer respondents are left after the filter is applied.  So it's very hard to say whether this poll is picking up that Remain supporters have genuinely become much more likely to vote over the last week, or whether the greater margin of error in the filtered numbers is causing an illusory Remain boost.  For what it's worth, YouGov have failed to detect any sudden reversal of the enthusiasm gap - their poll continues to suggest that Leave voters are more likely to be absolutely certain to vote.

*  *  *

SCOT GOES POP POLL OF POLLS

As a final adjustment to the time-frame covered by the Poll of Polls, from now on only polls that were at least partly conducted within the last seven days will be taken into account.  As I mentioned last week, I've made an executive decision that ORB's turnout-filtered numbers will be used for the rest of the campaign - in the last update that was boosting Leave slightly, but now it's helping Remain.

50/50 ONLINE/TELEPHONE AVERAGE :

Remain 45.4% (+0.8)
Leave 45.8% (-0.8)

ONLINE AVERAGE :

Remain 43.5% (+1.2)
Leave 45.7% (+0.3)

TELEPHONE AVERAGE :

Remain 47.2% (+0.4)
Leave 45.8% (-2.0)

(The Poll of Polls takes account of all polls that were conducted at least partly within the last seven days. The online average is based on six polls - three from YouGov, one from Opinium, one from Survey Monkey and one from BMG. The telephone average is based on five polls - two from Survation, one from ORB, one from BMG and one from Ipsos-Mori.)