Sensational Ipsos-Mori poll confirms that Remain's lead in telephone polling has dropped sharply

As you might remember, I had difficulty knowing what to make of the ORB telephone poll showing Leave in the lead, and the succession of ComRes telephone polls showing the Remain lead had slumped, because until now Ipsos-Mori's phone polling had failed to show any significant change.  That is no longer the case.

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

Remain 49% (-5)

Leave 41% (+5)

There have been some methodological tweaks since the last poll, so technically there's the possibility that the percentage changes listed above are not meaningful.  But in actual fact, one of the changes has worked in favour of Remain, not Leave.  If Ipsos-Mori hadn't introduced a 'squeeze' (an extra question pressing Don't Knows on how they're most inclined to vote) the Remain lead would be down to just six points.

What I find most extraordinary of all, though, is the revelation that if Ipsos-Mori had used the turnout-adjusted model which takes account of the polling errors at last year's general election, the Remain lead would have stood at just TWO points, essentially eliminating the previously huge divergence between telephone and online polling.  That flatly contradicts what we saw a few days ago from ComRes, who gave comfort to pro-European commentators by noting that their own turnout-adjusted model would have increased the Remain lead from seven points to fourteen (over the months, the ComRes model would have consistently boosted Remain).  I think all we can really say is that the pollsters have yet to reach a consensus on the correct lessons of last year's disaster - which may be just as well, because it was a self-reinforcing consensus that partly caused the problem in the first place.

The big health warning on today's poll is that the vast bulk of the fieldwork was carried out before the Brussels attacks.  The sole post-Brussels poll to date was an online ICM poll, which seemed to show a swing back to Remain, although as it's only one poll, the jury is still out on whether that was a real shift in opinion or an illusion caused by normal sampling variation.  On the headline numbers, Remain moved from being two points behind to take a lead of two points - but if it hadn't been for the introduction of a turnout filter, it would have been a five-point Remain lead (actually more like four points without a distortion caused by rounding).  On a like-for-like comparison, that's Remain's best showing in an ICM poll for many weeks, although admittedly it's still comfortably within the longer-term 'normal range' for the firm.

*  *  *

SCOT GOES POP POLL OF POLLS

The increase in the Remain lead in this update of the Poll of Polls is misleading - it's mostly caused by the quirk of the Leave-friendly ORB poll making up half of the telephone sample in the last update.

50/50 ONLINE/TELEPHONE AVERAGE : 

Remain 44.1% (-0.1)
Leave 40.4% (-2.2)

ONLINE AVERAGE :

Remain 40.6% (-0.3)
Leave 39.3% (-0.8)

TELEPHONE AVERAGE :

Remain 47.5% (n/c)
Leave 41.5% (-3.5)

(The Poll of Polls takes account of all polls that were conducted at least partly within the last month. The online average is based on nine polls - five from ICM, three from YouGov and and one from TNS. The telephone average is based on four polls - one from ComRes, one from Ipsos-Mori, one from ORB and one from Survation.)