Since the Electoral Commission intervened to make the EU referendum question wording more neutral, a string of polls conducted among volunteer online polling panels have shown the 'Leave' camp jumping by several points to effectively draw level. However, telephone polls have typically shown much stronger support for 'Remain' than online polls, and until today we hadn't seen a telephone poll which asked the new question. My firm expectation was that we would see the same boost in support for 'Leave', but that this would still leave 'Remain' with a significant lead. In fact, the new ComRes poll shows the huge lead growing further, albeit only very slightly.
Should the UK remain a member of the European Union, or leave the European Union? (ComRes, telephone fieldwork, 26th-28th September) :
Remain 55% (+4)
Leave 36% (+3)
On the face of it, then, the impact of the new question has been to widen the gulf between online and telephone results even more. However, I think the most likely explanation for the relatively static lead suggested by ComRes is that their previous wording was actually pretty similar to the new question. You never know, though - it could be that online respondents who see the question written down are more influenced by the nuances of language.
Either way, it's quite clear that the disparity between the two types of poll isn't going away any time soon. Of course, we've seen something very similar with recent results on Scottish independence, with telephone and face-to-face polls finding a lead for Yes, and volunteer online panel polls tending to show a slender lead for No. My inclination is always to assume that 'real world' polls are closest to the mark, but the reality is that we have no way of knowing who is getting it right. Britain could be poised to leave the European Union, or the 'Leave' campaign could have an enormous mountain to climb. Take your pick.
One thing that we have to guard against is the possibility of getting a misleading sense that there is less uncertainty than there actually is, due to online polls appearing more frequently (because they are cheaper to conduct). You might recall that one or two of the usual suspects in the comments section of this blog attempted to dismiss the Yes leads in the Ipsos-Mori and TNS independence polls as "outliers", but in truth the No leads in online polls might well look like the outliers if telephone polling was much more common.
Incidentally, if we could only solve the mystery of how Conservative voters feel about the EU, we'd be a lot closer to knowing the likely outcome of the referendum. The ComRes telephone poll suggests that Tory voters break almost exactly in line with the general population - 56% want to stay in the EU, 35% want to withdraw. But the YouGov online poll conducted earlier in September showed that Tory voters were significantly more anti-EU than the electorate at large, and that a clear majority of them wanted to leave (51% for 'Leave', 33% for 'Remain').
Annoyingly, YouGov didn't provide any geographical breakdown of their figures (I'm tempted to call them "obsessively secretive" again, just for the pleasure of watching Laurence Janta-Lipinski explode). But ComRes have published Scottish subsample numbers - as you'd expect, the lead for 'Remain' is bigger than across Britain, at 61% to 29%. But the unweighted Scottish sample size is a paltry 90, so treat with extreme caution.