I've been meaning to update the EU referendum Poll of Polls for quite some time, but because I had settled on a method that gives equal weight to telephone and online data collection, it wasn't possible to do it in the absence of any recent telephone polls. But it seems that phone polls are like the proverbial London buses - you wait months for one, and then two come along at the same time.
SCOT GOES POP POLL OF POLLS
Should the United Kingdom remain a member of the European Union or leave the European Union?
50/50 ONLINE/TELEPHONE AVERAGE :
Remain 50.1% (+1.4)
Leave 37.6% (+0.4)
ONLINE AVERAGE :
Remain 43.1% (-0.7)
Leave 41.7% (+3.4)
TELEPHONE AVERAGE :
Remain 57.0% (+3.5)
Leave 33.5% (-2.5)
(All polls conducted at least partly within the last month are taken into account. The online average is based on nine polls - four from ICM, two from Survation, one from ORB, one from BMG and one from YouGov. The telephone average is based on one poll from ComRes and one from Ipsos-Mori.)
As you can see, the trend is far from clear. If only online polls existed, we'd probably be halfway towards convincing ourselves that Leave have made significant progress in recent weeks and have practically drawn level. But the two new telephone polls suggest public opinion is moving in completely the opposite direction, so much so that in the 50/50 average, the online trend is more than cancelled out, and the Remain lead has slightly increased.
I'm fairly confident that the method I'm using is the best one available, because almost any alternative you can think of would give too much weight to online polls, which are conducted far more frequently. That might not be such an issue if the gulf between the results produced by the two types of poll was not quite so huge and consistent, but it is. However, the problem with what I'm doing is that each telephone poll is given a disproportionate amount of weight, and so if there's any specific problem with an individual poll, it becomes greatly magnified. Unfortunately, I'm a bit sceptical about the approach Ipsos-Mori are using for their phone polls, because they ask about referendum voting intention in two different ways, and they use the actual referendum question second, not first. Logically, I have to use the results from the real question, but I can't help wondering if the first set of results are more meaningful. If we used those instead, the average Remain lead in telephone polls would be significantly lower, at 54.5% to 35.5%, and the 50/50 average would be Remain 48.8%, Leave 38.6%. If that's a better reflection of reality, it could be that Leave have indeed made some modest progress of late.
UPDATE : I see from John Curtice's analysis that I've misunderstood Ipsos-Mori's approach slightly - they don't ask respondents the question two different ways, but instead split the sample in two and ask each half of the sample one of the two versions of the question. That makes the numbers for the actual referendum question much less problematical, but quite what the value of the exercise is escapes me. OK, it helps Ipsos-Mori make a comparison with the question they've been asking for years, but surely that consideration is more than outweighed by the detrimental effect of cutting the sample size in half.