Daily catch-up: The EU referendum isn't really 'on a knife-edge'

One opinion poll puts the 'Remain in the EU' vote 26 percentage points ahead. Another today puts 'Leave the EU' nine points ahead. What is going on?

Another picture of old London from Sir William Davenant. The old Palace of Westminster from the River Thames as it was after the great fire of 1834. Nothing changes. The people in the boats in the foreground seem to be taking photos of it on their smartphones.

• Two EU referendum polls on Monday, from ICM and Survation, suggested the outcome would be close. ICM had "Remain" ahead by one point; Survation had "Leave" ahead by two. The technical journalese term for this is "on a knife-edge". Yesterday two more polls, from ComRes and Ipsos MORI, suggested comfortable leads for the "Remain" vote, by 21 and 26 points respectively. Then we had another poll today, from Lord Ashcroft, suggesting that "Leave" is ahead of "Remain" by nine points.

What is going on? Part of it is down to question wording. Gideon Skinner of Ipsos MORI gave a good example of this from a related subject yesterday. If you ask people if they support "giving 16 and 17 year olds the right to vote" in the referendum, they support it by 52 per cent to 41 per cent. But if you ask if they support "reducing the voting age from 18 to 16", they oppose it by 56 per cent to 37 per cent.

In its referendum question, Ipsos MORI found more support for "Remain" when it asked the actual question that will be on the ballot paper (a 26-point lead), than when it asked the question it has asked in the past, "stay in" or "get out" (a 17-point lead for "stay in").

Lord Ashcroft has a completely different question, which was to ask people to put themselves on a scale of 0 to 100, where 0 means they would definitely vote for the UK to remain in the EU and 100 means they would definitely vote to leave. I have no idea why this should give a higher score for "Leave".

More important, though, is the difference between online and phone polls. Online polls (ICM, Survation and Ashcroft) give more anti-EU results than phone polls (ComRes and Ipsos MORI). Katharine Peacock, Tom Mludzinski and Adam Ludlow of ComRes have written a paper about the difference which ought to be read by anyone trying to understand polling in general and polling on the referendum in particular.

They note the sharp difference between the same EU referendum question, adjusted to the same weights, in their online and phone polls, between a dead heat online and a 21-point lead for "Remain" on the phone. This difference does not show up in general election voting intention. They deduce that the level of engagement of the online panel skews the sample against EU membership, a bias that does not affect party voting intention. They think phone polls are more accurate for the EU referendum, and that "Remain" currently has the advantage.

Of course, predicting votes is about politics, politics, politics. A lot depends on how David Cameron's negotiations go (I think he's winning), and the refugee crisis or the euro could blow up at any time. But at the moment, "Remain" is ahead.

The betting market, incidentally, implies a 62½ per cent probability of a "Remain" vote.

• And finally, thanks to Andy Kind ‏for this:

"It's that magical time of year where a new generation of parents gets to explain that it's 'kin', not 'king'."