I’ve got an article over on the YouGov website about the difficulty on polling on the Brexit financial settlement (or “Brexit divorce bill” as the more Eurosceptic elements of the press tend to call it). Brexit is obviously a very complicated issue – the Brexit deal will almost inevitably dominate the next year of British politics, yet the complexities of it mean it’s very hard to ask about until there’s actually a deal on the table.

The financial settlement between Britain and the EU should, on the face of it, be one of the more simple issues. On the face of it you might expect it to be fairly simple to ask people what sort of financial settlement the public would think was reasonable and what sort of settlement would have the public thinking Theresa May has struck a poor deal. In fact such questions give us a very poor guide, simply because most people are not particularly good at comprehending very large numbers.

If you ask a question about what a reasonable price is for, for example, a pair of shoes, it should work very well. Everyone knows roughly what shoes cost, and know the value of £10 or £30 or £100. The same does not apply for government spending – £50 billion is an unfathomably large amount of money… but then, so is £20 billion, or £10 billion or £5 billion. Most of us don’t really have any good yardstick for judging just how big or small these huge numbers are, nor whether they are a good or bad deal for Britain.

Nevertheless, if you ask people about a financial settlement people will still express opinions. Back in August there was an ICM/Guardian poll that found 41% of people though a £10bn settlement would be acceptable, up from just 15% in April. This seemed like a startling rise, but as both ICM and the Guardian cautioned, it could just be the way the question was worded. In April ICM first asked about the lower figure of £3bn, but in August £10bn was the lowest they offered.

This seemed like a more plausible explanation to me, but just to be sure we tested it at YouGov. We used a split sample – one half of the respondents got a grid of three questions asking about settlements of £5bn, £10bn and £20bn. The other half of the sample got a grid of three questions asking about settlements of £25bn, £50bn and £75bn.

On the first bank of questions 38% thought £5bn would be acceptable, 18% thought £10bn would be acceptable, 11% thought that £20bn would be acceptable. Looking at the other half of the sample, 29% thought that £25bn was acceptable, 9% thought that £50bn was acceptable, 6% thought that £75bn would be acceptable (full tabs are here.)

Taken as a whole we get the the rather perverse finding that while support generally falls as the size of the settlement increases, £25 billion is far more acceptable to the public than £20 billion. This is nonsense of course, and the reason is simple enough – people take their cues from the question itself. In the first half of the sample, £5bn was the lowest amount asked about, £20bn the largest amount, and many respondents presumably took this as an implication that £5bn was a low settlement, £20bn a high one. For the second half of the sample £25bn was the lowest figure asked about, so many respondents presumably took the implication that this was a low settlement. Whether people said a sum was acceptable or not was less about the actual number, more about whether the question implied that it was a low or high figure.

The point is that questions about what level of “divorce bill” will be acceptable to the public don’t really tell us much. People don’t have any good way of telling what is a good or bad deal and are really just expressing their unsurprising preference for a smaller settlement. When (or if) Britain and the EU do finally agree on a sum, it won’t be so much the particular figure that determines whether the public see it as a victory or a sell-out, but whether the media and political class present it to them as a good or bad deal.

Meanwhile, lastest GB voting intention figures this week are below – both show the parties pretty much neck-and-neck, neither show any obvious movement:
YouGov/Times (12th-13th Sept) – CON 41%(nc), LAB 42%(nc), LDEM 7%(+1), UKIP 3%(-1) (tabs)
ICM/Guardian (8th-10th Sept) – CON 42%(nc), LAB 42%(nc), LDEM 7%(nc), UKIP 4%(+1) (tabs)

Two new voting intention polls this week showing very similar figures. YouGov‘s latest poll was actually conducted last week, but was only released today and has topline figures of CON 41%, LAB 42%, LDEM 8%, UKIP 4% (full tabs are here.

The regular ICM poll for the Guardian, conducted over the weekend, has extremely similar topline figures – CON 42%, LAB 42%, LDEM 7%, UKIP 3% (full tabs are here).

ICM also asked about people’s attitudes towards Britain paying a financial settlement as part of our Brexit negotiations (a so-called “exit fee”). ICM asked similar questions back in April and found very little support – only 10% thought paying a £20bn settlement would be acceptable, 15% a £10bn fee and 33% a £3bn exit fee. This time the figures suggested in the question were changed to what are probably more realistic figures and with interesting results – now 9% think a settlement of £40bn would be acceptable, 11% a £30bn settlement, 18% a £20bn settlement, 41% a £10bn settlement.

On the face of it this one might think this is a startling change, a few months ago only 15% thought it would be acceptable to pay a £10bn settlement as part of Brexit, now 41% think it’s acceptable. I think it’s probably actually a good example of the importance of context in a question. Most people are really not that good at putting figures of billions of pounds in context – any sum that involves the words billion is a huge amount of money to begin with, so what would be a relatively small settlement? A moderate settlement? A huge settlement? The only thing respondents have to scale it by is the question itself. In April £3bn was implicitly presented as the small option and £10bn was presented as the medium option. In this poll £10bn is implicitly presented as the small option and £20bn or £30bn are presented as the medium options – hence why a £10bn settlement suddenly seems to be so much more paletable.

That’s not to say the question doesn’t tell us anything at all – there’s still an interesting increase. In April only 33% thought a “small” financial settlement would be acceptable as part of the Brexit deal; now that figure has risen to 41% (despite the actual figure quoted tripling!). It looks as if the public may be moving towards accepting that a financial settlement may be an inevitable part of Brexit.


Ipsos MORI’s monthly political monitor for the Standard has topline figures of CON 41%, LAB 42%, LDEM 9%. This is MORI’s first poll since the general election, and like other companies now shows Labour with a small lead over the Conservatives. Fieldwork was Friday to Tuesday. As far as I can tell, the methodology is back to MORI’s usual methods, as they were using before the election campaign. Full details are here.

To update on other voting intention polls earlier this week, ICM for the Guardian on Tuesday had voting intentions of CON 42%(+1), LAB 43%(nc), LDEM 7%(nc), UKIP 3%(nc). Fieldwork was over the weekend, and changes were from a fortnight ago. Full tabs for that are here.

Finally YouGov for the Times, which was released on Monday but conducted last week, had topline figures of CON 40%, LAB 45%, LDEM 7%, UKIP 2%. Tabs for that are here.

ICM have resumed polling for the Guardian. Topline figures for their first post-election poll are CON 41%(-3), LAB 43%(+2), LDEM 7%(-1), UKIP 3%(+1) – changes are from the election result.

In terms of methodology, ICM have dropped the turnout model that produced such large, but ultimately incorrect, Tory leads as well as their political interest weighting. This isn’t going all the way back to their 2015 methodology (ICM also made a change to how they reallocated don’t knows who refused to give a past vote and, of course, switched from telephone to online), but it’s a long way in that direction.

There is only one GB poll so far today – ICM’s weekly poll for the Guardian, their penultimate of the campaign. Voting intentions are almost identical to their poll for the Sun on Sunday yesterday, with topline figures of CON 45%(nc), LAB 34%(nc), LDEM 8%(-1), UKIP 5%(nc). Fieldwork was Friday to Sunday, so will have been largely before the terrorist attack in London Bridge. Full tabs are here.

Survation will have a telephone poll out later tonight (probably midnight judging by past weeks’ timings) for Good Morning Britain, delayed for a day because of the terrorist attack. Other than that I expect most companies will now be looking towards their final call polls tomorrow, Wednesday or (if MORI stick their usual timetable) Thursday morning.