Opinium have a new poll in the Observer today (I think it’s the only poll in the Sunday papers, at least, it seems to be the only voting intention poll). Headline voting intentions are CON 41%, LAB 37%, LDEM 8%, UKIP 6%. Fieldwork was Thursday and Friday and the full tables are here. The four point lead echoes the YouGov poll that came out on Thursday, which had toplines of CON 41%, LAB 37%, LDEM 9%, UKIP 4% (tabs.

As well as their usual trackers, the Opinium poll also had some questions on the Brexit deal and what comes next. Asked how likely they think it is that there will be a “satisfactory” deal by March 2019, 26% think it is likely, 50% think it is unlikely. Satisfactory is, of course, in the eye of the beholder – some people presumably think there will be deal, but that it will be an “unsatisfactory” one, as the next question asked what people think is the most likely outcome – 30% expect us to leave with a deal next March, 33% to leave without a deal, 16% that we will not leave in March 2019.

The poll also asked what should happen next if there is no deal, or Parliament does not approve a deal. In the event of no deal at all, 14% think there should be a general election, 23% a new referendum, 13% an extension in order to continue negotiations, and 32% that Britain should just leave without a deal. In the event that a deal is struck, but Parliament rejects it, 12% think there should be a general election, 10% a deal vs no deal referendum, 20% a deal vs remain referendum, 14% that the government should return to negotiations, 25% that Britain should just leave without a deal.


Party conference season is sometimes a period of volatile polling – each party typically gets its own week of media coverage which, if all goes well, they’ll use for some positive announcements. This year it also immediately followed the Salzburg summit and Theresa May’s Brexit statement that followed. Below are the voting intention polls since my last update.

YouGov (18-19th) – CON 40%, LAB 36%, LDEM 11%, UKIP 5% (tabs)
Opinium (18-20th Sep) – CON 37%, LAB 39%, LDEM 9%, UKIP 8% (tabs)
BMG (21st-22nd Sep) – CON 38%, LAB 38%, LDEM 10%, UKIP 4% (tabs)
ICM (21st-24th Sep) – CON 41%, LAB 40%, LDEM 9%, UKIP 4% (tabs)
YouGov (24-25th Sep) – CON 42%, LAB 36%, LDEM 11%, UKIP 4% (tabs)
ComRes (26-27th Sep) – CON 39%, LAB 40%, LDEM 9%, UKIP 5% (tabs)
Opinium (26-28th Sep) – CON 39%, LAB 36%, LDEM 9%, UKIP 6% (tabs)
BMG (28-29th Sep) – CON 35%, LAB 40%, LDEM 12%, UKIP 5%

They are a mixed bunch – the YouGov poll showing a six point Tory lead got some attention, and I’m sure the BMG poll out this morning showing a five point Labour lead will do much the same. As ever, it’s wrong to pay too much attention to outliers. Normal sample variation means that if the underlying average is a Tory lead of a point or two, random noise will occassionally spit out a 6 point Tory lead or a small Labour lead, without it actually signifying anything. Collectively recent polls don’t suggest a clear impact on voting intention from either the Salzburg statement (while YouGov showed a larger Tory lead, ICM did not), or from the Labour party conference (while BMG show an increased Labour lead, Opinium showed the opposite).


-->

There is plenty of new polling in today’s papers, including two polls proporting to show that large numbers of people would vote for new political parties. One by BMG for the Huffington Post, claiming 58% of people would consider backing a new party at the next election, and a ComRes poll for BrexitExpress, claiming 53% of people in a selection of Tory constituencies would consider voting for a single issue party campaigning to “conclude Brexit as quickly and as fully as possible”. There have been various other polls in recent weeks asking similar questions about how popular new parties would be.

These sound like large figures, but you should take them all with a huge pinch of salt – the reality is that quantifying the prospects of a new political party before it exists is an almost impossible task. Certainly it is not something that can be done with a single question.

First let’s look at the question itself. Polls tend to take two approaches to this question, both of which have flaws. The first is to say “Imagine there was a new party that stood for x, y and z – how likely would you be to consider voting for it?”. The problem with that as a question is that “consider” is a pretty low bar. Does thinking about something for a fleeting second before dismissing it count as “considering”?

An alternative approach is to say “Imagine there was a new party that stood for x, y and z. How would you vote if they stood at the next election?” and then prompt them alongside the usual political parties. This does at least force a choice, and sets the new hypothetical party alongside the alternative established parties, prompting to people to consider whether they would actually vote for their usual party after all.

There are, however, rather deeper problems with the whole concept. The first is the lack of information about the party – it asks people whether they would vote for a rather generic new party (a new anti-Brexit party, a new pro-Brexit party, a new pro-NHS party, or whatnot). That misses out an awful lot of the things that determine people’s vote. Who is the leader of the party? Are they any good? Do the party appear competent and capable? Do they share my values on other important issues? Can I see other people around me supporting them? Are they backed by voices I trust?

Perhaps most of all, it misses out the whole element of whether the party is seen as a serious, proper contender, or a wasted vote. It ignores the fact that for most new parties, a major hurdle is whether voters are even aware of you, have ever heard of you, or think you are a viable challenger. That is the almost insoluble problem with questions like this: by asking a question that highlights the existance of the new party and implies to respondents that it is a party that is worthy of serious consideration a pollster has ignored the biggest and most serious problem most new parties face.

That’s the theory of why they should be treated with some caution. What about their actual record? What about when people polled about hypothetical parties that later became real parties that stood in real elections? Well, there aren’t that many cases of large nationwide parties launching, though there are more instances of constituency level polls asking similar questions. Here are the examples I can find:

  • At the 1999 European elections two former Conservative MEPs set up a “Pro-Euro Conservative party”. Before that a hypothetical MORI poll asked how people would vote in the European elections “if breakaway Conservatives formed their own political party supporting entry to the single European currency”. 14% of those certain or very likely to vote said they would vote for the new breakaway pro-Euro Conservatives. In reality, the pro-Euro Conservative party won 1.3%.
  • Back in 2012 when the National Health Action party was launched Lord Ashcroft did a GB poll asking how people would vote if “Some doctors opposed to the coalition government’s policies on the NHS […] put up candidates at the next election on a non-party, independent ticket of defending the NHS”. It found 18% of people saying they’d vote for them. In reality they only stood 12 candidates at the 2015 election, getting 0.1% of the national vote and an average of 3% in the seats they contested.
  • Just before the 2017 election Survation did a poll in Kensington for the Stop Brexit Alliance – asked how they might vote if there was a new “Stop Brexit Alliance” candidate in the seat, 28% of those giving a vote said they’d back them. In the event there were two independent stop Brexit candidates in Kensington – Peter Marshall and James Torrance. They got 1.3% between them (my understanding, by the way, is that the potential pro-Europe candidates who did the poll are not the same ones who actually stood).
  • Survation did a similar poll in Battersea, asking how people would vote if a hypothetical “Independent Stop Brexit” candidate stood. That suggested he would get 17%. In reality that independent stop Brexit candidate, Chris Coghlan, got only 2%.
  • Advance Together were a new political party that stood in the local elections in Kensington and Chelsea earlier this year. In an ICM poll of Kensington and Chelsea conducted in late 2017 64% of people said they would consider voting for such a new party. In reality Advance Together got 5% of the boroughwide vote in Kensington and Chelsea, an average of 7% in the wards where they stood.

In all of these examples the new party has ended up getting far, far, far less support than hypothetical polls suggested they might. It doesn’t follow that this would always be the case, and that a new party can’t succeed. I suspect a new party that was backed by a substantial number of existing MPs and had a well-enough known leader to be taken seriously as a political force could do rather well. My point is more that hypothetical polls really aren’t a particularly good way of judging it.


The regular Ipsos MORI political monitor came out in today’s Evening Standard. Topline voting intention figures were CON 39%(+1), LAB 37%(-1), LDEM 13%(+3), UKIP 2%(-4). Fieldwork was Friday to Tuesday and changes are from MORI’s last poll in July (they take a month off for August).

As with other recent polls the Conservatives seem to have recovered a tiny lead since falling behind after the Davis & Johnson resignations. Worth noting is that 13% for the Liberal Democrats. This is the highest they have recorded in any poll since the general election. While one shouldn’t read too much into a single poll – especially one whose fieldwork overlapped with the Lib Dem party conference – the wider polling trend does suggest some uplift in Liberal Democrat support: six of the nine polls so far this month have the Liberal Democrats back up in double figures.

The poll also asked about confidence in the Brexit negotiations, finding predictably low figures. 28% of people said they were confident Theresa May would get a good deal for Britain in the Brexit negotiations, 70% were not.

There was, however, not much more confidence that alternative Prime Ministers would do any better. 28% were confident Jeremy Corbyn would get a good deal were he Prime Minister, 67% were not. If Boris Johnson was PM 33% would be confident he’d get a good deal, 64% would not.

Full details are here.


On Monday the government tabled the final recommendations of the boundary review. As usual, I’ve done updated notional calculations for what the results of the 2017 election would have been if fought on the new constituency boundaries. They are viewable in full on a google spreadsheet on the link below:

2017 notional election results on new boundaries

For those who have followed the process these recommendations are not much different from those at the revised stage. The Commissions have altered a number of proposed constituency names, and moved a few wards back and forth here and there, but in most cases the broad recommendations are very similar to the last lot – the most significant differences are in East Sussex and around Stockport. As such the party partisan impact of the proposed boundaries are also much the same. If the last general election had been fought on these new boundaries the result would have been something like Conservative 307(-10), Labour 234(-28), SNP 30(-5), Liberal Democrats 8(-4), Others 21(-3). All of the high profile seat changes are largely the same – Boris Johnson and Iain Duncan Smith still both see their seats become tight marginals, Jeremy Corbyn’s seat is still carved up and combined with half of Dianne Abbott’s seat.

A few caveats to those numbers. First, for the avoidance of doubt, they are not a prediction of what would happen now, it’s an estimate what would have happened if the votes cast in 2017 had been counted on these new boundaries. Secondly, they cannot take account of whether people would have voted differently if the boundaries had been different. My feeling is that always someone understates how well the Lib Dems would have done – someone in a Lab-Con marginal might have voted differently if their ward had been included in a Con-LD marginal. Thirdly, these are just one estimate. Rallings & Thrasher, who produce the official estimates that the BBC, Sky and other media outlets would use to calculate swings at the next election have already produced their own totals, which are similar to the ones I have (for good reason, I use pretty much the same method that Rallings & Thrasher do) – their totals are CON 308, LAB 232, SNP 33, LDEM 7, Other 20.

The most important caveat however is that these boundaries still have to be voted on by Parliament to actually come into force. The DUP may back them after all (the initial recommendations were very bad for them and good for Sinn Fein, but the revised and final recommendations retained a four seat arrangement for Belfast, meaning the DUP should retain all their seats), but that still leaves the government with a wafer thin majority. It will only take a handful of rebels (either worried about their own seats, or objecting to things like the seat crossing the Devon-Cornwall border, or opposed in principle to the reduction in seats) to block the changes. We won’t find out in the immediate future, as the government have said the vote will be delayed for some months while the necessary legislation is drafted.

For more on the boundary review, Keiran Pedley has done a nice interview with the great Professor Ron Johnston on his podcast here.

While I was playing with boundary data yesterday, there was also a new YouGov poll of London for Phil Cowley at Queen Mary University London that I haven’t had time to write about yet. Full details of that are here.