The election of a majority Conservative government means that the Parliamentary boundary review will presumably go ahead on the rules passed under the last government, but delayed by the Liberal Democrats (the review that was started in the last Parliament was abandoned before it was completed after the law was changed). There is no need for the government to pass any laws to implement this, it will start up automatically early next year once electorate numbers are available, though Parliament will still have to vote to implement the Boundary Commissions’ recommendations, and with a small majority that is not necessarily a given – last time round there were a couple of Tory MPs who said they were going to vote against the new boundaries, and the government doesn’t have much of a majority to begin with.

Anyway, a couple of people have asked me how this election would have looked had the revised boundaries proposed in the last Parliament gone through. I’ve done a rough rejig of my provisional boundary calculations using the result of this election, and had the new boundaries gone through the Conservatives would have won 322 seats, nine fewer than they did but enough to give them a healthy majority of 44 in a Commons of 600 MPs. Labour would have won 204 MPs (28 fewer), the SNP 50 seats (and would have pushed Labour out of Scotland entirely) and the Lib Dems just 4.

Of course, this is not necessarily a good guide to what the new review this Parliament will produce – electorate numbers will have changed since 2010 and given some of the discussion after the abandoned review I suspect the English Commission may be a little more open to splitting wards so the proposed changes are less disruptive (something that requires only a change of mindset, not a change of rules!), but we shall see.


I’ve just got back from the BBC after working all night (you may have seen my bald spot sat just to the left of Emily Maitlis’s big touchscreen last night) and am about to go and put my feet up and have a rest – I’ll leave other thoughts on the election until later in the weekend or next week, but a few quick thoughts about the accuracy of the polls.

Clearly, they weren’t very accurate. As I write there is still one result to come, but so far the GB figures (as opposed to the UK figures!) are CON 38%, LAB 31%, LDEM 8%, UKIP 13%, GRN 4%. Ten of the final eleven polls had the Conservatives and Labour within one point of each other, so essentially everyone underestimated the Conservative lead by a significant degree. More importantly in terms of perceptions of polling it told the wrong story – when I was writing my preview of the election I wrote about how an error in the Scottish polling wouldn’t be seen so negatively because there’s not much difference between “huge landslide” and “massive landslide”. This was the opposite – there is a whole world of difference between polls showing a hung Parliament on a knife edge and polls showing a Tory majority.

Anyway, what happens now is that we go away and try and work out what went wrong. The BPC have already announced an independent inquiry to try and identify the causes of error, but I expect individual companies will be digging through their own data and trying to work out what went wrong too. For any polling company, there inevitably comes a time when you get something wrong – the political make up, voting drivers and cleavages of society change, how people relate to surveys change. Methods that work at one election don’t necessarily work forever, and sooner or later you get something wrong. I’ve always thought the mark of a really good pollster is someone who puts their hands up to the error, says they’ve messed up and then goes and puts it right.

In terms of what went wrong this week, we obviously don’t know yet, certainly I wouldn’t want to rush to any hasty decisions before properly looking at all the data. There are some things I think we can probably flag up to start with though:

The first is that there is something genuinely wrong here. For several months before the election the polls were consistently showing Labour and Conservative roughly neck-and-neck. Individual polls exist that showed larger Conservative or Labour leads and some companies tended to show a small Labour lead or small Conservative lead, but no company consistently showed anything even approaching a seven point Conservative lead. The difference between the polls and the result was not just random sample error, something was wrong.

I don’t think it was a late swing either. YouGov did a re-contact survey on the day and found no significant evidence of this. I think Populus and Ashcroft did some on the say stuff too (though I don’t know if it was a call-back survey), so as the inquiry progresses other evidence may come to light, but I’d be surprised if any survey found enough people changing their minds between Wednesday and Thursday to create a seven point lead.

Mode effects don’t seem to be the cause of the error either, as the final polls conducted online and the final polls conducted by telephone produced virtually identical figures in terms of the Labour/Conservative lead (though as I said on Wednesday, they were different on UKIP). In fact, having a similar error with both telephone and online polls is evidence against some other possibilities too – unless by freakish co-incidence unrelated problems with online and telephone polling produced almost identical errors it means things that only affect one modeare unlikely to have been the cause. For example, if the problem was caused by more people using mobile phones, it shouldn’t have affected online polls. If the problem was caused by panel effect, it shouldn’t have affected phone polls.

Beyond that there are some obvious areas to look at. Given that the pre-election polls were wrong but the exit polls were right, how pollsters measure likelihood is definitely worth looking at (exit polls obviously don’t have to worry about likelihood to vote – they only interview people physically leaving a polling station). I think differential response rates is something worth examining (“shy voters”… though I think enthusiastic voters is just as risky!), and the make-up of samples is obviously a major factor in the accuracy of any poll.

And of course, it might be something completely unrelated to these things that hasn’t crossed our minds yet. Time will tell, but first some sleep.


-->

Most pollsters produced their final polls last night, ready to go in the first edition of whichever paper commissioned them. Today we have the final few companies – Ipsos MORI, who do polling for the Evening Standard so always publish on election day itself, Populus and Ashcroft, who do their polls on their own accord, so didn’t have to finish in time for a print deadline last night. We also have the final figures from ICM, who put out interim figures for the Guardian yesterday, but then continued fieldwork into the evening.

  • Lord Ashcroft’s final poll has topline figures of CON 33%, LAB 33%, LDEM 10%, UKIP 11%, GRN 6%. Full tabs are here
  • Ipsos MORI have final figures of CON 36%, LAB 35%, LDEM 8%, UKIP 11%, GRN 5%. Full details are here.
  • Populus have final figures of CON 33%, LAB 33%, LDEM 10%, UKIP 14%, GRN 5%. Tabs are here.
  • Finally ICM have published their final figures for the Guardian. Yesterday’s interim numbers were 35-35, today’s final figures shift only slightly to CON 34%, LAB 35%, LDEM 9%, UKIP 11%, GRN 4%. Tabs are here.

I said on Tuesday I’d revisit my final prediction in light of the final polls. My earlier prediction was based on Con and Lab being neck and neck, so no change there. The final few Scottish polls have shown slightly smaller leads for the SNP – between 20% and 23% – so while Labour are still neck-and-neck nationally, perhaps they are doing a little better in Scotland and a little worse in England than I predicted. We shall see.

As was the picture yesterday, all the polls are essentially showing a neck and neck race – they’ll either all be about right, or all be wrong. The only company showing a gap of more than one point between Conservative and Labour is Panelbase, who have a two point Labour lead. Over the past few weeks there has been some comment on the apparent difference between phone polls and internet polls, whether phone polls were showing a Conservative lead and online polls not. If this ever was a pattern, rather than just co-incidence, it’s not present in the final results, the average for the final telephone polls is CON 34.5%, LAB 34.3%; the average for the final online polls if CON 33.0%, LAB 33.0%. You’ll note that online polls have both Lab and Con lower – that’s because there is a significant difference between the pollsters on how well they think UKIP will do – telephone pollsters all have UKIP on 11-12%, but online pollsters vary between 12% from YouGov, Opinium and BMG right up to 16% from Survation and Panelbase.

And, that’s it. The next poll will be the broadcasters/NOP/MORI poll at 10pm. I’ll be working on the BBC election coverage through the night so won’t be posting any analysis here overnight, but feel free to stay and chat in the comments section if you want. In the meantime, good luck to all standing and campaigning. Good luck to all pollsters on getting it right. And good luck to those poor souls who keep or lose their jobs tonight based on a public vote.


FINAL POLLS

The day is almost upon us. Tonight we will be getting final call polls from most of the companies – though MORI and Lord Ashcroft’s figures won’t be released until tomorrow morning. In total eleven companies have polled during the campaign, I’d expect to see figures from everyone:

  • BMG already published their final call for May 2015 yesterday – results were CON 34%, LAB 34%, LDEM 10%, UKIP 12%, GRN 4%. (tabs)
  • TNS‘s poll this morning – which I assume is their final one – had topline figures of CON 33%, LAB 32%, LDEM 8%, UKIP 14%, GRN 6%. (here)
  • Opinium released their final poll this morning, with topline figures of CON 35%, LAB 34%, LDEM 8%, UKIP 12%, GRN 6%. Full details are here.
  • Finally so far, and most surprisingly, ICM have released interim figures from their final poll showing CON 35%, LAB 35%, LDEM 9%, UKIP 11%, GRN 3%. ICM’s last five polls showed Conservative leads, so a tie is a slight surprise. Their fieldwork is still ongoing – they will release updated figures tomorrow morning – but given three-quarters of their fieldwork is done they are unlikely to change much. (tabs)

Updates to follow as polls appear…

  • UPDATE1: Panelbase have released their final poll, with topline figures of CON 31%, LAB 33%, LDEM 8%, UKIP 16%, GRN 5% – the first of the eve-of-election polls to show a Labour lead. Note that Panelbase did a separate England & Wales poll and Scotland poll and then combined them together, so their Scottish figures are from a full size Scottish survey, weighted down to the correct proportion of a GB poll. In Scotland voting intentions were CON 14%, LAB 28%, LDEM 5%, SNP 48%. Full tables are here.

UPDATE2: Tonight’s final three polls from today are out (remember, MORI, Ashcroft and the full ICM figures are tomorrow). Details are below:

  • The final YouGov poll for the Sun and the Times has topline figures of CON 34%, LAB 34%, LDEM 10%, UKIP 12%, GRN 4%. Like Panelbase, the YouGov GB poll contains a separately weighted full size Scottish sample, and in this case also a Welsh one. Scottish figures were CON 14%, LAB 28%, LDEM 7%, SNP 48%; Welsh figures were CON 25%, LAB 39%, LDEM 8%, UKIP 12%, Plaid 13%. Tables are here: GB, Wales, Scotland
  • Survation in the Mirror also have neck-and-neck figures – they’ve released figures using the normal method and their new ballot paper with candidates names method, but their headline figures are the latter, which produces topline figures of CON 31.4%, LAB 31.4%, LDEM 9.6%, UKIP 15.7%, GRN 4.8%
  • Finally ComRes for the Mail and ITV are almost neck and neck – a one point Tory lead, with topline figures of CON 35%, LAB 34%, LDEM 9%, UKIP 12%, GRN 4%. Full details are here

Looking at the eight final polls we have so far the figures are very close in terms of the Conservative vs Labour race. Four companies (YouGov, ICM, Survation and BMG) have dead heats between Labour and Conservative, three have one point Conservative leads (TNS, Opinium and ComRes) and Panelbase have a two point Labour lead.

Levels of Lib Dem support are also quite similar across pollsters, with numbers all between 8% and 10%. There is a little more variance with UKIP support – the difference between pollsters isn’t as huge as it once was, it’s still there in the final eve-of-election calls – varying from 11% with ICM to 16% in Survation and Panelbase. I’ll update with the final final polls tomorrow morning.


It’s a truism that there isn’t one election on May 7th, there are 650. However, the brutal reality is that lots of them will behave much the same in terms of swing, and that in lots of them the outcome is a virtual certainty and they won’t matter. A good 450 or so seats we can be pretty confident won’t change hands this Thursday unless the polls are very wrong. We can actually boil down the election to four battlegrounds. Here’s where they are and what we know about them from the polling so far.

Conservative vs Labour battleground

Conchart2

The main battleground in determining which party will lead the government is that between Labour and Conservative. It’s also by far the largest – it’s true to say that the political geography of Britain has become ever more diverse since the days when almost every race was just Lab-v-Con, but the biggest single chunk of winnable marginals is still just that.

It’s also the battleground where good old uniform national swing remains a fairly good guide. It won’t predict individual seats – there will always be some seats with much bigger swings, some with more smaller ones – but in aggregate it should give a good picture. Overall current polls show a swing of about 3% or 3.5% from Conservative to Labour. In the Con-Lab battleground that should win Labour roughly forty seats.

However, there are two important caveats to this. The first is that almost all the Con-Lab battleground is in England & Wales, and GB polls are distorted by the completely different swing in Scotland. Labour’s vote is up by around 5 or 6 points in England & Wales, down by about 15 to 20 points in Scotland. If you look at the data in just England & Wales you find a Con>Lab swing nearer 5 points, which would win Labour around sixty seats.

The second is whether the swing in the marginal seats is the same as the swing in England and Wales as a whole. Looking at the historical data there is good reason to expect it won’t be. The vast majority of the Con-Lab battleground seats are being fought by first time Conservative incumbents who won the seat in 2010, this means they will be gaining an incumbent advantage they didn’t have last time, while in many cases Labour will be losing an incumbent advantage they enjoyed in 2010. Looking at data from past elections this impact is pretty consistent even if it is worth only a couple of percentage points (it’s worth far more for the Lib Dems). There is some evidence to support this – the recent ComRes poll of Con-Lab marginals found a swing of 3.5%. Looking at the broad sweep of Lord Ashcroft’s polls in these seats and adjusting the older Ashcroft polls to account for changes in the national polls since they were done the average swing comes out around 3.8%.

In practice this means the swing in the Con-Lab marginals may well be similar to that in the national polls, but only because Labour’s over-performance in England & Wales is cancelled out by Conservative over-performance in Con-Lab marginals. That means Labour gains from the Tories of around 40 seats, if the national polls are neck-and-neck (if the Conservatives are a point or two better, the gains will obviously be less)

Of course there will be variation between seats, so not all Con-Lab marginals with majorities below 7% will fall, there have been some constituency polls suggesting good chances of Conservative holds in marginals like Loughborough, Worcester or Kingswood. Equally though there will be some seats with larger majorities that do fall – London constituency polls in particularly have shown larger swings, so watch for places like Ealing Central & Acton or Finchley & Golders Green.

The SNP Landslide

SNPchart

The second biggest focus on election night will probably be the Scottish seats. What the story will be in Scotland is not in dispute, it will be a SNP landslide. The question is only the scale of that landslide. All the polling evidence gives the SNP a very large lead, varying between 20 and 35 points. The questions are where it ends up in that range, how accurate it is and how it translates into seats.

To deal with the overall polls first, I can well imagine that some polls in Scotland will overestimate SNP support. There have been huge shifts in party support since previous elections (and probably significant changes in the drivers of voting intention in Scotland) making it hard to model and weight Scottish samples. Equally SNP support is extremely enthusiastic – I can well imagine differential response rates becoming a problem. That said, polling error in Scotland probably won’t cause much of an upset because of the sheer size of the SNP lead – to put it bluntly, if polls give a party a 5 point lead and it turns out its actually a draw then it makes a huge difference. If polls give a party a 25 point lead and it turns out that lead is actually only 20 points it is not, in practice, such a big deal, even if the scale of the error is the same. The difference will only be between “vast landslide” and “huge landslide”. I cannot see the polls being so wrong that the SNP don’t get a crushing victory.

So how will the SNP landslide in votes translate into seats? Well, with a swing of this scale Uniform National Swing really does break down completely. UNS assumes parties shares of the vote go up and down by the same amount in each seat, but Labour cannot lose 20 percentage points in every seat in Scotland, it would give them a negative share of the vote in nine seats. The same applies to the Liberal Democrats. As a result of this floor effect, Labour and the Liberal Democrats must be losing more support in seats where they had more to begin with – their vote has fallen too much to be evenly spread across all of Scotland. This means that Labour and the Lib Dems could lose even more seats than suggested by uniform swing (and means even if the national share of the vote for the SNP isn’t as good as polls suggest, they could still get the sort of landslide in seats that the polls suggest).

The scale of the SNP surge is such that very few seats have any realistic chance of withstanding it. The most plausible ones are the very largest Labour majorities, the Glasgow North East, Kirkcaldys of the world, Jim Murphy in Renfrewshire East, the Lib Dem stronghold of Orkney & Shetland and perhaps the border seats (if the SNP don’t take Berwickshire, it is also a marginal between the Lib Dems and Conservatives).

Liberal Democrat Defence

LDchart

Given the Conservative party’s most viable coalition partner is the Liberal Democrats how many seats change hands between the two parties doesn’t make much difference to the electoral maths after the election. It is still obviously important for negotiations, party morale, the psychologically and politically important issue of who is the biggest party (and, of course, for who is the MP in those seats!). Liberal Democrat battles against Labour are far more important in terms of the hung Parliament maths.

The Liberal Democrats’ ability to win and hold seats has a famously limited relationship with their national vote share. In 1992 they got 18% of the vote and won 20 seats, in 1997 their vote went down to 17% but they more than doubled their number of seats to 46. In 2010 they gained votes, but lost 5 seats. How many seats they win has always been largely reliant upon their ability to harness tactical and personal votes in their areas of strength. That said, it’s not realistic to expect a party to lose half their national support and emerge unscathed. While I’ve seen a few claims for potential Lib Dem gains that aren’t completely ludicrous (Watford or Maidstone & the Weald, for example), generally speaking the Liberal Democrat election aim is to limit their inevitable losses as much as they can. This depends upon the demographics and political opponents in their seats, and the incumbency and entrenchment of their individual MPs.

In England and Wales the Liberal Democrats have 46 seats. In eleven Labour are the second placed party, in thirty-four the Conservatives are second placed (though in at least two of them, Sheffield Hallam and Cambridge, Labour are probably the bigger threat) and in Ceredigion Plaid Cymru are second placed. In the vast majority of the seats we have individual polls from Lord Ashcroft to give us an idea of how the race is looking. There are two extremely obvious trends – one is that the Liberal Democrats are collapsing where their main challenger is Labour, but holding up well where the main challenger is the Conservatives. The second is the sheer variation between seats, even within the LD-Con battleground and the LD-Lab battleground.

Ashcroft has polled all the LD/Con marginals that might feasibly change hands. The average swing in these seats was just over 2 points from LD>Con, enough to take about seven seats. However the swings ranged from ten percent LD>CON in Chippenham, to swings of seven percent from CON>LD in Eastbourne and Sutton & Cheam, and in practice this meant ten of the constituency polls had the Conservatives ahead – but these are just snapshot polls with margins of error, so many of these seats are in play. Note also, that many of the polls were last year and the Liberal Democrats have recovered slightly since then.

Looking at the LD-Lab battleground the average swing was a crushing 12 points from LD>Lab, meaning many of these seats are almost nailed on certainties for Labour. The exceptions are Birmingham Yardley, where John Hemming polled surprisingly well, Bermondsey where Simon Hughes was protected by a huge majority, Cambridge and Sheffield Hallam where Labour are coming from third and I expect the Lib Dems will benefit from tactical voting (Ashcroft showed Clegg behind in Hallam, but more recent ICM polling has him ahead). Hornsey and Wood Green is also interesting – the Lib Dem own polling has them doing better there and both the Lib Dems and Labour seem to be targetting it heavily, so it may be much more of a toss up than Ashcroft suggested.

UKIP Targets

UKIPchart

There is no easy way to come up with a list of UKIP targets – demographics, local and European election results can give us a steer, so can some of UKIP’s published statements about which seats they are targeting. Realistically though when a party has more than tripled their vote it is hard to accurately judge where their positions of strength and weakness are. The seats below are my best guesses of their most plausible gains (there are other seats where they have strength like Waveney, Great Yarmouth or Redcar that are in the Con-Lab battleground list… but I don’t think they stand much chance of actually winning any others, and constituency polling in some of those seats has shown them on the wane. As to how they will do in these seats – I don’t think any are necessarily easy to call. Everyone assumes Douglas Carswell will hold Clacton given his margin of victory in the by-election, Mark Reckless in Rochester looks more vulnerable. Thurrock looks too close to call, as does Thanet South with its contradictory polling. Great Grimsby was a plausible UKIP gain, but recent polling has Labour with a healthy lead. Polling commissioned by UKIP donor Alan Bown gave them a stonking lead in Boston & Skegness last year, but this year an Ashcroft poll found the Tories ahead. My own guess is that Clacton will probably be a hold, and they have a chance in these other seats… but they won’t strike home in all of them.

And the rest

That leaves a few other interesting seats that don’t fit into any of the main battleground categories, but could change hands. Two are the seats held by smaller parties – I expect the Greens to hold on in Brighton Pavilion (but not gain anywhere else), how George Galloway will do in Bradford West is anyone’s guess. Watford appears to be the only Con-Lab-LD three way marginal that is still a three way marginal for the three parties – it could go either of the three ways.

And so, a prediction

I generally hold to the pollsters’ maxim of snapshot not prediction, so I avoid predictions like the plague for most of the Parliament as the polls may yet change. In 2010 I waited until after the final polls were done before getting off the fence, but it gave me very little time to actually write anything, so this year I’ve done it up front. Obviously if Wednesday’s final polls do show the Conservatives eeking out a small lead I’ll reconsider and make my prediction more Conservative – when the facts change, I changed my mind. As it is though, my personal best guess is Conservatives around 277 seats, Labour around 267, the Lib Dems around 29 and the SNP around 52. I’ll revisit those once we have the final polls.