Given the success of the approach at the 2017 election I expect we’ll see several MRP seat models this time round. The first one to emerge however is one constructed by Focaldata, using data from mixed sources, including YouGov, that Best for Britain have used to drive a tactical voting website. It has caused some controversy – particularly on the comment pages of the Guardian – with people arguing over the validity of its recommendations. I won’t get too far into that (vote for whoever the hell you want), but thought it was probably worth making a few comments about MRP itself, considering it will crop up again through the campaign.

What is MRP?

First, we need to understand what MRP is. It stands for multilevel regression and post-stratification, which almost certainly doesn’t help. There is a academic paper by Ben Lauderdale and his colleagues who run the YouGov MRP that explains it in great detail here, however the short version is that it’s a modelling technique aimed at producing robust estimates for small geographic areas from large national samples. In the context of elections, that means coming up with estimates of vote share in single constituencies based on a big national sample.

Using traditional techniques even very, very large samples don’t contain enough respondents to be a good guide to individual seats. If you had a huge sample of 50000, divided by 632 seats it would still give you less than 100 people a seat – which wouldn’t be enough to produce decent data. I’ve seen this as a naive criticism of the Best for Britain MRP model (there are only 70 people per seat!) but in fact that is exactly the problem that MRP is intended to solve.

MRP works by modelling the relationship between demographic and political variables and voting intention (the multilevel regression part), and then applying that to the demographics and political circumstances in each individual constituency (the post-stratification). So in this case, an MRP model would look at how demographics like age, gender and education relate to vote intention, and how that differs based on political variables (Is there an incumbent MP? Is it a remain or leave area?). That model is then applied to the known characteristics each seat. What that means is the projection in an individual seat is not just based upon how respondents in that seat say they would vote, it’s effectively also based on how respondents with the same demographics in seats with similar political circumstances say they would vote.

How well does it work?

At the last election YouGov had an MRP model that performed very well – correctly predicting the hung Parliament and some of the more unusual election results like Canterbury and Kensington. Clearly, given the accuracy of the YouGov model, it is possible to use MRP successfully to produce decent seat level estimates from a big national sample.

Best for Britain’s defence of their tactical recommendations relies heavily on how well the YouGov MRP model did in 2017. However, not all MRP models are necessarily equal. It isn’t one single model, it’s a technique, and it’s possible to do it well or badly. It is not certainly not a magical guarantee of accuracy. If we look back to 2017 the YouGov MRP model got all the attention, but it wasn’t the only MRP model out there. Lord Ashcroft also commissioned an MRP model, but that wrongly predicted a Tory majority. Just as some polls have been more accurate than others in recent years, some MRPs may be more accurate than others.

The things that drive the quality of a MRP model should be the quality of the data that’s going into it, and the quality of the model itself – have those designing it picked demographics and political factors that allow them to accurately model voting intentions? As an external observer however, it is quite hard to judge that. For the YouGov model there is its track record from 2017. From other MRP models, we’re driving a bit blind. We know it is a technique that can be very successful if done well, but we won’t really know if it is being done well until it’s compared to actual election results.

Are tactical voting recommendations based on an MRP model sensible?

In principle, yes. MRP is obviously not perfect or infallible – nothing is – but it is an established technique for producing estimates of support in small geographical areas from a larger national poll. Certainly it should be better than using a crude uniform swing, or just basing recommendations on what the levels of support were at the previous election and assuming nothing has changed since.

In practice, of course, it depends on the quality of the model and the tactical decisions that people make based upon them – I certainly don’t intend to get into that debate, especially since I expect there will be various rival tactical voting sites with different recommendation, and perhaps different aims and motivations.

543 Responses to “MRP models and tactical voting”

1 9 10 11
  1. I have two options for beacon lighting, though I doubt either would be in sight of The Malverns.

    Beeston Castle (see Civil War) is on a pimple over the Cheshire Plain and could bring the bad (or good) news to the citizens of Chester and the jumble of pipes and chimneys that is Runcorn and Widnes.

    In the opposite direction and with car access, so easier on the knees, is Mow Cop, overlooking what might, in reality, be called the mess that is The ex-Potteries.

  2. Farage has pulled out of the Sophy Ridge show on Sky News this morning. ??

  3. ‘Corbyn snubs Festival of Remebrance

    I’m surprised this isn’t getting a lot of coverage

    Out of the election book of “how to drop %10 in the polls.”’

    I doubt it would have the slightest impact on anyone who is not already a committed Tory / BXP.

    I suspect he is continuing to be up north where all the floods are which is just as important in many people’s eyes. And will certainly count more up there…

  4. More proof that Poppies & Remembrance Sunday has now sadly been wholly coopted by nationalists who are completely contrary to the original “never again” message.

    Stopped buying poppies after both Blair (Iraq) and Cameron (Libya) didn’t even blush about wearing them.

  5. @Jack

    The floods are a small area of Yorkshire, (which I think of as the Midlands not the North) Bad though it is and my thoughts are with those affected.

    The thoughts of The Nation will be with those that made the ultimate sacrifice to defend our country. (or at least they should be)

  6. “The thoughts of The Nation will be with those that made the ultimate sacrifice to defend our country. (or at least they should be)”

    Thought police?

  7. Not a fan of Corbyn but he was visiting the residents and emergency services in Doncaster in relation to the floods. I do hope this year people do not try and politicise the remembrance services this year, it is very distasteful in my eyes.

    You might as well hope for people not to attack Johnson for articles about letterboxes. Corbyn has made himself vulnerable to this kind of charge because of his past actions, of course it’s going to be used against him. What’s surprising is that people on his team weren’t wise enough to know that missing anything to do with Remembrance services would be a big old own goal. Visiting flood victims is undoubtedly the right thing to do, but it would have been very possible for him to get in a car back down to London in time to get to the Albert Hall. That he hasn’t is either foolish, or suggests that he was content to have a reason to miss it. Whichever way, one of the cardinal rules of campaigning is surely to avoid giving your opponents ammunition to use against you. Both the parties are making a stunning number of unforced errors, and the campaign has hardly got underway.

  8. I think @Alec’s general analysis is right and we are on course on current trends for a comfortable Tory win even if the Remainer tactical voting is more effective than the Leaver counterpart.

    I’m not sure the manifestos are likely to change that much either as Labour’s is mostly known and the Tories won;t make 2017 style gaffes. I suspect only a great big scandal or an implosion in one of the TV debates will stop Johnson now

  9. Yougov tables are out. A lot of supplementary questions on there, interesting to see that people think the Russia report should be published before the GE by 62% to 15%. As might be expected from those figures, every demographic wants it to be published, conservatives by 53% to 25% for instance.

  10. @ Garj

    “That he hasn’t is either foolish, or suggests that he was content to have a reason to miss it.”

    I don’t think it’s foolishness or forgetfulness. Those in the campaign team must have thought about it and decided on this course of action. Now what the logic of that choice was, I have no idea.

    One thing I would like to point out. Just because some of us don’t display riduculously big poppies, or go to the Cenotaph, doesn’t mean we don’t appreciate the sacrifices made by so many people. We just prefer to mark it in our own way. Such as opposing any new conflict.

  11. @Jim Jam, would like to know how you can be so sure of that, lib dem vote share has gone up a lot since election 2017 and Battersea had one of the highest remain votes in the country so it would seem likely that they have risen substantially here too. Ok, so the lib dems only got 8% at the last election (although there was an independent who was a bit of a lib dem who was on 2%), and the lib dems have no history of seriously challenging in this seat, but projections showing it as pretty close between the big three parties do not seem unrealistic IMO

  12. Let’s remember that Labour won a hugely impressive 40% of the vote in 2017 – an even more incredible 48.9% in Wales.

    Replicating that, never mind going forwards, was always going to be difficult.

    For the Tories, replicating their 2017 performance only in terms of seats would be advantageous if the Gaukeward squad are no more.

    As Alec says, Scotland will be interesting as the defining issue of Remain / Leave there is at a different flavour to rUK.

  13. NICKP

    Have to say I’ve come to share your view and your regret at the hijacking of remembrance and the poppy symbol in particular by the extreme nationalist tendency.

    I could just about stomach Blair and Cameron. Politicians are always thus. But I think it has manifested itself most distastefully in the last straw sense for me in the hounding of Irish footballers who have reservations at the recent habit of forcing them to wear things in English football matches. I may not share the footballers’ interpretation of Irish and British history but I defend their right not to have someone else’s rammed down their throats.

    For the first time in many years I will be quietly walking the dog in an hour’s time rather than participating collectively. I resent being put in that position.

  14. @ JIM JAM – “no Labour MPs will want to deny Lab voters the opportunity to vote Labour”

    Which is excellent news for CCHQ, Boris’s chances of winning a majority and UK leaving the EU with Boris’s deal.

    HAD LDEM and LAB agreed a seat split pact then it would have been along the lines of LDEM-Green but this time LAB would have been by far the dominant partner and enjoyed by far the largest number of seat gains/holds.

    Several LAB MPs have commented on the issue but I’m very used to being told I’m wrong so I’d encourage folks to look into it themselves.

    As for polls narrowing then I’d change that to “converging” (as BMG have converged from the other side). Not all the new polls are up on wiki yet but they will be soon and they are “converging” around a 10% lead for CON (although IMO the “intra” splits are more important than the CON lead over LAB)

  15. Old Nat

    “Best for Britain” – what an odd title for a “tactical voting” site that has no recommendations for anywhere in NI or Scotland.

    I’m surprised that you seem unaware of who Best For Britain are, as their campaign, and their polling efforts have been mentioned repeatedly by various posters on this site over the past couple of years.

  16. There are also, of course, remembrance services held in Yorkshire.

  17. Remembrance has been hijacked as yet another nauseating exercise in nationalistic pomposity. It used to be a rather dignified affair, allowing people to go quietly about their business, showing reverence and respect in their own way, on their own terms. Pious twits constantly trying to outbid each other has caused me to dread this whole spectacle.

  18. @ TW

    “As for polls narrowing then I’d change that to “converging””

    Yes, I’d agree with that. In fact it looks suspiciously like herding in the four most recent polls. I was also hoping for a new ComRes and ORB to see if they were herding too. I guess we’d have heard by now, so they can’t be running weekly.

  19. Many of the new polls ask who is best at each important issue and quite a range between the different companies.

    Opinium are generally the best of a bad bunch for LAB (eg LAB still ahead on Health / NHS)

    YG are in the middle (CON ahead on everything and tied on NHS from -5 last week)

    Deltapoll are best of a good bunch for CON

    Note they all use slightly different wording and Boris/CON do best in polls that the increase in emphasis on Boris v Corbyn (and decrease the emphasis on party loyalty).

    Implication? Boris will want to keep the focus on “Vote ABC = Get PM Corbyn” BUT Corbyn has a lifeline with the TV debates and given his approval ratings are rock bottom then he will have more to gain IF he can “win” those debates. Can he make lightening strike twice and narrow the CON lead during campaign? Well Boris has Corbyn’s playbook and is showing up at least so TBC…

  20. Oldnat,
    “Those wishing to give Johnson an overall majority should probably make themselves aware of the massive extension of the executive power of the UK Government”

    Oh dear, not another needle hidden in the brexit haystack which the lords might stab itself on and revolt? Another issue to surface and create legitimate opposition to prevent the bill passing, yet the issue wouldnt be directly about Brexit?

    What was that theory about trying to stop brexit while not opposing brexit?

  21. That YouGov poll, projected by Flavible, is absolutely grim for Labour in Wales. No seats at all in the North.

  22. Expect a doctored video to be issued later today by the Tory campaign team showing Corbyn apparently urinating on the Cenotaph this morning. Obviously he didn’t, but the video will airbrush his wreath out and make it look as if he is about to open his flies instead. A fuzzy image will then be superimposed over his groin region. This will go viral on social media by the end of the day.

    When questioned on this by Piers Morgan on tomorrow morning’s ITV Good Morning show, the Tory Chairman, James Cleverley will say. “OK, I accept he wasn’t actually urinating on the War Memorial, but it’s clearly what he wanted to do. We know that from his past comments. He hates Britain and always sides with terrorists…..”

    This could be the game-changer we’ve been waiting for.


  23. Electoral Calculus seem to be including many (but not all) of the known “pacts” (and candidates that have pulled out) for NI and gives a seat by seat run down:

    Anyone who has their own views of the impact of “pacts” and wants to compare them to other models might want to take a quick look at some of the GB seats NOW and compare the current numbers to the numbers AFTER Martin “zeroes out” parties in seats where a candidate has pulled out

    (eg seats like Exeter, Portsmouth South and Watford)

    A bit GEEKy perhaps but I’m very keen to see how other models adjust for “pacts” (I just hope he doesn’t update the overall regional numbers at the same time)

  24. Glancing through the yougov questions this one stood out to me:

    “Thinking about Britain’s relationship with the EU after Brexit, do you think we should…

    Aim to retain the right for EU citizens and British citizens to live and work in each others’ countries: 67%

    Aim to end the right of EU citizens and British citizens to live and work in each others’ countries:15%”

    Sadly I’d say these numbers are flattered by the inability of the public to understand a simple question. I’m sure a chunk of the respondents thought they were being asked about people already living in the UK/EU.

    I’ve sort of lost the will to even follow it all anymore. I can’t stand Johnson, Corbyn makes my skin crawl, the DUP are Neanderthals, Sturgeon rabbits on like a Duracell bunny, the greens are in cloud cuckoo land and every single time I listen to a member of the public struggling to articulate (or even find) the reason why they’re going to vote one way or the other it just shakes my faith in democracy even further.

  25. Flavible’s seat projection based on last nights Yougov poll:

    CON 386 (+68)
    LAB 157 (-105)
    SNP 49 (+14)
    LDEM 35 (+23)
    PC 4 (-)
    GRIN 1 (-)

    Labour are down to 1 in Scotland, Tories down to 5.

  26. Back to comparing this year’s polls with a similar period in the 2017 GE, we come to the weekend after the local/mayoral elections which were so disasterous for Labour. With very approximate averaging of weekend polls, we get:

    Con 47/40
    Lab 29/29
    LIbD 9/16
    SNP 4/4
    GRN 3/3
    UKIP/BXP 6/7

    The only real difference is the -/+ 7% in the Con/LibD figures. However, this was pretty much the low point for Labour, and if there’s to be a similar story in 2019, the VI needs to start shifting now. At the moment I don’t see it, but few saw it in 2017 either.

  27. Labour seem to have gained about 1% in the week looking at the latest polls but the Tories have edged to 40-41% in three out of the last four polls. Brexit edging down and LibDems static.

    As TW posted the Tories do well in all the issues questions, ahead on education and equal with Labour on NHS and 18% clear on the economy.. Corbyn seen as much more extreme than Johnson.

    If things stay like that i will be happy.

  28. I don’t think that Corbyn’s absence from the Albert Hall last night will have any effect; he was at the Cenotaph this morning. Interestingly, the Greens were missing.

    On another election issue,.I found it surprising that people are now complaining about their Commons seat being occupied by the Speaker:

    There is more to politics than just voting in the Commons (and anyway the system with a Seaker and three deputies divided equally between the two largest parties ensures that their votes cancel each other out). It is rather like objecting to pairing.

    It seems to me that these people are just looking for an excuse to be angry.

    On the subject of Jo Swinson, I thought that at the beginning she was too strident in what she said about Jeremy Corbyn. She could have simply said that she did not believe he could achieve a majority even if supported by the LibDems. That would have had exactly the same effect without offending Labour voters who might vote tactically in Con/LD marginals.

  29. CB11
    I’d be happy to be counted as part of your coalition of you’d be prepared to have me, fortunately in Northampton South there would be no need this time around to don a nose peg and vote lib since they only achieved about 5 percent of the vote last time. I’d be happy to vote green, and have done in the past, or PC if they were to branch out this far.

    Don’t think we’ve got a beacon here, shinning up the lift tower might be a bit tricky, and will have it’s light replaced by the lit up Christmas tree arrangement it sports every year by then.

    I confess to feeling hugely disheartened and pessimistic about the prospects of Johnson not getting a majority, even more so than 2017, although cling to a glimmer of hope that result might be repeated. Johnson seems bullet proof, he starts from such a low moral bar I simply can’t imagine what he would need to do for something to cut through.

    I agree totally about remembrance Sunday, but I’ve been refusing to go along with it for about forty years as it’s been hijacked by the right for as long as I can remember. Things do seem to have moved on in so far as I rarely get pressured into apologising for holding pacifist views any more, although it may just be that I’m well past the age where I particularly care if people agree with me, or have any expectation of changing anybody else’s mind.

  30. I’ve updated my model and this is what it looks like:

    Con 355 (+37)
    Lab 194 (-68)
    SNP 51 (+16)
    LD 26 (+14)
    PC 4(-)
    Grn 1 (-)

    A little different to Flavibles.

    The suggested overall VI:

    Con 37
    Lab 24
    LD 17
    SNP 4

  31. Meanwhile the BBC reports another Tory candidate has stood down for ten year old racist posts on Facebook, why these people have historically said things on a public forum they claim never to have agreed with I find difficult to fathom. The Guardian doesn’t seem to have picked up on this but is running at least two major articles on anti Semitism in the labour party.

    At least when the manifestos get published the media will have something concrete to talk about, I’m finding the whole campaign on all fronts the most depressing political experience I can remember in my lifetime, including results night in 1992.

  32. Brave New Thread

  33. Alec,
    “The Lib Dems going full belt for revoke seems to have lost them support amongst Cons. It was a terrible move,”

    I dont altogether agree. They did have to have a remain position, so what could it be? “2nd ref is simply labour’s position, whereas they wanted to try to keep themselves more remain than labour.

    Libs are very unlikely to win this election. Presumably they accept that, so they arent going to have to live with the consequences of being elected on a manifesto promise. What I think is their main chance is to be the most remain party in a field where labour fails to demonstrate its ability to deliver for remain (for whatever reason, either electorally now or losing a 2nd referendum). That will place them well to become the remain party in the future.

    If Brexit happens, does labour become a rejoin party or a make the best of it party? If the latter, then libs have a crowbar to use to pick up rejoin voters for the future. With brexit going badly, that represents a route to majority lib government.

    Any better ideas how to get a liberal government within the next ten years?

  34. Danny – depends on Tory majority and if enough Tory back-benchers are left who would prefer something more than a FTA with the EU,

    Labour would be arguing for a particular future relationship in the first year of Tory Government, no point coming out for rejoin so early. In fact I would be one of those saying stability is key and that a rejoining call so early would be wrong.

    For a GE in 2028/30 perhaps?

    NB) Corbyn would go and the leadership contest and outcome would shape the policy of course.

    Not a done deal yet as a 4% narrowing which is quite possible Perhaps not likely but possible) as ABT DKs firm up, Greens get squeezed as Labour will be targeting left leaning leavers with measures attractive for them beyond Brexit.

  35. It’s the Spanish General Election today (not another one!)

    It will be interesting to see how close the final polls get to the ultimate outcome. For reference the final poll by GESOP/El Periòdic was as follows:

    PSOE (centre left) 26.7% (116-122 seats)
    PP (centre right) 19.4 (83-88)
    VOX (far right) 14.9 (47-53)
    Unidos Podemos (socialist left) 14.3 (39-44)
    Ciudadanos (Originally centre-centre right, now radical/anti establishment right – leaning far right) 7.9 (14-18)
    Mas Pais (socialist left) 2.8 (2-4)
    Esquerra Republicana Catalana (left wing Cat Nationalists) 2.6 (12-13)
    Junts Per Catalunya (right of centre Cat Nats) 1.4 (6-7 seats)
    CUP (far left Cat Nats) 1.1 (3-4 seats).

    176 seats are required for an overall majority.

    Spain follows a regional PR method that exaggerates the vote in rural areas (hence the PP always getting a higher seats/votes ratio than PSOE).

    The Catalan parties vote shares and seat numbers should more properly be presented separately to those of the national parties (as in the UK with the SNP).

  36. jim Jam,
    notwithstanding more pro con spin on here in the last day, nothing is decided yet and the election is very wide open.

    Your post suggests you think labour would be a make the best of leaving party. Just as there is a proportion of people who would pay any cost to leave the EU, so I think there is one for being a member. They will also be floating voters.

    If libs can pick up 10% of hard rejoin voters, they are in contention for serious numbers of seats. They wont do it this time (barring a labour disaster), but next time labour might give them these voters.

    I suspect libs think there is an outside chance of a landslide win this time. And they are right, but its very outside.

  37. Agree Danny,

    Lib Dems being rejoin in 2024/5 (FTPA will be repealed surely)would draw some votes from Labour but it doesn’t make it the wrong thing for Labour to do in essence.

    As above the size of the Tory Majority, hence what future relationship we end up with and the leadership race would shape but I am not writing this GE off just yet; and a majority for ref 2 is still possible.

  38. Temperatures of -7c in Braemar last night. We’ve had constant sub zero temps for 2-3 nights. No snow or ice worth mentioning though. Flooding in some parts of the UK.

    And five weeks to go…I expect lower turnout once folk get into the routine of staying close to home once the worse weather is more constant.

    Having said that, November weather can be more wintry and Decembers have been mild.

  39. Remembrance ceremonies a noticeably white/WWC phenomenon in this part of W London.
    Other notable feature today: in my afternoon canvass Corbyn mentioned as a negative several times. I’m no Corbynista but I find it staggering that he can be seen as a negative when the alternative is Johnson.

  40. Guymonde,
    ” I’m no Corbynista but I find it staggering that he can be seen as a negative when the alternative is Johnson.”

    i think that might be because on the whole tory MPs are a lot more supportive about their own leader than labour MPs are about theirs.

  41. MOG
    My two penn’orth for what it’s worth. I think it adds up ok.

    Con 344
    Lab 183
    LD 50
    SNP 49
    Grn 2
    PC 3
    DUP 8
    UUP 1
    SF 8
    Ind 1
    SPKR 1

    I think the LDs may regret standing down in places like Canterbury, if that’s what happens. If I was a remainer Tory (and there will be plenty of them in Canterbury) I would be looking to lend them my vote but no way would I give it to Labour under its current toxic leadership. I would either not vote (unlikely) or stay Tory. For many remainer Tories, they fear a Corbyn govt far more than they fear leaving the EU.

    So, just as you can’t expect all BXP supporters to vote Tory if they have no candidate of their own, the same applies to LD voters and Labour. Trying to fix the result might just backfire on them.

  42. Robertinnewark,
    “If I was a remainer Tory (and there will be plenty of them in Canterbury) I would be looking to lend them my vote but no way would I give it to Labour under its current toxic leadership. I would either not vote (unlikely) or stay Tory.”

    I think you have faulty logic here.

    First, the libs will not win in canterbury. So anyone voting for them knows this, and any con deciding to vote lib knows it is a wasted vote already.

    If he has already made that decision, why would he not simply waste his vote by not voting, if libs withdraw and there isnt anyone else to vote for? In choosing lib he has already rejected both parties who stand a chance of winning. Why would he go back to them?

    Second, choosing to waste your vote by voting lib is already accepting that in doing so it increases the chance of lab winning, because you are not voting con. It means anyone who would do this has already decided it is more important to vote against brexit even if that means helping lab.

    It is quite rare, where libs are in second place, that a former con choosing to support them is making a step to support a party with any chance of winning.

    The implication is pretty much every remain ex tory now supporting lib would be open to voting lab if given no lib alternative.

1 9 10 11