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 7 8 9 10 11
  1. @Oldnat / Prof

    Not sure the Greens really had to try hard with regards P&NP. They only sat once in 2015, and came 5th.

    It’s most likely going to come down to whether P&NP residents are Brexiteers or Remainers to decide it.

  2. @David Carrod

    Not partisan at all. Nosiree Bob.

  3. First leaflet through the door here is a Labour one, although Plaid were canvassing this week a few streeiaway.

  4. Streets away. automatic correction.

  5. ON asks; How do LD supporters on here view Swinson’s performance?

    I think the LDs were very short on leader material in Parliament, hence some consideration given to the option of a leader who is not an MP. I’d have preferred Vince Cable to take us into this election, despite the age thing likely being thrown at him. To my mind, Layla Moran is the most obviously talented, but could understand why as a new MP in a very marginal constituency, she didn’t want to put herself forward.

    It’s too early to say with Jo Swinson. She doesn’t come across that well in short sound-bites, although seems to present better in person and some of that might show, were she allowed to participate in the leader’s debates,which even managed to make Nick Clegg look good. While it’s clear what the LD position is on Brexit, there is a need for her to expand on other policies, which might attract those not deciding their voting purely on Brexit.

  6. “The ‘Remain alliance’ could accidentally help Boris Johnson win a majority and force through Brexit”

    Errr… not in the DIRECT way that they say!

    The Greens are the “Unite to Remain” (U2R) pick in Exeter and Stroud and with no disrespect to their chances that will probably help Bradshaw (Exeter) and Drew (Stroud) keep their seats as LDEM will not be standing (it’s also possible the Green candidates stand as “paper candidates” in both cases).

    We’ve discussed Portsmouth South and Cardiff Central on UKPR before and perhaps the article has a point in those two cases but see below for why the INDIRECT implications might be more important.

    However, the LDEM-Green-PC (U2R) alliance should directly help PC win Ynys Mon, give Greens a better chance on Isle of Wight (still a long shot) and help LDEM keep their GE’17 seats and take 6+ from CON.

    Net DIRECT benefit of Unite to Remain (U2R) Alliance is IMO 5+ seats

    However, there are a few INDIRECT implications of U2R

    1/ LAB will not be taking part
    a/ Since LDEM and LAB generally have very different CON marginals (either as holds or targets) then CCHQ will be breathing a HUGE sigh of relief that Corbyn-McDonnell have totally ruled out pacts.
    b/ LDEM can hence then claim to be THE Remain party (and if/when Boris wins they can blame LAB)
    c/ Above also means we might see bitter infighting between LAB and LDEM going into the GE (ie getting the blame in early!!)
    d/ If voters see LAB and LDEM will never work together then does that hinder any grassroots or individual tactical voting recommendations and GOTV for both of them (who wants a hung parliament with parties that won’t work together?!?)
    e/ Having Greens back LDEM makes LDEM look “Greener” than they really are (and hence Green “lent votes” that have previously helped LAB might now help LDEM – which in most cases helps CON of course as it lowers the threshold in LAB-CON marginals)

    2/ Leavers might think they have to be more tactical
    a/ More pressure on Farage and BXP candidates to “stand down”
    b/ More awareness amongst BXP VI to switch to CON to ensure “Remain” alliance doesn’t say B*llox to Brexit

    Now part of the reason why I think polling suggesting 370ish CON seats might be wrong is due to potential of 20+ seat gains from Remain alliances/tactical voting (and 0 from Leave side).

    So if Remain are only gaining 5+ DIRECTLY (and hindering themselves INDIRECTLY) then either B4B’s MRP analysis or Electoral Calculus, Flavible, my, others, etc model predictions are going to closer to the actual result (unless we see a significant narrowing in CON’s lead)

    NB Obviously a slightly biased opinion so keen to hear a counter argument from Remain side.

  7. Also add into point 1/ above, the increased chance of internal LAB factional infighting as some 2nd ref and Remain LAB MPs will be furious that LAB HQ has refused to take part in any alliance.

    If voters see LAB’s internal war flaring back up then that doesn’t exactly help encourage GOTV for LAB VI.

  8. Rather embarrassingly I only got 4/10!

    Really just a campaign tool but still a bit of fun!


  9. For those that didn’t get the Number Cruncher Politics email then it is worth signing up. Matt points out the “INTRA” regional issue is important.

    This is something I’ve mentioned before.

    – “Bristol” can distort SW England
    – “Uni towns” in a region can distort the region
    – London and Scotland have “intra” differences

    He also provides a link to a site that is showing


    It’s pure coincidence that every approach is showing SNP at “50” and please note these numbers are in all cases “central” numbers, based on a huge list of assumptions and caveats. IMO you should add a “confidence limit” range of at least a +/-30 (maybe 50) to CON seat number.

    However, even taking 50 off the lowest number will still mean CON win most seats and make it virtually impossible for Corbyn to form a govt (unless Swinson totally capitulates and wants to send her party back into the electoral wilderness in the GE after this one)

  10. David Carrod

    I am quite happy for you to make such comments on here.

    It’s always hard to know what other folk are really like, so clarification from you about yourself is useful.


  11. Statgeek

    The point about the putative SGP in P&NP wasn’t that they could win (or even come remotely close).

    It was that their intervention might prevent the seat staying SNP (2017 majority 27). It’s all about positioning within the current Scottish Parliament and the 2021 Holyrood elections.

  12. Peter Cairns

    You did better than me!

  13. Trevor,

    ”some 2nd ref and Remain LAB MPs will be furious that LAB HQ has refused to take part in any alliance”

    Is just not true no Labour MPs will want to deny Lab voters the opportunity to vote Labour.
    There are of course divisions within Labour but your notion is simply inaccurate.

  14. Jim Jam

    “no Labour MPs will want to deny Lab voters the opportunity to vote Labour.”

    True – in a way. By denying Labour members in NI the right to stand Labour candidates, automatically you preclude them from having been “Labour voters”.

    Isn’t the definition of such an argument – “sophistry”?

    ”Not partisan at all. Nosiree Bob.”

    No more partisan than the majority of Lab supporters on this site.

    ”It’s always hard to know what other folk are really like, so clarification from you about yourself is useful.”

    I’ll let others be the judge:

  16. Alec
    “I’m wondering where you feel your political home is now?”

    I’ve never really had a political home. I tend to vote for whatever I consider the least damaging party at the time. That’s often the Tories, but by no means always.
    (Later post)
    Sorry I didn’t know what a Sure Start centre was till I googled it. It appears to be some kind of State Nursery School?
    OK, I have very little knowledge of the benefits system. I did go on something or other for the last 6 months before I reached retirement age, after suffering a serious illness, but never had any problem. I bow to your greater experience.
    (Later post)
    I do have a lot of sympathy for those in genuine poverty through no fault of their own, though some of the modern definitions of poverty are ridiculous.
    Peter Cairns
    I got 5/10 on the Scotland quiz.

  17. @Edge of Reason

    ” If you take a starting point in the 1950s then it’s quite unusual.”

    Well, I thought I’d be as current as I could – like in my lifetime.



    Thanks for your post and, typically, it’s a very reasoned and deeply thought out one. I think you and often think alike on politics, as we occasionally remind each other, and there’s much that I agree with you about in relation to Labour’s electoral difficulties under Corbyn’s leadership. I’ve spelt out my misgivings about him many times and without rehashing all of them, I think they may be very similar to yours. He’s very much a throwback to 1970s Labour, and probably trapped in that time warp still, seeing modern political and social challenges very much though that refracted and timeworn lens. I think he chooses political confreres from that era too who see the world much as he does. That has been one of his most serious mistakes as leader in my view. He’s led from a much more narrow base than he needed too and denied himself and his party the talents of big hitters like Cooper, Burnham, Benn and Miliband from the social democratic wing of Labour. His front bench has accordingly looked either callow (Long-Bailey, Butler, Burgin, Chakrabati) or, as you rightly observed, somewhat old fashioned and anachronistic (Trickett, McDonald, Gwynne, Abbott, Healey, Gardner, Lavery etc.). Take Starmer, McDonnell, Ashworth and Thornberry out, and I see no real political stars there, not great proponents or proselytisers for the cause . Many potential ones seem to be lurking sullenly on the back benches or on their way out of the Commons.

    It’s why I don’t think Labour can possibly win this election with Corbyn as their leader and putative PM. but I’m desperately hoping, like you and John33 I think, that a sort of brave rearguard action can stop Johnson returning to Downing Street. We need a little help from our “friends too, as they say. Sturgeon, Farage, Swinson, Plaid Cymru and Ms Lucas and for us Labourites to rally round the old stricken party again. I thinks an essentially defensive damage limitation campaign lies ahead.

    We are where we are and it is what it is. No good wishing it was otherwise. Too late for that.

  18. @ Colin your comments on my posts are always welcome. I always read yours and gain much from them. I would comment on them if I felt I could contribute but you are often talking of technical matters of which I know little although I am still interested in them.

    on the point at issue I have never voted conservative but I found much to respect in the ‘traditional’ one nation conservative party. So too did my socialist father-in-law who would have been given a court martial and shot as an early pacifist in the first world war but for the intervention of a conservative peer. This cannot have been a popular move at the time and I wonder how many of the conservative MPs now would go out of their way in such a cause.

  19. @jim jam

    “no Labour MPs will want to deny Lab voters the opportunity to vote Labour.”

    obviously not. They should, however, be ruthlessly selective and target their efforts on those seats they need to hold or are within a hair’s breadth of winning.


    Thank you.

    What an interesting story. And your final question indicates to me that you feel the Conservative Party has moved to the right-just as I am convinced Labour has moved far to the left.

    It seems you & I are clinging on to our basic political beliefs in the hope that the Party which we both imagine embodies them-actually does so at present , and will show some sign that they do !

  21. Oldnat,

    “ You did better than me!”

    I always do!


  22. Peter Cairns


  23. I suspect polls will be like buses tonight. Nothing for ages then several at once. First one now in view on Election Maps:

    Westminster Voting Intention:

    CON: 41% (-1)
    LAB: 29% (+3)
    LDM: 15% (-1)
    BXP: 6% (-3)
    GRN: 2% (=)

    Via @OpiniumResearch, 6-8 Nov.
    Changes w/ 30 Oct-1 Nov.

    Let’s see what the others look like. So far, it seems to me, both CON and LAB have been benefitting from squeezing the other parties, but the gap between them hasn’t changed much.

  24. Opinium Poll Observer
    4 points gap narrowing seems quite a bit for one week, 25%, from 16 to 12.
    Surprised Brexit so low, that will help Tories.
    See what the others come up with.
    Fairly optimistic for Labour considering bias and smears
    Good positive campaign going well so far,

  25. Opinium Poll Observer
    4 points gap narrowing seems quite a bit for one week, 25%, from 16 to 12.
    Surprised Brexit so low, that will help Tories.
    See what the others come up with.
    Fairly optimistic for Labour considering bias and smears
    Good positive campaign going well so far,

  26. No wonder Johnson tried to rush through his Withdrawal Bill without scrutiny!

    HoL Constitutional Committee have done a preliminary analysis of it.

    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 being proposed, and consider whether that is actually what they want for the long term, as well as their immediate political needs.

  27. OOps?

  28. I’m sure both Labour and the Lib Dems are at a local level concentrating on the areas they need to win and focus resources in such a way and this will to some degree affect the dispersal of votes. This would be due to having limited resources to fight the campaign both in terms of personnel and finance. For instance in London Lib Dem and Labour would swamp activists where they think they can win.

  29. @David Carrod

    “No more partisan than the majority of Lab supporters on this site.”

    Be sure to post up which Lab supporters referred to tin pot parties. I can’t remember who it was. As mentioned already, it’s good to see this level of honesty from some in the voter pool.

    For what it’s worth, I don’t want to rob you of the party and government you deserve. I just don’t want you and your ilk imposing it on me.


    My friend in Canterbury says there is a great and very energetic ground operation by the Labour party there.

  31. CROSSBAT 11
    Hello to you; hoping for a win tomorrow v Brighton, to go back to 7th place.
    I think Corbyn is a throwback to the 1970’s far left, I agree.

  32. @Peter Cairns

    4/10 too. I knew the first three, and the rest had me flummoxed. A lucky 4th guess.

    I should have had a read of this first:

  33. While waiting for more polls, I quickly totted up the new voter registrations since the election was called. It’s now got past 1 million (just), but don’t be too impressed. There were 3 million new registrations after TM announced the 2017 GE. I imagine it will eventually reach roughly that level (last time, 600k registered on the very last day), but I don’t expect it to be very different to 2017 in the end.

  34. Trig Guy

    Are these numbers just hits on the site, before they are transmitted to the local authorities?

    Is there any data on how many are just registering a change of address, or unsure if they are on the roll so re-registering to make sure?

  35. Battle has been joined in Brighton Kemptown.

    Two out cards today -Labours goes heavy on Remain and mps voting record on Brexit,Tory ignores Brexit ,doesnt mention Johnson and splashes on nhs,police and schools. is the progressive capital of Enngland.

  36. @Earwiggle – Pam

    “Opinium Poll Observer
    4 points gap narrowing seems quite a bit for one week, 25%, from 16 to 12.
    Surprised Brexit so low, that will help Tories.
    See what the others come up with.
    Fairly optimistic for Labour considering bias and smears
    Good positive campaign going well so far.”

    Some on hear believe (and I hope they are right) that Lab-BXP voters may return if BXP leans towards the Tories.

  37. @Batty

    Wholly agree with your anaysis of Corbyn and the Labour front bench.

    A heavy defeat and the party may yet go to the Benn/Cooper wing of the party for a new leader. A reasonable defeat and a Starmer or Thornberry leadership will almost certainly increase the base of the shadow cabinet, but probably not to the extent of bringing in Benn or Cooper (Burnham is already in a better job, and I doubt Milliband wants to come back anyway). A Corbyn win, or near thing, would bring a Long-Bailey leadership, gawd elp us.

    I think we have to hope for the middle option, with the SNP and (holding noses) the LibDems doing enough to deny Johnson a majority.

  38. @ChrisLane1945

    Good luck tomorrow. I saw Brighton a week or so ago and they’re a much improved side. It will be a difficult game for this fitful United side. Their coach Potter is someone to keep an eye, I think.

    As for Corbyn, I think we agree. However, now’s not time for post-mortems. Plenty of time for those another day perhaps. Let’s see if he can confound me again!

  39. Think the 16 in last week’s Opinium may have been on the high side while within MOE so a 4% narrowing probably overstating.

    I would be surprised if most polls do not show some narrowing as there was always some launch boost for the Tories, plus more DKs on the non-Tory side now firming. Expect more and that the real gap as we start properly is closer to 10% meaning Labour have work to do to get close enough to stop a Tory OM but it is possible at least.

  40. Matt126

    “I’m sure both Labour and the Lib Dems are at a local level concentrating on the areas they need to win and focus resources in such a way and this will to some degree affect the dispersal of votes.”

    I think that has always been the case for most parties (unless they have access to a bottomless pit of “dark money” – possibly from Russian oligarchs, according to a number of reports).

    For Scotland and Wales there is an additional dimension, in that parties have to plan their budgets to include the devolved Parliament elections in 2021.

    Sensible parties in Scotland also have to anticipate the possibility of an Extraordinary General Election in 2020, if the political circumstances dictate that course of action.

  41. @ ON

    Good questions. I believe they are actual registration apllications, though I guess not all applications are successful. So they are more than just site visits (which are probably also counted). I’m sure, as you suggest, many are just a change of address or re-registration. For those percentages, I guess you’d have to ask someone in the know.

    However, while these numbers give no idea about the real numbers of new voters (however you define “new”), at least they are fairly directly comparable to the 2017 GE situation*, since we have exactly the same information then and now. I was interested in knowing if those initial newspaper headlines suggesting record voter registration meant anything. I suspect they were meaningless, and it’s just business as usual.

    * The only real difference is that there was a boost due to the EP elections in mid-2019, but this was a relatively small blip compared to 3 million.

  42. Trig Guy


    “I guess you’d have to ask someone in the know.”

    Sadly, that probably excludes all the journalists who write stories about it! :-)

  43. No further polls so far tonight. I could be wrong but normally (predominately Conservative supporting) press is first out of the traps if it’s a Labour stinker

  44. It is surprising and depressing that a successful barrister in England does not know where the border between England and Scotland runs.

    His guess tells us that he belongs to those existing on hearsay.

  45. On helping decide where to cast my vote in WAK, I have been waiting for some leaflets from the SNP.

    But so far none, compared to 5 received from LibDems and Tories.

    With a new candidate, the SNP workers should have been out despite the cold and wet, to tell if he compares to the young and energetic Andrew Bowie.

    Not saying that it will have much affect but this could be the first issue arising from having a election in December that could affect some of the Northern constituencies (although bad weather could happen any time of the year)

  47. Opinium has shown similar swings before. For example, their polls since mid-September have shown Con leads of 12%, 15%, 12%, 15%, 13%. 16%, 16% again and now 12%. And those are in chronological order, so for a while it even seemed they were alternating between 12s and 15s.

    Whether other pollsters also show the gap narrowing will give a clue as to whether the narrowing in Opinium is just noise / normal variation.

  48. Regarding the opinium poll two points of interest:
    1) A bit of an uptick in Labour support amongst Leave voters. Apparently more leave supporters now back Labour rather than the brexit party. The remain support seems flat
    2) The Lib Dem support amongst remain voters as dipped slightly whilst Tory Support as increased by a similar amount. This suggests slight move Lib Dem to Tory amongst remain voters

  49. Davwel

    Since you are politically sophisticated, I’m surprised that you find yourself dependent on leaflets [1] to help you determine your vote.

    Your previous comments have implied that your ABT vote will be determined by the prospects of success by another party – which seems sensible.

    [1] As a matter of interest, how many of those SCon and SLD leaflets were hand delivered, and how many arrived via Royal Mail or other commercial bumf distributers?

  50. Britain Elects

    2 minutes ago

    CON: 41% (+1)
    LAB: 29% (+1)
    LDEM: 16% (+2)
    BREX: 6% (-5)

    via @DeltapollUK
    Chgs. w/ 02 Nov

1 7 8 9 10 11