J. L. Partners

[borisminor]

A new pollster using a predictive model when they haven't sampled in any election, with only two voting intention polls in the last 18 months before today, seems to be a rather "brave" (to quote Sir Humphrey) decision.

That said it might be that they are picking up local factors better than other pollsters which are not picking up. This has the same vote share as the YouGov MRP which had a crushing defeat for the Conservatives.

The fieldwork date is important - it came before the national service and pensions announcements which other polls released today have picked up.

In addition, their unpredictive sample gives a 16% Labour lead with 39% vs 23% for Labour and the Conservatives.

[johng]

May 28, 2024 17:27:13 GMT adlai52 said:

JLP say that they use a specific mechanism for reassigning undecideds - is there a simple explanation for what this is and how its different from the other pollsters?

Yes, they are using different mechanisms to deal with don't knows.

Most pollsters either ignore them or ask a push question.

However, JL Partners and Opinium use modelling to include don't knowns in their polls. That's why they have considerably lower Labour leads that other pollsters.

Will Jennings has an interesting Twitter thread on it. I think he's basically calling them out for cooking the books

I think we will have to wait until the votes are counted to see who is right though.

[borisminor]

May 28, 2024 19:40:56 GMT johng said:

May 28, 2024 17:27:13 GMT adlai52 said:

JLP say that they use a specific mechanism for reassigning undecideds - is there a simple explanation for what this is and how its different from the other pollsters?

Yes, they are using different mechanisms to deal with don't knows.

Most pollsters either ignore them or ask a push question.

However, JL Partners and Opinium use modelling to include don't knowns in their polls. That's why they have considerably lower Labour leads that other pollsters.

Will Jennings has an interesting Twitter thread on it. I think he's basically calling them out for cooking the books

I think we will have to wait until the votes are counted to see who is right though.

Yes, I think they have been too clever by half here. Trying something completely untested in an election campaign does not seem a particularly wise thing to do for confidence in their polls. I'm sure JL Partners think they are right. I am just not convinced they are.

[iain]

May 28, 2024 19:56:33 GMT borisminor said:

May 28, 2024 19:40:56 GMT johng said:

Yes, they are using different mechanisms to deal with don't knows.

Most pollsters either ignore them or ask a push question.

However, JL Partners and Opinium use modelling to include don't knowns in their polls. That's why they have considerably lower Labour leads that other pollsters.

Will Jennings has an interesting Twitter thread on it. I think he's basically calling them out for cooking the books

I think we will have to wait until the votes are counted to see who is right though.

Yes, I think they have been too clever by half here. Trying something completely untested in an election campaign does not seem a particularly wise thing to do for confidence in their polls. I'm sure JL Partners think they are right. I am just not convinced they are.

Isn’t this James Johnson’s company? In which case the idea that this is a purely partisan exercise can’t be completely ruled out.

[batman]

Instinctively I don't think Labour's lead is as modest as that. Canvassers may not be seeing wild enthusiasm for Labour, but all over the place they are seeing very few people, in the seats that matter, appearing likely to vote Conservative. That isn't to deny that come polling day it might end up being as narrow as that, but at this particular point I don't think the other polls are quite as wrong as J.L.Partners imply that they are. There is a tendency to come up with clever formulae to distribute don't knows. But the assumption, which certainly some pollsters seem to have but I'm not quite sure that this pollster does, that don't knows will split the way they did in previous election is a pretty grand one. Times are different, the don't knows are not necessarily the same people as before, the Tories are looking a great deal less attractive to the electorate than they have for a generation. Even in a general election, of course, quite a lot of don't knows will end up not voting at all, question is how many.

[manchesterman]

I too instinctively feel this margin to be too narrow; however we should remember that a poll is a snapshot of opinion today, not a prediction of the future. So, whilst the other polls may have been over-egging the Labour lead, I think this one is clearly under-egging it!

However, it may be a useful tool for Labour (ironically) to help avoid complacency both amongst its party workers and its potential voters. If all the polls were showing leads of 23/24/25 points, then come election day, many Labour-inclined voters may decide they have something better to do that day. Labour voters staying away in their droves would be about the only thing that could prevent a certain Labour majority.

[Deleted]

The don't knows we've been getting in canvassing almost all fall into one of three camps.

The first is Tory voters in Tory areas complaining that "I don't know any more, they're all the same". Some of these will go to Reform, some will go back to the Tories, many will not vote. A few are on a journey towards voting Labour this time round as a reaction to the seer awfulness of the government but are finding it a difficult journey.

The second is "generic anti Tories" who will vote for one of Lab, LD or Green. None of these will go Tory, most will go Labour eventually though it will vary drastically from seat to seat.

The third are people who haven't thought about it until we've talked to them on the doorstep (part of the c. 30-40% who vote in general elections but no others). These are generally swing voters. Most will vote though I suspect only a few will vote Tory once we get to the end of th campaign and they're having to make their minds up. A lot of this group though have already decided to vote Labour - the don't knows seem to be a pretty small proportion of the "don't think about it until polling day" vote

[Adam in Stroud]

May 29, 2024 10:52:44 GMT @europeanlefty said:

The don't knows we've been getting in canvassing almost all fall into one of three camps.

The first is Tory voters in Tory areas complaining that "I don't know any more, they're all the same". Some of these will go to Reform, some will go back to the Tories, many will not vote. A few are on a journey towards voting Labour this time round as a reaction to the seer awfulness of the government but are finding it a difficult journey.

The second is "generic anti Tories" who will vote for one of Lab, LD or Green. None of these will go Tory, most will go Labour eventually though it will vary drastically from seat to seat.

The third are people who haven't thought about it until we've talked to them on the doorstep (part of the c. 30-40% who vote in general elections but no others). These are generally swing voters. Most will vote though I suspect only a few will vote Tory once we get to the end of th campaign and they're having to make their minds up. A lot of this group though have already decided to vote Labour - the don't knows seem to be a pretty small proportion of the "don't think about it until polling day" vote

Tangential to this, the presumption that DKs will default to their previous vote is plausible where nothing arises in the campaign to shift them. As far as the air war goes, that's probably the default probability, subject to e.g. Starmer saying something mad in the debates or someone coming up with a great cut-through policy (or a massively unpopular one - hi Teresa.)

But the ground war could shift them and on that front I think the Conservatives are not in great shape. Furthermore, the places where that is most likely to happen is in the target seats.

I can imagine seats where we (LDs) are second, Labour weak, and the Tory vote is cratering towards DK, but there aren't resources for either LD or Lab to fight it. Those could easily be retained by Conservatives, whatever the people at Electoral Calculus tell you. If they aren't, it'd probably be down to the air war pushing Lab above LD to sneak it, but that'll take a really piss-poor Conservative campaign. But those are seats the Conservatives should be holding anyway.

[The Bishop]

This poll was relentlessly hyped up by the likes of Cole and Staines, so they had obviously been tipped off about it well in advance. Now who would do that?

[johng]

May 29, 2024 11:24:15 GMT Adam in Stroud said:

But the ground war could shift them and on that front I think the Conservatives are not in great shape.

From what I have seen, they really aren't.

[hullenedge]

[hullenedge]

[YL]

Curiously this one wasn't hyped by "Guido Fawkes".

[cogload]

Reversion.

[iain]

[hullenedge]

[cogload]

Unadjusted:

[redtony]

WHY HAS A 22 LABOUR LEAD BEEN REDUCED TO JUST 17

[batman]

Jun 17, 2024 21:45:52 GMT redtony said:

WHY HAS A 22 LABOUR LEAD BEEN REDUCED TO JUST 17

it is not actually untoward. Almost all posters make adjustments to the raw figures for a number of valid reasons, e.g. the so-called shy Tory factor, but also taking into account the necessity that the poll is properly representative of the population as a whole. In the past pollsters did not make the necessary adjustments which was part of the reason why they got it so wrong in 1992, but also in 1997 when they greatly overestimated Labour’s lead, big though it actually was. Raw figures are sometimes interesting but adjusted ones are almost certainly more accurate.

[redtony]

yes but it is a huge correction
in 2017 most pollsters did not pick up the swing to labour po;;s are not predictions of how voters will vote at the election