Talk:Opinion polling for the 2024 United Kingdom general election

Polls started on last election day

"Opinion polling for the next United Kingdom general election is being carried out continually by various organisations to gauge voting intention. [...] The dates for these opinion polls range from the 2019 general election on 12 December to the present day."

To me, this reads as if one or more polls for the next election were carried out actually on the last election day, 12 December 2019, which seems a bit unlikely. If in fact this is not meant then perhaps the wording could be improved. 2A00:23C8:7B0C:9A01:543C:A619:5528:AB57 (talk) 18:33, 25 March 2024 (UTC)

New seat prediction

Electoral Calculus has a new seat predicition (at the 29th of March): https://www.electoralcalculus.co.uk/prediction_main.html

I don't know how to insert it correctly, so maybe someone else can help. Tulpenliebhaber (talk) 19:48, 6 April 2024 (UTC)

Labour Together? Really???

Get this poll off this site forthwith. We should not be including internal polls or party-sponsored polls in any form. 2603:7000:603A:342B:4964:A243:AFC1:71EF (talk) 03:11, 2 May 2024 (UTC)

The name makes clear its origin. Readers can decide how much weight to give their results, which are in line with others. The layout of their tables is almost exactly like YouGov. I wouldn't be surprised if it was contracted out to them. Speaking of YouGov, their founders were Conservatives, Nadhim Zahawi and Stephan Shakespeare, an owner of ConservativeHome. I doubt we can ever exclude all political bias in polls, especially given the ownership of the UK press. --Cavrdg (talk) 08:20, 2 May 2024 (UTC)

It is not an internal poll; it was done using standard polling methodology. It is not a party-sponsored poll; Labour Together are a think tank and not part of the Labour Party (although they are Labour supporting). We include all polls done by pollsters who are members of the British Polling Council. Bondegezou (talk) 08:22, 2 May 2024 (UTC)

Labour Together are not a member of the BPC. 38.96.180.251 (talk) 00:01, 4 May 2024 (UTC)

May 2024 elections

My personal preference is to include the following as the line for last Thursday's elections:

United Kingdom local elections, Police and crime commissioner elections in England and Wales and the Blackpool South by-election

This shows: local elections in England, the PCC elections and the Blackpool by-election.

I must object to the phrase 'Local elections in England and Wales' there were no local elections in Wales and I can't see how huge police area elections can be classed as anything remotely local. We have a specific results page complete with the usual info box and breakdown for the PCC elections. Kalamikid (talk) 10:45, 8 May 2024 (UTC)

Our article at 2024 United Kingdom local elections includes the PCC elections. The BBC page on the local elections covers the PCC elections. They are elections for local areas, albeit relatively large ones. I think it is unnecessarily nitpicky to separate them out in a note on a polling table. Bondegezou (talk) 10:49, 8 May 2024 (UTC)

You and I are clearly going to disagree given the edits we've both made. How about a compromise of "2024 United Kingdom local elections" given that's what the wiki page is called? Kalamikid (talk) 12:14, 8 May 2024 (UTC)

I prefer the form as it is currently (as edited by Bondegezou). It is descriptive and succinct. --LukeSurl ^t c 12:20, 8 May 2024 (UTC)

Northern Ireland Life and Times Survey

@Ralbegen and Unionofpeopleni: Just to say, you were right the NILT survey was general election polling - it just wasn't clear from the link posted (it looked like a party ID question, as opposed to a voting intention question). I've added a note including the question asked and a link to the questionnaire for clarity.

More generally: what's the policy on removing undecided voters (the 'none of these' or 'I don't know' responses)? Most polls only include people giving a voting intention in the headline figures, so removing them would make improve the ability to compare with other polls. The figures appear to be APNI 28; SF 24; DUP 19; UUP 13; SDLP 9; Grn 5; Oth 4 without them. Clyde1998 (talk) 12:39, 16 May 2024 (UTC)

We normally present numbers excluding undecided voters. However, normally those numbers are provided by the pollster. We can, as per WP:CALC, calculate our own numbers excluding undecided voters. I tried that here, but I got confused as to which of their categories should count as undecided and which as other party. Bondegezou (talk) 15:34, 16 May 2024 (UTC)

I've modified it using {{#expr:x/80 * 100 round 0}}%, on the basis 'None of these'; 'Other answer'; 'I don't know' should be removed; the other options are all clearly relate to voting intentions (only 'Other answer' is ambiguous to me - as that could mean a VI for an independent). Makes it a bit easier to calculate, can be adjusted easily should we need to and makes it a bit clearer where the figures are coming from. Clyde1998 (talk) 16:03, 16 May 2024 (UTC)

I'm not certain 'Other answer' should be excluded. I'm not even certain what's best with 'None of these'. Bondegezou (talk) 20:28, 21 May 2024 (UTC)

Graph correction

The 16 May People Polling poll has released corrected figures after an error was found in their original figures. This has been corrected in the table, but it also needs correcting in the graph please. Bondegezou (talk) 11:29, 21 May 2024 (UTC)

This is maintained by Ralbegen LukeSurl ^t c 18:36, 21 May 2024 (UTC)

Thanks for flagging, it should be fixed now :-) Ralbegen (talk) 14:08, 22 May 2024 (UTC)

Polling aggregation graph.

Hi! @Ralbegen

I made a graph visualizing the distribution of opinion polling between the beginning of the year and the calling of the election. I sourced it from the main table of nation-level voting intention.

The intention is to capture the political environment, which has been by and large stable, immediately before the campaign period. Pre-election campaigns tend to shift political attitudes, which is the rationale behind this piece of visualization. The different intensity of colours shows the range distribution/variation of each party’s support level since January.

The idea of this graph is similar to a histogram: the more polls having a party’s support level at a given percentage, the more solid the shade of colour.

Do you think this is a meaningful/informative piece of visualization? If so which part of the article it belongs to? Thanks for sharing your thoughts. 沁水湾 (talk) 22:21, 24 May 2024 (UTC)

Hi @沁水湾:, thanks for pinging me! It's a neat visualisation :) I think something like this could be interesting to include, though I am concerned that most of the text isn't readable without zooming in. I wonder whether it might be clearer if you removed the high/low value text and the 50% bar, and used vertical gridlines with a conventional x-axis? If it were going to fit somewhere on the page I think it would be most comfortable at/near the top of the 2024 national polling section. Ralbegen (talk) 22:34, 24 May 2024 (UTC)

Thanks! The high/low numbers are indeed secondary information. It’s nice to include(as it shows off the extent of the range) but not strictly necessary. The font size for the average/2019 result numbers are also quite small. However, they are somewhat more essential. Do you think we should size them up or just ditch them?

I’m sorry but I didn't quite get what exactly you were referring to with “ vertical gridlines with a conventional x-axis.” Can you link me to an example? Thank again!

I also noticed several mistakes in the graph. I’m not with my computer. I’ll be correcting them later. 沁水湾 (talk) 22:59, 24 May 2024 (UTC)

It's up to you but I would suggest removing the high/low numbers while either sizing up or removing the last election/polling average labels.

By vertical gridlines I mean vertical lines in the background at (eg) 10%, 20%, 30%, 40%, 50%, and by a conventional x-axis I mean with axis labels at the top or bottom saying "10%", "20%", etc. That would let users have a sense of what the values are without needing to use so much text and could also make it a little easier to see differences between parties. Ralbegen (talk) 23:42, 24 May 2024 (UTC)

Hi! @Ralbegen

I've just uploaded a newer version of the graph. I've omitted the high/low numbers, increased the font size of the remaining text slightly, and adopted a layout akin to my byelection diagrams: grey dash lines of 5% intervals. What are your thoughts? 沁水湾 (talk) 16:51, 25 May 2024 (UTC)

I cannot understand that graph at all & I guess that most people will not either. Its also positioned belo the tracker graph nd about the data tables. This is very inconvenient as its now not easy to switch from one to the other. If I had my way I would delete it altogether. 2.103.106.203 (talk) 21:14, 27 May 2024 (UTC)

I tried to place it in the article in the least distruptive way (for both desktop and mobile). However I'm not convinced it is necessary as it doesn't actually convey any information that isn't already in the tracker plot. I lean on the side of removing it. LukeSurl ^t c 07:56, 28 May 2024 (UTC)

I think it's time to remove this. Although well-presented there is no information here that isn't on the main plot (with more temporal detail). Jan-May 2024 is a somewhat arbitrary time period to highlight, and its relevance declines as it drifts further into the past. LukeSurl ^t c 10:43, 12 June 2024 (UTC)

Very large collection of data points

This is the largest article on Wikipedia right now.[1] It may be too detailed. I counted 1877 links with "https". I don't know who would read this top-down. It has value as a database, but Wikipedia is not a database.

Problem is, the raw data points seem to be almost all of the article. Has anyone besides WP done any coverage on the polling trends themselves? What form should this article take? Not to mention others like it.[2][3][4] Wizmut (talk) 22:27, 24 May 2024 (UTC)

WP:NOTSTATS uses a polling article being split from a main election article as an example of good practice. Almost every poll receives secondary reliable source coverage and poll trends receive dedicated coverage from most large media organisations, including with full tables of results. For previous UK elections when the article has got very large, there have been splits. US American elections often have multiple dedicated polling articles split by subject (national vs statewide, etc). Ralbegen (talk) 23:30, 24 May 2024 (UTC)

A split is probably in order for load-time considerations. Easiest to implement would be to create Opinion polling for the 2024 United Kingdom general election in Scotland and Opinion polling for the 2024 United Kingdom general election in Wales, and, potentially Constituency opinion polling for the 2024 United Kingdom general election. LukeSurl ^t c 18:51, 25 May 2024 (UTC)

In 2015 the article had two splits-out: sub-national opinion polling and constituency opinion polling. This time we have many fewer constituency polls and more other geographical splits. If there's a view to split the article I think a single sub-national opinion polling page including all non-UK/GB polls in the same format they are on this page would make most sense. Splitting out individual short sections would not make much difference to the page and would create a nebula of very short polling articles. Ralbegen (talk) 23:17, 28 May 2024 (UTC)

Hmmm, I think you're right. Probably best to just wait till July 5 and there are no more additions and reconsider then. LukeSurl ^t c 09:14, 29 May 2024 (UTC)

Possible addition to the article

Hi @Ralbegen

I am considering whether adding the change between polls of the same poller (i.e. whether the lead has widened or narrowed) after the lead is a good idea or not, and I just wanted to get your opinion (and the opinion of anyone else) on whether this is a good idea or not.

I think this is a good idea because then people can then easily compare polls, but I do realise that this probably will increase the page size a lot and thus negatively impact load times.

Feel free to ask for clarification if any of this doesn't make sense, I have the feeling that I haven't really explained it that well but I hope you get the general gist of what I mean. SuperGuy212 (talk) 11:49, 29 May 2024 (UTC)

In practice this can already be done by sorting tables by pollster and inspecting the trends. I don't think adding commentry to specifically state this would be workable. It would complicate the tables, make these extremely long article even longer, make tables that are already difficult to view on mobile wider, and would be a huge amount of work to add and verify for all existing polls. LukeSurl ^t c 13:36, 29 May 2024 (UTC)

A new opinium poll

Sorry I have disability but want to help. There is new poll announced by sky news on the tv. Can be added to the article?

https://x.com/OpiniumResearch/status/1796980059829387317?ref_src=twsrc%5Egoogle%7Ctwcamp%5Eserp%7Ctwgr%5Etweet

thank you for reading the message and i hope to help again!! ☺️ 2A02:C7C:DE32:DF00:D80B:797E:CA79:CEE (talk) 19:11, 1 June 2024 (UTC)

Another new one!!

https://news.sky.com/story/election-latest-news-uk-sunak-starmer-tories-labour-live-12593360?postid=7772460#liveblog-body 2A02:C7C:DE32:DF00:945D:9FD1:71E:747F (talk) 20:00, 4 June 2024 (UTC)

New Poll for Article

2A02:C7C:DE32:DF00:E027:A9E1:5751:6A5D (talk) 18:57, 6 June 2024 (UTC)

You don't say what you want to add. Nor does this require intervention by an administrator.--Bbb23 (talk) 00:58, 7 June 2024 (UTC)

Hi @Bbb23 and thank you very much for taking the time to respond to me.

I am sorry if I caused trouble, I didn’t mean it!

I would like to add the opinion pools! 2A0C:B381:5FA:2400:1429:5F72:5E70:D4C6 (talk) 16:57, 8 June 2024 (UTC)

20–27 May 2024 seat prediction

Checked the link on this poll as it is a significant outlier, the previous editor has added the 'tactical voting' seat tally not the regular more reasonable tally? The methodologies are not well defined and the more extreme values have been used. Should this be kept, switched to the standard values or removed entirely? 2A02:C7C:9457:5200:7A55:943A:45B:7ABE (talk) 10:32, 2 June 2024 (UTC)

The pollster is being a bit cheeky really, publishing both "with tactical voting" and "without tactical voting" numbers. They're the pollster, it's their job to predict how much tactical voting will happen! Anyway, it seems like both sets of figures have equal-ish priority so I'm making a version of the table that displays both but is clear that they are from the same underlying polling. LukeSurl ^t c 09:56, 3 June 2024 (UTC)

New YouGov method

The next YouGov poll will change methodology and result in a smaller Labour lead than the previous method.

This is going to lead to confusion when trying to compare change in movement.

They said the first poll they release under the new method will show both VI under old and new method. Should we put both in? 2603:7000:603A:342B:D551:F1FC:7F7D:8012 (talk) 13:12, 5 June 2024 (UTC)

Campaign period graph

CoaxAndBotany has added a new main graph (first image, white background) to the page which covers the period from the election being called on 22 May to election day on 4 July. I think a graph covering that period is a sensible thing to include. I've also given a lot of thought to how best to visualise data for this page. I'd like to offer something more like this one (second image, grey background) if editors are interested, using the same code I've used to maintain the main graph on this page for the last few years.

I think advantages include:

• No vertical text or repetition of the same year
• Direct labels, which are considered best accessibility practice
• Significantly less text overall, making it more to-the-point and possible to caption in multiple languages
• A single type of graph rather than compositing multiple types graphs that can't be separated
• LOESS span calculated with cross-validation (fairly recent addition to the code, which I can override if users would prefer more wiggling than the method thinks is justified)
• Consistency with the longstanding graph

I appreciate it's not perfect and am happy to take feeback. I can make changes if people agree with the principle of sticking with purposefully minimal, clean, data visualisation for this page in these weeks when the most eyes are on it. For example, I like having the election line without an in-graph label, which I think can live in a caption and therefore be more language-independent.

I also appreciate the time and effort that other editors put into producing graphs! Of course it's completely valid if editors prefer an alternative data visualisation. But I'm pleased with how my code has ended up and would like to offer this option for your consideration. Hope you like it too! Ralbegen (talk) 00:16, 11 June 2024 (UTC)

If we could, it would be nice to make the graphical summary as similar as possible to the 2019/2017 pages? Personally I prefer leading with the 'big chart' (since last election) as those other pages do, then the 'campaign period' chart, then the aggregation. I think if they're more neatly formatted (similar sizes), and have titles, we don't really need the 'switcher' and can display all 3.

Appreciate this is personal preference but aids consistency between polling pages. 91.125.229.176 (talk) 10:08, 11 June 2024 (UTC)

CoaxAndBotany's graph is very nice, but I think I agree with Ralbegen's principles, so would prefer that chart. Bondegezou (talk) 10:20, 11 June 2024 (UTC)

I don't mind the campaign graph but I don't think we should remove the 'overall' graph which begins in 2019. I think it's reasonable to have both. — Czello ^(music) 10:26, 11 June 2024 (UTC)

This is a good addition. I do think that 3 days is too short for a rolling average - it will be subject to a lot of variation due to which pollsters (with differing house effects) published in this short period. 7 days seems more apposite. LukeSurl ^t c 14:35, 11 June 2024 (UTC)

A small point on the two graphs showing more recent polling: they currently display UK-wide vote share for 2019 GE in comparison to current polling. Typically, the opinion polls only report GB vote share. For example, Conservative vote share is reported as 43.6%, the UK figure, whereas the more applicable GB vote share was 44.7%. Small difference, but I think it should be consistent between the two. CometCruiser (talk) 15:59, 11 June 2024 (UTC)

Add "Events" to Seat Predictions data

Key events appear in the main national polling and regional polling data tables - there seem to be clear rules to limit what gets included e.g. Leader changes, election announced etc which all seems sensible. However, the same hasn't been included in the seat predictions which seems a significant oversight. Especially since Swinney and Farage came in very recently. Please can someone authorised add those key events in? It will really help to make the article consistent and more reader friendly. I don't have the authority to make the edit. Many Thanks WestminsterWhistleblower (talk) 15:37, 11 June 2024 (UTC)

Pollsters

What are the standards for deciding which pollsters to include in this list? Tompw (talk) 21:06, 11 June 2024 (UTC)

Generally they're BPC members or conduct themselves in a similar way (scientific polls with published tables). Ralbegen (talk) 21:48, 11 June 2024 (UTC)

Thanks! Tompw (talk) 13:00, 12 June 2024 (UTC)