Talk:Opinion polling for the 2024 United Kingdom general election/Archive 5

Archive 1

←

Archive 3

Archive 4

Archive 5

Archive 6

Archive 7

Archive 8

Note on methodology changes?

The Opinium poll out today is the first to be conducted and published under their new methodology (which better accounts for and factors in respondents choosing 'don't know', to clarify), as detailed here by their head of political polling Chris Curtis. Is it worth putting a note in the article somewhere explaining this, maybe in anticipation that casual readers of the article may be bemused as to why Opinium's poll leads for Labour may come in lower than other pollsters (as aforementioned; even though Labour's lead would have actually gone up from 5% to 10% if the previous methodology had been retained)? --Phinbart (talk) 20:13, 12 February 2022 (UTC)

Definitely worth making clear, maybe even in the prose before the table? Ralbegen (talk) 20:39, 12 February 2022 (UTC)

Yeah, I was thinking there. Something along the lines of "The pollster Opinium conducted its research under a different methodology from its poll dated 9-11 February 2022", with a reference to the voting intention page they make clear the changes on, and maybe also saying "therefore making comparisons with their earlier polls redundant"? Would it also be worth giving a brief outline of the changes made, or would just that do? --Phinbart (talk) 20:54, 12 February 2022 (UTC)

How about The pollster Opinium changed its sampling and turnout weighting methodology in February 2022, so their polls before and after this change are not comparable. For their first poll, the new methodology showed a Labour lead of 3 percentage points where their previous methodology would have shown a Labour lead of 10 percentage points.? It's not like there's a lot of prose on the page and I think it'd be best to avoid alluding to something that can be spelled out without too many words. Ralbegen (talk) 22:06, 12 February 2022 (UTC)

That seems alright. Maybe I'd say the second sentence should be worded with the reference to the change in Labour lead being For their first poll, it [the new methodology] lead to a smaller Labour lead than had previous methodology been used instead, though I think being explicit with what the difference would've been exactly may be more helpful.... And on a related note, would it be any use to say what the lead would've been with said previous methodology for forthcoming Opinium polls, as they have said that to basically settle in to this new methodology they'd be reporting, for the next few polls at least, the results under both the new methodology and (estimates for) the old? Maybe as a note in the 'lead' column? --Phinbart (talk) 22:43, 12 February 2022 (UTC)

I agree it needs noting. Really the chart should reflect this too - not sure how. A vertical line at the change point maybe? Btljs (talk) 10:40, 13 February 2022 (UTC)

I don't think the chart needs cluttering too. The chart covers polls by lots of companies, using different methods, and indeed all of whom revise their methods from time to time. Bondegezou (talk) 15:02, 13 February 2022 (UTC)

Sorry, I didn't read this before adding my large paragraph about this today. Worth someone checking my para as the change is complex to summarise. It seems quite a large addition on a detail where it is positioned, but I couldn't see where else it could go. Rwendland (talk) 16:02, 13 February 2022 (UTC)

Yeah, it looks fine, really. Explaining it as fully as possible rather than just glossing over or not even mentioning the exact changes is probably the best way to do it. --Phinbart (talk) 17:08, 14 February 2022 (UTC)

FindOutNow poll

Hi, I'm a little confused over which numbers should be put into the table for the FindOutNow poll from 14-18 February. The voting intention numbers as tweeted out by Britain Elects and included in the table seem to be the predicted vote shares as displayed here, with a footnote saying "the predicted vote share percentages differ slightly from the basic voting intention from the poll, as the regression techniques and models operate differently from classic polling analysis. This modifies the parties' vote shares slightly."

The tables have said 'basic voting intention', and different numbers. The "headline Voting Intention, adjusting for likelihood to vote, excluding don't knows and refusals" has the Tories lower at 31%, Labour higher at 39%, and Lib Dems a chunk lower at 10%, among other differences. Which should be inputted into the table? --Phinbart (talk) 14:26, 25 February 2022 (UTC)

I think we should go with their headline figures (predicted vote shares). Different pollsters adopt different methods. It is now our job to critique these (for that would be WP:OR). We report their best estimate, and we provide a citation for those who want to go delve into the details. Bondegezou (talk) 14:52, 25 February 2022 (UTC)

New boundary projections

Hi all, the 11-23 JL Partners poll conducted 2 seat projections, one using 2019 boundaries, and another using the proposed boundaries. They haven't proposed the full results for the latter yet (tables yet to come out), though would it be worth adding a note or a new section for these projections? Quinnnnnby (talk) 16:15, 27 February 2022 (UTC)

If we wanted to include it, then having a new column for which boundaries we're concerned with (2010–2019, initial proposals, revised proposals, etc...). As there are 2019 boundaries for this poll, and I'm pretty sure every other poll, it would make sense to me if we kept things as they are for the time being and monitored whether more pollsters start to use the initial proposals. When the commissions produce final boundary proposals and they become law, we should hopefully see a widespread move to using them! In that case, it would probably be best to have two different tables for seat predictions: one for the old boundaries, and one for the new boundaries. But for the time being I'd suggest we carry on as we are! Ralbegen (talk) 20:21, 27 February 2022 (UTC)

Extending polling graph to next election

Ralbegen I've looked at polling graphs for other nations and most if not all adjust the graph to leave blank space for polls between now and the next election. Since its very unlikely a snap election occurs, would it be possible to do the same for the UK graph. It also shows more visually how far in the future the election and puts the polling into context a bit more. Obviously, any other views on it would be nice to have as well. Thank you. EnglishPoliticalPerson (talk) 17:55, 25 March 2022 (UTC)

@EnglishPoliticalPerson: I don't think the date of the next election is certain at all—the last two elections were early and the FTPA is no longer law. A general election could be called now and held in May! I don't think it would be useful to include blank space in the graph, and suspect it would fall foul of WP:CRYSTAL, sorry. Ralbegen (talk) 22:20, 25 March 2022 (UTC)

Yeah, thats fine. Was just looking at every other nations wiki which very rarely extend the graph as new polls are added. Even Spain's polling graph is done this way which had 3 election held early so id find it unlikely it conflicts with WP:CRYSTAL as it would just be showing when the election is scheduled to be held before which we do know. EnglishPoliticalPerson (talk) 22:38, 25 March 2022 (UTC)

Partygate article

Westminster_lockdown_parties_controversy#Public_opinion has a small graph of polling to show the impact of Partygate. This seems reasonable with extensive RS noting the link. I'm just wondering if the good people here could help improve the graph that's been done. It would be good if it had a bit more of a baseline before the first Partygate news, included all the parties included here, and did better smoothing of its lines. Bondegezou (talk) 20:43, 16 January 2022 (UTC)

Good idea: I'll take a swing at that tonight. It'll need a bit of adaptation from the usual code in order to include events but that's probably something worth having on hand anyway! Ralbegen (talk) 11:47, 17 January 2022 (UTC)

Hi Ralbegen. I wondered whether you had had a change to look at this? The current graph at Partygate#Opinion_polls seems sub-optimal to me. Bondegezou (talk) 20:27, 5 February 2022 (UTC)

Hi Bondegezou, I took a couple of swings at it at the time but couldn't get it quite how I wanted it—the changes I tried caused other things to break. However, after your ping today I realised that I can just tinker with it in a vector editor instead of trying to get everything just-so in R. You can see my first swing at it to the right. I'm happy to take feedback on it: I think it responds to your three points (by focusing on a smaller time period it's also a better smooth than the main graph on this article!). I've also chosen initially to only include one labelled vertical line, because I don't see how the other lines in the existing graph on the page justify their inclusion! Ralbegen (talk) 21:44, 5 February 2022 (UTC)

I like it and, yes, it tackles some of the problems with the current graph, particularly it’s line-fitting approach. Bondegezou (talk) 07:56, 6 February 2022 (UTC)

Update required. The polling image in the Partygate article only goes up to the end of January. Is someone able to update it? Ghmyrtle (talk) 14:23, 12 May 2022 (UTC)

Survation decimal places

When I added the Survation poll today, I included them, but was reverted by two different users. I think the additional resolution is informative and useful. The margin of error on opinion polls is greater than one percentage point (though it depends, if I recall correctly, on the value in question) but an extra significant figure gives more information about where that margin of error is centred. If you look at the graphs I maintain on this page you can see the impact of rounding forcing the points into discrete percentages, except for a couple of Survation polls where I've recorded the additional decimal place—my untested suspicion is that if all polls were reported to 1dp, we would see a tighter range of polls that trace a clearer trend, with extreme values more often than not being rounded towards the trend (like today's 45% being rounded from 44.6%). On some other matters here we have agreed that we should take our lead in presentation from pollster write-ups, which do not include this additional resolution. However, those discussions were about the inclusion of partes as individual columns or as part of an expandable others column. Information was always included, just allocated prominence according to due weight. Rounding from 1dp to 0dp removes information. What are users' views about (a) whether we should report an additional decimal place when it's included in the pollster's data tables and (b) whether I should use the additional decimal place in the CSV file I use to produce the graphs on this page? (I suspect I might be in a minority on this one!) Ralbegen (talk) 15:41, 8 July 2022 (UTC)

While the additional decimal place may prove useful in some cases I'm against any change from 0dp as it's a false precision leading readers to assume polls are more precise than they actually are. I do think we should again follow the pollster write-ups here too however, so should they decide to include additional precision in any poll's write-up (maybe larger MRP polls? though if I recall they still used 0dp there) then we could include that precision on that occasion. I know on some older polls for past UK General Elections there are a few .5 numbers in there.

With regards to using the additional dp in the chart I have no preference there as I don't believe the misleading of precision would apply. BeeEdits (talk) 16:12, 8 July 2022 (UTC)

Plaid Cymru?

Is it really neccesary to add Plaid into the polling table? They will never poll above 1% and as polling in the UK *near* always rounds to the nearest whole number the value of adding it is tentative. EnglishPoliticalPerson (talk) 15:38, 26 August 2022 (UTC)

I have reverted the edit as it meant all Plaid data was removed from the table rather than put back into the Others column. To answer your query, for one, the inclusion of Plaid into the table does mean we're not left with loads of the collapsible information in the Others column, and it all looks a lot neater and more easily readable and navigable. I do agree that given it's either polling at 0% or 1% with only one (that I could see through my skim read) variation of 2% in the past year, it seems a bit pointless, though. Then again, if we are including its Scottish equivalent (SNP), you could say it's only fair Plaid gets a column of its own. It's a tricky one; interested to hear what others think. --Phinbart (talk) 23:02, 26 August 2022 (UTC)

I think, seeing as virtually all polls include a Plaid Cymru figure (and many of the ones that don't also don't include figures for Reform) it helps keep the table tidier and makes it easier to use to include a Plaid column.Mozartnut (talk) 00:53, 27 August 2022 (UTC)

I would keep it. In general, the more the figures are broken down and the 'Other' column reduced the better, and if ever the Plaid vote does vary significantly we will know about it. Dudley Miles (talk) 08:33, 27 August 2022 (UTC)

The arrangement of columns to not include Plaid with an individual was agreed with wide engagement previously, and Reform were added after engagement and agreement. Reform were being reported in pollsters' write-ups more prominently than they were at the time of the first discussion, and their average briefly eclipsed the SNP's. Plaid Cymru can never do that in UK-wide polls. If they were as successful as the SNP, they would be polling at 1-2% instead of 0-1%. They only compete in Wales, which represents less than 5% of the UK's population. Recording them in a separate column would be undue weight compared to their prominence in reliable sources, and can give no useful information to the reader. Included with other parties, their performance is still available at an appropriate degree of weight. It is also sensible to keep the parties included in the chart the same as those represented with their own columns. A line draped along the x-axis with no hope of rising from it would not be an improvement. Ralbegen (talk) 18:05, 27 August 2022 (UTC)

I see where you're coming from; I did not think, in any case, it would be worth adding Plaid into the polling chart to complement its featuring in the table, for precisely the reasons you spell out, in that the limited variation would mean just a constant dark green line along the near-bottom of the chart and it would just add to the clutter of the other low-polling parties' displayed trajectories too. Plaid could probably go back into the Others column if needs be, but I think I'd expand the width a bit more to compensate for the less tidy table created as a result of expanding the Others 'show/hide'... thing. Phinbart (talk) 00:44, 28 August 2022 (UTC)

Im of the opinion that including a party who's polling extremely rarely surpasses 1%, with often tenatious sub-samples for the region of the UK it represents making it's number even more irrelevant, is just unneccesary and clutersome. I have my own greivances with including the SNP however the variation with them is far greater and their polling performance is far greater. Even though almost every poll prompts them, it doesn't make it any more reasonable to give them there own column; if one wanted to see how they we're doing they'd look at Welsh specific polls which give a very good idea of their performance where national polls can tell you, frankly, nothing. At least for the SNP you can generally assume if they poll over 5% it's typically about 45 - 55% of Scotland whereas the difference between 0% and 1% in wale's is not even worth investigating further as it will always be MoE. EnglishPoliticalPerson (talk) 11:24, 28 August 2022 (UTC)

Poll data template for flexibility?

If the Plaid Cymru change might be the kind of change done in the future, perhaps it is worth (for 2023?) having a template for the data, which could be changed to a different output style easily. Maybe something like:

{{opcldsgrpo|31|^44|12|4|4|2|1|2|lab|13}}

Exact form (and practicality) would need some research. Could go full hog and include the whole row's data so one line per poll? Is that a good direction to go? I did opdrts so I'd be willing to try to do this in Nov when I have more spare time, if there is a consensus to use it. Rwendland (talk) 17:18, 31 August 2022 (UTC)

Would the idea be to have maximal data recorded in the template, which could then be displayed according to a parameter? My initial thoughts are that it would be easier to have named parameters than a baked in order; that adding new parties like the Brexit Party or the Independent Group for Change in the last Parliament would be more difficult than it is currently; and that including parties from tables that routinely get zero (like the WEP for YouGov), or are less frequently interesting (like UKIP for Deltapoll and maybe one or two others) would result in less frequent and less auditable inclusion unless you could template-automate the show/hide text. (I'm not opposed to templatisation on principal and I think that opdrts was a very useful innovation!) I wonder if there's the potential for an election box analogue for polling tables, with an {ophead}, {opresult}, {opelection}, {opevent}, and so on. I suspect that might be beyond feasibility! Ralbegen (talk) 20:14, 31 August 2022 (UTC)

Restoration of removed information

Ralbegen, within this edit you removed some factual clarifications I added without giving any reason for doing so. What I wrote is clearly all factually true and I don't see any good reason to remove it. Also, consensus does not need to be formed prior to adding something to the page. Helper201 (talk) 15:34, 8 September 2022 (UTC)

What I added can be seen here. All I did was paragraph some information that was already there and clarified the difference between Great Britain and the UK and noted that the SNP and Plaid don't stand candidates across GB or the UK and only in Scotland and Wales respectively. I see nothing contestable in any of that. Helper201 (talk) 15:39, 8 September 2022 (UTC)

Hi Helper201, sorry about that. I'm afraid I didn't notice the change when I made the edit - I was editing an old revision, adding in polls for the intervening period with the condensed version of the columns. I don't have strong views one way or the other on the inclusion of your additional material and I'm happy for it to stay; I wouldn't have deliberately removed it. Ralbegen (talk) 15:48, 8 September 2022 (UTC)

Ralbegen, no problem. Thanks for the reply and your work on this page. I understand how these sorts of things can happen. All the best. Helper201 (talk) 16:05, 8 September 2022 (UTC)

Savanta 23/24 July

Please can someone add the Savanta results for 23/24 July 2022. Ta. Bernard Naish (talk) 12:16, 1 August 2022 (UTC)

Please can the Savanta poll released today that had been displayed till a few minutes ago be put back. Bernard Naish (talk) 17:15, 12 September 2022 (UTC)

Should the death of Elizabeth II / accession of Charles III be mentioned?

I noticed in the revision history that several IP editors as well as User:Labonio added Elizabeth's death to the table. This was reverted by both User:Czello and User:Bondegezou, arguing that it is not relevant to polling, unlike political events relating to elected officials.

I have no strong opinions on this. On the one hand, I feel that it's intuitive to include it. On the other hand, I kind of see their point; I don't expect that her death will have a significant impact on the polls, even if it is a highly noteworthy event. I was wondering if any other editors have any thoughts on this.  Vanilla  Wizard 💙 22:27, 8 September 2022 (UTC)

I do feel it is worth mentioning, and potentially, concomitantly, the news parties may be cancelling their autumn conferences (as per journalist/Britain Elects runner Ben Walker on Twitter). There could likely be an impact on polls from this. I see the argument for removing the death of Prince Philip (which I added last year, for it to be removed), but I do believe there is plausibility in adding this to the table. This is complicated as we do not have precedent for this, although you could argue there is in the fact the pandemic and the ensuing lockdowns haven't been included in the table at all (which would add context into how the Conservatives rose to above 50% in the polls and up to 26-point leads, but that's another issue), and they were also of, but obviously not of equal importance, great significance to public life. --Phinbart (talk) 23:25, 8 September 2022 (UTC)

• Yes- as one who is primarily a reader of this article, it would seem odd if the death of The Queen not mentioned and rather insignificant by-elections are. At the very least, it adds to navigability by anchoring a specific date that everyone will know into a very large table. The Monarch also has a constitutional role, and the Commons will be looking radically different over the next week, as tributes are paid and new oaths are taken. Both Liz Truss and Keir Starmer may, through their oratory, rise or fall in the polls. schetm (talk) 04:10, 9 September 2022 (UTC)

• No Speculation that this will affect polling constitutes WP:OR and WP:CRYSTAL. There are many polling articles on Wikipedia. I’ve not seen any of them add a line for the death of a non-political head of state. This is a table of polling results: it is not a timeline. Bondegezou (talk) 06:18, 9 September 2022 (UTC)
  • In April 2013, Queen Beatrix of the Netherlands abdicated. Here is the Dutch polling article: Opinion polling for the 2017 Dutch general election. You will note there is no mention of the abdication. Bondegezou (talk) 06:21, 9 September 2022 (UTC)
  • In 2014, King Juan Carlos of Spain abdicated. Again, the polling article for that period, Opinion polling for the 2015 Spanish general election, has no mention of this. Bondegezou (talk) 06:25, 9 September 2022 (UTC)
  • There is a table of polling results in the 2014 Belgian federal election article. No mention there of King Albert II’s 2014 abdication. Bondegezou (talk) 06:29, 9 September 2022 (UTC)
  • The Japanese Emperor notionally has a greater role than the British monarch. Opinion polling for the 2021 Japanese general election has many commentary rows, but nothing on the abdication of Emperor Akihito. Bondegezou (talk) 06:32, 9 September 2022 (UTC)
  • The Queen was also the Queen of Canada and of Australia but Opinion polling for the 45th Canadian federal election and Opinion polling for the next Australian federal election have not noted her passing. Bondegezou (talk) 06:45, 9 September 2022 (UTC)
• No I think the event-issue was well discussed back at its time, and the compromise solution was to adopt a minimalist approach (i.e. showing only events such as relevant leadership changes or elections). The death of Queen Elizabeth II, notwithstanding its scale as a historical event and as a key turning point for the UK, is not relevant for opinion polling. Impru20^talk 06:38, 9 September 2022 (UTC)
• No. It is not relevant to this article. Dudley Miles (talk) 07:37, 9 September 2022 (UTC)
• No. It could affect the polls - how well the leaders rise to the occasion; the suspension of current debates - but so could many other things. It is an important event that will be covered extensively elsewhere (for example, 2022 in the United Kingdom). --Wavehunter (talk) 07:57, 9 September 2022 (UTC)

• Yes- This is not a matter of whether it affects the polls, that's not what the "notable events" are about. However, there is a new head of state and this should be marked. The death of Queen Elizabeth the second should not be included but the transfer of position of the head of state to King Charles the third should be. By the same token there should be no mention of the funeral or coronation. So for the 8th of September it should simply say "King Charles III becomes head of state". The rule for notable events on these pages has always been the change of official positions, such as change of party leaders, and elections ONLy (Boris' resignation intention back in July should not be included by the way, based on your own rules). This is a key part of the mechanism of the transfer of power in a constitutional monarchy. 8:06, 9 September 2022 (UTC) — Preceding unsigned comment added by 146.90.147.221 (talk)

Yes, it is a matter of whether it affects the polls. This is an article on opinion polls. I should remind everyone here that opinion polling tables are not listings of events, for that we already have 2022 in the United Kingdom. There was already a strong debate on whether events should be added to the table (most opinion polling articles don't add them, and most of those that do, do so at the unilateral leisure of single editors without any consensus), but definitely a change of a non-political official position is not a relevant change here (a state secretary resignation would be even more relevant, and those are not added either). Impru20^talk 10:30, 9 September 2022 (UTC)

• No We're in new territory here. We have no idea if a change of the head of state in itself would affect opinion polls. Until and unless any research can be found to show that it does, I think we should assume not. ChrisTheLemon (talk) 11:32, 9 September 2022 (UTC)

• No for the reasons mentioned previously. We're not doing a timeline of British events here, the things listed should be directly related to politics or political events that might affect polling. — Czello 12:09, 9 September 2022 (UTC)

• No, as others have said it shouldn't be added just because it might possibly be relevant; if it becomes clearly relevant, then it can be added. IMO who the head of state is has little relevance to politics. AlmTec (talk) 12:50, 9 September 2022 (UTC)

Yes, Despite it's lack of relevancy to political polling it's still important to note as it is the change in the head of state in the nation. I don't feel effect on polls is neccesarily too important in this case due to the role the Queen plays in British culture and the fact alot of things must now be done due to her death e.g. MP's pledging there allegiance to the new king, New currency needing to be printed, Parties reducing there activities, I believe there's also going to a weeklong recess for parliament. It's knock on effect will definitely effect the nation and possible by extension polling anyways.

Tl;dr

- New head of state (Big change in the structure of the UK's institutions)

- Plays a huge role in British culture (potentially more than the examples Bondegezou suggests)

- Major Knock on effect's EnglishPoliticalPerson (talk) 14:58, 9 September 2022 (UTC)

Again, this is not a page on British culture or whatever, but on opinion polling. This is not an article to discuss gossip or historical events. We have lots of articles for that, starting with Death of Elizabeth II, 2022 in the United Kingdom, Monarchy of the United Kingdom, etc. This article pertains just to opinion polling. The effect on currency, parties' activities or parliament recesses are outside the scope of this article. Impru20^talk 16:11, 9 September 2022 (UTC)

Yes If we're mentioning changes in Prime Ministers/party leaders, surely we should mention something as big as a change in reign.Mozartnut (talk) 15:51, 9 September 2022 (UTC)

• Yes There can't be a bigger event than the death of a monarch and it has huge implications on society and the framework of British politics. 82.26.31.56 (talk) 22:10, 9 September 2022 (UTC)

  Could you name the implications on 'the framework of British politics'? As others here have said the head of state doesn't have a political role at all, and this article is about polling - not society or culture. AlmTec (talk) 16:37, 10 September 2022 (UTC)

  The King does have a political role, the government is his government, and he can dissolve it as the Queen did in Australia in 1975 Australian constitutional crisis Vahvistus (talk) 08:59, 18 September 2022 (UTC)

• No We rightfully include only a minimal number of events, which could be reduced even further. We should not be expanding the scope of non-opinion poll material in the tables. Ralbegen (talk) 09:05, 10 September 2022 (UTC)
• No This article by its name is strictly about political opinion polls in the United Kingdom, thus it only includes political polls, political elections and changes of party leadership. Many things can have an effect on the polls (even significant events like the current energy crises, covid, the death of a much loved queen) but we have no idea of their actual effect on the polls. Plus if we start including stuff, where do we stop, the article would become very long and ... arguments for events to be included way way longerrrrrrzzz. Best keep this article simple and straight forward. UK Events of notable importance should be included in the article set up for that purpose, 2022 in the United Kingdom, as well as their own articles. ~ BOD ~ ^TALK 20:23, 10 September 2022 (UTC)
• No If we were still using the old rules for including news in the tables, I'd lean towards yes but seen as many major political events aren't getting included here now it seems odd to include a nominally apolitical event. There is plenty of coverage of the Queen's passing on Wikipedia now much of which will have been a lot easier for a casual reader to find over the last few days than this article.--Llewee (talk) 23:27, 17 September 2022 (UTC)

Yes The pollsters ask about a general election which can only be called by the head of state. It is a convention that it is at the request of the Prime Minister, but this convention was not followed in 1975 Australian constitutional crisis. Plus, the titles of the positions being elected to have also changed. The implied question being asked by pollsters is now "if the King called an election for tomorrow, who would you vote for to form His Majesties government?"Vahvistus (talk) 08:54, 18 September 2022 (UTC)

The fact that we need to bring an extremely exceptional (and not repeated ever since) break of convention 50 years ago in Australia to attempt to justify the event inclusion here is more an argument against it than it is for it, really. It should be noted that articles on Australian opinion polls, which are slightly (though not much) more open to event inclusion than UK ones, do not include the death of Elizabeth II or changes in the Governor-General as events. Impru20^talk 17:30, 18 September 2022 (UTC)

As you say it's a convention, so not a law which made it straight forward to break in 1975 and so would be easy to break again. This time Dissolution and Calling of Parliament Act 2022 has an ouster clause which bars the Supreme Court from challenging the decision. Parliament is dissolved at the will of the King. Like his mother before him King Charles can legally dissolve parliament on his own authority and he appoints a prime minister to form his majesty's government. Vahvistus (talk) 09:44, 19 September 2022 (UTC)

Again, your whole argument is based on a 'what if' situation (WP:CRYSTALBALL, someone), based on a precedent in another country, that does not even address the fact that the King's powers or other regalisms are not issues that are within the scope of this article (the fact that when dissolution happens neither the Monarch nor the PM are mentioned—as seen in Opinion polling for the 2019 United Kingdom general election— proves just how off-topic this argument is). Further, as said (and left unaddressed), this is not even slightly mentioned in Australian opinion polling articles (the country from which the precedent has been taken), which do not mention whether there is a change of Governor-General either (and these are, again, slightly more permissive with events than the UK page is). Impru20^talk 10:33, 19 September 2022 (UTC)

Yes the succession caused a pause in politics which has a knock on effect for polling. I would not favour including a coronation, but I would favour including the accession. Kalamikid (talk) 14:02, 22 September 2022 (UTC)

No The head of state is purely ceremonial and we have no reason to believe polling was impacted. We might as well start including the weather report if we want to include things with no obvious impact or relevance to the article in question. There are already pages specifically for notable events, this one is for polling information. Cactuslunch (talk) 07:21, 28 September 2022 (UTC)

No Exceptional event, but no impact on the polling. There is no need for informative explanation. See contrasting discussion below regarding the September mini-budget Jw2036 (talk) 13:19, 1 October 2022 (UTC)

Adding the mini-budget to table

Seeing this being added to the table, I'm personally on 50/50 on it. Consensus is typically not to add events like this, but this could be seen as an exception. I've undone the revisions as of right now. EnglishPoliticalPerson (talk) 23:03, 30 September 2022 (UTC)

Additional to this, we haven't got barnard castle in the table so I dont know if this should qualify. EnglishPoliticalPerson (talk) 23:07, 30 September 2022 (UTC)

I'm not sure why I'm chiming in given I've twice given support to adding things to the table and that never helped them remain in the table, but I do believe this is significant enough to warrant inclusion. If Black Wednesday was included in the pre-1997 opinion polls, I think the comparable set of events here also deserve mention. It would certain allow readers some context as to why Labour's leads have gone (mostly) stratospheric seemingly all of a sudden. I just wonder whether its inclusion depends on whether Labour's boosted performance is sustained long-term; after all, it could be a reactionary blip, and Labour returns to leads in the low teens or high single figures relatively soon... Phinbart (talk) 23:43, 30 September 2022 (UTC)

Indeed, its a fair assessment. I'm not too sure on it, just think it should go through the appropriate channels first. EnglishPoliticalPerson (talk) 12:24, 1 October 2022 (UTC)

I am in favour of adding it. Whether or not the lead is sustained, it is crucial context for a dramatic change in the polls. Dudley Miles (talk) 12:36, 1 October 2022 (UTC)

My view is that events which have a significant bearing on the polling situation should be added. This makes this page informative, rather than a collection of data. There is no doubt the mini-budget has had a dramatic effect. I see the need to be restrictive, but big political events that dramatically change the polling should be there Jw2036 (talk) 12:43, 1 October 2022 (UTC)

As an example of precedent "The Fuel protests" are listed in the 2001 election polling, because they made a sudden, otherwise inexplicable change in the polls Opinion_polling_for_the_2001_United_Kingdom_general_election. Having that listed is extremely informative.

Another precedent: "Black Wednesday" in Opinion_polling_for_the_1997_United_Kingdom_general_election. Jw2036 (talk) 13:02, 1 October 2022 (UTC)

Jw2036 (talk) 12:55, 1 October 2022 (UTC)

• If we include it, what's the reliable source basis for doing so? What's the threshold RS coverage needs to meet to merit inclusion for other events? Unless we have clear answers to those we are editorialising. This is a particularly strong example but you can go through the graph and tables and see various events aligning with poll changes that potentially are related: COVID, Starmer's election, Barnard Castle, vaccination beginning, the Paterson scandal, Partygate, the Pincher scandal, and the mini-budget. Some people also want the Queen's death. We need a razor to determine whether to include an event or not, and current consensus is for only party leadership changes, local elections, and parliamentary by-elections to be included. If we want to broaden that, we need a robust definition for what goes in. I think we should either leave it as is, or remove events from the tables altogether. Ralbegen (talk) 13:00, 1 October 2022 (UTC)

No, I don't agree we are editoralising at all. Events that cause dramatic swings in the polls are very rare: I give two examples above. None of the rest you list produced any significant deviation. What we are doing is informatively labelling what would otherwise be an inexplicable, sudden deviation in the polling. What this does is aids understanding to those reading the page, who can pretty clearly see that "something" happened on e.g. Black Wednesday or Sept 2001, but otherwise might not know what without a link to the relevant article Jw2036 (talk) 13:14, 1 October 2022 (UTC)

I'm not sure, but certainly I think it would need a reliable source to say that the mini-budget (or 'fiscal event') caused the change in the polls. Otherwise, perhaps it was Italy beating England in the football that made people turn away from the incumbent party. Who are we to say? --Wavehunter (talk) 13:39, 1 October 2022 (UTC)

I think the mini-budget should be added. It does seem to be the catalyst for a monumental shift. But we do, as Ralbegen mentions, need a razor. I'm sure we could find RS saying 100 different events caused shifts in polling in the last year alone. We certainly can't add every single event. This change will also have been affected by Labour's Conference, which was received well in most coverage, and conference season often does have some impact on polling, but they're not included in the tables. So RS existing alone isn't enough. BeeEdits (talk) 14:34, 1 October 2022 (UTC)

We have a good cite used in the last para of September 2022 United Kingdom mini-budget#Political, The Times explaining the big change in the 23–25 Sep YouGov survey which it commissioned, basically just preceding the Labour Party Conference, for the content: 'which The Times attributed to "voters turning against Kwasi Kwarteng's tax-cutting budget" with just 19% of those polled considering the mini-budget to be "fair".' This shows the big change started before the conference. Rwendland (talk) 14:46, 1 October 2022 (UTC)

In the text in that article, the link between the budget and the polling change is (correctly) explicitly attributed. If we include the mini-budget as the only event to have its own line outside our normal fare, we are linking it to the polling change in Wikipedia's voice. If the threshold is a reliable source, including a broadsheet newspaper, links an event to a polling change, far too many events would be included. Ralbegen (talk) 15:50, 1 October 2022 (UTC)

As for proposal of other events, consensus on events in UK polling election articles is clear: the events that are to be included as of current are already in excess of what should include an opinion polling table. Note that even if we could agree on the mini-budget announcement being a "catalyst for a monumental shift", election is still two years away unless otherwise established. This shift will be pointless by then when countless other events may (and will) impact opinion polling, so stop seeing opinion polling tables as listings of events, because they are not (nor should they be). Any noteworthy event in 2022 should be listed at 2022 in the United Kingdom, this article's scope does not cover that. Period. Impru20^talk 17:48, 1 October 2022 (UTC)

I can see the argument that we included Black Wednesday in 2001 polling table. However I agree, if we're going of consensus for the past few years then this is not worthy of being added to the table - Especially as we don't include any other budget announcements let alone mini-budgets. There's been many major scandals in this term, none of which we've included so why start now? EnglishPoliticalPerson (talk) 17:54, 1 October 2022 (UTC)

None of the other scandals and major events of this parliament term, or of any recent parliament term, have had as dramatic an effect on the polls as this mini-budget has. People might come back to this page in the future and wonder why Labour suddenly got a 30% lead out of nowhere. And this is comparable to Black Wednesday which as mentioned before was also included on its respective page. 82.26.31.56 (talk) 21:10, 1 October 2022 (UTC)

Fine, but this article is not for that. Also, you should consider that most of the polling changes happened a few days after the mini-budget, when the economic impact of it was evident. Thus, you would have to add other events as well (the depreciation in sterling's value, the BoE intervention, etc.), which is one of the obvious reasons why adding events to these tables is problematic, as typically it's not just one event that impacts events, but a chain of events. Once again, this is not the scope of this article. Further, a single row is not enough to fully explain the context of that event, meaning the addition of it is merely for decorative purposes since you won't be able to explain the causes or the consequences of any other event that is added. 2022 in the United Kingdom already addresses these and other events, and it is the article that readers should be redirected to if they want to know more context into the pre-election timeline. Impru20^talk 06:40, 3 October 2022 (UTC)

If people want to know what caused it they can look that up? This page isn't for a list of event, regardless of the effect on polls. Ralbergan makes a fair point by asking what we'd classify what is considered big enough to be included. I'd argue you either go all or nothing if I'm being fair. We should be consistent in our implementation of the consensus found in previous discussions. EnglishPoliticalPerson (talk) 21:32, 1 October 2022 (UTC)

You don't need to go all or nothing, you just need to include events that have dramatic effects on the polls such as this. As said elsewhere, Black Wednesday and the 2000 fuel protests were similar events that had huge effects on the polls and they're included on their respective pages, so how is this different?82.26.31.56 (talk) 22:54, 1 October 2022 (UTC)

I'm kind of 50/50 too, if it causes a change in general polling trends it should 100% be added - but as others have said this it could just be a temp poll boost. Looking fairly dramatic so far though. AlmTec (talk) 23:38, 1 October 2022 (UTC)

I'm plumping for no, don't add it. As others have said, a reliable source is necessary but not sufficient. We could give a cut-off - for example, if either major party moves by more than 10 percentage points in a seven-day period, then we include it as the cause of a dramatic move. But then there would be discussion about why 10%, why seven days, and how to measure this with so many polling companies all saying slightly different things. Simplest and cleanest to leave it out and keep events on the events page (2022 in the United Kingdom). --Wavehunter (talk) 08:55, 2 October 2022 (UTC)

From a BBC John Curtice article today: "The swing to Labour was similar in size to that seen on Black Wednesday in 1992, the first time the policies of a Conservative government had produced turmoil on the money markets.

If voters remembered the events of the past week when they go to the polls in two years' time, Labour could be on course for a three figure majority, said Sir John, even if Ms Truss's policies work as intended and produce economic growth." 82.26.31.56 (talk) 19:04, 2 October 2022 (UTC)

https://www.bbc.co.uk/news/uk-politics-63110539 82.26.31.56 (talk) 19:05, 2 October 2022 (UTC)

Major political event that has shaped British politics. How can you NOT include it??? Weird to even have the discussion. — Preceding unsigned comment added by 92.184.98.70 (talk) 21:29, 2 October 2022 (UTC)

Because this is an article on opinion polls, not a list of "major events shaping British politics". It falls out of the scope of this article to add events that someone considers to be important. Plus, adding one row barely mentioning the event is not even enough to fully explain such event and provide context for it, which proves that the point of adding events here because of their alleged relevancy is quite moot. You have 2022 in the United Kingdom for that. Impru20^talk 06:43, 3 October 2022 (UTC)

So why was Black Wednesday included in the 1997 opinion poll article? No-one's explained why that can be included but not the mini-budget. 82.26.31.56 (talk) 09:33, 3 October 2022 (UTC)

That's a good question. It looks like someone unilaterally added it on their own volition in November 2019, which if I remind correctly would be after the current consensus on events was reached (meaning it was an edit against consensus). Nonetheless, I see you have subsequently removed it from the table, which is nice. Impru20^talk 21:46, 3 October 2022 (UTC)

Black Wednesday was quite a significant turning point event, and I think events like it should definitely be kept; at the very least put it in a paragraph somewhere... You can't just claim that something is against consensus because the consensus isn't your opinion, clearly the fact that it survived three years of scrutiny by people who are not you shows something. AlmTec (talk) 23:31, 9 October 2022 (UTC)

We do have an explicit agreed consensus on what goes in these tables and Black Wednesday doesn’t qualify, so, in that sense, its inclusion is against consensus. Having a paragraph talking about the polling where Black Wednesday is mentioned, that sounds like a very good idea to me. I’m all for reliably sourced discussion of the polls, just not with turning a table of polling results into a broader timeline. Bondegezou (talk) 06:25, 10 October 2022 (UTC)

Exactly. Aforementioned arguments against addition aside, people don't seem to realize that adding & explaining events in one row is generally a bad idea, since much needed context or explanation will be missed. For example, in the case of the mini-budget it wasn't just the announcement that affected opinion polling (possibly, the announcement itself had zero impact), but the immediate economic aftermath, the sterling crash, the interest rate sky-rocketing, the mortgage cost crisis, then the Labour and Conservative conferences showing the former's unity and the latter's division. There's a lot to explain there to properly provide context por polling, which is something you won't do with one row.

Also, AlmTec, the addition of Black Wednesday was against consensus then and now. It doesn't matter whether it went unnoticed for three years (note that Wikipedia is a work in progress, you don't need to notice it immediately), it is still against consensus, so its removal is due, then and now. Again I should remind that its addition was unilateral and undiscussed, and the fact that its presence was even being used as an argument for attempting to add the mini-budget here seems how harmful it was at attempting to circumvent consensus. Unless consensus changes, that event is not due in that table. Impru20^talk 06:43, 10 October 2022 (UTC)

Against - Deciding which events are electorally/ sentimentally significant and which not is editorialising. Causation is always a subjective matter (Bradford_Hill_criteria). Leave the data. There are lots of lists of events. Readers can make their own analysis. They'll disagree because there's a "generous ambit within which reasonable disagreement is possible". Aliis Exterendum. Cutler (talk) 21:01, 3 October 2022 (UTC)

What might be nice would be if readers could see the polling table and UK events page side by side and construct their own narrative. I'll leave that one to the developers. Cutler (talk) 21:11, 3 October 2022 (UTC)

So this article is about opinion polling, but editors seem to have a narrow interpretation of what the science of opinion polling encompasses. It is not simply how an election is likely to pan out if one was held tomorrow or a load of numbers, but rather understanding how the population reacts to various events in the political sphere. So many reliable sources have talked about the effect the mini-budget had on the opinion polls, as political scientists have seen how the Tories have dropped in the polls, and identified a link with the mini-budget. So I think the mini-budget should be referenced to, because it is a part of what this article is about, and maybe include some prose of how the mini-budget is linked to the numbers pollsters have found with reference to reliable sources. However it seems that many events are automatically included in the list of polls, such as all by-elections, even if no link to opinion polling can be made or no reference can be made to the UK population's opinion. For instance, the 2022 Southend West by-election, which was triggered by the murder of David Amess, has had no determinable impact on the population's opinion and was neither indicative of that opinion (the Tories were the only main party standing), so why is that included in the table? So the article should include political events with a connection to the population's opinion or opinion polling, with cited discussion on how events have impacted polling results, and not include certain types of event automatically (i.e. without evidence there is a link to opinion polling). I also note that political scientists may also discuss why some events had minimal impact on polls, so I don't agree with what some editors have suggested where they must be a 10pp change in the polls to include an event. --TedEdwards 22:41, 3 October 2022 (UTC)

Clap. Clap. Yes, this exactly. If I knew this discussion would result in the ridiculous decision to remove Black Wednesday from the 1997 polling article I wouldn't have contributed to the discussion and effectively got the ball rolling. I've added what's in the 2020 and 2021 sections to the 2022 section, directing users to relevant articles that will clarify and provide context for polling outcomes. Phinbart (talk) 12:01, 7 October 2022 (UTC)

Well, actually, the "ridiculous decision" was to unilaterally add Black Wednesday on the 1997 opinion polling article against the then-established consensus, let it go unnoticed by a couple years, then use it to justify another breach of consensus in this article. Events are not added based on their impact on opinion polls; rather, current consensus is a compromise solution, since many people actually argued against adding any events at all. Anyway, adding links to the proper listings of yearly events in each section is a sensible and welcome approach: if users want to know what events may have had an impact on opinion polling, then let them access to the full list of events (which is, in any case, more throughly explained that what could ever be done here). Impru20^talk 12:13, 7 October 2022 (UTC)

• Oppose: we have a consensus for what to include in the tables. Let’s stick to that. This article gives a table of polling. Stop trying to make it into something else. Polling articles for other countries don’t do this: why should we be different? If you want to have some well-referenced text on how polling has shifted, then have that as accompanying text to the table. Bondegezou (talk) 12:22, 7 October 2022 (UTC)

Chart updating?

Is the chart updating? I feel like it should show October... Joe (talk) 09:07, 13 October 2022 (UTC)

Having now properly looked, I think I'm being silly... Joe (talk) 09:20, 13 October 2022 (UTC)

It should be up-to-date! I want to look at making the date breaks a bit clearer because you're not the first person to have double-taked at them. Ralbegen (talk) 11:26, 13 October 2022 (UTC)

You're probably already aware of this, but it would be helpful if the minor vertical lines represented new months rather than 1.5 month intervals. Gbuvn (talk) 12:27, 13 October 2022 (UTC)

That's exactly what I was thinking of changing! I'll just need to find a way to make it a higher integer when relevant - for instance, I use almost the same code for the graph on Republicanism in the United Kingdom which covers a much longer periodm and even this page might cover more than five years by the end. It shouldn't be too hard, I just need to get round to it. Ralbegen (talk) 12:33, 13 October 2022 (UTC)

There's a poll with the Conservatives on 19% but I don't see any 19% blue dot on the graph? I think it's not updating. Datawiki2 (talk) 17:49, 17 October 2022 (UTC)

This is fascinating. Using Electoral Calculus and working on the 19% figure for the Conservatives, this shows them (in theory) reduced to a single seat (South Holland and The Deepings)! I do of course agree that a few Conservatives based on their personal standing might save their seat on local reputation, but probably not more than a handful. Rif Winfield (talk) 13:39, 21 October 2022 (UTC)

That's indeed interesting. https://www.independent.co.uk/news/uk/politics/tory-poll-liz-truss-labour-starmer-b2204553.html There're also polls where they would have 20 seats. If it's done by a legitimate polling company and there're precise number, they should be added at "Seat predictions", as it ends at September 2022. Davide King (talk) 13:16, 24 October 2022 (UTC)

Smoothing the curves fitted to the data

Looking at the ripples in the lines of the four parties at the bottom of the chart, it seems likely that sampling error is playing a part, so I wonder how much that is true for the other two; that's one justification for increasing the smoothness. Another is that wanting to look at this chart could be from a historical perspective (when less smoothing would be wanted, and the ability to correlate party support with events) but it could also be from the point of view of trying to look into the future, and predicting where the trends seem to be going. - - - - - - - - - - - - - - While I recognise that ANY line that's fitted to this polling data makes assumptions and loses information -- unless it simply joins the dots -- I find the smoothing shown in this chart to be not smoothing enough. Although I recognise that the small peaks and troughs correspond to events, I would prefer to see a smoother set of lines, or even the option of affecting the coefficients that go into the LOESS arithmetic. How difficult would it be to give readers smoothing options? - - - - - - - - - - - - - - When plotting charts using spreadsheet software, various options are available, such as the number of points to be considered as local at each point, the degree of polynomials, and ideally, readers could avail themselves of all that, although on the other hand, the options could be choosing between two .SVG files or three, with a default one displayed, and a button to click for, say, a smoother option. Nick Barnett (talk) 09:09, 24 October 2022 (UTC)

I try to keep the span at a level that makes the curve visibly reflect changes in the polling figures—I'm doing it visually rather than using cross-validation. This normally means reducing it by a small amount every so often; less often as the time period covered by the graph grows. Any significant deviation from where I've set it (currently .060) looks worse and either ignores clear changes (for higher values) or over-responds to short-term changes that are clearly attributable to release schedules (for lower values). For the sake of the article, I don't think it'd be an improvement to have multiple span options for the graph. Ralbegen (talk) 16:01, 24 October 2022 (UTC)

Adding Partygate

I recently added the Partygate allegations to the table as an event, as this appears to be a major turning point in the Conservatives' opinion polling, and the decrease in their popularity. It was reverted by Ralbegen who has said that consensus is against events are generally excluded. I'm opening this thread to propose that we add the first Partygate allegations (30th November 2021) given how significant this has been to opinion polling.[1][2][3][4] The Partygate scandal appears it'll be career-defining to Boris Johnson and this parliament, and so it seems only appropriate it's added here. I also added the Autumn budget as I believe this was when the opinion polls first started to turn against Johnson, though I self reverted when I saw Ralbegen's edit - I think this would also be good to include, but I think it's less pressing than Partygate. — Czello 09:43, 7 June 2022 (UTC)

This is a table of polling figures. It is not a timeline of events. If you want to improve the section on polling in the Partygate article, that would be really good, but we’ve long agreed that this article is not for detailing events.

The other problem is that Partygate did not occur at one point in time. There have been multiple reports and consequences on multiple dates, so which would we put? Bondegezou (talk) 11:56, 7 June 2022 (UTC)

No, but certain events have an impact on polling figures - we include other events, such as by-elections or new leaders, for example. My original edit was for the initial allegations of Partygate coming out. — Czello 13:10, 9 June 2022 (UTC)

Personally, I'm losing confidence in the value of including by-elections, local and devolved elections, and new party leaders as events in the table, but there's longstanding practice and agreement to include them. There are plenty of events which ahve been linked by reliable sources to changes in opinion polls, including budgets, scandals, conferences, and all sorts else. If we include all of them, or, say, all of them mentioned by a certain number of quality of reliable sources, the tables might easily become cluttered with them and their pertinence wouldn't be clear to readers. I think that kind of material is better covered in prose, including in the background section of Next United Kingdom general election, where there's a lot more flexibility in how to present things. If there were a need for a great abundance of material specifically on the link between political events and polling, more than would be appropriate in an election article, a prose section in this article would be a better way of including it than as rows in the tables. But at the moment there are better places for that material: the polling section of articles about scandals etc, and in the background sections of election articles. Ralbegen (talk) 13:22, 9 June 2022 (UTC)

I don't think any event should be included because of its potential impact on the polls, that's post-hoc. However, an event should be included if it changes what the polling options mean. If Johnson is PM/ Con head, then asking people whether they would vote Con, is asking if they'd vote for Johnson. If Truss is head, then asking the same question means asking if they'd vote Truss. In that sense, by-elections are completely irrelevant. However, resignations of PMs are relevant, because as soon as Johnson resigned, people saying they'd vote Con are not saying they'd vote Johnson anymore. The table should make clear at any point what "Con", "Lab" etc means, and in UK politics, that is the leader of the party and government. Thedudesupreme (talk) 21:38, 24 October 2022 (UTC)

(as such, neither partygate nor the mini budget are relevant either, as they leave intact what the choice options stand for) Thedudesupreme (talk) 21:41, 24 October 2022 (UTC)

Most polling articles for other countries don't include any such commentary rows. I'd happily get rid of all of them here too, although by-elections or other national elections are at least directly relevant. I entirely agree with Ralbegen that we can cover discussion of how polling changed in reaction to events in prose. We can have prose in this article, we can have prose at Next United Kingdom general election, we can have prose at Partygate#Opinion_polls. The last of those particularly needs updating. Bondegezou (talk) 13:57, 9 June 2022 (UTC)

Can I again request input from Czello and anyone else, if you have time, to help improve Partygate#Opinion_polls. There's a graph there that is now very out of date. The ideas and citations suggested above could work there well. Bondegezou (talk) 11:04, 14 June 2022 (UTC)

Liz Truss resignation

Liz Truss's resignation on 20 October should surely have its own line in the polling table just like all other Prime Minister's such as Boris Johnson and Theresa May. The fact Liz Truss's premiership was short and one of the worst in British history doesn't affect its significance anymore than that of Eden's or Chamberlain's. Billionth Customer (talk) 18:02, 25 October 2022 (UTC)

Worst is an opinion, not a fact. But I'm not sure what you're talking about. Boris Johnson's resignation is not listed, only "Liz Truss becomes Prime Minister...", and the same is done for Truss/Sunak. Augusthorsesdroppings10 (talk) 18:05, 25 October 2022 (UTC)

Possibly - as I remember it - Boris Johnson announcement of his intention to resign was listed and then removed at a later date. As it seems you're right I will drop my point. Billionth Customer (talk) 18:14, 25 October 2022 (UTC)

JL Partners Polls 19–21 Oct & 22 Aug - 1 Sep

There's been a poll added to the table from JL Partners sourced to a tweet from @polit_pro_polls. The tables for that poll are here. JL Partners don't seem to have published this as usual voting intention and instead just part of a poll on the Leadership of the Conservative Party where current VI is used for the crosstabs. The usual "Weighted by likelihood to vote, with Don't knows removed" is not in the tables. The tweet sourced seems to use the All Respondents table and recalculated to remove don't knows. (Incorrectly, Conservative should be 25%)

Should we be including this poll in the table or not? It's by a BPC pollster, but I don't believe designed strictly to be voting intention. If it should be included there is an additional poll here from August/September that should be included. BeeEdits (talk) 09:50, 26 October 2022 (UTC)

Opinium

As we know Opinium have changed their methods and say it now underestimates the Labour score by 6 points. Its now very apparent that their results are consistently out of line with all the others. Should we exclude their result - post them after adding 6 points or what? Why did we take down the note about the Opinium method change? Bernard Naish (talk) 21:55, 29 October 2022 (UTC)

We definitely shouldn’t be adding arbitrary numbers. We cannot carry out WP:OR. But we can include notes when cited to reliable sources. Returning the note about the method change sounds good to me. If there’s any RS discussion we can refer to, let’s consider that here. Bondegezou (talk) 07:37, 30 October 2022 (UTC)

Still there as cite note 'd' for me - am I missing something? Opinium said the method change reduced the Labour score by 6 points, which isn't the same as underestimating. If they thought it underestimated, they wouldn't have done it - they did it because they believed that their old method was overestimating by 6 points under the conditions of the time (which doesn't mean it's now 6 points - could be more, could be less). It does give different results to the other polling companies, but until there's an actual GE we won't know if that's because Opinium are wrong or because everyone else is. After the election there might be some RS discussion as part of wider assessments of polling quality, and if Opinium do end up being notably right or wrong it'll probably be dissected then. 2A02:C7F:9977:9A00:A7C:28C5:9078:E81 (talk) 14:44, 31 October 2022 (UTC)

Chart grid lines

A small thing: the labels of the chart and major grid lines are every 3 months. Could the minor grid lines be set to 1 a month so there are 2 between each major grid line? A division of one and a half months is not very useful. Btljs (talk) 06:13, 22 November 2022 (UTC)

Chart missing labels and fixed default size

The chart maintained by @Ralbegen in the Graphical Summary section has been updated recently with a format adjustment, however this change has also included removing the labels for each line that state which party it represents, which and it also seems to have added a default width and height to the svg (720pt and 432pt respectively) which stops the svg from scaling to best fit the screen when the zoom is at 100%. Could these aspects of the change be reverted?

The labels were helpful in making it clear in the image itself which party each colour represented, without relying on information external to the graph.

The lack of a default width and height it the graph were useful in allowing the svg to be opened and automatically scale to make use of all the space in the window, instead of being confined to a certain area and requiring additional steps to best fit the screen. Danyaal99 (talk) 08:57, 2 January 2023 (UTC)

Hi Danyaal99, I've restored the previous format for now! Ralbegen (talk) 16:13, 3 January 2023 (UTC)

Thanks @Ralbegen, that looks good to me! Danyaal99 (talk) 19:44, 3 January 2023 (UTC)

Rounding issue

The table shows the FindOutNow poll as representing a Lab lead of 17 pts, but FindOutNow themselves describe it as a Lab lead of 16 pts. I'm guessing that this is because the figures in the poll are rounded, and if the Lab lead is calculated from the unrounded figures prior to being rounded, it may be 16 rather than 17. Alternatively, it could be a misprint in FindOutNow's summary (since they have after all mislabelled 'Feb 2022' as 'Feb 2021'!). See https://twitter.com/Channel4News/status/1575202541763280914 (the same link cited in the article).