Talk:Opinion polling for the next United Kingdom general election

From Wikipedia, the free encyclopedia
Jump to navigation Jump to search
WikiProject Elections and Referendums (Rated Future-class)
WikiProject iconThis article is within the scope of WikiProject Elections and Referendums, an ongoing effort to improve the quality of, expand upon and create new articles relating to elections, electoral reform and other aspects of democratic decision-making. For more information, visit our project page.
Future-Class article Future  This article has been rated as Future-Class on the project's quality scale.
 
WikiProject Politics of the United Kingdom (Rated Future-class, Low-importance)
WikiProject iconThis article is within the scope of WikiProject Politics of the United Kingdom, a collaborative effort to improve the coverage of Politics of the United Kingdom on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
Future-Class article Future  This article has been rated as Future-Class on the project's quality scale.
 Low  This article has been rated as Low-importance on the project's importance scale.
 

Error in the chart[edit]

Pertinent to what I have been saying about the new local regression rather than moving average, we now have the situation where the Conservatives are shown as currently polling higher than Labour, which is simply not apparent in any of the polling. It would probably be helpful to decrease the amount of smoothing, but also for us to know what calculation is being made. Onetwothreeip (talk) 11:06, 22 June 2019 (UTC)

I agree this is a problem. Loess has problems at the ends of the range of the data. I think it's technically possible to fix the end point to some recent moving average, thus avoiding this problem, but quite how you'd do that in practice, no idea. Bondegezou (talk) 11:56, 22 June 2019 (UTC)


I also think this is a problem. We also have spurious trends showing LibDem and Green currently rising fast, and Brexit about to fall off a cliff, when in fact all three are flattish after the Euro Elections. I'd continue to plead for a return to simple moving averages graphed at the average date of polling.RERTwiki (talk) 12:04, 22 June 2019 (UTC)

I agree, we really should just return to a moving average of some sort. Using regression is using a sledgehammer to crack open a nut. Onetwothreeip (talk) 12:12, 22 June 2019 (UTC)
I concur with those concerns, but I do approve of the smoothness of the loess plot compared to a wiggly moving average. Can we have a kernel-weighted moving average (based mostly on near values, but somewhat on further away values)? Bondegezou (talk) 12:30, 22 June 2019 (UTC)
Except for the fact that a moving average shows essentially the exact same growth. https://i.imgur.com/8Fa1zZH.png (This graph is only among 2019 polls so the growth looks less steep). The only significant difference is that Brexit is "falling off a cliff" (barely moved downward in loess) yet it's still shown as ahead of Labour and the Tories. Filinovich (talk) 12:43, 22 June 2019 (UTC)

I think one reason that the moving average looks ugly is that multiple polls on a single day produce multiple datapoints on that day, with a jump between them. The effect can be reduced a little by changing the date ascribed to the moving average from the day of publication of the poll to an average date of all the polls in the average. That would also (correctly) show that the moving average is a little behind the times. Lastly, you might be able to press the 'smoothed line' button in Excel and see if that helps, if you think appropriate.RERTwiki (talk) 10:30, 25 June 2019 (UTC)

I've got this watchlisted. I've changed the moving average so it uses a rolling 28-day window instead of the previous one which was based on the n most recent polls. Mélencron (talk) 15:45, 25 June 2019 (UTC)
Could this be weighted by recency? If there are only a dozen polls within 28 days, it currently makes a big difference when a poll drops off the end. Really the effect of new polls should outweigh that of those dropping off. Btljs (talk) 11:56, 26 June 2019 (UTC)
Whatever method is used, I think that it should be clearly stated in the article e.g. if the calucation is to sum each party's percentages for the last 28 days and divide by the number of polls, just say that. Jontel (talk) 06:57, 7 July 2019 (UTC)

Removing UKIP, ChUK, PC, SNP columns[edit]

Given the collapse in polling of UKIP and Change UK, I think July 1 would be a most appropriate time to restart the table without their results given in their own columns. I would also move Plaid Cymru and SNP, not only for low polling particularly with PC, but because their national results are very much irrelevant and instead should be considered with the perspective of polling results in Wales and Scotland respectively. Onetwothreeip (talk) 09:03, 27 June 2019 (UTC)

I agree, though would keep SNP and PC. --Comment by Selfie City (talk about my contributions) 14:48, 27 June 2019 (UTC)
I'd keep CHUK and UKIP but not SNP and PC, for reasons I've already given in the last discussion. Mélencron (talk) 16:06, 27 June 2019 (UTC)
I'd prefer to remove SNP/PC over ChUK and UKIP as the former are often grouped together or excluded entirely from headline voting intentions while the latter are often listed (albeit with recent low proportions). Filinovich (talk) 16:10, 27 June 2019 (UTC)
But almost no one supports Change UK and UKIP anymore. They are little more than minor parties. (To be clear, as of June 30 I do not mean that we should remove them from past tables; they should just not be added in current or future tables.) --Comment by Selfie City (talk about my contributions) 16:23, 27 June 2019 (UTC)
If we remove Plaid Cymru and the SNP from the table, it should be throughout. Their noise-level national VI figures aren't new. I'd also support a new table that doesn't include Change UK and UKIP in principle—I'd be interested to see whether reliable sources have changed how they report headline figures. We should take our lead from those (though all else being equal, I support your proposal with the addition of removing PC/SNP figures from the UK/GB tables throughout). Ralbegen (talk) 17:05, 27 June 2019 (UTC)
I'm going to preface by noting that all pollsters still report results for all parties in their tables. That having been said, here's a table of pollsters (listed in reverse chronological order from most to least recent) for which parties mentions from the presentation of results for their most recent poll. Those with only pre-Euros numbers are listed below the dark bar.
Removing a party with 35 seats and potentially more makes absolutely no sense. If the argument is that they are not national then neither are labour or conservative - think Northern Ireland. If the argument is that theey are a small persentage then do not include ChUK, Ukip and any others that poll low. Be consistent - not partial.--Subsea (talk) 08:21, 1 July 2019 (UTC)
Significant numbers of users will continue to want separate listing for SNP and PC. I don't understand why separate listing would present serious inconvenience to users who do not want this information. Francis Waring (talk) 13:41, 1 July 2019 (UTC)
Pollster SNP Plaid Cymru UKIP Change UK
YouGov No No No No
Ipsos MORI No No Yes No
Opinium No No Yes Yes
Survation No No Yes Yes
ComRes Yes No Yes Yes
BMG No No Yes Yes
Deltapoll N/A (tables only; all parties included)
Number Cruncher Politics No No No No
Panelbase N/A (tables only; all parties included)
Hanbury Strategy Yes No No Yes
Kantar Yes Yes Yes Yes
OnePoll Yes No Yes Yes
ORB Yes No Yes Yes
Sky Data Yes Yes Yes Yes

Mélencron (talk) 17:38, 27 June 2019 (UTC)

To clarify, I would remove SNP and PC columns for the entire national polling table, and start a new table without UKIP and ChUK columns. We could make that new table right now using existing data, but we can at least start it at July 1. Onetwothreeip (talk) 21:58, 27 June 2019 (UTC)

I’m okay with that. --Comment by Selfie City (talk about my contributions) 22:50, 27 June 2019 (UTC)

I'd prefer changes happen at year end. UKIPs disappearance has only been a month long, and surely nothing has changed since the decision to include SNP&PC? What's different with those now? CUK should probably not have been added in first place, and a good example of why we should wait longer before making changes. 86.183.35.158 (talk) 16:03, 28 June 2019 (UTC)

Well, Change UK could have done a lot better than it did. Those of us who made the tables couldn’t have predicted it’s performance. --Comment by Selfie City (talk about my contributions) 16:26, 28 June 2019 (UTC)
We should follow RS (so thanks to Mélencron for table). I also note that opinion polling tables for other countries often include large numbers of parties.
Looking at that table, we should drop Plaid Cymru (although we'll need footnotes for when the SNP figures is actually an SNP+PC figure). I can see an argument for dropping UKIP and ChUK, but pollsters are still persisting in highlighting them, so we need to too... at least for now. Happy for this to be watched closely and to move to excluding them when we see some more pollsters drop them (as seems likely will come in time). Bondegezou (talk) 16:29, 28 June 2019 (UTC)
I would also prefer changes to happen at the end of the year, but the middle of the year is the second best option. If UKIP or ChUK come back we could always restore the column.
The decisions to include SNP and PC were made differently. SNP was included because they were polling at a level similar to other minor parties that we were including at the time, but this was surely a mistake given the context of SNP polling. Plaid Cymru was introduced after barely any discussion because some of its results were combined into SNP. We can undo both of these actions.
Given that Change UK was polling above 10% multiple times, I have no regrets that we included them in the table. Onetwothreeip (talk) 22:05, 28 June 2019 (UTC)
Although there are other polling articles that have wider tables, this is normally because of seat projections and not reported polling responses, particularly where there is proportional representation. We should consider how polling organisations are presenting their information, but we are by no means bound by that and we can choose to display information the way we think is best for an encyclopaedia. I think we can agree this article should be simpler to read than what polling organisations publish. Onetwothreeip (talk) 22:05, 28 June 2019 (UTC)
I looked at some other polling tables and counted columns for parties with "reported polling responses". I am excluding seat projections. I looked at the European Union for roughly comparable democracies:
  • Austria: 7
  • Belgium: Flanders 8 and Wallonia 9 (+ 2 summary columns for those in government vs. out)
  • Bulgaria: 9
  • Croatia: 15 + undecided
  • Cyprus: 10
  • Czechia: 13 (+ 2 summary columns for those in government vs. out)
OK, bored now. It is apparent that many other polling tables on Wikipedia happily support large numbers of columns. With SNP, Plaid, ChUK and UKIP, we're on 10. This is comparable with the above. I don't see any design argument that we need to reduce.
Most countries use PR of some sort, which is weakly associated with higher numbers of effective parties. Greece only uses a hybrid, semi-PR system, so I looked at that as well: 14 party columns. India, using FPTP, only has 3 columns, and the US only 2. But it's not the voting system that matters: it's the number of parties having an impact. There are only two significant parties in the US; there are more in the UK.
I suggest we are bound to reflect how reliable sources present information, as per WP:BALANCE, WP:NPOV and WP:RS. I do not agree that "this article should be simpler to read than what polling organisations publish", at least not simpler than the summary level Mélencron looked at in the analysis above. Not showing a party is taking a position on that party's significance and we must be neutral. Bondegezou (talk) 16:21, 29 June 2019 (UTC)
I also agree that Wikipedia should reflect how RS treat which parties to include and also be given due weight, and that I don't agree with the editing philosophy in general of removing columns on the basis of how low some party happens to be polling, as long as they are still being included in most polls and their results are being reported on. For now, that situation does not appear to have changed with either UKIP or CHUK, and the only basis on which they should be removed is if pollsters stop prompting them entirely. The SNP/Plaid case is ambiguous (and I've argued both ways in the past) but I currently lean towards exclusion/retaining values in "others", because it is basically not possible for there to be significant movements in their vote share without massive changes in polling in either Scotland or Wales, though pollsters are mixed in whether to list them or not in their summaries (though all but one the most frequent pollsters do not do so). Mélencron (talk) 17:04, 29 June 2019 (UTC)

───────────────────────── I think from Melencron's table above (thank you!) it looks like there's an RS case to remove SNP and PC from the tables (into the "other" column). I think it's also worth noting that the three most prolific political pollsters (YouGov, Opinium and Survation) don't include SNP/PC in their write-ups. Ralbegen (talk) 17:55, 29 June 2019 (UTC)

Okay, here's a summary of positions in this discussion so far [chart redacted from public view by SelfieCity due to concerns over canvassing]. @Absolutelypuremilk, Chessrat, Jontel, and TedEdwards:, as participants in past discussions as to whether to add parties, would you care to offer any input here? I think there's a fairly clear consensus to remove one or both of SNP/Plaid right now, but still divided on whether to remove UKIP/CHUK in the future. Mélencron (talk) 18:43, 29 June 2019 (UTC)

I'm happy to remove both SNP and Plaid. I think UKIP & CHUK should be removed at the end of the year, given that they were polling pretty highly earlier on in the year that's notable in itself. In contrast SNP & PC have been pretty steady throughout the whole Parliament. Absolutelypuremilk (talk) 18:47, 29 June 2019 (UTC)
One the basis that PC/ SNP are regional parties with inevitably low poll ceilings at national level, I would support putting them in All others at national level, but broken out for their own areas. CHUK and UKIP had a past which may be of interest to some. I think for now it would be helpful to locate the columns of the four major parties next to each other on the left, followed by the Greens, followed by CHUK and UKIP. That moves them out of the way. Perhaps, then, we can have a rule of dropping national parties from the table after a certain period of less than 5% scores. Jontel (talk) 19:19, 29 June 2019 (UTC)
As far as I'm concerned, the SNP is polling above CUK/UKIP, and Plaid is polling at the same level as them, so I'd support removing either all of them or none of them. It would look odd to have a table where parties on 0%/1% are included in the main columns, but a party on 4% (the SNP) is relegated to "others". My preferred solution would be new table on July 1 with only the five main parties by polling (Con, Lab, LD, Brexit, Green) in, leaving the current table for January to June as-is. Second-best option would be just continuing with the current table/columns. Chessrat (talk, contributions) 21:54, 29 June 2019 (UTC)

I really want to make it absolutely clear that this is about removing UKIP and ChUK from future tables, not tables that have previously existed. If we are to remove UKIP and ChUK by the end of the year, we might as well make a new table starting July 1 2019 which is surely the most natural way to split these tables after by calendar year. I don't mind whether or not we remove the SNP/PC data from already existing tables, but it's at least easier to set a new policy going forward. I think we can at least set up a new table format starting in July to see how it works. If the UKIP/ChUK parties make a resurgence in the polls, we could always create those columns again.

@Bondegezou and Mélencron: Counting how many columns other articles have is surely WP:OTHERSTUFFEXISTS. Those articles may very well be doing them wrong, and the United Kingdom general election polling series is the most prestigious we have. The main difference of course is the proportional representation. I agree that we should present information as reliable sources do, but when it's only polling that is giving any notice of UKIP or Change UK, that strongly indicates that they are unimportant parties on the basis of the non-polling media. Whatever poll results that we perceive as low are highly correlated to how the media considers those parties in importance. When we look at how reliable sources treat these parties, we should be looking at the entire news media landscape, not just polling organisations which take a statistical view. Onetwothreeip (talk) 02:51, 30 June 2019 (UTC)

@Mélencron: Just to be clear, I don't think that the case to remove UKIP/Change UK from future tables is here yet, though like others in the thread I think it's something to keep an eye on. So I'm a yes for that column for now. Ralbegen (talk) 10:45, 30 June 2019 (UTC)
Okay, thanks for clarifying. I think I agree that it's probably something that can be done in the future, but it hasn't been that long since the Euros yet and most pollsters are still including both of them, but I wouldn't be surprised if they get dropped by pollsters eventually if it's clear either/both have become nonentities or are not capable of standing candidates everywhere. Given the current views in this discussion, an outright majority seems to support removing columns for SNP/Plaid, so I'll probably take a look at that later – it should probably take a couple of hours or a little more to check for any potential errors. Mélencron (talk) 12:41, 30 June 2019 (UTC)

Absolutely astonished to see that the SNP and Plaid Cymru have been removed and lumped into an "others" category. This is plainly perverse given that the SNP in particular routinely poll significantly higher than two of the parties that have not been removed (UKIP and Change UK). I discovered a reference in the edit history to the supposed "strong consensus" in favour of removing the SNP and Plaid - well, please regard this contribution as weakening that "consensus". Nobody in this discussion has yet offered a single credible reason for removing the SNP while parties with far less support are not removed. The only plausible explanation for such a self-evidently irrational decision being rammed through is unthinking Anglocentricity. I'm not going to get involved in a flame war, but I look forward to this decision being reversed, as it inevitably must be sooner or later. Sofia9 (talk) 23:31, 30 June 2019 (UTC)

I fully agree there's a greater argument for including the SNP than for including UKIP+Change UK, and the former but not the latter being removed is... odd. There should be consistency in whether parties with minor polling results are included or not. Chessrat (talk, contributions) 01:59, 1 July 2019 (UTC)
Sofia9 This probably should have been explained better. SNP and Plaid Cymru polling results will remain in tables for results in Scotland and Wales respectively. There is no conspiracy to minimise the Scottish and Welsh sentiments of United Kingdom politics. Is this honestly a serious contribution to the discussion? It's actually in the interests of the SNP and Plaid Cymru for their polling results to be more clearly displayed in the regional tables, since 40% of Scotland is a more notable figure than 4% of Britain. Ralbegen I agree that we shouldn't remove UKIP and Change UK from the table, it's just about making a new table and not including them in that one. The new table can come into effect for results after July 1 2019. Onetwothreeip (talk) 08:54, 1 July 2019 (UTC)
I don't think that anybody's made a convincing RS-based case to create a new table excluding UKIP and Change UK yet, sorry. I think if Survation and Opinium stop including them in their result write-ups, that'd be a good time to stop including them. Ralbegen (talk) 09:07, 1 July 2019 (UTC)
Ralbegen Reliable sources are largely disregarding UKIP and Change UK. The only sources that are showing any regard for those parties are the polling organisations. I don't think anybody has argued that the results for UKIP and ChUK should be removed from the table altogether, just placed into the others column. Onetwothreeip (talk) 09:39, 1 July 2019 (UTC)
Which reliable sources? Media sponsors in their write-ups of the polling they've commissioned and other media coverage of polling? From Mélencron's excellent table of pollsters' presentations of results, only YouGov and NCP don't include them. If you look over newspaper coverage of polling results and it's clear that the Conservatives, Labour, Liberal Democrats, Green Party and the Brexit Party have their polling results covered substantially more than Change UK and UKIP, please present those results here and I'll happily be part of a consensus to start a new table which doesn't include columns for them. I think we have to follow reliable sources, so that has to be the bar. Ralbegen (talk) 09:59, 1 July 2019 (UTC)
The usual reliable sources such as newspapers and television news. Very little coverage of UKIP and Change UK, and almost no coverage of their polling results. The actual polling organisations are of course going to publish their own results, but that's not surprising. I could find you examples but just from memory I can remember news stories about Brexit Party leading in the polls, but I don't think there have been any stories focused on UKIP polling at 1% or 0%. Onetwothreeip (talk) 11:15, 1 July 2019 (UTC)

Five-column version mockup[edit]

I've made a quick demonstration of what a five-column version of the table would look like for June. (Of course, if people did decide on this option, it would make more sense to start from July rather than June). Chessrat (talk, contributions)

Pollster/client(s) Date(s)
conducted
Area Sample
size
Con Lab Lib Dem Green Brexit Other Lead
YouGov/The Times 24–25 Jun GB 2,059 22% 20% 19% 10% 22% 7% [a] Tie
Ipsos MORI 21–25 Jun GB 1,043 26% 24% 22% 8% 12% 6% [b] 2%
Opinium/The Observer 19–20 Jun GB 2,009 20% 26% 16% 6% 23% 9% [c] 3%
Survation/The Mail on Sunday 19–20 Jun GB 2,016 24% 26% 18% 6% 20% 7% [d] 2%
YouGov/The Times 18–19 Jun GB 1,641 20% 20% 21% 9% 23% 6% [e] 2%
YouGov/The Sunday Times 13–14 Jun GB 1,672 21% 21% 19% 9% 24% 5% [f] 3%
YouGov/The Times 9–10 Jun GB 1,702 17% 19% 22% 8% 26% 7% [g] 4%
ComRes/The Daily Telegraph 7–9 Jun GB 2,017 23% 27% 17% 5% 22% 7% [h] 4%
BMG/The Independent 4–7 Jun GB 1,520 26% 27% 17% 6% 18% 7% [i] 1%
6 Jun Peterborough by-election
YouGov/The Times 5–6 Jun GB 1,670 18% 20% 20% 9% 26% 6% [j] 6%
YouGov 4–5 Jun GB 1,663 18% 19% 22% 9% 25% 7% [k] 3%
4 Jun Anna Soubry becomes leader of Change UK after six of its MPs leave the party[1]
YouGov 31 May–1 Jun GB 1,644 18% 19% 23% 10% 23% 7% [l] Tie
  • Oppose Restore the SNP and Plaid Cymru. Sofia9 (talk) 23:35, 30 June 2019 (UTC)
  • Oppose. Restore SNP. Perverse to exclude a party that is the third largest in the House of Commons. Jmorrison230582 (talk) 06:34, 1 July 2019 (UTC)
  • Oppose. The SNP might be marginal in the minds of the collective anglosphere, but they are still the third largest party at Westminster and have a membership comparable to the Conservative party. Strange decision to limit the data in this way. SymonBanMacaulay (talk) 07:13, 1 July 2019 (UTC) Note: An editor has expressed a concern that SymonBanMacaulay (talkcontribs) has been canvassed to this discussion.
  • Oppose. As above - if these are England only polls then fine, but to lump the 40% or whatever of Scottish SNP respondents into other is very misleading. There is no strong reason for reducing the number of columns. Btljs (talk) 07:18, 1 July 2019 (UTC)
  • Oppose.Restore the SNP and Plaid Cymru. It makes no sense to remove the SNP which currently holds 35 seats, and Plaid which holds more seats than either UKIP or Change UK.DACN (talk) 07:20, 1 July 2019 (UTC) Note: An editor has expressed a concern that DACN (talkcontribs) has been canvassed to this discussion.
  • Oppose. Restore SNP. How can the UK's third largest party by seats and probably second largest party by members be classed as 'other'? This is embarrassing on the world stage. I would have thought that an opinion poll page should cover all opinion and any party that polls any figures should be included to give a full picture. There might be a new party formed next week that grows in popularity over the next 10 years. This page should reflect that and show the rise over time, otherwise the page is pointless. Crepusculous (talk) 08:27, 1 July 2019 (UTC) Note: An editor has expressed a concern that Crepusculous (talkcontribs) has been canvassed to this discussion.
  • Oppose. Restore SNP. As the UK's third largest party by seats, the SNP are an important part of the story of opinion polling for the next UK general election. 91.151.10.54 (talk) 08:14, 1 July 2019 (UTC) Note: An editor has expressed a concern that 91.151.10.54 (talkcontribs) has been canvassed to this discussion.
  • Support of course. Where did these these !votes in two hours from editors I've never seen before come from? Did some Scottish National Party forum get notified of this?
The proposed table does not include UKIP or Change UK, so there's no argument there. The polling tables for Scotland still include SNP, so the 40% of Scotland responding with support for SNP is very much represented on this article, but the 4% of Britain that does so is not particularly useful or relevant. It would actually be more misleading to have the SNP column in the main table, as its political influence is much greater than what any other political party with 4% of the national vote would have. The multiple instance of an argument about members is also quite suspicious, but also irrelevant. Onetwothreeip (talk) 08:54, 1 July 2019 (UTC)
"Editors I've never seen before". I've been an editor on here for many years, including on polling related articles. As for the substance of your argument: "its political influence is much greater than what any other political party with 4% of the national vote would have". Well that's tough. MPs are elected by constituency, and there are 59 in Scotland. As the SNP are currently the leading party in Scotland, they are likely to win most of those 59 seats. You might not like the electoral system that the UK has, but that's the reality of the position. In the Canadian federal election polling article that the Bloc Quebecois are included in the federal polls, even though they are a a "regional" party who presently control 10 seats in the Canadian House and are polling about ~5% for Canada as a whole. So where's the consistency? Jmorrison230582 (talk) 09:03, 1 July 2019 (UTC)
Likewise, the polling article for this year's Spanish election includes several "regional" (mainly Catalan) parties. The biggest of which got 3.9% of the vote (ERC). What rationale is being used for excluding the SNP? Jmorrison230582 (talk) 09:11, 1 July 2019 (UTC)
Your username is actually familiar to me, so I don't claim that I haven't seen you before. We're not accountable to how the Canadian and Spanish articles are designed, but I've argued that the Spanish polling articles should have fewer columns. The polling tables aren't about how many seats the party has, if any. It would be more reasonable to have a column for SNP in a table of seat projections rather than proportion of votes. I think we could have a column where SNP and Plaid are combined, but what's most important is that UKIP and Change UK columns are not restored upon making a new table starting with July results, so I would also agree to a table proposal that only omitted those two parties. Onetwothreeip (talk) 09:17, 1 July 2019 (UTC)
NO changes should be made unless the is consensus to change.
ATM the is NO to change, so dont change. ~ BOD ~ TALK 09:24, 1 July 2019 (UTC)
Are you saying that the RS-based discussion above isn't a consensus because a bunch of canvassed !votes have appeared? Responding to off-wiki partisan canvassing as if it's legitimate consensus-building rewards disruptive behaviour. There are editors with valuable contributions to make who weren't involved in the discussion above, including those who disagree with the proposed changes, like you, and those contributions are really valuable. But we should bear no heed to comments made following instructions on a blog post. Ralbegen (talk) 09:51, 1 July 2019 (UTC)
No, i cant speak for others. I have not involved in discussion above due to having very limited time due to external personal life issues, and i am afraid i am unable to respond further atm. ~ BOD ~ TALK 10:20, 1 July 2019 (UTC)
Like 182 others i watch this page. I would like to know what this accusation of canvasing is based on. More importantly my main concerns are ~ firstly that the discussion, both above the Five-column version mockup, together with the equally valid shorter responses & votes below it has only taken place over only 4 days, and that changes were being made without giving other editors sufficient time to contribute. Secondly I 'strongly oppose' the removal of the SNP from the main table (because not only that they are currently the 3rd largest party in parliament, they may could very possibly also have a significant role in any existing or future minority government situation. I also think it is too early to remove the other parties. ~ BOD ~ TALK 16:05, 1 July 2019 (UTC)
@Bodney: The accusation of canvassing is due to this publication of this blog post telling people to come and object, with instructions about how to do so, followed by a lot of activity opposing the changes, including from SPAs and single-purpose IPs. Ralbegen (talk) 07:47, 6 July 2019 (UTC)
  • Oppose Removal of SNP and Plaid Cymru from main table as per above. Onetwothreeip Wikipedia is open to all interested editors, this page does not belong to anyone. Lets have a proper RfC. ~ BOD ~ TALK 09:14, 1 July 2019 (UTC)
Bodney This page is not owned by any editor or by a group of editors either. Onetwothreeip (talk) 09:21, 1 July 2019 (UTC)
Correct, I am very glad you agree with me Onetwothreeip. This discussion only started on the 27th June, that is only 4 days ago. Big changes like removing the 3rd largest party from these tables, should only be made after a reasonable period of time, so that all interested editors, even those who have not been involved in the discussion previously may add their responses. Sudden changes make a page unstable. ~ BOD ~ TALK 14:08, 1 July 2019 (UTC)


  • Oppose I believe any party that is polled regularly should be included. It's not like there are too many columns anyway compared to other countries. --Gbuvn (talk) 09:17, 1 July 2019 (UTC)

Clearly what should happen now is that the !votes should be withdrawn and this should be reverted to a simple discussion. I gave support to this proposal in the face of blatant canvassing but there are certainly potential changes that could be made to this proposal. It seems to be the good faith act of SelfieCity that scared somebody and turned what should have been a discussion into a survey. Onetwothreeip (talk) 09:30, 1 July 2019 (UTC)

If you take a look through this category on opinion polling for future elections, you will see that for almost every article there are 6 or more parties included in the main graph and/or table. For the articles that include less than 6 parties, you will see that the parties which hold the smallest share of voting intentions almost always hold a smaller share than the SNP in the article in question. I say almost, but in fact, I could not find a single party not included in the main graph and/or table which holds an equal or greater percentage of voting intention than the SNP. Also, if we are going to add separate polling for certain countries, we should add a section for England as well. They are as much a country as any other. TagPro129 (talk) 10:03, 1 July 2019 (UTC)

Regarding England-only polling. There is some academic work (e.g. the "Future of England" survey, conducted by YouGov) done on an England-only basis, but commercial (i.e. for the media) polls are rarely done that way. I think it would only happen if the Scottish party system became completely separate, as in Northern Ireland, but the companies would likely then just poll on an "England and Wales" basis. Jmorrison230582 (talk) 11:01, 1 July 2019 (UTC)
I ran a few searches and couldn't find any England-only polling. In that case, I agree it would be pointless to add this section.TagPro129 (talk) 11:35, 1 July 2019 (UTC)
  • Oppose. As previously, plenty of polling tables have lots of columns and seem to work. UKIP, Change UK and SNP are still regularly covered in the reporting of polling, and in broader political reporting. That may change for ChUK and UKIP, but I don't see a need for change yet. There clearly isn't a consensus to drop any of the parties... with the exception perhaps of Plaid. I'm OK with dropping Plaid. Bondegezou (talk) 11:16, 1 July 2019 (UTC)
  • Oppose per all the previous arguments. Other polling tables have way more columns without problems, and they are all relevant parties. —Nightstallion 11:28, 1 July 2019 (UTC)
  • Oppose. Restore SNP. There really can be no justification for removing the data for the third largest party in the UK parliament.≈≈≈≈ — Preceding unsigned comment added by Bdw1948 (talkcontribs) 15:02, 1 July 2019 (UTC) Note: An editor has expressed a concern that Bdw1948 (talkcontribs) has been canvassed to this discussion.
  • Oppose - The SNP are the third largest party in the UK parliament, perhaps we should lump the LibDems in with the "others". I can't see any value in deleting detailed stats and aggregating into an "Others" column for the UK. The table as it stood was perfectly readable. Teh klev (talk) 20:06, 4 July 2019 (UTC)

The arguments given to remove the SNP and PC from the UK/GB table are that:

  • Most pollsters don't include them in their results write-ups, including the three most prolific pollsters.
  • Variation in the UK/GB polling results for the SNP and PC don't tell the reader much about support for those two parties. Whether the SNP gets 2%, 3%, 4% or 5% doesn't actually reflect the level of support the SNP has, whereas their results in Scotland-only polls do. Ditto for Plaid Cymru getting 0% or 1% compared to their performance in Wales-only polling.

The argument given to include them (mainly specifically the SNP) are:

  • The SNP has more MPs than other parties that are included in the table.
  • The SNP has more members than other parties that are included in the table.
  • Opinion polling articles for future elections in other countries have lots of columns, so there's no need to reduce the number of columns here.

I hope that's a fair enough summary. I find the arguments to remove them from the table, which I amongst others have made in the discussion above, more convincing. Particularly the case that we need to follow reliable sources. I don't think that the number of MPs or members that a party has should be something we consider beyond how reliable sources report polling based on that kind of thing. I find the points about other countries' future election polling pages, made by Bondegezou and Jmorrison230582, are useful for telling us that there's no imperative to reduce the number of columns in the UK/GB table. But beyond that, I feel like it would be an WP:OSE argument to say that wide tables used elsewhere in the project means we shouldn't remove columns here where there's an otherwise good case to do so. Ralbegen (talk) 11:41, 1 July 2019 (UTC)

I'd have to agree with this assessment. I would just like to note two other points I've seen. SNP and PC are sometimes lumped together in even the results tables by RSes, which would somewhat suggest that these sources don't think these parties should have numbers reported separately. On the other side, I've seen the argument that the decision to remove is purely because of Anglocentrism gone mad, although this argument is largely without merit as PA. Filinovich (talk) 13:05, 1 July 2019 (UTC)
Ralbegen, I can't agree that you've summarised the arguments for inclusion fairly, because you've omitted the most important one of all - that the SNP's percentage support in GB-wide opinion polls is high enough to warrant inclusion, irrespective of how many MPs or rank-and-file members they have. Even if UKIP and Change UK are eventually removed, it should still be remembered that the SNP have often outpolled the Greens across GB in the past, and it's entirely possible (and arguably likely) that they will again at some point in the future. Sofia9 (talk) 13:39, 1 July 2019 (UTC)
@Sofia9: You're right, I did forget to include that, sorry! I continue to think that we should give columns to parties whose polling results are emphasised by reliable sources rather than setting a Talk page threshold for polling results. Whilst that may be necessary on some other pages (I think it happens fairly routinely on US polling articles?), I feel like that would be unnecessarily WP:ORy here. For the UK/GB right now, based on Melencron's table of polling company summaries above, UK/GB results for the SNP or PC are significantly less widely covered than the other parties, particularly accounting for the most prolific polling companies which comprise a majority of UK/GB polls on the page (YouGov, Survation and Opinium). I think any argument to include the SNP and Plaid Cymru in the UK/GB table has to be based on reliable source coverage. Ralbegen (talk) 14:02, 1 July 2019 (UTC)
@Sofia9: I would just note that if RSes start to exclude UKIP or ChUK from headline figures, I would be in favor of excluding them from the next table. As for SNP outpolling Greens, it's a bit WP:Crystal. Filinovich (talk) 14:42, 1 July 2019 (UTC)
In that case you'll be relieved to hear that we don't need a crystal ball - the SNP have outpolled the Greens within this calendar year. (I eagerly await you directing me to a "Wikipedia is not a history book" link or a "Wikipedia is not about what's going on in the present" link or a "look, just take it from me that Wikipedia isn't what you think it is" link.) Sofia9 (talk) 15:01, 1 July 2019 (UTC)

As long as the numbers are available and well sourced I see no reason why they shouldn't be reflected on the table. Anyone consulting the list should be able to find those numbers. Not including them just makes things more difficult for the end users of the article who will have to locate the numbers elsewhere. David Hastings 13:31, 1 July 2019 (UTC) — Preceding unsigned comment added by DRHastings (talkcontribs) Note: An editor has expressed a concern that DRHastings (talkcontribs) has been canvassed to this discussion.

@DRHastings: That isn't a reasonable argument on its own. As an admittedly weak analogy, YouGov often asks about WEP, but other sources tend not to, and it often isn't reported in the headline figures YouGov posts. While SNP and WEP definitely poll at very different proportions, it isn't worth getting into what threshold should be used to include/exclude a party. It's best to follow the format that RSes take. Further, because most sources do ask about National parties, footnotes have been made in the Others column that lists what proportion they received. The end user still has access to the values. There's still a graph for the lay user with SNP. Filinovich (talk) 14:43, 1 July 2019 (UTC)

The SNP and PC should be included separately. It seems ridiculous that the 3rd largest party (SNP) are grouped with ‘others’, this does not reflect accurately results of polling with them omitted. — Preceding unsigned comment added by Peeablo (talkcontribs) 14:40, 1 July 2019 (UTC) Note: An editor has expressed a concern that Peeablo (talkcontribs) has been canvassed to this discussion.

After hearing all these arguments, it's a decent point that polling tables don't need to be restricted to a smaller number of columns (comparing to other elections throughout Europe)... and the SNP is the third-largest party by number of seats... so I think I'd change my position to yes, include SNP and PC. Chessrat (talk, contributions) 16:09, 1 July 2019 (UTC)

  • Support removing SNP & PC - The SNP are only contesting 8.7% of the available constituencies and PC only 6.1% and these are either UK or GB polls. while they are a significant deal in their countries they are not significant overall. if the Yorkshire Party put up candidates in each of the Yorkshire constituencies they'd be fielding more than PC and a few short of the SNP. would we start including them (assuming they win a seat or 3)? There is a section which deals with Scotland and a section which deals with Wales which when dealing with the SNP and PC which are far more informative than the national polls.=> Spudgfsh (Text Me!) 18:43, 1 July 2019 (UTC)

There is no need to remove the SNP, Scotland is still part of the UK so no need to leave the third-largest party in the UK out Wilbo101 (talk) 19:04, 1 July 2019 (UTC) Note: An editor has expressed a concern that Wilbo101 (talkcontribs) has been canvassed to this discussion.

the point of this article is not about the number of MPs that a party has, it is about the level of support overall in the UK. The inclusion of the SNP and PC (and I'd say the same for Change UK) is that they, due to their regional nature, don't read across very well to overall VI polls.=> Spudgfsh (Text Me!) 19:11, 1 July 2019 (UTC)
Who decided this only one point of the page. Why should it be limited that point. The tables can be viewed in several ways. ~ BOD ~ TALK 20:31, 1 July 2019 (UTC)
  • Oppose removal of SNP and PC: the SNP are the third largest party in the commons, they cannot be excluded on any reasonable measure. I note that Bloc Québécoise are included at Opinion polling for the 2019 Canadian federal election. The template has plenty of space for multiple parties, although I can see some sense in removing UKIP and Change UK if we feel that it has become too large. Grinner (talk) 19:51, 1 July 2019 (UTC)

The arguments for excluding the SNP are ridiculous, such as IF there were a Yorkshire party they should not be included either. Well, there IS no 'Yorkshire Party', but IF there were one and it had Members of Parliament, then it should be included. ANY party that has Members of Parliament is in effect a 'national party' since Westminster is the NATIONAL parliament of the UK. I question including UKIP which HAS no MPs, however. But I see no legitimate reason for excluding the SNP or Plaid Cymru except pure Anglocentric bias. Wikipedia is supposed to be better than this. — Preceding unsigned comment added by 50.53.38.200 (talk) 21:01, 1 July 2019 (UTC)

There is a Yorkshire Party. Regardless, I don't see any reason why the number of MPs a party currently has should have any effect on whether it is included in a table of polling for a future election. By this argument, parties which are rarely, if ever, included in national polls, such as DUP, Sinn Fein, or even named Independents must be given a column in the table purely because they were able to get a seat sometime between the last election and now. I'm honestly tired of all the comments here saying there are no legitimate arguments on the other side when the commenters themselves clearly haven't even entertained any of the actual arguments. This isn't about Anglocentric bias. I couldn't care less which nations in the UK are most prominent. SNP and PlaidC should be excluded from the table, not because of some inherent prejudice I have, but because Reliable Sources don't report them in headline figures and will occasionally lump the two parties together as National. Filinovich (talk) 01:32, 2 July 2019 (UTC)
While we're correcting spurious inaccuracies, I should mention that Scotland and Wales are part of the Anglosphere too. Onetwothreeip (talk) 01:42, 2 July 2019 (UTC)

An Editor Fears that one side of the debate is trying to discredit the other side by repeated spurious claims of canvassing, and has given no substantial evidence to back this up. ~ BOD ~ TALK 08:18, 2 July 2019 (UTC)

The link to the blog post has been posted previously where the author asks readers to comment on this discussion, followed by several new or inactive editors commenting on the discussion, with the same perspective, purpose and arguments. Onetwothreeip (talk) 08:27, 2 July 2019 (UTC)
You'll notice I've only marked specific posts as canvassed. My only goal was to note which editors were likely to have come from that blog post due to incredible lack of edits. If they had more than a couple recent edits, I probably haven't tried to "discredit" them. Filinovich (talk) 10:31, 2 July 2019 (UTC)
Apologies not spurious, and i did not think to check recent activity. I still do not believe it is possible to know who came because of the blog and who came because they were watching. So personally I dont think they should be branded and disgarded as such. When the is a moot point like a RfC editors are often invited from several relevant Projects to get involved, because the more contributions can help make better decisions. I am unsure if a massive misdeed has been done here. ~ BOD ~ TALK 16:13, 2 July 2019 (UTC)
No worries. If the editors that I've marked object to it and can show they weren't canvassed, I have no issue removing the note. It's just to note that there is some concern that the person joined the discussion as a result of inappropriate canvassing. I doubt any users that have actually been canvassed will return to defend any of their arguments as most of them will have just come here to make the one post to vote and then let it slip from their minds. I imagine those who have come from other projects to participate in discussion will continue to productively do so. Filinovich (talk) 17:27, 2 July 2019 (UTC)
"If the editors that I've marked object to it and can show they weren't canvassed..." - That's an absurd statement. You'd be asking them to prove something that is literally impossible to prove. (Or are you going to require them to provide complete CCTV recordings of their movements over the last 48 hours?!) Sofia9 (talk) 01:07, 3 July 2019 (UTC)
@Sofia9: Perhaps I should have been more clear. Given that the only evidence I've seen to suggest most were canvassed is a lack of activity, I'd personally be satisfied by a reasonable explanation in most cases (e.g. User:Example messaged me, among other users, to participate in discussion) or even an actual attempt to contribute to discussion rather than rattle off a talking point that's been made already. I'm not asking for undeniable proof. Apologies for not clarifying that. Filinovich (talk) 11:29, 3 July 2019 (UTC)
I am not sure if anyone can request such confirmations from other editors, its not something I have witnessed before. ~ BOD ~ TALK 12:00, 3 July 2019 (UTC)
  • Oppose Bad idea for lots of reasons previously stated. But thanks. --Dweller (talk) Become old fashioned! 10:48, 2 July 2019 (UTC)
  • Oppose SNP has too may MPs to eliminate. In addition to reasons given above, 1% UK changes to SNP vote have a large impact on result of election. I ran + and - 1% change to SNP vote through the Electoral Calculus User-defined Prediction engine and roughly -1% SNP change results in -34 SNP MPs (thus others increase) and +1% SNP results in +19 SNP MPs. So by eliminating SNP we make our data much less useful in predicting possible election outcome. I personally think we should leave the table alone to the end of this year rather than changing mid-year. Rwendland (talk) 11:34, 2 July 2019 (UTC)
Electoral Calculus doesn't have the option for +/- 1% SNP at the national level, only in Scotland. Filinovich (talk) 11:50, 2 July 2019 (UTC)
You can create the effect by ticking the "Scotland prediction" box and upping/lowering the SNP "Scotland Support" by about 30% (the default has 3.1% national SNP support so 30% generates a rough 1% UK-wide change, which you can see in the Electoral Calculus prediction as it outputs SNP as % of UK) with corresponding rough opposite other party adjustments - I did say roughly above. My point wasn't to get accurate Electoral Calculus predictions, but to show SNP going up or down is very significant for the next GE outcome given our electoral system, so we should keep it in. Rwendland (talk) 14:16, 2 July 2019 (UTC)
I stand corrected. Though I would just note that much of the change in support for SNP between national polls is a result of sampling variance. There's no evidence from the current data that SNP's support has changed significantly since 2017 at the national level. I'm still in favor of listing the Scotland-only polls with SNP as the most prominent party because differences in numbers of those polls are often significant enough to rule out natural variance, SNP is never lumped with PC in those polls, and because of their large support in Scotland, they're always listed in headline figures by the pollster and reliable sources.
@Onetwothreeip: I wasn't aware of any canvassing - I've been a member since the mid noughties with a few thousand edits. I have been in a few of these types of discussion over the years and they tend to drive people into more and more entrenched positions. I agree that the number of people voting on one side or the other shouldn't necessarily dictate decision making. However, strength of feeling is valid and should be recognised - particularly when you are talking about representation of a group of people, the opinions of members of that group must surely carry more weight? Btljs (talk) 12:45, 2 July 2019 (UTC)

In the event that we decide to remove the SNP and PC (which I still think is the best idea) I do feel that this discussion is enough to show that there will be a constant battle to keep them off the table. This wikipedia is full of pages where a consensus was reached on a subject only for editors who weren't involved in the discussion or disagreed with the result to keep reverting the content. Regardless of my personal opinion on the matter I think removing the SNP and PC from these tables will be more trouble than it's worth. => Spudgfsh (Text Me!) 17:17, 2 July 2019 (UTC)

The battle seems only to be over the SNP, who we could conceivably keep in. There isn't such concern for Plaid Cymru, UKIP or Change UK. Onetwothreeip (talk) 06:36, 3 July 2019 (UTC)
Seriously? I really do not think anyone can draw that conclusion at this stage. Just because someone has not expressed an opinion either way on parts of the discussion, they can not be treated as a vote to support the removal of those parties. After treating no opinion is expressed as neither as a Support or a Object, the numbers supporting the status quo or no change are clearly higher than those wishing change. To make my position clear, I am for no removal of any party for a minimum to 2020, and certainly not for Plaid until after the next election. ~ BOD ~ TALK 12:00, 3 July 2019 (UTC)
The proposal isn't to remove Change UK or UKIP, in fact I am firmly against doing so. The proposal is to make a new table starting from July and not include columns for them, but to have them in the Other column. We shouldn't be cursed by happening to be in July rather than January, a split in the middle of the year is perfectly reasonable. SNP and Plaid are different, but I would merge them universally. Onetwothreeip (talk) 12:08, 3 July 2019 (UTC)
The time of year is irrelevant :) Maybe I should have been clearer, the is currently *no consensus* to remove UKIP & Change (or Plaid) columns and place their votes in a Others column. ~ BOD ~ TALK 13:54, 3 July 2019 (UTC)
when it comes to CHUK and UKIP I think the outcome of the next general election will make them totally irrelevant to the polls and they stop getting put onto the tables. That is especially true of CHUK who will disappear completely. I don't think it'll make too much of a difference if we wait until then or not. Some of these discussions may be better had after then. (and I expect that to be at some point this year)=> Spudgfsh (Text Me!) 18:27, 3 July 2019 (UTC)


OPPOSE taking out SNP and PC. Astonished to see someone advocating that and in the same breath saying leave CHUK and UKIP in. I can see an argument fir leaving those two out, but only after several months of very low polling. What I think is much more important than this discussion is having the Brexit Party column next to the main 3. It’s very annoying to have to scan across to that party’s column all the time. Almost hard work on a smartphone (be aware that most users of the net do so on smartphones now.) Boscaswell talk 23:59, 6 July 2019 (UTC)

The smartphone issue is my biggest concern. It takes much more effort to move data from a column into the Others column, than it takes to simply start a new table under new conditions. I think this is why we should be quite hasty in making new tables without the ChUK and UKIP columns. Onetwothreeip (talk) 02:41, 7 July 2019 (UTC)

Oppose - keep the status quo on the existing 2019 chart. UKIP and Change-UK were "big" once and that data is relevant to the context. Remove them when we (arbitrarily!) start a new table in 2020 or after any election later this year.Cutler (talk) 08:23, 9 July 2019 (UTC)

Error in chart II[edit]

Brexit Party appears to be overstated and Conservatives seem to be understated. Onetwothreeip (talk) 06:36, 3 July 2019 (UTC)

Brexit Party: (22+12+23+20+23+24+26+22+18+26+25+23+24+26+22)/15 = 22.4
Conservatives: (22+26+20+24+20+21+17+23+26+18+18+18+20+17+19)/15 = 20.6
Mélencron (talk) 12:48, 3 July 2019 (UTC)
You’re good at this! --Comment by Selfie City (talk about my contributions) 13:22, 3 July 2019 (UTC)
The chart has changed since I raised the concern. Onetwothreeip (talk) 13:33, 3 July 2019 (UTC)
Yeah, I added the latest two polls – when I self-reverted to the older version including the SNP/Plaid the last two polls got taken out and I didn't re-update the chart until now. Mélencron (talk) 14:57, 3 July 2019 (UTC)

Notes[edit]

  1. ^ "Change UK loses six of its 11 MPs". BBC News. 4 June 2019. Retrieved 4 June 2019.


Cite error: There are <ref group=lower-alpha> tags or {{efn}} templates on this page, but the references will not show without a {{reflist|group=lower-alpha}} template or {{notelist}} template (see the help page).