Talk:Opinion polling for the 2019 United Kingdom general election/Archive 1

Remove UKIP and Greens

UKIP and Greens both gained under 2% in the last general election now. I believe this is not sufficient to include them in the headline parties in this page.

If they are good g to continue being on the headline Plaid Cymru should also be included as they are the only other major party in the UK not to be included (keeping in mind most polls are GB only) — Preceding unsigned comment added by 109.175.185.120 (talk) 13:31, 10 June 2017 (UTC)

"Plaid Cymru are listed under Scottish National Party in most polls." Please read the main article before posting. Apache287 (talk) 13:43, 10 June 2017 (UTC)

We follow what reliable sources, in this case the pollsters, do. That's the Wikipedia way. If the pollsters stop listing UKIP, we drop them. If they keep including them, then we include them. Let's wait for some actual,polls before making any decisions! Bondegezou (talk) 14:59, 10 June 2017 (UTC)

Agreed. The situation in Scotland, Wales, and North Ireland is even more dire- the Greens and UKIP both got 0.2% of the Scottish vote, the Greens got 0.3% of the Welsh vote, and the TUV got 0.4% of the Northern Irish vote. But none of these parties should be removed from the table until we have evidence that most pollsters don't list their vote shares any more. Chessrat ^{(talk, contributions)} 22:31, 10 June 2017 (UTC)

As the Greens have an MP, one of only 4 parties in England to be so represented, to exclude them would be rather strange. Boscaswell talk 21:33, 11 June 2017 (UTC)

Results for the GE

Have they been calculated and the percentages rebased to exclude NI? If so, please undo the edit I made which changed all the figures. Thanks. Boscaswell talk 10:12, 11 June 2017 (UTC)

The GB figures keep being changed. For the record, the GB shares are: Con 43.5, Lab 41.0, LD 7.6, SNP 3.1, UKIP 1.9, Grn 1.7, Oth 1.2. That's worked out by UK figures minus NI figures. Unless people have an alternative calculation can this be left as it is now? FriendlyDataNerdV2 (talk) 15:08, 11 June 2017 (UTC)

No worries. Thanks. Boscaswell talk 21:30, 11 June 2017 (UTC)

Date of poll to put on the graph

Hi all, what date should I put on the graph for the date of the poll? For the last election I put it as the first day of the fieldwork, but in the table they were listed according to the last day of the fieldwork, which got a bit confusing for me. Any thoughts? Absolutelypuremilk (talk) 17:23, 12 June 2017 (UTC)

I'd go with the last day so that the graph and table will match up. Gaia Octavia Agrippa ^Talk 20:04, 12 June 2017 (UTC)

Grenfell Tower fire tragedy

This is a pivotal event in this country. The political implications are massive. Even Laura Kuennsberg is hinting at this. Much, much more significant than the last terror attack. I'd edit it in myself as a bar on the grid, but there might be disagreement, so I'll leave the idea here for now. Boscaswell talk 08:46, 16 June 2017 (UTC)

No. This is not a list of pivotal events for the country, nor a list of events with massive political implications. Consensus so far has been to keep notes to a minimum. There are many other articles that are better suited to tracking such significant and tragic events, and their impact on politics. Bondegezou (talk) 10:59, 16 June 2017 (UTC)

Please don't put this in now. If it turns out, in hindsight, that the fire plays a unique and genuinely pivotal role in British politics with a significant effect on polling, then there might be an argument for including it. But the only reason I can see that anyone would want to include it at this stage is because they want it to be perceived as a political game-changer. It might well be, but can't we wait and see? 31.108.74.94 (talk) 11:58, 16 June 2017 (UTC)

I agree, we have previously kept consensus to events which directly affected campaigns or the leadership of parties, e.g. the terror attacks which stopped campaigning. Absolutelypuremilk (talk) 12:53, 16 June 2017 (UTC)

I think we should wait and give it time, I think it quite probable when later polls out that we will see an affect on the polling, the reaction to the disaster seems to be quite negative towards to current government. The question is if it will have any long term affects on the polls and government. BSOleader (talk) 15:49, 16 June 2017 (UTC)

I bough to your superior judgement, oh guardians of the status quo. 🙄 You know, it's funny. And sad. Even those who have on the past been wise (about forecasting election results) who are the consensus of opinion on politicalbetting.com, and those pollsters such as ICM and IPSOS (whose opinion polling weightings were spectacularly wrong for June 8) simply could not accept that there was a massive groundswell of public opinion that would transform the political landscape of the UK. Reactions here to my suggestion reflect the same refusal to understand, to accept. "...the disaster seems to be quite negative towards (the) current Government." *coughs, splutters, shakes head* Boscaswell talk 12:24, 17 June 2017 (UTC)

I'm not prognosticating on the significance of the even, nor what may happen to the polls. By all means, add reliably sourced text on the political implications to the relevant articles (the one for the fire itself, perhaps Theresa May). However, it is not the purpose of the opinion polling table to interpret: it is merely a table of results.

Take a look at Nationwide opinion polling for the United States presidential election, 2016 for a comparison: there were numerous very significant events during the period covered by that polling (when the primaries were won, pussygate, James Comey's final email announcement). Count how many are noted as rows in the tables given. Bondegezou (talk) 14:06, 17 June 2017 (UTC)

Somehow I don't think having a go at the consensus on the page is going to see people switch to your point of view now is it. People don't agree with adding it because it's not an event with direct political consequences unlike the terror attacks during the campaign. If you can't accept that then that's fine, just don't be insulting about it. Apache287 (talk) 16:19, 17 June 2017 (UTC)

I think you have failed to understand what the page is about. It is not a list of significant events in UK politics. It is not a discussion of what might or might not change voters' opinion. It is a list of opinion polls conducted within the UK, both on voting intention in the next general election and opinion on the performance of the leaders of the major parties. The only logical reason the change in leaders is included at all is that polling information on voters' opinion on the leaders is included in the page, so it makes sense to also contain information on when the leaders have changed. If you wish to compile a list of events significant in UK politics between the 2017 GE election and the next GE, then you should make a page specifically for that. This page is simply for listing information contained in opinion polls. 192.41.131.250 (talk) 10:35, 21 June 2017 (UTC)

Graph of polls since 2010

As there's a disagreement over the graph that is on the page, I think it's best to take it to a discussion here. My personal preference would be to remove it. In my view this page is designed to help inform people of the political landscape post-June 8th. As a result, a graph which only shows polls prior to that date isn't very informative to that end. FriendlyDataNerdV2 (talk) 15:07, 25 June 2017 (UTC)

I agree that any polling graph for this article should only include post-8 June polls. Mélencron 15:18, 25 June 2017 (UTC)

I agree, maybe this graph could be re-inserted in a smaller form somewhere further down the article, but I don't think it should be re-inserted where it was. Absolutelypuremilk (talk) 15:18, 25 June 2017 (UTC)

As creator of the graph I would suggest perhaps keeping it up until we can create a graph for the next election. I feel if we keep it up it can provide a useful background regarding the LibDem decline, UKIP bump and reemergence of 2 party politics. It can be useful in detailing Labour's surge and may help provide context for the polls which come out following the 2017 general election. Views?TomPumpkin69 (talk) 17:15, 25 June 2017 (UTC)

I agree with keeping it until enough polls emerge for the current parliament to make it worth creating another, then relegating it to a lower section or removing it entirely. Apache287 (talk) 22:39, 28 June 2017 (UTC)

I do not think it should be the headline graph, this article is about Opinion polling for the **next** United Kingdom general election, not the last two. If it is used I agree with "maybe this graph could be re-inserted in a smaller form somewhere further down the article <clearly described>, but I don't think it should be re-inserted where it was".-- BOD -- 23:04, 28 June 2017 (UTC)

I don't know if it would be possible to have a Opinion polling for United Kingdom general elections page that covers general opinion polling for general elections and to enable polls from before 1992 to have a location on Wikipedia without having to create a separate page for them - quite similar to what I did with the Opinion polling on Scottish independence page. It may be the most suitable location for a graph like this. Clyde1998 (talk) 14:12, 8 July 2017 (UTC)

Creating Opinion polling for United Kingdom general elections would be fine with me. Absolutelypuremilk (talk) 16:53, 8 July 2017 (UTC)

Ok, does someone want to arrange scaling it down and putting it in somewhere smaller? If so where would be the best place?TomPumpkin69 —Preceding undated comment added 15:39, 30 June 2017 (UTC)

@TomPumpkin69: The seven-year and 25-year graphs are a good idea, but appear to have a variable time scale on the x-axis. I think it would be better to have a constant time scale, with the years labelled (as all other opinion poll graphs are); would you be able to change this please? Chessrat ^{(talk, contributions)} 13:24, 8 July 2017 (UTC)

@TomPumpkin69: I second this. If we take the 25-year graph, it shows 1992 to 2010 as roughly the same size as 2010-15. I think that it gives too much emphasis on that time period, as well as giving too much emphasis on polls from the end of the campaign (as that's when there's the most amount of polling conducted). I don't know if there would be any benefit to adding a line for 'other parties' to the graph, should this information be available as well. Clyde1998 (talk) 14:12, 8 July 2017 (UTC)

@TomPumpkin69: I agree, the time axis on this plot seems very strange. Absolutelypuremilk (talk) 14:10, 10 July 2017 (UTC)

Hi all, would you be able to tell me how to do some sort of axis on Excel please? im very new to Excel and tbh it is a miracle I came out with this graph in the first place! Thanks!TomPumpkin69 (talk) 15:11, 25 July 2017 (UTC)

Including NI parties in main table

I separated the Northern Ireland parties from the Other parties in this edit– other UK parties are very different from Northern Irish parties, and I don't think it makes much sense to conflate the two into one figure.
But, this was then reverted in this edit with the edit summary "Not something we have really done before so please raise it in the Talk page".
Well, of course it's not something that has been done before because polls have never included Northern Irish parties before, so we're in an unprecedented situation. That being said, I didn't expect my edit to be at all controversial, but I'm raising it on the talk page anyway as requested. So, what do you think? Should the Northern Ireland parties be separated from the Other parties? Chessrat ^{(talk, contributions)} 16:08, 3 July 2017 (UTC)

I don't think it is helpful - the % figure doesn't tell us anything about who is likely to win seats. It also doesn't tell us anything, apart from the ratio of turnout between NI and the rest of the UK, because (independents aside) NI voters cannot vote for GB parties so the figure shouldn't change too much. Absolutelypuremilk (talk) 16:54, 3 July 2017 (UTC)

It does tell us something: how popular other Great Britain parties are. With the current table layout, the latest Survation poll says that 6% of voters support "Other", whilst the latest Opunium poll says that 1% of voters support "Other". The former figure includes Northern Ireland while the latter does not, explaining the large discrepancy. But a reader cannot work out from the table what proportion of Survation's 6% is from Northern Irish parties and what proportion is from other Great Britain parties. Thus, with the current table layout the "Other" column is rather useless, as the Other statistics from one pollster can't be compared with the Other statistics from another pollster. Chessrat ^{(talk, contributions)} 19:08, 3 July 2017 (UTC)

I understand Chessrat's position and see the logic to it. However, I presume that, over time, the Survation polls including Northern Ireland will be very much in a minority, as with the previous article, and a separate NI parties column will then be overkill. We need some sort of note to explain this anomaly, but not a whole column, I think. If I'm wrong in that presumption, I reserve the right to change my mind!

Isn't there some way to extract GB-only data from Survation's polling tables?

I also note that the Conservatives and the Greens do stand in NI, even if they get very few votes, and UKIP used to very recently. Bondegezou (talk) 22:04, 3 July 2017 (UTC)

Regarding GB-only polling data: yes, it can easily be calculated from the detailed tables. But I'm not sure whether including such figures would be a violation of WP:OR. Anyway, here's the GB-only data. Chessrat ^{(talk, contributions)} 23:03, 3 July 2017 (UTC)

Date(s) conducted	Polling organisation/client	Sample size	Con	Lab	Lib Dem	SNP	UKIP	Green	Others	Lead
28–30 Jun	Survation	989	42%	41%	7%	2%	2%	2%	3%	1%
17 Jun	Survation/Good Morning Britain	977	42%	45%	6%	3%	2%	1%	2%	3%
10 Jun	Survation/Mail on Sunday	1,025	39%	45%	7%	3%	3%	2%		6%

Thanks for raising it as requested, it's appreciated. I agree it's an issue but thought people might want to weigh in first.

Bondegezou there is indeed a way to extract the GB figures, as Survation issue tables showing the individual sample breakdowns for each nation of the UK. So it's a simple matter to just go into the tables, subtract the NI figures from the overall figures, and recalculate the percentages. I can do that for Survation polls going forward if that's what people think works? (So for example, the most recent poll would change to Con 42%, Lab 41% - virtually unchanged from the June 8th figures, if a slight swing to Labour)

Currently the impression is given that there has been a decline in both major parties' support (or a Labour backwards shift) when the reality is that within Great Britain, which is what we measure in the polling table, the landscape is virtually unchanged. It is an eternal source of frustration that some pollsters include NI and some don't, because it results in small but annoying alterations to the headline figures. Easy to solve though if people are ok with me adjusting the table figures and adding a note explaining that. FriendlyDataNerdV2 (talk) 23:06, 3 July 2017 (UTC)

Oh just seen the latest edit. Nice work, I'd be happy with that. FriendlyDataNerdV2 (talk) 23:06, 3 July 2017 (UTC)

Presuming Survation and their inclusion of NI remain in the minority, I'm in favour of calculating GB figures from the Survation polls for comparability and I think that comes under WP:CALC. Another idea would be to include a supplementary table just of the full, UK Survation results, with an additional column or columns for NI parties.

If the main table remains dominated by Survation, then I think Chessrat's solution is best. Bondegezou (talk) 08:07, 4 July 2017 (UTC)

So... are we going with giving GB-only figures for Survation? Bondegezou (talk) 22:25, 8 July 2017 (UTC)

A bit of a non-problem, in my view; it's something that's easily dealt with using a footnote and there isn't a good reason to change it or recalculate and use figures which quite literally aren't published. The difference resulting from the inclusion and exclusion of NI parties seems too minor to necessitate this. Mélencron (talk) 23:18, 8 July 2017 (UTC)

There is good reason, which is that if we don't do it we aren't comparing like-with-like. If people are very firm on including Northern Ireland in the polling then they should be shown in a seperate table to GB-only polls. How can we be indicating the GB results at the bottom of the table, only to show a UK-wide poll in the table itself? It gives the impression that parties are up or down in GB when they actually aren't. It's misleading. FriendlyDataNerdV2 (talk) 23:24, 16 July 2017 (UTC)

I think it's more misleading to use figures which aren't actually explicitly published and distributed by pollsters – the figures aren't necessarily perfectly one-to-one, but the differences produced by the inclusion or exclusion of Plaid Cymru, of parties in Northern Ireland, etc. are insignificant enough that they can essentially simply be ignored. The recalculation of figures already pushes the boundaries of WP:OR (given that only Wikipedia would be publishing these as the topline numbers), and also because of the fact that these are hardly the greatest methodological inconsistencies which would result in divergences or significant house effects between pollsters (as we saw in June). The simplest solution is to use the published toplines and stick with them, honestly – as Survation is the only pollster that includes NI parties, just append an {{efn}} whenever they appear in the table, and maintain the previous solution when it comes to YouGov figures including Plaid Cymru. There isn't a need to make this needlessly complex. Mélencron (talk) 23:42, 16 July 2017 (UTC)

I'm with Melencron on this. I'll add that NI respondents cannot be neatly removed from the sample, as both UKIP and the Conservatives compete in NI. Thus removing those that expressed intention to vote for the NI parties w/o removing NI UKIP and Conservative voters might result in over-estimating GB-wide UKIP and Conservative support (it's probably not a very serious concern given the low support that UKIP and Con enjoy in NI, but it is enough to trigger OR concern). Rami R 09:17, 17 July 2017 (UTC)

We do remove NI Con and UKIP, that's the point...all NI respondents are excluded. FriendlyDataNerdV2 (talk) 19:46, 17 July 2017 (UTC)

And the point I'm trying to make is that we are including the Great Britain results in the table...and then inviting readers to compare them to the latest poll, which might not be a GB-only poll. If we're going to put polls covering different geographic areas in the table, without adjusting them, we can't have that set of results because people will assume they represent changes from that GB result when they simply don't. FriendlyDataNerdV2 (talk) 19:50, 17 July 2017 (UTC)

Indeed. There's no OR concern: we're removing the NI data from the polling.

We should avoid confusing the reader and mixing UK and GB polls does that. As I've said, I'm in favour of showing GB-only results for UK polls. If we don't, we should clearly mark out UK polls, or perhaps put them in a separate table. Bondegezou (talk) 22:36, 17 July 2017 (UTC)

Thanks Bondegezou, I agree. To give an example, the UK-wide Survation figures for 30th June put the Tories on 41% and Labour on 40%, -1% since the election compared to the GB figures for both parties. But compared to the UK figures, it's unchanged (with a slight decline in the Tory share). In GB only, Labour is on 41%, unchanged since the election. In other words, presenting UK wide figures as GB figures has the effect of saying that Labour's vote share has fallen back since the election - when in fact it has remained unchanged. Another example: Survation on 16-17 June has Lab 44%, Con 41% UK-wide, but in GB only it's Lab 45%, Con 42%. Due to rounding it can mean the difference of a percentage point for the two biggest parties, and in a finely balanced 2-party system, that matters. Bondegezou is right - let's keep taking the NI respondents out. It's not hard and the tables being released makes it really easy. A note is visible to explain it. FriendlyDataNerdV2 (talk) 00:38, 18 July 2017 (UTC)

Should we add Plaid Cymru to the table?

The table currently shows six parties – Con, Lab, LD, SNP, UKIP, Green – plus "other". But I've just noticed that all of the pollsters seem to include seven options – those six parties plus Plaid Cymru – in their tables. So, if we're going to be consistent with the data that reliable sources use, would it not make more sense to add Plaid Cymru to the table? I thought I'd raise this issue on the talkpage before actually editing the article. Chessrat ^{(talk, contributions)} 00:46, 4 July 2017 (UTC)

Agreed. Bondegezou (talk) 08:08, 4 July 2017 (UTC)

If all the pollsters do it, then yes we should include it. So far Panelbase, Opinium and Survation have posted polls including PC. BMG, ICM and Ipsos also did it in previous polls. However, YouGov and Comres did not have a separate figure for PC, so it might be confusing including it for some but not others? Maybe we should wait and see whether this changes in the new polls. Absolutelypuremilk (talk) 09:57, 4 July 2017 (UTC)

On the previous article, we didn't include Plaid. We added a footnote for those results from, e.g., YouGov where Plaid and SNP were combined: that is, we listed the figure under the SNP but with a note saying it include Plaid too. Bondegezou (talk) 13:08, 4 July 2017 (UTC)

As yet all the polls thus far have included a separate figure for Plaid. That said, it's always below 1% - unsurprising given they only stand in Wales. Do we want to show it to 1 decimal point or show as 1%/0% depending upon the rounding down or rounding up? FriendlyDataNerdV2 (talk) 15:46, 4 July 2017 (UTC)

I think rounding to 0% is odd. Better to say "<1%". Bondegezou (talk) 15:53, 4 July 2017 (UTC)

Another option would be to use at least one significant figure for all figures, thus including a decimal for figures less than 1%, such as 0.6%. This might give an impression of false accuracy, though. Chessrat ^{(talk, contributions)} 16:39, 4 July 2017 (UTC)

Due to Yougov's polls combining SNP and PC figures, it would probably be best to put SNP and PC next to each other in the table. It goes against the rule we normally use, of ordering parties by their vote share in the previous election, but it would allow the SNP and PC columns to be combined for Yougov's polls, thus avoiding the need for "also includes Plaid" footnotes. Like this: Chessrat ^{(talk, contributions)} 16:39, 4 July 2017 (UTC)

Date(s) conducted	Polling organisation/client	Sample size	Con	Lab	Lib Dem	SNP	PC	UKIP	Green	Others	Lead
	Hypothetical YouGov poll					[Combined SNP and PC data]
8 Jun	General Election results (excluding Northern Ireland)	–	43.5%	41.0%	7.6%	3.1%	0.5%	1.9%	1.7%	0.7%	2.5%

I like the column solution. I fear one decimal place is misleading percussion. Bondegezou (talk) 10:34, 5 July 2017 (UTC)

(That should, of course, say "precision"...) Bondegezou (talk) 21:04, 5 July 2017 (UTC)

Reconsideration

With respect to all the contributors so far, I think we should look again at this issue. When I first saw the revised page, I thought that it looked rather strange and out of keeping with the page's normal look. Plaid Cymru is a significant party in Wales, but it is very small in GB terms. Its vote share is around 0.5%, which is smaller than both UKIP and the Green Party and less even than the residual "Others". The problems stated above are genuine: some pollsters don't report it separately, and those that do might just show a "*" (eg ICM poll 6-7 Jun). On the page's table it will probably always show a value of either 0% or 1%, which contains very little useful information for readers of the page. And its position next to the SNP has the unfortunate consequence of breaking the party ordering in terms of vote share at the last general election. I would suggest that we consider whether the Plaid Cymru polling results are best displayed in the Wales-specific polling section where they would be meaningful, and not shown in the GB table where they are not meaningful. Mwbaxter (talk) 08:22, 10 July 2017 (UTC)

Agree with the above – given its miniscule vote share nationally, it makes more sense to report it only in Wales-specific polls. The old footnote describing the combined SNP+PC vote share should suffice for YouGov polls. Mélencron (talk) 13:33, 10 July 2017 (UTC)

I agree, there isn't much point showing either 0% or 1% for every poll which includes them. Absolutelypuremilk (talk) 14:09, 10 July 2017 (UTC)

I agree it looks a bit odd, but as all pollsters contain Plaid Cymru figures in their GB-wide polls, shouldn't we follow their practice? Chessrat ^{(talk, contributions)} 13:24, 14 July 2017 (UTC)

Ultimately, I agree: we should show what the pollsters report. But it's a marginal point: I can live with either answer. Bondegezou (talk) 14:31, 14 July 2017 (UTC)

Chessrat: YouGov don't show them, so it's not 'all'. FriendlyDataNerdV2 (talk) 00:07, 16 July 2017 (UTC)

Pollsters are divided on this question. It's not true that "all" pollsters show Plaid. Plaid is not shown separately by three pollsters (YouGov, Ipsos-MORI, ComRes) though it is shown by four others (ICM, Survation, Opinium, Panelbase). The pollsters' tables contain a lot of information, including the BNP and Mebyon Kernow. In my opinion it is valid for the Wiki page to select which poll data is relevant to users. Why should we clutter the page with a partially complete column which can realistically only show zero or one? Mwbaxter (talk) 13:43, 31 July 2017 (UTC)

I think we should generally err on the side of inclusiveness and follow who the pollsters report. However, Mwbaxter makes a good point. I can't see any point including the BNP or MK either. For comparison, I note that Opinion polling for the German federal election, 2017 does not include the Pirates Party, even though some pollsters do report them, which supports Mwbaxter's position. Bondegezou (talk) 14:16, 31 July 2017 (UTC)

Approval ratings

Is there any interest in adding approval ratings for Corbyn and May? Its a different question than preferred lead, a positive/negative question, but it is still a type of opinion polling. Gaia Octavia Agrippa ^Talk 00:16, 5 July 2017 (UTC)

I'm a bit concerned about the idea – I've had similar ideas regarding approval polling but in the end decided not to because of significant differences in question wording which might produce considerably different results. Mélencron 01:06, 5 July 2017 (UTC)

I support the addition of approval ratings to the article. If there are any significant differences in wording between pollsters, then the article can explain the details of this. Chessrat ^{(talk, contributions)} 08:52, 5 July 2017 (UTC)

Many commentators warn of the different wording used for approval ratings. If included, I'd rather they were split by polling company so we didn't suggest to readers that they can directly compare numbers from one question to another. Bondegezou (talk) 10:36, 5 July 2017 (UTC)

Ah I was going to raise this! Great minds think alike and all that. As some have said above though, questions differ from pollster to pollster. There are generally 5 types of questions asked:

• Well/Badly (YouGov up until 1st June 2017)
• Favourable/Unfavourable (YouGov 12th June 2017, ORB, ComRes,
• Good job/bad job (Panelbase)
• Approve/disapprove (Opinium, GfK)
• Satisfied/Dissatisfied (Ipsos MORI)

I'll draft some ideas for how that might look and post them here. Also worth saying that some pollsters ask approval ratings not only for Corbyn/May, but also for the other party leaders. Though as UKIP and the LDs currently have an interim leader this doesn't apply just yet FriendlyDataNerdV2 (talk) 17:08, 5 July 2017 (UTC)

I also wonder whether a spinoff article would be more appropriate. This one will soon fill up. Bondegezou (talk) 17:32, 5 July 2017 (UTC)

Thank you all for your input. Thank especially to FriendlyDataNerdV2! Is there anyway of having Corbyn and May in the same table? That would take up less space and allow easier comparison. Gaia Octavia Agrippa ^Talk 22:38, 6 July 2017 (UTC)

I have a couple proposals below as to how to display them within combined tables and not isolate them by pollster or party leader – my own preference would be for the second of these proposals, being more compact. Mélencron (talk) 00:16, 7 July 2017 (UTC)

I like the 2nd proposal a lot, Mélencron - though my personal preference would be to separate out the 2nd proposal into May tables and Corbyn tables. That way we're not stretching out the screen too much and giving readers too much information at once (people will want to see a quick figure for May and Corbyn separately, comparing them will only require scrolling down) but we're not sorting them by pollster. So like this (see below - "FriendlyDataNerd's 2nd proposals") FriendlyDataNerdV2 (talk) 00:36, 7 July 2017 (UTC)

The main reason that I wanted to include both main parties' leaders within the same table was to serve as a point of comparison, but I can't really see a more compact solution, so it might be better to isolate the tables after all. I've created a third version below separating it into two tables. Mélencron (talk) 00:44, 7 July 2017 (UTC)

I love it! Awesome work on this. FriendlyDataNerdV2 (talk) 01:44, 7 July 2017 (UTC)

@Gaia Octavia Agrippa, Chessrat, and Bondegezou: Any thoughts? I'm pretty happy with the result. Any objections to adding either version to the article? (btw, @FriendlyDataNerdV2:, if you want to reformat it to the standard wikitable formatting on this article, that would be fine – I just reformatted it so it was easier for me personally to deal with, but it could be adjusted for the sake of consistency.) Mélencron (talk) 15:36, 7 July 2017 (UTC)

I'm not keen on combining Good/Bad, Approve/Disapprove and Favourable/Unfavourable in one table because they are not directly comparable. I would rather have separate tables: all the Good/Bads in one, all the Approve/Disapproves in another, and so on. Bondegezou (talk) 17:03, 7 July 2017 (UTC)

I think that's a bit beside... beyond?... the point, though – even for apparently similar questions (say, good/bad and approve/disapprove), the question wording can still vary and therefore produce similar results; I threw this together based on polls in France, for example – though some approval questions seem identical on the surface ("Favorability of actions" and "Positive/negative opinion", for instance, or "addressing problems" versus "resolving problems"), differences in question wording and pollsters' house effects still have enough of an impact that if you're concerned about comparability, it's necessary to essentially create a table for every pollster, which is a needless waste of space. I think the method I proposed (noting question wording in the table with a footnote for the specific question) suffices to resolve this, but I would say that, wouldn't I? I don't have a better solution. Mélencron (talk) 18:15, 7 July 2017 (UTC)

I too am concerned about the issue of grouping the differently worded questions. While they might all be aiming for positive/negative, they are still different questions and different answers. For example, if a poll asked if you a favourable or unfavourable opinion of Manchester United is that really comparable to a poll asking if you think the players are currently doing a good or bad job? I would say no. I would support a version of Mélencron's second proposal, which combines both May and Corbyn, but with a separate table for each question. Gaia Octavia Agrippa ^Talk 18:30, 7 July 2017 (UTC)

I support using Mélencron's first proposal. It's good at showing all the data in one table, while clearly distinguishing the different question wordings. Chessrat ^{(talk, contributions)} 21:21, 7 July 2017 (UTC)

I would support Bondegezou's suggestion of a spin-off article for consideration of the leaders. If this parliament lasts a full five years, we're going to have an awfully long article. If left in, I suggest we need to rewrite the introduction as it says the page is about voting intention. --Wavehunter (talk) 08:49, 8 July 2017 (UTC)

FriendlyDataNerd's proposals

[Original proposal removed to avoid cluttering up the discussion] FriendlyDataNerdV2 (talk) 00:37, 7 July 2017 (UTC)

FriendlyDataNerd's 2nd proposal

Theresa May

Date(s) conducted	Polling organisation/ client	Sample size	Question wording	Theresa May
				Approve	Disapprove	Neither	Don't know	Net approval
16–21 Jun 2017	Panelbase	5,481	Good/Bad	28%	45%	27%	—	–17%
11–12 Jun 2017	YouGov	1,729	Favourable/Unfavourable	29%	63%	—	8%	–34%

FriendlyDataNerdV2 (talk) 00:36, 7 July 2017 (UTC)

Jeremy Corbyn

Date(s) conducted	Polling organisation/ client	Sample size	Question wording	Jeremy Corbyn
				Approve	Disapprove	Neither	Don't know	Net approval
16–21 Jun 2017	Panelbase	5,481	Good/Bad	45%	27%	28%	—	+17%
11–12 Jun 2017	YouGov	1,729	Favourable/Unfavourable	46%	46%	—	8%	0%

FriendlyDataNerdV2 (talk) 00:37, 7 July 2017 (UTC)

Mélencron's proposals

First version

Date(s) conducted	Polling organisation/ client	Sample size	Theresa May									Jeremy Corbyn
			Fav	Unfav	Good	Bad	Approve	Disapprove	Neither	Don't know	Net approval	Fav	Unfav	Good	Bad	Approve	Disapprove	Neither	Don't know	Net approval
30 Jun–3 Jul 2017	ICM/The Guardian	2,044			28%	54%					–26%			44%	35%					+9%
27–29 Jun 2017	Opinium/Observer	2,010					31%	51%	18%		–20%					42%	38%	20%		+4%
16–21 Jun 2017	Panelbase/Sunday Times	5,481			28%	45%			27%		–17%			45%	27%			28%		+17%
11–12 Jun 2017	YouGov	1,729	29%	63%						8%	–34%	46%	46%						8%	0%

Second version

Date(s) conducted	Polling organisation/ client	Sample size	Question wording	Theresa May					Jeremy Corbyn
				Approve	Disapprove	Neither	Don't know	Net approval	Approve	Disapprove	Neither	Don't know	Net approval
30 Jun–3 Jul 2017	ICM/The Guardian	2,044	Good/Bad	28%	54%			–26%	44%	35%			+9%
27–29 Jun 2017	Opinium/Observer	2,010	Approve/Disapprove	31%	51%	18%		–20%	42%	38%	20%		+4%
16–21 Jun 2017	Panelbase/Sunday Times	5,481	Good/Bad	28%	45%	27%		–17%	45%	27%	28%		+17%
11–12 Jun 2017	YouGov	1,729	Favourable/Unfavourable	29%	63%		8%	–34%	46%	46%		8%	0%

Third version

Theresa May

Date(s) conducted	Polling organisation/ client	Sample size	Question wording	Approve	Disapprove	Neither	Don't know	Net approval
30 Jun–3 Jul 2017	ICM/The Guardian	2,044	Good/Bad	28%	54%			–26%
27–29 Jun 2017	Opinium/Observer	2,010	Approve/Disapprove	31%	51%	18%		–20%
16–21 Jun 2017	Panelbase/Sunday Times	5,481	Good/Bad	28%	45%	27%		–17%
11–12 Jun 2017	YouGov	1,729	Favourable/Unfavourable	29%	63%		8%	–34%

Jeremy Corbyn

Date(s) conducted	Polling organisation/ client	Sample size	Question wording	Approve	Disapprove	Neither	Don't know	Net approval
30 Jun–3 Jul 2017	ICM/The Guardian	2,044	Good/Bad	44%	35%			+9%
27–29 Jun 2017	Opinium/Observer	2,010	Approve/Disapprove	42%	38%	20%		+4%
16–21 Jun 2017	Panelbase/Sunday Times	5,481	Good/Bad	45%	27%	28%		+17%
11–12 Jun 2017	YouGov	1,729	Favourable/Unfavourable	46%	46%		8%	0%

Graph

Here is the graph, doesn't seem to be worth putting it in yet. Absolutelypuremilk (talk) 22:00, 6 July 2017 (UTC)

Opinion polling for the 2022 UK general election. The moving average is based on the last ten polls.

I would leave it a month or two tbh. G-13114 (talk) 23:19, 6 July 2017 (UTC)

Is there a way of the markers on the chart showing the actual dates of the poll (as opposed to publication date)? So if field work was 3-5 July it would show a dash covering those three days. Btljs (talk) 10:55, 7 July 2017 (UTC)

Not that I know of! Absolutelypuremilk (talk) 11:52, 7 July 2017 (UTC)

Thanks for doing the graph. I agree it doesn't seem worth putting it in yet. Let's get to at least 10 polls...? Bondegezou (talk) 12:17, 7 July 2017 (UTC)

Would it be possible to scale the graph to the latest possible date of the next general election (i.e., 5 May 2022), in the vein of something like this? Mélencron (talk) 02:39, 9 July 2017 (UTC)

This was how I had it at the start of the 2015-2017 polling article, but there was a consensus (which I agree with) that this made it harder to see the trends and that the empty space didn't add anything. Absolutelypuremilk (talk) 07:24, 9 July 2017 (UTC)

I'll add the new graph once the full Opinium figures come out. Absolutelypuremilk (talk) 09:23, 20 August 2017 (UTC)

Hi, I've looked at the graph and I see that Labour is listed at 40% in the last two polls even though it has only been at 40% once recently. Is this correct? AureIio (talk) 18:58, 1 October 2017 (UTC)

Thanks for spotting this - I had 42 for Cons and 40 for Lab instead of the other way around! Now fixed. Absolutelypuremilk (talk) 19:16, 1 October 2017 (UTC)

Looked at the graph again and it has Labour at 37% when it hasn't polled at that level since June 2017. :) AureIio (talk) 19:35, 16 December 2017 (UTC)

Looks like 41% to me. Mélencron (talk) 19:39, 16 December 2017 (UTC)

There is a red dot at 37%. I think Con/Lab are swapped from the BMG poll AureIio (talk) 20:23, 16 December 2017 (UTC)

Yeah, that looks to be the case. @Absolutelypuremilk: could this be resolved? Mélencron (talk) 20:27, 16 December 2017 (UTC)

Thanks for spotting this, now fixed. Absolutelypuremilk (talk) 10:23, 17 December 2017 (UTC)

England polls

In the 2015 article, some users helpfully uploaded the English-only results from many GB-wide polls. This was eventually taken out of the 2017 article because (as I argued) barely any polls had been added to it during the entire two year period of 2015-17 and so it wasn't informative at all to readers, and the amount of work required to go through all of the polls was beyond most people. However, we can start again with this one and I think it'd be useful: the samples for England are sizeable. However, as it's not consistent for us to include them, I thought I'd check here. For context, it would look like this. (Thus far, only YouGov have not published English data). FriendlyDataNerdV2 (talk) 15:46, 13 July 2017 (UTC)

Date(s) conducted	Polling organisation/client	Sample size	Con	Lab	Lib Dem	UKIP	Green	Others	Lead
30 Jun–3 Jul	ICM/The Guardian	1,331	41%	45%	6%	4%	3%	0%	4%
28–30 Jun	Survation	626	42%	44%	7%	3%	2%	3%	2%
27–29 Jun	Opinium/Observer	1,146	40%	47%	5%	5%	2%	0%	7%
16–21 Jun	Panelbase/Sunday Times	3,616	42%	47%	6%	3%	2%	0%	5%
17 Jun	Survation/Good Morning Britain	740	45%	45%	6%	2%	1%	1%	Tied
14 Jun	Tim Farron resigns as leader of the Liberal Democrats<ref>{{Cite web|url=https://www.theguardian.com/politics/2017/jun/14/tim-farron-resigns-liberal-democrats-fresh-face-struggled-with-questions-over-his-christianity|title=Tim Farron resigns: Liberal Democrats' fresh face struggled with questions over his Christianity|last=|first=|date=|website=|archive-url=|archive-date=|dead-url=|access-date=}}</ref>
10 Jun	Survation/Mail on Sunday	686	41%	47%	7%	4%	—	1%	6%
9 Jun	Paul Nuttall resigns as leader of the UK Independence Party
8 Jun	General election results (England only)<ref>{{cite web |url=http://www.bbc.co.uk/news/election/2017/results |title=Results of the 2017 General Election |publisher=BBC |accessdate=10 June 2017}}</ref>	27,165,789	45.4%	41.9%	7.8%	2.1%	1.9%	0.9%	3.5%

That's fine with me! Absolutelypuremilk (talk) 16:01, 13 July 2017 (UTC)

I am wary (which seems to be what I normally say here!). These are not polls of England. They haven't, as far as I understand it, been demographically matched at the England level the way the poll is at the GB-level. They are sub-samples, albeit large sub-samples. If included, they should be clearly introduced for what they are. Bondegezou (talk) 17:07, 13 July 2017 (UTC)

I don't support the inclusions of subsamples which themselves aren't necessarily demographically representative. Mélencron (talk) 17:15, 13 July 2017 (UTC)

I am fine with attaching a note to explain that it is a sub-sample, but England represents 86.5% of the people surveyed so any demographic changes are unlikely to make a big difference, certainly at the precision these polls work to. Absolutelypuremilk (talk) 17:32, 13 July 2017 (UTC)

I take the point that as subsamples that are a large majority of the original sample, there's less of an issue (not like including Scottish or Welsh subsamples), so I'm not going to die in a ditch on that point. But, if included, we need sufficient explanation to introduce them. Bondegezou (talk) 14:30, 14 July 2017 (UTC)

Fair point - maybe if we add this explanation before the table:

"Some opinion pollsters include England-only data in their data tables. Unless indicated otherwise below, these are not England-only polls but only the English data from UK-wide surveys". FriendlyDataNerdV2 (talk) 14:41, 15 July 2017 (UTC)

Leadership approval opinion polling for the next United Kingdom general election

Per the discussion above, this now exists: Leadership approval opinion polling for the next United Kingdom general election

"TBC?"

For several parties, the term "TBC" is showing up in the data table. In most contexts, this means "to be confirmed," but I have no idea what it means here. Absent of a footnote explaining it, other readers might be similarly confused. Is there any way that the term can be defined, and that an appropriate explanatory footnote (or similar device) can be placed in the article? schetm (talk) 02:23, 27 September 2017 (UTC)

I believe TBC is included in cases where a newspaper reports some of the results of a poll but omits details for the minor parties, and the poll itself is not available yet. 88.108.74.48 (talk) 18:37, 3 October 2017 (UTC)

Yes, but what does it actually mean? What do the letters "TBC" stand for? schetm (talk) 00:35, 4 October 2017 (UTC)

'To be confirmed'. The data for those parties has yet to be confirmed. Not that strange. FriendlyDataNerdV2 (talk) 12:24, 6 October 2017 (UTC)

Subscripts

Subscripts should be included as in my opinion they are polls. As the polling agencies ask people from different areas who would they vote for and then they make a calculation for uk wide. Also is used in regional polling for England on Opinion polling for 2017. 86.130.188.95 (talk) 14:13, 9 October 2017 (UTC)

Sub samples are not proper polls - they are not weighted demographically or politically and are not reported by media outlets for that reason. The 2017 regional polling from YouGov was different in that YouGov themselves reported them as proper polls. They are not doing so with these. Only full, weighted polls of regions (e.g. London polling) should be included here. FriendlyDataNerdV2 (talk) 17:37, 9 October 2017 (UTC)

Cabinet Minister resigning

While they certainly resigned in controversial circumstances, I don't think they warrant inclusion on the opinion polling table, what does everyone else think? Tidus mi2 (talk) 09:53, 9 November 2017 (UTC)

I tend to agree. If we add events like this it becomes difficult to draw the line on what are significant enough for inclusion. We had this problem in the 2017 election version, with additions and reverts, which settled down to only including events that suspended campaigning, which was a clear line. I suggest we continue with this kine. (Other election related events also included, like by-elections, major local elections, and calling of general election.) Rwendland (talk) 11:13, 9 November 2017 (UTC)

I also agree and have gone ahead and reverted the changes. If an editor wants to add events other than leader changes then they need to justify it here (as with suspension of election campaigning prior to the election). The consensus has for a long time been just for leadership changes, because this directly affects the polling questions asked, and this is a page for poll results and not a list of political events. — Preceding unsigned comment added by 89.242.241.41 (talk) 16:33, 9 November 2017 (UTC)

Perhaps a substantial amount of speculation in reliable sources that a particular event caused a change in support for political parties could justify inclusion of individual events. The only times I expect we might see that are when there is a clear change in polling results, and then I'd expect the speculation to come some way after the fact if not longer. For instance, the Comey letter in the most recent US Presidential election would meet this criterion. I've not seen anything such speculation over Fallon's or Patel's resignation. Ralbegen (talk) 18:21, 9 November 2017 (UTC)

If enough regular editors feel this is an unwarranted addition I will accept it, however if edits are going to be reverted and a talk page discussion opened it would be best to notify the user whose edits are being reverted that a discussion is taking place. :) --RevivesDarks (talk) 21:59, 9 November 2017 (UTC)

I don't think this event should be included in the table. Absolutelypuremilk (talk) 11:44, 11 November 2017 (UTC)

How did you generated the graphical summary?

Hey all. Sorry for asking this Question which is not relevant to the Article, but I have been trying to generate a graph for Cypriot Elections 2018 at the Greek version of Wikipedia, and it is a little bit confusion. How can someone create such a graph? Is there a tool within the WP or should someone create it in an extra-WP page and upload the picture afterwards? Thank you. Τζερόνυμο (talk) 09:31, 22 November 2017 (UTC)

Hi, I created the graph in Numbers (the Apple version of Excel) by typing in the values by hand. I'll discuss the specifics on your talk page so we don't clog up this article. Absolutelypuremilk (talk) 09:41, 22 November 2017 (UTC)

Kantar Public

I suggest we remove that joke of a poll that supposes that the 18-24 turnout would be only 19% (down from 57% at the last election according to a YouGov post-election survey). --81.98.195.164 (talk) 23:35, 24 November 2017 (UTC)

As much as the poll is most likely completely wrong due to the turnout figure It should stay because this page is here to document what polling organisations are currently suggesting if an election is taking place today. To remove the poll would also suggest a bias against the organsation and Wikipedia is here to give impartial information. JDuggan101 (talk) 12:22, 25 November 2017 (UTC)

That would be original research. Kantar Public are a member of the BPC, so I don't think there's any reasonable inclusion criterion that would exclude them from the article. Ralbegen (talk) 18:47, 25 November 2017 (UTC)

You seem to be rather confused. Kantar does not "suppose" that the 18-24 demographic turnout would be 19% - they report that only 19% of the 18-24 year olds they polled said they would definitely or probably vote in the next election. They are simply reporting the responses of the people questioned. Presumably you want the poll removed because you do not like the result - but that is not how this page works. — Preceding unsigned comment added by 89.242.242.219 (talk) 09:42, 3 December 2017 (UTC)