Jump to content

Talk:Opinion polling in United Kingdom constituencies (2010–2015)/Archive 1

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia
Archive 1

26 New Constituencies Polled

Here's the link to the individual polls - http://lordashcroftpolls.com/2014/05/conservative-labour-battleground/ Guyb123321 (talk) 15:34, 24 May 2014 (UTC)

More polls

These're individual constituency polls, not a group poll...they're just reported & comissioned together. http://lordashcroftpolls.com/2014/07/con-lab-battleground-swing-tories-drops-ukip-pick-labour-votes/ — Preceding unsigned comment added by Iliekinfo (talkcontribs) 22:25, 22 July 2014 (UTC)

Survation, Sep 2014

Some more: [1] Bondegezou (talk) 18:27, 27 September 2014 (UTC)

New polls

The individual constituency polls (Survation) had a sample size of figures just under 600.

Boston & Skegness poll (Survation): UKIP - 46% CON - 26% LAB - 21% GRN - 3% LDEM - 2% Fieldwork: 5-9 Sept. Thanet North poll (Survation): CON - 33% UKIP - 32% LAB - 24% LDEM - 6% GRN - 5% Fieldwork: 25-26 Sept. Rotherham poll (Survation): LAB - 48% UKIP - 37% CON - 6% LDEM - 4% GRN - 1% Fieldwork: 23-25 Sept.


Watford poll (Lord Ashcroft): LAB - 29% (+2) CON - 27% (-8) LDEM - 25% (-7) UKIP - 14% (+12) Berwick upon Tweed poll (Lord Ashcroft): CON - 33% (-4) LDEM - 30% (-14) UKIP - 17% (+14) LAB - 16% (+3) Cheadle poll (Lord Ashcroft): LDEM - 34% (-13) CON - 30% (-11) LAB - 19% (+10) UKIP - 13% (+10) Chippenham poll (Lord Ashcroft): CON - 39% (-2) LDEM - 24% (-22) UKIP - 16% (+13) LAB - 14% (+7) Eastbourne poll (Lord Ashcroft): LDEM - 46% (-2) CON - 25% (-16) UKIP - 18% (+15) LAB - 7% (+2) Oxford West & Abingdon poll (Lord Ashcroft): CON - 38% (-4) LDEM - 30% (-12) LAB - 18% (+7) UKIP - 9% (+6) Eastleigh poll (Lord Ashcroft): LDEM - 40% (+8) CON - 25% (-14) UKIP - 21% (+17) LAB - 12% (+2) Mid Dorset & Poole North poll (Lord Ashcroft): CON - 38% (-7) LDEM - 32% (-13) UKIP - 19% (+14) LAB - 8% (-13) North Cornwall poll (Lord Ashcroft): LDEM - 33% (-15) CON - 33% (-9) UKIP - 20% (+15) LAB - 10% (+6) Solihull poll (Lord Ashcroft): CON - 37% (-6) LDEM - 28% (-15) UKIP - 16% (+14) LAB - 12% (+3) Somerton & Frome poll (Lord Ashcroft): CON - 41% (-4) LDEM - 27% (-21) UKIP - 17% (+14) LAB - 9% (+5) St Austell & Newquay poll (Lord Ashcroft): CON - 27% (-13) LDEM - 26% (-17) UKIP - 25% (+21) LAB - 13% (+6) St Ives poll (Lord Ashcroft): LDEM - 32% (-11) CON - 31% (-8) UKIP - 18% (+12) LAB - 11% (+3) Sutton & Cheam poll (Lord Ashcroft): LDEM - 45% (-1) CON - 27% (-15) UKIP - 14% (+12) LAB - 11% (+4) Taunton Deane poll (Lord Ashcroft): CON - 34% (-8) LDEM - 30% (-19) UKIP - 15% (+11) LAB - 14% (+9) Torbay poll (Lord Ashcroft): CON - 30% (-9) LDEM - 30% (-17) UKIP - 21% (+16) LAB - 15% (+8) Wells poll (Lord Ashcroft): CON - 35% (-8) LDEM - 28% (-16) UKIP - 16% (+13) LAB - 13% (+5) Bermondsey & Old Southwark poll (Lord Ashcroft): LDEM - 36% (-12) LAB - 35% (+6) CON - 14% (-3) UKIP - 9% (+9) Cardiff Central poll (Lord Ashcroft): LAB - 36% (+7) LDEM - 24% (-17) CON - 17% (=) UKIP - 9% (+7) Hornsey & Wood Green poll (Lord Ashcroft): LAB - 43% (+9) LDEM - 30% (-17) CON - 14% (-3) Redcar poll (Lord Ashcroft): LAB - 44% (+11) UKIP - 23% (+18) LDEM - 18% (-27) CON - 12% (-2) Cambridge poll (Lord Ashcroft): LAB - 33% (+9) LDEM - 32% (-7) CON - 19% (-7) GRN - 8% (=) UKIP - 7% (+5)

anybody add :)145.52.143.140 (talk) 12:09, 1 October 2014 (UTC)

Shorten the Title

The title to this page, which amounts to 12 words, seems quite long, any ideas of how it could be shortened?Guyb123321 (talk) 11:38, 3 May 2014 (UTC)

Does anyone have any objections to moving this page to "Opinion polling for the next United Kingdom general election in individual constituencies" - that way the page would show up when people started typing Opinion polling for the next United Kingdom....Guyb123321 (talk) 09:53, 10 May 2014 (UTC)

No objection to the re-arranged title from me. Bondegezou (talk) 16:27, 10 May 2014 (UTC)
Think title is still too long, and mis-describes content. Removing the word 'individual' helps reduce the mis-description. How about "Opinion Polling in UK constituencies and constituent countries"? That way it becomes more appropriate to include Scottish Parliament and Welsh Senedd polls (which do really need to go somewhere), polls of groups of marginals. I am also suggesting removing the "next...general election" since some of the polls were clearly directed at by-elections, and further, it will still be useful to see the relationship between a poll and the subsequent result in a particular constituency in the run up to the following (2020?) general election. DrArsenal 85.133.27.17 (talk) 07:18, 16 October 2014 (UTC)
Polls for the Scottish Parliament and Welsh Assembly should be somewhere else, I would have thought. The polls on this page are Westminster polls of Scotland and Wales. Bondegezou (talk) 07:16, 17 October 2014 (UTC)
I would tend to agree that a different page would be better for those, as well as the Constituent Nation (and any English regional) Westminster polls, but I don't think that stops "Opinion Polling in UK constituencies and constituent countries" being a suitable shorter title for this page in the mean time, which was what I was really suggesting. DrArsenal 85.133.27.17 (talk) 13:20, 18 October 2014 (UTC)
Just edited Rochester & Strood, in the process discovering that both are explicitly by-election polls, and not Gen Election polls. Looking at the page, many of the polls for individual constituencies are by-election polls. My guess is less than half the data reported is "Opinion polling for the next United Kingdom general election in individual constituencies". Really think the page name needs changing again. DrArsenal 46.208.137.165 (talk) 22:54, 24 October 2014 (UTC)
What about "Opinion polling in individual constituencies, 2010-5"? Bondegezou (talk) 22:07, 26 October 2014 (UTC)
Must include 'UK'. Apart from that, the big issue I would have is it doesn't fit the groups of constituencies - at the moment the Constituent Countries and the multiple marginals. It might be that we are getting towards enough on Constituent Country polls for them to be on a separate page, but even if they did go on a separate page, we still have multiple marginals. DrArsenal 46.208.137.165 (talk) 07:07, 28 October 2014 (UTC)

"Opinion polling in United Kingdom constituencies, 2010-5"? Bondegezou (talk) 08:23, 28 October 2014 (UTC)

Seems good to me, works with current content (is better than current title even for groups of constituencies, countries, etc). I haven't got a clue how you change a page title, though. DrArsenal 46.208.137.165 (talk) 15:22, 29 October 2014 (UTC)
You go to the "More" bottom in the top right-hand corner and "Move" is listed there, but you have to be logged in to do it.
Does anyone else have a comment on the article name? Bondegezou (talk) 09:07, 2 November 2014 (UTC)

When to include Green Party?

Bondegezou deleted my addition -to the Clacton section- of the Green Party. I included it on the basis that beating the LibDems in the By-election had made the Green share unusally relevant in polling for that constituency. It is worth thinking, for example, whether this is a constituency in which LibDems will be in 5th position at the next GE (having been 3rd at the last). Bondegezou deleted, commenting "Best Green performance in the constituency is under 2%; best polling is 3%". So, what should the criterion be for including the Greens? Clearly if there is no opinion polling data about a party beyond a general 'others', it isn't worth having a column. But if there is, is there ever a good reason to not include on this page? If there is, what is it, and when does it justify excluding some of the data available? DrArsenal 46.208.137.165 (talk) 18:10, 19 October 2014 (UTC)

The LibDems did terribly in the Clacton by-election. Thus, beating the LibDems in that election doesn't seem much reason to include the Greens. That the Greens beat the LibDems is a sign of how bad the LibDems did, not how well the greens did. Look at the %ages the Greens get in this table: they've never done better than 2% in an actual election and their best polling performance is 3%. So Why the desire to note their performance in this particular constituency?
I'm generally in favour of including what data there is. It feels to me in this case that including the greens adds nothing. Bondegezou (talk) 19:05, 19 October 2014 (UTC)
It's good to know that you are "generally in favour of including what data there is". Would you feel better about Greens being included in the Clacton table if Greens were added to more of the other ones? DrArsenal 46.208.137.165 (talk) 23:01, 19 October 2014 (UTC)
I think I would feel better if consensus was established on this Talk page before anything else was done on this matter. Bondegezou (talk) 07:49, 20 October 2014 (UTC)
I'm not in a hurry to add Greens to a lot of existing tables any time soon (it's a blooming lot of work for not much gain, even if we agree on a policy to include, which I would need to be confident it isn't completely wasted). With that assurance, do you have any further answer to my question of last night - is the basic problem for you how low the Green share is in Clacton, or is it 'why include it here but not in other places'? It would help us work out criteria if you could answer (as it would, indeed, if 'us' wasn't just you and me!). I have already given part of my answer to your "Why the desire to note their performance in this particular constituency?" - I don't think there needs to be a reason for a particular constituency, since we agree on 'including what data there is'. But at the same time this constituency was worth being one of the first with the data added because, as you point out, the LibDems did so badly in the bye-election. That makes it worthwhile to ask whether the LibDems are likely to do so badly at the General Election that they will be in 5th place then, too. My thought is adding Greens to Clacton tells us at least as much about the LibDems in Clacton as about the Greens. DrArsenal 146.227.109.248 (talk) 17:05, 20 October 2014 (UTC)
With the main opinion poll article, I have argued that every party listed in the pollster's data tables should be reflected in the table, so I can't argue here any different! But if doing that, I'd like to see that for every constituency herein. I don't see the Green performance in Clacton as telling us anything significant. Bondegezou (talk) 20:00, 20 October 2014 (UTC)
So, nobody is talking to us, Bondegezou. It feels like us two (at least) could agree on every party listed in the pollster's data tables being reflected in the table for every constituency. But I can't do them all at the same time. If I do start that 'blooming lot of work for not much gain', do you suggest any particular order? DrArsenal 46.208.137.165 (talk) 18:44, 22 October 2014 (UTC)

Nobody loves us, DrArsenal. :( Oh, well. I would be happy to see you add every party listed in the pollster's data tables for every constituency. It feels presumptuous to tell you in what order to do such a blooming lot of work! However, if you want a suggestion, I'd start with those with the highest "other" vote in the polling. So, only two seats see >5% for others in polling: Boston & Skegness; Eastleigh. Clacton, Dudley N, Great Grimsby, Great Yarmouth, Heywood & Middleton, Rochester & Strood, Thurrock and Wythenshawe & Sale East have also seen >5% for others in actual elections. Bondegezou (talk) 15:28, 23 October 2014 (UTC)

Thinking about it, as Rochester & Strood is the most actively polled at the moment, I'd probably want to start there, but could certainly do Boston & S and Eastliegh next after that. Does that sound reasonable? DrArsenal 46.208.137.165 (talk) 22:19, 23 October 2014 (UTC)
Go for it. Bondegezou (talk) 12:25, 24 October 2014 (UTC)
Eastleigh is a pig, that I'm doing bit-by-bit. Meanwhile, I have added several new constituencies, and in the process noticed that those that were here before I started adding were overwhelmingly the good constituencies for UKIP. More needs to be done (adding further constituencies, and I would suggest, removing Synth and Original Research to reduce the resultant bias. DrArsenal 46.208.137.165 (talk) 15:28, 29 October 2014 (UTC)
Data not available for Dudley N or G.Grimsby. DrArsenal 46.208.137.165 (talk) 23:07, 31 October 2014 (UTC)
Bondegezou IP editors have twice removed all the painstakingly added Eastleigh detail. I have twice done 'undo' with an edit summary asking them to come here to discuss. As far as I am concerned, I'm not too bothered myself whether fringe candidates in the by-election are listed or not (indeed, I suspect few if any will be General Election candidates, and they haven't appeared in GE-related polling -cf current page title). BUT, you asked me to add details of all parties who pollsters had recorded. DrArsenal 46.208.137.165 (talk) 23:17, 1 November 2014 (UTC)


DN & GG are here http://lordashcroftpolls.com/wp-content/uploads/2014/05/Lord-Ashcroft-battleground-presentation-May-2014.pdf I think the dodgy cherry picking of some ashcroft polls and grouping the others as if they were regional polls is partially responsible for the apparent bias you noted. Iliekinfo (talk) 23:21, 1 November 2014 (UTC)

Edit conflict....FYI Dr Arsenal I also removed your detail....it's just standard practice across wiki to only include the major parties (for that constituency election) in these sorts of tables. As per RS. E G Eastleigh_by-election,_2013. Really don't want to put you off your v. good work, but don't think they belong. It's not even slightly clear what most of those abbreviations stand for, and I don't think they pass notable....do they all even have articles? Iliekinfo (talk) 23:21, 1 November 2014 (UTC)

Iliekinfo, was it you who removed the detail once, twice, or all three times? I think we need to get a response from Bondegezou, though. Is it really necessary to include parties like OMRLP who will not contest the constituency at the General Election, if we are to have Greens listed in Clacton polling?
BTW The Eastleigh_by-election,_2013 page is NOT RS, and all but one of the parties I added here DO have articles (couldn't you check?). I think, though, Iliekinfo, if Green Party were added back in to Eastleigh, the two of us would agree, but I don't know whether Bondegezou would, and I don't know if others who haven't yet joined us in discussing would. DrArsenal 46.208.137.165 (talk) 23:52, 1 November 2014 (UTC)
DN & GG may have been polled by Ashcroft, but beyond that presentation there is no info from Ashcroft that I can find about those polls. As he isn't (AFAIK) a BPC member, presumably he doesn't feel he has to abide by its rules.

The bias hasn't got anything to do with grouping polls, btw. It was there (and worse) before most of the grouped polls were added. If anything, adding that mitigated the bias a smidgeon: adding further constituency polls, as I have repeatedly been doing, is a better way, though. DrArsenal 46.208.137.165 (talk) 00:05, 2 November 2014 (UTC)

If, Iliekinfo, we exclude minor parties in the table, then what's the cut-off; should we even include the Greens for this constituency (where they've never polled well)? I think we should include everyone listed by the polling company as a safer rule. However, if a party is only listed by one poll and we have a table of several polls, I'm happy to see that party omitted from the table (which would exclude several of these minor parties). Bondegezou (talk) 09:11, 2 November 2014 (UTC)

Multiple Marginal Polls

In most cases those polls listed do not exist, they are groups of individual constituency polls which should be (and in some cases have been) entered below as individual constituency polls, not added up as a 'poll of marginals' in the normal way such polls are usually done (eg the comres one). Also, worse...there was a 'lead' column, comparing completely different areas with the general election & each other, that's just madness. I appreciate it sounds like semantics, and most 'Reliable Sources' haven't distinguished in a clear manner, but these are groups of individual constituency polls, not polls of groups of constituencies.Iliekinfo (talk) 23:11, 1 November 2014 (UTC)

Iliekinfo, I added several of them. However, I didn't do any adding up: that was done by Ashcroft. I accept it is silly having a 'lead' column comparing apples with peaches with bananas, but unless the marginals are representative of constituencies GB-wide, it is always going to be, and if they are, then it is not marginal polling, but ordinary national polling. Psephologists want to know what marginal polling is showing in case marginals behave differently, but how can you do a representative poll of 'marginals'? What is a sensible, consistent criterion? If you do what Ashcroft and Comres did in May, all you discover is that the 'biting point' has moved since the last General Election, when what you want is the 'biting point' for the next GE. And if you go in search of the biting point, as Ashcroft has claimed to be doing, you inevitably are looking at constituencies that aren't comparable.
However, it is still worth having the rest of the listing, apart from the 'lead' column. Readers may get real value from seeing in summary form that when Ashcroft polled 11 Con-held seats, Lab were polling at 38, Con at 32 etc in those seats, even if they aren't comparable to the 5 Libdem-held seats where the aggregate was Con 15% Lab 38% Ukip 28% LD 10% Green 6%, which itself is VERY interesting: indeed so interesting I am going back to look at it again now. Further, the links help indicate to potential editors where to find a load of polls to help cure the incompleteness. DrArsenal 46.208.137.165 (talk) 23:34, 1 November 2014 (UTC)
If Ashcroft presents them as grouped polls, then I think it's reasonable for us to present them as such, although perhaps a footnote is in order to say that the individual constituencies are also detailed below. The "lead" column is ridiculous though for this table: let's drop it forthwith. Bondegezou (talk) 09:13, 2 November 2014 (UTC)
Looks like consensus for the 'lead' column being dispensed with. Meanwhile, I have noticed and corrected transposed LD and UKIP scores in several rows. While checking others, though, I notice that the poll(s) for 10 Sep–3 Oct 2014 are reported here on the basis of ordinary general election question, whereas for others it is on the basis of 'in your constituency'. I think I have been consistent in the individual constituency tables (using the 'in your constituency' numbers), but I honestly don't know which is the better set of numbers for this table. DrArsenal 46.208.137.165 (talk) 10:49, 2 November 2014 (UTC)
I would favour the 'in your constituency' question, which I think is more predictive. A footnote explaining might help. Bondegezou (talk) 15:03, 2 November 2014 (UTC)

Constituent Country Polls

I know I am not the only person who will welcome Clyde1998's addition of a "Constituent Country Polls" section. It is an important and worthwhile addition to Wikipedia's coverage of UK opinion polling (especially since the same demographic behaves very differently either side of the border). I recommend you look at the talk page for "Opinion polling for the next United Kingdom general election". Relevant issues discussed there include how to deal with polls for the Scottish Parliament (2 sorts) as well as Westminster, and smaller parties. Dr Arsenal 46.208.137.165 (talk) 15:36, 15 October 2014 (UTC)

Thanks 72.83.48.214 for adding NornIrn. DrArsenal 46.208.137.165 (talk) 22:42, 23 October 2014 (UTC)
Northern Ireland is now repeated: it's covered here, but it's also covered at Opinion_polling_for_the_next_United_Kingdom_general_election#Northern_Ireland_polling with a bit of context added. We should avoid repetition, but I am ambivalent as to which place is better for the material. However, I do think the contextual notes should be preserved. Bondegezou (talk) 12:24, 24 October 2014 (UTC)
Wow: Clyde1998 has done a great job adding a load of further Scottish polls! DrArsenal 46.208.137.165 (talk) 22:14, 26 October 2014 (UTC)
The section has disappeared from the page: I know that most of the content is now on Opinion polling for the next United Kingdom general election, but the inclusion there is subject to ongoing debate, partly tied up with issues around whether to include Greens in the main table there, which haven't themselves been resolved. I suspect it would be wiser, for the time being, to keep the section on this page too. DrArsenal 46.208.137.165 (talk) 22:00, 5 November 2014 (UTC)

Rochester & Strood

This article is about opinion polling for the next general election. That's what it says in the title. Ashcroft has released a poll for Rochester & Strood: this includes two questions -- one about the forthcoming by-election and one about the general election. The former figures are included in this article. That seems wrong. We should include the general election figure here instead (or, at least, as well). The by-election figure can be included in the by-election article. Were I better at handling Wikipedia tables, I'd make the change now! Bondegezou (talk) 11:03, 12 November 2014 (UTC)

Perhaps we should match the Clacton precedent, and list both sets of numbers? DrArsenal (talk) 19:23, 12 November 2014 (UTC)
Sure. Bondegezou (talk) 12:28, 13 November 2014 (UTC)
Dun.
Seems to me like you need to improve your skills at editing existing Wikipedia tables, Bondegezou. The only way is to practice - why not start by trying to see if you can sort out (or just delete) the wrong shade of yellow for the Inverness, Nairn, Badenoch and Strathspey opinion poll lead? There is preview, and it won't matter too much if you muck it up: even if you click 'save page' you can always go to history and click 'undo'. Nobody will see it except you and me, and I will understand (after all, I keep pressing 'save page' and then realising that I have made a mistake that needs another edit quickly after. DrArsenal (talk) 22:54, 13 November 2014 (UTC)
I've had a go at tables before, and with some success, but, my, it is fiddly. Bondegezou (talk) 15:37, 14 November 2014 (UTC)
I thought you must have got some useful techniques, as I think you started this page, didn't you? But yes, editing the blighters can be fiddly. Thats why I talked of a "blooming lot of work" above. DrArsenal (talk) 20:20, 14 November 2014 (UTC)

London

I came across a couple of London polls: because of the date, they call themselves London Mayor http://cdn.yougov.com/cumulus_uploads/document/dj7eq6ky59/YG-Archives-LondonMayor-190112.pdf & http://cdn.yougov.com/today_uk_import/yg-archives-pol-yougov-londonvimayor-210611.pdf but actually include (p1) Westminster VI. I expect there will be more such polling out there, and will be more produced before the GE: I think they need to be somewhere. Does it live on this page? It is as much 'individual constituencies' as the ComRes or earlier Ashcroft multiple marginal polls, for example. Do we start a new 'regional polls' section (with a note that Scot/Wales/NI are on Opinion polling for the 2015 United Kingdom general election)? DrArsenal (talk) 09:03, 3 December 2014 (UTC)

You could rename the section "Constituent country and regional polling" in the main article to include polling from London and the other English regions - assuming there are some. London has 73 seats, so it's an important seat of polling data - it should be included somewhere. I've found another poll for the London area. Clyde1998 (talk) 19:52, 14 December 2014 (UTC)
It seems they are being collected on that page. DrArsenal (talk) 19:45, 19 December 2014 (UTC)

Categorising constituencies

This page has now grown to 47 separate Westminster constituencies and is showing no signs of slowing down. For organisation, now is the time to place them into categories, if not sooner. The only question is how they should be grouped, either by current party or by region. Unless anybody has any other ideas, I think it would best to group these results by current party. 123.2.85.195 (talk) 06:28, 17 December 2014 (UTC)

You are right that the growth of the page isn't slowing down. I doubt we will have caught up before Michael Ashcroft releases another batch. I would tend to disagree, though, about categorisation - the ordinary user (if there is such) is more likely to come to the page looking for a poll about their specific constituency - alphabetical order is the easiest for them to understand and find such constituency. Political aficionados may want a different order, but are more likely to be able to find alternative sources. Mind you, we should be adding links to such lists (lists of constituency held by party and marginality, links to sources that try to predict constituency results in other ways, etc). DrArsenal (talk) 09:22, 17 December 2014 (UTC)
I agree with the regional categorisation idea, as it would give more context to the seat - Conservatives likely to do better in the South and Labour more likely to do better in the south. Also, it would be able to help people understand where the seat is in the country - I, personally, don't know where "Bermondsey and Old Southwark" is (but judging by the size of the Lib Dem vote - I would guess the South West). Alternatively, each constituency could have a short information set with it to explain where the seat is and the size of the electorate. (For example) Clyde1998 (talk) 16:38, 17 December 2014 (UTC)
I agree with DrArsenal: an alphabetical listing is easiest for the reader to find what they want. I think the only categorisation that would make sense is the four nations (England, Scotland, Wales, NI). Every seat is linked to an article, so the information Clyde1998 wants is only a click away. Bondegezou (talk) 16:44, 17 December 2014 (UTC)
I just had a look and the "Bermondsey and Old Southwark" seat is in London... Right now, from my quick scan, there is only four seats outside of England listed on the article. I still feel that there should be some sort of regional categorisation for the English seats - even if it's just South (SW+SE+London), Midlands (WM+EM+East) and North (NE+NW+Yorks). Clyde1998 (talk) 16:55, 17 December 2014 (UTC)
User:DrArsenal, any categorisation would still mean that readers can find their constituency in the contents. A simple listing in alphabetical order is not always the best idea, as it certainly isn't the best way if we were listing 650 different constituencies. User:Bondegezou, there are very few seats outside of England to justify categorisation by constituent country alone. There should at least be categorisation by the three non-England countries as well as commonly used regional descriptions like "London", "South West" or "North England". Further, if we are to cater for the user which only wants their own specific constituency, then surely they would know whether their constituency was in London, Wales or the South East. 123.2.85.195 (talk) 00:09, 19 December 2014 (UTC)
The best way would probably be to group them by European constituencies, which is 8 in England and 3 outside England, for a total of 11 categories. This is the way BBC News ([2]) has organised the 650 constituencies (North West, North East, Yorkshire and the Humber, East Midlands, West Midlands, South West, South East, Eastern, London, Wales, Scotland and Northern Ireland), and these groups have real defined borders which do not cross through any constituencies. Onetwothreeip (talk) 02:03, 19 December 2014 (UTC)

I'm interested in this topic and I would have difficultly knowing with some constituencies which Euro-seat or region they were part of. I still feel an alphabetical listing is better. Does Wikipedia's various style manuals not have any suggestions about how to organise long lists? Most listy things I've seen on Wikipedia do it alphabetically. Bondegezou (talk) 20:55, 19 December 2014 (UTC)

After further research, I have realised that the European constituencies of the United Kingdom are actually based on official statistical divisions from the Nomenclature of Territorial Units for Statistics which can be found here. These boundaries are not arbitrary; they
  • are described with common names of UK geography,
  • are used as constituencies in European elections,
  • are official statistical boundaries; and
  • are used by media sources such as the BBC to categorise Westminster constituencies.
Within these categories, it should be perfectly right the constituencies be listed alphabetically within their groups. The problem lies within listing more than 50 constituencies on one page, but the constituencies should still be listed so that they can be found by name in the contents. Onetwothreeip (talk) 23:00, 19 December 2014 (UTC)
I never had doubt that there are reliable sources that list constituencies by region, and the only regions that have consistent support in RS are those you list, Onetwothreeip. However, there is still the problem that Bondegezou identifies: there are a lot of places (especially near boundaries) where people wont know which side of the boundary a given constituency is - how many people know correctly which region Grimsby and Scunthorpe are in? Or Aldridge, Ashfield, Bassetlaw, Beckenham, Bedford, Bedfordshire (NE,SW & Mid), Blackley & Broughton, Bosworth, Bournemouth (E&W), Burton, Bromley & Chislehurst, Braintree, Carshalton & Wallington, Chingford & Woodford Green, Christchurch, Colne Valley, Daventry, Dorset (N,S,W,&Mid), Eddisbury, Herefordshire (N&S), High Peak, Hornchurch & Upminster, Milton Keynes (N&S), New Forest (E&W), Orpington, Pendle, Poole, South Holland and the Deepings, Spelthorne, Swindon (N&S), Tewkesbury, The Cotswolds, Wantage, Wellingborough, Worcester, etc. (I could add more to the list, but I think it's long enough to make the point).
Of that list, I think I would have put 11 constituencies in the wrong region if I hadn't just compiled the list, and I know what the regional boundaries are, unlike most people. People who get their TV from a different region from the ones you would expect on the official list often think they are in a different region from the one where they vote for European Elections, too (eg Watford, which is also inside the M25, and despite being to the NW of London is in the 'Eastern Region'). I also note that the List of United Kingdom Parliament constituencies is simply alphabetical, and that there has been a little very mildly relevant talk there. DrArsenal (talk) 09:28, 20 December 2014 (UTC)
I see onetwothreeip has re-ordered the article, putting English constituencies first, with Wales, Scotland and the Multiple Marginals list at the end. I think this was premature, given the discussion here was ongoing, and I was sorely tempted to undo/revert, but have resisted that temptation for now. I don't think any of those changes is unreasonable (indeed, I think probably the multiple marginals list is better at the end). On balance I would prefer Scottish and Welsh constituencies to take their place in the alphabetical list, and even if outsiders might have trouble working out why (if listed) Berwickshire is in a different list from Berwick, at least people will know which side of those borders their own constituency is on! DrArsenal (talk) 10:10, 20 December 2014 (UTC)
You bring up very valid concerns, which is why I would disagree with removing the constituency names themselves from the contents, to ensure that they can easily be found by name. My point is precisely that people don't really know which constituencies are in which NUTS-1 region, as User:Clyde1998 said. It would give greater context to the average user to know where the listed constituencies are in the United Kingdom. I can tell you the average user is not looking at their own constituency, because 50 out of 650 is nowhere near half.
However, the List of United Kingdom Parliament constituencies article is not listed alphabetically. Aberdeen North is almost at the bottom of the page, as it is listed by constituent country. This page is not giving away much information either, as the data for each constituency on this page is much more than a list of constituencies. Each constituency on that page is also not listed in its contents, which makes it harder to find a specific constituency than it would be on this page.
If you'd like to see a terrible list of constituencies, I'd recommend Results of the 2010 United Kingdom general election by parliamentary constituency, but even that monstrosity does have some organisation by NUTS-1 division. The good news is that there won't be any more individual constituency polling for some time which would allow us to catch up to the new ones. Onetwothreeip (talk) 10:02, 20 December 2014 (UTC)
The fact that over time the average (modal?) user within the UK won't find their constituency on the page doesn't mean that won't be what they are looking for. The easier we can make it for them to determine whether or not, the better. Listing in 'contents', however, becomes more problematic the more constituencies there are listed, and the less obvious the ordering is - if (as I illustrated with my glimpse at List of United Kingdom Parliament constituencies) it looks like it is alphabetical, people will assume it is a single alphabetical list. If I'm right, they will look down the list until they get to where they expect, and perhaps draw the conclusion that if it isn't there, it isn't listed. Another problem is the number of people who don't know the name of their constituency - people often get the name of their council confused with the name of their constituency. Ideally I would want to include a postcode search, but I suspect we couldn't implement one that worked with the limited (and constantly increasing) number of constituencies listed.
Certainly for the moment, the typical user won't be looking for their own constituency, but rather will be a political anorak of one colour or another, but as the General Election gets closer, the balance can reasonably be expected to change, as more of the population become (at least mildly) interested in politics. After the election, the data will be of interest to a much smaller audience who are much more likely to be tolerant of any oddities of organisation.
While List of United Kingdom Parliament constituencies article is not listed completely alphabetically when you first get there, it is within each of the constituent countries, and it is alphabetically sortable. The luxury of being sortable as the user wishes between countries and UK-as-a-whole alphabetically (also on Results of the 2010 United Kingdom general election by parliamentary constituency, which BTW I wouldn't call a monstrosity, so much as a good source of data) isn't open to us. DrArsenal (talk) 10:36, 20 December 2014 (UTC)

While I'm all for being bold, perhaps we could have waited for a consensus before making big changes? Bondegezou (talk) 11:50, 20 December 2014 (UTC)

The readership of this article is not for people who want to know about the polling in their own constituency, and that'll be the case throughout the election campaign. This article is read by people who want to research the opinion polling in comparison to the different areas of the United Kingdom.
If people are still looking for an individual constituency only, and the constituency is against probability on this page, and the person knows the correct name of the constituency, then they will still be able to find that in the contents. What you are doing now is simply speculating on how a few people may or may or not view this article.
Even if we follow a similar system to the main list of constituencies, then I would agree that constituencies within their regions should be listed alphabetically.
Anybody who doesn't say Results of the 2010 United Kingdom general election by parliamentary constituency is a monstrosity must have a wider screen than my reasonably sized laptop. There's no need for the vast majority of those constituencies to have blank data for SNP, PC, SF, SDLP, DUP and UUP results. Onetwothreeip (talk) 12:16, 20 December 2014 (UTC)
I don't think we know who the readership of this article is. To answer that would require a considerable research effort: I know, I've been doing some research on Wikipedia! (See here - plug, plug!) I have been imagining that readers of this article are interested in what is going on in their constituency, as well as others interested in the broader picture. Bondegezou (talk) 17:06, 20 December 2014 (UTC)
Bondegezou, while I agree that we can't know who the readership is, I think you will agree that good usability design requires paying attention to various use cases, and how to facilitate such use with maximum intuitiveness. We agree that one use case we should cater for is readers interested in what is going on in their constituency, another sensible use case is Onetwothreeip's "people who want to research the opinion polling in comparison to the different areas of the United Kingdom." Are there any more? DrArsenal (talk) 21:00, 20 December 2014 (UTC)
Good perspective, yes, agreed! Other use cases... There are other ways of categorising constituencies as well as doing so by region, e.g. Which party currently holds the seat, type of seat (urban, rural, coastal), marginality, &c. One can't satisfy all of these at once very easily. Bondegezou (talk) 21:10, 20 December 2014 (UTC)
Even if people come to this article looking for their own constituency or any random constituency, we are unable to cater for that since only some constituencies are listed. Bondegezou brings up a very good point, and we should give a basic description of the constituency with each one (such as urban, rural, suburban etc.) Onetwothreeip (talk) 23:29, 20 December 2014 (UTC)
I'm not saying we should give that information. We need to be careful of WP:SYNTH. Bondegezou (talk) 22:38, 21 December 2014 (UTC)

How is that synthesis? It's not synthesis in other place. Onetwothreeip (talk) 04:38, 22 December 2014 (UTC)

Bondegezou didn't say it was synthesis, merely that we need to be careful. I am confident if we are careful we can avoid the problem in those respects by making use of reliable sources such as ukpollingreport. The problem though, would be that it still wouldn't be sortable. I can see it would be beneficial (on another page) to have a comprehensive list of constituencies that included such descriptive words in a sortable form with links to this page where they are on this page. I have tried asking on Talk:List of United Kingdom Parliament constituencies - no response - and Talk:Results of the 2010 United Kingdom general election by parliamentary constituency with one, negative, response. DrArsenal (talk) 10:45, 22 December 2014 (UTC)

These constituencies aren't representative.

I added the following text to the lead:

These constituencies are not representative, to the contrary, they have been selected for polling because somebody felt they were unusually interesting (typically a marginal, a party leader's constituency, or one where there was a by-election). As such it is unwise to infer from them what the situation might have been at the same date in other constituencies which are not listed (neighbouring constituencies can be politically very different from eachother, for example).

twice Onetwothreeip has deleted it, and at the same time made other unhelpful changes to the lead at the same time.

In the 'summary text' of this second time round deletion Onetwothreeip says it is 'non-encyclopaedic'. I can provide reliable source back-up, but if it keeps getting deleted as quickly as if it had been vandalism, it is not practical to do so. DrArsenal (talk) 14:52, 22 December 2014 (UTC)

I think we can all agree that "because somebody felt they were unusually interesting" or "it is unwise to infer from them" is not a sentence that belongs in an encyclopaedia. What you're trying to say would be better placed in an article about opinion polling in general, as I'm not sure why it's here. Wikipedia is also for not stating the obvious, such as that constituencies are different to other constituencies. We'll let the readers of the article make their own conclusions about the polling.

The other edits were because;

  • the polls are not just for the run up to the election, they are throughout the entire parliament.
  • the groups of marginal constituencies polled are polled individually.
  • I can't think of an article where the introduction says that a part of the article is "lower down".
  • Lord Ashcroft in this case is not a person, it is a polling entity.
  • the opinion polling of individual constituencies for the next election does not have an indeterminate end. They'll end at the next election.
I'm sure if you asked other editors, they would agree that there are parts of the introduction which are simply not quite right for the article. Onetwothreeip (talk) 22:48, 22 December 2014 (UTC)
Onetwothreeip says "I think we can all agree...". Er, no. We can't all agree on that. I don't think what I am saying in that paragraph is obvious to the typical reader - which I guess takes us back to the difference between us about who we think will read the article. I felt it should be on this page because it is more relevant to this page than a page that talks about more typical opinion polling. As for "We'll let the readers of the article make their own conclusions about the polling", well I don't see any problem with warning them not to draw erroneous conclusions, if reliable sources also make such warnings (I know I can find such warnings).
I'm glad we are agreed on the need for simplicity. I prefer 'In the run up to' to 'Preceding' precisely because it is simpler English, but so far as I can see conveys the same meaning.
Not all of the polls at #Multiple constituency polling are polled individually - ComRes ones and Ashcroft's 2013 weren't.
There needs to be something to indicate to readers to look for #Multiple constituency polling below the long list of constituencies, if that is where it has been put.
Ashcroft releases polls called the 'Ashcroft National Poll', and the BPC invited "him" to join. It is absurd, therefore, to make a distinction between Michael Ashcroft the person and 'Lord Ashcroft' the pollster, at the same time as linking to a wikipedia page titled "Michael Ashcroft" that doesn't mention any polling since 2010, anyway.
No opinion polls that will be published between now and the date of the election are on the page, for an obvious reason. When we get to polling day, then your wording will be correct. In the mean time, it isn't.
Where I could see value in your edits, onetwothreeip, I have tried to retain them this time. DrArsenal (talk) 23:41, 22 December 2014 (UTC)
No matter if you think it's valuable or not, saying that " because somebody felt a certain way" is not an objective fact or something that is found in an encyclopaedia. Why opinion polling is conducted for elections is not what belongs on this page.
You're using the introduction to give your own notes to the readers instead of an objective summary. Many readers would rightfully assume that a result in one constituency could be similar in an unpolled similar or neighbouring constituency, so clearly there are means for readers to make their own conclusions instead of treating all these constituencies separately from each other. It doesn't mean that should be in the article either, since we should let the readers decide what to do with the information they've given. The Wikipedia entries for each Star Wars movie does not have a warning to tell readers that each movie is different and that erroneous conclusions shouldn't be made about others.
Likewise, I've never seen "by the way, the section about casting for this movie is down the bottom, you just have to scroll down a bit".
On the very .pdf files that Lord Ashcroft publishes, they are styled as being published by "Lord Ashcroft". When it comes to the polling data, there is no mention of the Michael Ashcroft. It's also not as if it's Michael Ashcroft who is calling the constituencies and asking them for their voting intention. There are people who do that for him, much like how Mr. Survation and Mr. Y Gov aren't one person who do the polling.
I have only been editing parts which could be improved. The rest I have retained. Onetwothreeip (talk) 23:59, 22 December 2014 (UTC)
I think it would be useful context to say something about how these constituencies are chosen and that they are not representative per se. Sources to support that should not be too hard to find. I'll have a look if sourcing is the main concern. Bondegezou (talk) 00:07, 23 December 2014 (UTC)
I wouldn't object to saying that most of these constituencies are marginal or holding by-elections, but the problem is the amount of irrelevant and condescending non-information. Onetwothreeip (talk) 00:40, 23 December 2014 (UTC)
Bondegezou, perhaps you could draft some words that hopefully Onetwothreeip won't object to (since it appears I haven't hit the right tone)? At one stage I wondered if Onetwothreeip's objection was that there was too much in the lead, and came to the conclusion that a short section (perhaps 'Selection and representativeness of constituencies'?) before the first poll would be a better place for such text than the lead.
Onetwothreeip says "I've never seen ...'the section about casting for this movie is down the bottom'". Of course not. Articles about films all take the same form, so people looking at them get to know where to expect various elements, and their 'contents' don't have more than 50 sub-sections before getting to mention of that section in the contents. At the most closely linked page, with only 5 sub-sections, it says in the lead "Separate polls covering constituent countries of the UK and English regions are reported further below", and further down (before 2014) it also says "Detailed poll results that break down 'others' for some dates are available in a second table, below."
Read what Michael Ashcroft says about his own polls As for "Mr. Survation and Mr. Y Gov aren't one person who do the polling" - we agree on that, but the difference is that both of them are (parts of) businesses with separate legal identities to the individuals involved. I can see no evidence that "Lord Ashcroft Polls" and "Ashcroft National Poll" have any legal identity other than that of Michael Ashcroft. It isn't relevant who conducts the fieldwork (indeed Ashcroft keeps this secret): the polls are published by an individual, not an incorporated organisation of any kind. DrArsenal (talk) 10:47, 23 December 2014 (UTC)

Yet-to-be-added constituencies. And 'Voting Intention' vs. 'Constituency Voting Intention'.

Here's a list of Ashcroft constituency polls, many of which haven't been added to this Wiki page yet: http://may2015.com/ashcroft/

I would add them myself, but I don't know how.

Also, I've noticed something. On this Wiki page, some of the percentages taken from the Ashcroft polls are for 'Voting Intention' and some from 'Constituency Voting Intention'. I would recommend uniformly using the 'Constituency Voting Intention', which Ashcroft himself regards as more accurate.

Renren8123 (talk) 00:56, 7 January 2015 (UTC)

Thanks Renren8123. We do need to keep a note of the gaps we haven't filled yet. I came across a constituency poll by another pollster and meant to add it, but have lost it now (aagh!).
Yes, we are trying to standardise on 'Constituency Voting Intention': please when you spot one that is based on the wrong one, please correct it (or at minimum leave a note here on the talk page).
And PLEASE do try to add a constituency poll or two yourself. I would suggest finding amongst the constituencies already listed one that was won by the same party in 2010 and which has the same party in the lead in the opinion poll, and copying that from the 'edit' window, opening the 'edit' next to #Constituency Polls, and scrolling down to the appropriate point in the alphabetical sequence for the constituency you want to add. Then paste, and start editing what you have just pasted to modify it to be suitable for your added constituency. Don't worry if you get stuck part way, forget something, or something doesn't quite work - we have all done that, and can help sort it out! DrArsenal (talk) 10:48, 7 January 2015 (UTC)

Page move

Thanks, Chessrat, for sorting out the 2010-15 instead of 2010-5: after I had moved the page (to '2010-5'), I realised the mistake, but decided to delay correcting it for a little, at least. This is a better format for the number. DrArsenal (talk) 21:56, 9 January 2015 (UTC)

I've been through and added links to this article from each constituency article. Well, it was kinda boring and I whizzed through it, so I may have missed some! I used Template:See also as it seemed most appropriate. Bondegezou (talk) 13:28, 4 February 2015 (UTC)

Scotland polling...

...is complete: http://lordashcroftpolls.com/2015/02/scottish-battleground/

The second voting intention question - page 6 in the constituency data - are the headline figures. Hopefully, I'll get time to start adding some later today... Clyde1998 (talk · contribs) — Preceding unsigned comment added by Clyde1998 (talkcontribs) 13:34, 4 February 2015 (UTC)

There are now 27 polls that haven't been listed here, if anyone has the time to list them?Guyb123321 (talk) 11:52, 7 July 2014 (UTC)

I've added Brighton Pavilion, there's a lot to get through people. It's not too bad once the table is in place. Owl In The House (talk) 22:04, 27 August 2014 (UTC)

Southampton Itchen, which is being cited as the most likely Conservative gain from Labour, appears to be absent from the article. There was an Ashcroft poll that put both parties neck and neck. Dovea (talk) 21:41, 19 February 2015 (UTC)

Categorising constituencies - again.

82 different constituencies now. We have to start categorising them before it is too late, especially before the pre-election campaigns really start and the polling increases. I still think regional is the best way forward, and if not that then by incumbent party. 123.2.85.195 (talk) 04:48, 14 February 2015 (UTC)

I don't see what's wrong with alphabetical as it currently is. I look at articles like List of MPs elected in the United Kingdom general election, 2010 and Same-sex marriage law in the United States by state and an alphabetical list seems to serve fine. Bondegezou (talk) 14:04, 14 February 2015 (UTC)
I would agree with anonymous IP, it is getting unwieldy now. G-13114 (talk) 14:29, 14 February 2015 (UTC)
I also agree with anonymous IP. Eamonster (talk) 00:04, 15 February 2015 (UTC)
I'm sorry Bondegezou, but the first article you cited is clearly a list article, and it would be impossible to emulate that model successfully in this article. As for the second article you cited, it's quite obvious that even fifty items is ridiculously long, but this article is currently at 82 and will have anywhere up to 650 by May this year. I do agree with you that we should keep the lists alphabetical, although within their respective subcategories.
I would like to begin sorting them into their NUTS-1 subdivisions (nine regions of England + Scotland/Wales/NI), as is standard by the UK government, used as constituencies for European elections and regions designated for the general election by BBC News election coverage. However I don't think I'll have the time to sort them all out by myself. 123.2.85.195 (talk) 06:48, 17 February 2015 (UTC)
This is a list article too. I cannot see anyone reading it through constituency by constituency! I find it very easy to use as it is: I know the name of the constituency I want, I know the alphabet, bingo. I do not know the NUTS-1 subdivisions: your suggested change would immediately make it harder to find what I want, and I suspect the same is true of many readers.
I note that List of United Kingdom Parliament constituencies also just uses the alphabet and makes no mention of NUTS-1 (and, indeed, I note NUTS-1 doesn't link to what you think it links to!). I note that the BBC 2010 Election site uses an alphabetical listing of all constituencies and doesn't give NUTS-1 subdivision anywhere on the individual constituency result pages. If other editors really insist on complicating an alphabetical list, I'd suggest just splitting into the 4 nations, each of which has a distinct political culture with different parties standing.
I looked at the Wikipedia guidance on this subject, notably WP:LEADEMBEDDEDLIST. This isn't particularly helpful: it says, "Although lists may be organized in different ways, they must always be organized. The most basic form of organization is alphabetical or numerical [...] When using a more complex form of organization, (by origin, by use, by type, etc.), the criteria for categorization must be clear and consistent."
This article is never going to have near to 650 entries. Not even Lord Ashcroft has the money to waste on polling that many. I don't expect vastly more constituencies to be polled: polling is more likely to return to many of these seats as the key battlegrounds. We do have a long list, but, as I said, I find it an easy to use article because it has a straightforward simple structure. Bondegezou (talk) 11:15, 17 February 2015 (UTC)
I think the anon IP was referring to the Regions of England plus Scotland, Wales and N Ireland. Which would make sense to me, and make things easier to find if you know which area of the country you want to look for. It would also allow comparison of the different seats in each region. G-13114 (talk) 19:36, 17 February 2015 (UTC)
Bondegezou, nobody is arguing against listing the constituencies alphabetically. What we're going to be doing is to split them into around ten simple widely known divisions, i.e. South West England, Yorkshire and the Humber, London etc. as well as the other three constituent countries. Or as G13114 says, the Regions of England plus Scotland, Wales and Ireland. The article you mention, List of United Kingdom Parliament constituencies is not comparable to this article. If such a simple layout was possible (all the data for one constituency appearing in one row), then it would be the obvious choice. Since there are at least two rows needed for every constituency, they require separate paragraphs. Categorising regions will certainly assist anybody wanting to find any constituency, especially since constituencies do not always have assumed names. I agree that they should be ordered alphabetically, within the categories themselves. Onetwothreeip (talk) 06:23, 18 February 2015 (UTC)
I contest the idea that the English regions are "simple [&] widely known". English regions are not well established in the minds of most readers. PoliticalBetting this morning lists 4 new Lord Ashcroft polls: Boston & Skegness; Castle Point; South Basildon & East Thurrock; North East Cambridgeshire. As a test, I thought to myself, "Do I know which English region each of these are in?" The answer is, "No, no, probably & probably." With a purely alphabetical list, I can find each of these constituencies immediately. With a regional subdivision, I can't. The alternative situation you posit, as I understand it, is that someone doesn't know the name of a constituency, but they know what region it's in and then they can more easily scan through a list: constituency names can be confusing, but I am unconvinced by that use case myself.
The other examples I can find -- of Wikipedia articles, or other websites (e.g. BBC, Election Forecast) -- just go with an alphabetical listing. I'm not trying to be difficult, but I'm afraid I really don't see what great advantage is achieved by splitting constituencies by region. Bondegezou (talk) 12:06, 19 February 2015 (UTC)

Actually, BBC News and ElectionForecast.co.uk also seperate constituencies by these regions. The reason for this is not to make finding a constituency easier or harder, but for the easier navigation of the page when there are between 80 and 650 elements which contain more than one row of data (which rule out the kinds of lists you are talking about that are used on other Wikipedia article, Election Forecast and Electoral Calculus). The question is not whether you know which English region these constituencies are, the question is whether you know which region your own constituency is in. For any other constituencies, they can easily still be searched in the contents especially with the ability of Control-F (or Command-F) to search for specific words. Onetwothreeip (talk) 01:26, 20 February 2015 (UTC)

Actually I noticed that the BBC website covering the 2010 election categorises seats by region see West Midlands for example. So there we have an example of this categorisation being used by a major organisation. G-13114 (talk) 02:44, 20 February 2015 (UTC)
As a first step to tidying it up I have sorted it by country as I don't think anyone disputes, England, Scotland, Wales and Northern Ireland are well understood. I would favour sorting by region within England as well given evidence of RS using regions for electroal purposes. I'm not sure consensus is there yet though so I've left England purely alphabetical for now.Andrewdpcotton (talk) 09:27, 20 February 2015 (UTC)
I think it is better now (sorted by nation) but I don't think I'd support splitting it down further into region. There is a lot of sense in grouping all the Scottish/Welsh/NI seats together as they have different parties contesting. But I don't see that a regional split really adds anything. Frinton100 (talk) 11:03, 20 February 2015 (UTC)

And incidentally, on the discussion about the BBC's use of regions - they are inconsistent. Yes, on their 2010 results map they do use the Govt. Regions, but in other contexts they don't, they use their own TV regions, which do not match up. See this article for example by the BBC Political Editor for the East of England - http://www.bbc.co.uk/news/uk-england-30987153 She includes Northampton and Milton Keynes as part of the "east" as they get BBC East TV, but they are in East Midlands and South East respectively. Conversely, she does not include Watford or Thurrock which are part of the East of England but come under BBC London. So I don't see the beeb's practices really help us decide one way or the other.Frinton100 (talk) 11:20, 20 February 2015 (UTC)

Thank you for the edit, Andrewdpcotton. However the article is still quite disorganised as one of the categories has over 60 constituencies and could have up to 533. On the BBC News election coverage, Frinton100, they seem to use these standard regional divisions for organising results but they aren't uniform on news articles. Still, we have a precedent of a major news organisation using these regions to sort Westminster constituencies. Onetwothreeip (talk) 22:15, 20 February 2015 (UTC)
Agreed, it's simply too big and unwieldy to not be subdivided. G-13114 (talk) 00:15, 22 February 2015 (UTC)

Minor Parties

It seems we also need a consensus on how we display minor parties in these polls. I think we should stick with showing polls exactly as the polling company have done.

The danger of doing otherwise is that we end up with inaccuracies. In Boston & Skegness for example, we cannot simply assume that the combined polling score of a party on 1.0% and one on 0.4% is automatically 1.4% (or indeed 1.3% as it seems to have been changed to now for some reason). http://survation.com/wp-content/uploads/2014/09/Boston-Skegness-Results.pdf

We also cannot simply perform the calculations ourselves, as in a weighted sample even the number of responses (not just the %) is rounded. On Table 5 of the B&S poll, we have an unweighted base of 328 and a weighted base of 342 - this means there were 328 phone calls/online responses, but these have been deemed to represent 342 people. As the number of responses for each party are increased/decreased by the weighting, the number of responses becomes a non-integer and is rounded for publication in the table. The original, non-rounded number is then used to calculate the percentage. This can clearly be seen by adding up all of the parties in the first column - it actually comes to 341, not 342. Or, look at "Some other party" and we see a total of 16, but apparently consisting of 12 men and 3 women, in reality this will be, for example 12.4 men and 3.3 women - these round down to 12 and 3 respectively, but the total is 15.7, which rounds up to 16. Similarly, if you calculate the AIFE percentage using the rounded figure of 3, you get (3/342)x100 = 0.87719....., which rounds to 0.9, yet is reported as 1.0%. Clearly a number greater than 3 (e.g. 3.3) has been used to reach a figure of 1.0%.

So adding up rounded totals supplied by pollsters will always give these sort of inaccuracies. It seems to me we shouldn't be trying to re-invent polls to make them look neater on Wikipedia, but we should display them as published by the polling organisations. Frinton100 (talk) 23:07, 20 February 2015 (UTC)

I agree: we should follow what the source says and report what parties the source reports. (In some cases, sources are not entirely consistent within themselves, so I am not averse to simple calculations, duly noted, were necessary.) Bondegezou (talk) 01:02, 21 February 2015 (UTC)
If you're referring to the Survation poll on Boston and Skegness, it seems to be very much flawed. Either way, we're not reporting the polls exactly how the polling organisations are reporting them. Nobody is suggesting that we represent the polls how the polling organisations do. We take the data from the weighted results, and turn these into percentages. The question at hand here is if we actually want to report on the data for every single political party that comes up, even if it isn't in the main set of data that the polling companies show. There's absolutely no reason to show the polling results for AIFE, a party which none of us have any reason to care about. There is however much reason for us to consider polling information much more carefully, especially in this special case where there is an absurdly small sample size.
We should only be including:
  • the major parties per consensus from the main polling article; Conservatives, Labour, Liberal Democrats, UKIP and Greens (for Great Britain).
  • Plaid Cymru and Scottish National Party in Wales and Scotland respectively.
  • any party that has polled at 5% or higher in the constituency, since (and including) the general election.
It is not our place to report all information that reliable sources give us, as Wikipedia article are not for dumping all possible data (per WP:NOTREPOSITORY and WP:RAWDATA). We can't keep listing every single result for every single political party in every single constituency, we should only take the relevant information that is useful including an "others" column. Onetwothreeip (talk) 08:57, 21 February 2015 (UTC)
Except that we do display the polls as the polling company reports - using the headline, weighted percentages, i.e. 45.5% for UKIP in the Survation B&S poll - this has been lifted directly out of the data table, and not calculated independently of the poll. In every other constituency poll we should be taking the percentage directly out of the data tables provided by the pollster.
For the most part we are not going to stray outside of those parties you list above, as any other parties will be lumped in with "others" by the polling companies. However, there are a few instances - notably by-election polling - where local circumstances mean other parties are listed separately. In the case of B&S, there are particular local factors around the inclusion of AIFE, and given the poll was done for a UKIP donor, I suspect he wanted to see what impact that party were having locally. It is a potentially significant party in the B&S contest - if only that they may take votes off a party that has a good chance of winning. What is even more significant in this poll is that AIFE appears to have been prompted rather than being lumped in with "others", while the Greens are listed under "others".
I will confess that I have my doubts about the Survation poll (and indeed many Survation polls), though not in terms of the detail of the data tables, more in the overall methodology that seems to have given such a starkly different result to Ashcroft just 5 months later. But the poll is what it is, and Survation are a member of the BPC, so I feel we should either display the headline figures for each party (not the huge amount of other information in the poll) or nothing at all - if only to prevent inaccuracies creeping in as I described above. Frinton100 (talk) 12:05, 21 February 2015 (UTC)
It's well known that Survation has a major mathematical bias towards UKIP, but I think these polls should be kept. I think a good addition to the main polling article would be how the main polling companies are different to the polling average. As for this poll in Boston and Skegness, I would completely agree to keep AIFE if they were significant or relevant, but the fact is that they only polled 1%, making them neither significant nor relevant. If we include them in this poll, there would be no reason to exclude any others, which would give the results too much irrelevant data for an encyclopaedic article. Onetwothreeip (talk) 21:48, 21 February 2015 (UTC)
I wasn't suggesting for one minute removing Survation polls altogether - as I say they are a member of the BPC so we should include their results. It is significant in the context of B&S that AIFE are polling so low (it might have been expected they would do better in that seat), and even a few hundred votes that might otherwise have gone to UKIP could be significant in this constituency. But the fact remains that we cannot accurately calculate an AIFE+Others percentage, so we should leave the data in the form it is churned out by the polling company. It is not overly detailed - no one is suggesting we add in the full gender/age/socio-economic breakdown that the data tables give, just one or two extra columns in a small number of polls. In the vast majority of cases it isn't going to be an issue, since most of the constituency polls include only those parties you listed above. It seems so far we have 2 in support of including all parties, with 1 opposing. Frinton100 (talk) 22:48, 21 February 2015 (UTC)
I think minor parties should be included in the tables if they are reported on by the pollsters. So regarding B&S, AIFE should be included. Eamonster (talk) 23:52, 21 February 2015 (UTC)

I have a very simple solution for your problem, Frinton100. In all but a small handful of poll results, they are rounded to the nearest integer. In this Survation poll, they are rounded to the nearest 0.1 decimal. If we simply have one standard format, using only whole numbers for the poll results (retaining .01 decimal for election results), then we completely avoid the trouble of having conflicting calculations for Others. When we're talking about AIFE, we're talking about one percent of 300 respondents which is only three people*. We shouldn't be disclosing all the possible data, and even the Survation poll (as flawed as it is) doesn't show AIFE as a headline figure. Onetwothreeip (talk) 09:57, 22 February 2015 (UTC)

I wouldn't agree with that approach, as it means further manipulating data from the pollsters which we should not be doing - we should be displaying data as provided. Again, inaccuracies will slip in if we do this - the UKIP score in the B&S poll is a good example - it is reported by Survation as 45.5%, which ordinarily would round to 46%, but if in fact the figure is 45.47% (which rounds to 45.5% to 1dp), then to the nearest integer we should round down - 45%. It may not seem like a big deal when a poll shows a 20-point lead, but could be if the lead was narrower.
I have explained previously that the 3 AIFE respondents in the unweighted sample have actually been subjected to a certain weighting (the formula for which is not described in the data tables), and so have become something like 3.3 respondents in the weighted sample, which has been rounded to the nearest integer. Try the calculation I described above - 3 out of 342 is not 1.0% to 1dp, but is actually 0.87719.....%, which rounds to 0.9.
In any case, it seems we have a consensus beginning to form - we so far have 3 people in favour of including minor parties as reported by polling companies, and 1 opposed.Frinton100 (talk) 11:23, 22 February 2015 (UTC)
Either way, we would be changing how the data is presented to fit the boxes that we have made as standard, otherwise we might as well just paste each .pdf file into this article. It's not manipulation, we just need to agree on uniform standard on how we present the data so that each poll isn't represented differently.
As for the lead being misrepresented after rounding, we have a precedent from the main polling article. The polls in that article are all to the nearest integer, even if they are reported with decimal places. For one poll, Conservative and Labour were both 34%, but the box for Labour was coloured in and a note was supplied showing that the opinion poll reported Labour was ahead 34.3 to 33.8 for the Conservatives. This is what we should be doing if this situation applies.
We eliminate the need for rounding to either 0.9 or 1.0 if we eliminate the decimal place and just make it 1%. This should go for every result, even if reported in decimal places, to maintain a uniform standard for the article, per style.
There is an unopposed consensus for reporting minor parties in the appropriate cases, but reporting minor parties in all cases (or more than a small amount) has been rigorously opposed by many editors of UK polling articles, which is something I am not happy about since I've always been for expanding the polling data to more parties. Onetwothreeip (talk) 02:00, 23 February 2015 (UTC)

Hornsey & Wood Green/LibDem polling

The LibDems have done a large number of constituency polls. These are proper quota sampled polls with fieldwork carried out by Survation. However, they are private, not routinely published, and differ in method to Survation's regular polling (whether those differences are valid or relevant is debatable). The LibDems have now released the data tables for one of these polls, in Hornsey & Wood Green: see here. Should we include this in the article? UK Polling Report and Political Betting have both discussed the case and are both ambivalent: UKPR ultimately is more negative, PB more positive.

I don't know what the answer is. Unable to decide whether such a poll should be included or excluded, I guess I'd go with a compromise of including it, but with an explanatory note added, perhaps citing the UKPR article. Bondegezou (talk) 11:10, 11 March 2015 (UTC)

Order of parties

As things stand, this page is pretty haphazard. Each constituency seems to use its own scheme for the listing of parties, with no uniformity from one electorate to the next. I'd like to suggest that the order be standardised in one of two ways:

  1. by total seats then votes at the last general election (e.g., for England: CON, LAB, LD, GRN, UKIP, BNP, Other)
  2. the same as (1), but with the party holding the seat at the last election moved to the first position

It would make reading the list a bit easier, and would make the page considerably neater. Grutness...wha? 10:44, 31 March 2015 (UTC)

I'm not certain. Having the party holding the seat in the first position feels right to me. After that... we could do by order that the parties came in at the last election in that seat? That sort of seems to be how several of the tables have been done. Bondegezou (talk) 13:42, 31 March 2015 (UTC)
That would work too. It just seems pretty haphazard right now. Grutness...wha? 00:06, 1 April 2015 (UTC)
I'd suggest we order them by performance at the last election, and afterwards new parties alphabetically. These issues and others such as categorising the eighty-five constituencies of England should be discussed thoroughly in preparation for an article covering the 2015-present article after the election. 123.2.85.195 (talk) 10:28, 14 April 2015 (UTC)
I imagine we'll have plenty of time to work on a 2015- article! Probably won't be a rush of constituency polling straight after an election. Bondegezou (talk) 10:59, 14 April 2015 (UTC)

Six more marginals

Ashcroft released six new constituency polls on 25th April. Some of them have already been added but others (like High Peak, Bristol North West and the update for Colne Valley) don't seem to be up. I'd add them myself but that table wikicode is an unpenetrable mess. http://lordashcroftpolls.com/2015/04/six-more-marginals/ Little Professor (talk) 11:09, 27 April 2015 (UTC)

Table wikicode... the bane of all our existences! Bondegezou (talk) 11:19, 27 April 2015 (UTC)
OK, I bit the bullet and added High Peak and Bristol North West myself. Some other kind soul has done Colne Valley. Little Professor (talk) 20:19, 28 April 2015 (UTC)