Talk:Opinion polling for the 2018 Turkish general election

Meral Akşener's party

As you can see, I've added a column for Meral Akşener's hypothetical future party. The party itself is controversial, and I'm sure adding it here – before it has even been legally established – is as well. However, unless the Turkish government is somehow able to block the formation of her new party, which is of course yet to be named, it's almost certain that it will be formed in the period September—November 2017, with most recent sources setting the date for late October. More and more polling firms are including her party, and I believe this is more than enough of a reason to include it here as well. We could perhaps even add footnotes explaining the party's current status, if some contributors remain skeptical. — Μαρκος Δ 22:24, 15 September 2017 (UTC)

Presidential polling table restructuring

I'd like to suggest that we restructure the table of opinion polls for the presidential election along the lines on the Brazilian one. Since few parties have announced their candidates yet, the candidates presented to respondents in opinion polls vary, which leads to an excessively wide table. If we instead just reduce it to one column per party, with candidate names in parentheses, that would eliminate said problem. Once all main candidates have been announced, we can perhaps again begin using the current, more standard layout. — Μαρκος Δ 17:47, 22 November 2017 (UTC)

The Brazilian article takes from the French article where it was first implemented but changed because it was annoying to compare different candidates if they were all within the same column. I think Impru, being image-oriented, changed all of them to portraits, which I would normally disagree about because of accessibility reasons (prefer text to images whenever possible), but might work better in this case. I think this discussion is premature at this state, though, since there's only been two polls so far. Mélencron (talk) 23:30, 22 November 2017 (UTC)

Yes, I was the one who originally suggested it for the French one as well. Especially in countries like Turkey, where party tends to come before person, it would make sense to organize the columns by party. Even more so when some pollsters apparently don't even present candidate names in their surveys (instead going with generic CHP'nin adayı and similar). That said, I get your point. And I guess you're right that a decision doesn't have to be made before more polls have been added. — Μαρκος Δ 17:36, 23 November 2017 (UTC)

I've now switched the table to the format I suggested. If you have objections, I suggest the following compromise: we leave this format in one table for the pre-official nomination period, and then use candidate-based columns in a separate table for the post-nominations period, in 2019.— Μαρκος Δ 00:06, 11 January 2018 (UTC)

New poll to be added

Would a Turkish speaker add this PIAR poll? [1] = [2] -- 2A1ZA (talk) 15:37, 4 April 2018 (UTC)

Graph needs better time indicators

it's not clear nov 15 nov 16 nov 17 , that these are years, and also where for instance march or last year may is .. the graph needs better time indicator each 3 month written like here --31.202.25.100 (talk) 16:02, 5 April 2018 (UTC)

You are literally the only one who does not understand it. I got a message from someone asking to explain the graph a while back, so when you were so confused now, I actually thought "hey, maybe it just needs a better layout, if it confuses so many people". But you know what? I went back and checked who it was that left that message I had gotten; it turns out it was you. I even explained the graph to you back then, but obviously you still couldn't grasp it. Seriously, if you neither understand the graph itself (which really should not be hard), nor the thorough explanation of it that I have given you in the past, then it is obvious to me that you are the problem, not the graph. "Nov-15", "Nov-16" etc. are perfectly acceptable ways of displaying months and years, though I'll grant you that it is indeed generally not preferred on Wikipedia.

But sure, I can add more axis ticks the next time I update the graph, if it's so hard for you to understand. I'm sorry if I'm being harsh here, but I really disapprove of your intransigent way of handling this, and your cursing in Turkish in your edit summaries. Your behavior is what leads to edit wars, and your edits themselves are bordering on vandalism.

On another note, I recommend that if you want to keep editing on Wikipedia, you actually create a user. You can't go editing for several months straight without being logged in, because then we can't communicate with you directly, which again can lead to confusion whenever you feel like editing articles you don't understand. — Μαρκος Δ 22:06, 5 April 2018 (UTC)

Hummmmmm... I said the periods between months are "not clear" as the span was too short without naming ..... did I say, I do "not understand" it? .. poor you, maybe it's the right time for you to learn some English to understand it correctly. --31.202.25.100 (talk) 10:23, 23 April 2018 (UTC)

Considering you left that message explicitly asking me what it meant, I think it's pretty safe to conclude that you did not understand it. I know it must be embarrasing, but that's no reason for you to come back and lash out three weeks later. I suggest you drop it. — Μαρκος Δ 20:05, 27 April 2018 (UTC)

what I said was very concrete, ″it's not clear nov 15 nov 16 nov 17 , that these are years,...″ scroll up, read it again, and learn some English, I don't come here often, that doesn't change the sentence, funny correlation..--31.202.25.100 (talk) 11:35, 1 May 2018 (UTC)

Yet on my user page you left the following message: "Why are the dates November 15, 16, 17 on the table?" (source). After another user tried to explain to you that it was not November 15th, 16th, and 17th, but years, you proceeded to claim that "... it was updated last week means nothing, still showing november, dear [son of a bitch] ..." (source). Even in the sentence that you yourself just quoted, you conveniently left out the part where you asked "... and also where for instance march or last year may is". It's hard to tell due to the your broken sentence structure, but I assume you meant to ask where March or May were on the diagram? They were there, you were just unable to read the diagram properly. This shows that you were clearly mistaken, and clearly just misunderstood it. Any normal person would just admit their mistake and move on. Instead, what you did was to demand a change to the diagram, as if it was its fault that you couldn't read it. I even agreed to change it (and changed it weeks ago), but it seems you got too hung up on the part where I pointed out your mistake to even notice. Regardless, I think it'd be best for you to stop trying to lecture me when it comes to English; first of all, based on what you've written so far, you seem like you struggle to cobble together even remotely sensical sentences. Secondly, you don't seem to know that the original date format (MMM-YY) is perfectly acceptable in the English language; if you can't even grasp that, then perhaps you could benefit from some English lessons.

But let me repeat what I said; I have no interest in discussing this with you further. You clearly missed the entire point of my original message, and have instead just gotten completely hung up on the part where I said you didn't understand it, as if that's somehow the most offensive thing one could say; I clearly hit a nerve. This is the pettiest non-issue I've ever been forced to discuss. Come on, grow up, be the bigger man, and just drop it. Surely you must have better things to do with your time than arguing over this. — Μαρκος Δ 13:44, 1 May 2018 (UTC)

New Table for Alliances

Could we create a table for polls explicitly showing alliance vs alliance numbers? E.g. for the PIAR poll conducted on 1 May, we could use the scenario where the Cumhur alliance is on 44.7, Millet on 40.2 and HDP on 14.9? I think this would be more useful seeing as the election will not be fought party v party, but rather, alliance v alliance... Masterpha (talk) 13:11, 6 May 2018 (UTC)

Μαρκος Δ any thoughts? Masterpha (talk) 13:12, 6 May 2018 (UTC)

Question: Is it known what the voter actually votes for? Do the ballots ask you to mark a party and hint at their alliances, or do they ask you to mark an alliance and hint at the parties in whatever way? -- 2A1ZA (talk) 15:03, 6 May 2018 (UTC)

Yes, the parties are listed separately on the ballet. This is what the ballot will look like. Voting for a party in an alliance automatically counts as a vote to the entire alliance, so there's no panachage. As for the suggestion of splitting up the alliance polls into their own section, I can support this. I'll try to work out a draft for it that makes sense, and take some inspiration from similar articles elsewhere. Good suggestion! — Μαρκος Δ 15:09, 6 May 2018 (UTC)

Thanks for the links. I think polls should be presented same structure as on the ballots, in party columns, with alliances clustered. -- 2A1ZA (talk) 15:15, 6 May 2018 (UTC)

No problem. I assume you mean something like this? With some tweaks and updates, that solution could absolutely work, and would prevent the fragmentation caused by splitting alliance polls into an entirely new subsection. On the other hand it triggers other issues. Since Turkish polling is so inconsistent, and often even shows multi-scenario results in a single poll, from what point do we show the alliances? We can't show alliance names at the top of a single table as it is now, since that would give the impression that the alliances have always been in place, and that would be misleading. The proposal would therefore force us to split up the table, and create that fragmentation anyway; I cannot support the splitting up of the current table. — Μαρκος Δ 15:32, 6 May 2018 (UTC)

Yes, that "Croatia example" is what I had in mind. As to your question, maybe different background colors for alliance time? If necessary at all, as there are breaking rows for changes in alliance membership already, just like in the "Croatia example". -- 2A1ZA (talk) 15:52, 6 May 2018 (UTC)

Maybe something like this could be considered too. Looks better in my eyes. --Gbuvn (talk) 17:18, 6 May 2018 (UTC)

@Gbuvn, that would frankly be my favorite option, were it not for the fact that the Turkish coalitions were formed at different times. In Italy, "center-left" and "center-right" are sort of an integrated part of the political landscape and discourse, and pollsters usually include their totals themselves. Conversely, in the case of Turkey, the allegiances have shifted over the past few years, and there is therefore no way for us to include an entire table for coalitions in such a neat way. Specifically, since Cumhur came before Millet, I fear we couldn't include numbers for Millet without it constituting original research. Please see the proposal I posted right under this section; I'd like to hear if you think that is at least better than the status quo. Μαρκος Δ 17:46, 6 May 2018 (UTC)

Remove second round presidential election pairs with irrelevant candidates

In the "second round" subsection on the presidential election, readers get confused and at least annoyed by the mix of relevant pairs with irrelevant pairs. And technically speaking, these irrelevant pairs to not even concern the topic of the article, the election. Seeking consensus to remove the irrelevant pairs. -- 2A1ZA (talk) 14:59, 6 May 2018 (UTC)

No matter what, you are not allowed to simply remove the irrelevant pairs. This article lists opinion polling for the 2018 Turkish general election, and those qualify as such; you have no right to remove them outright. That being said, solutions have been offered to this issue on other polling articles. A good one could be to simply hide the irrelevant and now-impossible pairings under a "hypothetical polling" section. Coupled with the suggestion for a new section explicitly for alliance polling, we have to put some work into a restructuring. I'll make some changes soon, so see what you think then. But let me repeat: you are simply not allowed to remove polling from this article outright, regardless of any users' opinions on the matter. — Μαρκος Δ 15:06, 6 May 2018 (UTC)

Sir, the talk page is for discussion and not for you giving orders. I would not mind if the pairs with irrelevant candidates get moved to the end of the subsection (hidden or not), but something must be done, the current state of affairs is wrong. If you implement a solution, fine with me, but if you do not, you have to present arguments on substance why it should not be done by somebody else (which you do not seem willing to do, as you appear to even agree on substance). -- 2A1ZA (talk) 15:11, 6 May 2018 (UTC)

That was not an "order". That was me informing you that no user is allowed to censor or destroy collections of information because some "readers get confused and at least annoyed" by it, whatever that means. You requested a consensus for removing the information – which in this case is not something you can ask for a consensus for – and not for moving the information, which is naturally allowed. — Μαρκος Δ 15:20, 6 May 2018 (UTC)

I've made the change now, 2A1ZA. I'll grant you that it absolutely does clear up the article a great deal, though it also raises the issue of having too many subsection layers. What's your opinion? — Μαρκος Δ 15:45, 6 May 2018 (UTC)

Perfectly fine with me that way. -- 2A1ZA (talk) 15:47, 6 May 2018 (UTC)

Overhaul of parliamentary and seat projection sections

Berkaysnklf, Mélencron, Nub Cake, Panam2014, 2A1ZA,

Hello! Multiple users have voiced concerns and wishes for how this article should be structured, and there have been many good suggestions and contributions. One of the opinions that it seems most of us share, is that it could be helpful to include coalition totals for both the parliamentary voting intention and seat projection sections, though this issue could be difficult to resolve in practice. I propose that we go with the simple layout used in, for example, the case of Denmark.

Based on our discussions so far, I have made a draft version of a new table, which includes coalition totals going back to 18 April, when the election was called. However, more could be added later so it goes back further, but maybe not further than to the gray "2018" row, since the coalitions were only officially founded very recently, and that would perhaps qualify as original research. I suggest that in multi-scenario polls, we still keep in all the different scenarios in the "parties" columns, and just calculate totals ourselves in the "coalitions" columns. In other words, the "coalitions" columns are not meant to replace content in the "parties" columns. I have also slightly reduced the font size in order for it all to fit; presently, the seat projections in the article are too wide. Lastly, I've removed some of the innumerable "comment" rows, which contained information I consider trivial. The only ones I have kept are those marking years, changes in government, and the ones that are directly elected to the conduct of the 2018 election. Party and coalition changes have been removed, since they clogged up the table, and are not even related to opinion polls.

I'd like for all of you to have a look at it in my sandbox; if you give it the green light, I'll just go ahead and insert it into this article. Though if you have concerns or objections, please voice them here so we can try and resolve them. Meanwhile, the small details are not important now, since the table can naturally be modified later; the most important thing is that everyone is okay with the restructuring itself. I'll be awaiting your responses. Μαρκος Δ 17:40, 6 May 2018 (UTC)

Another potentially good move could be to move the year rows into the "date" columns, as that would make the table even cleaner. That's what I have done with the seat projection tables. Μαρκος Δ 18:04, 6 May 2018 (UTC)

I think the Danish model is ok, but I would find the Croatian model which you linked above better suited, easier to read. -- 2A1ZA (talk) 18:29, 6 May 2018 (UTC)

In your draft for implementing the Danish model, you would have to remove the many numbers from the AKP column which actually are Cumhur numbers. -- 2A1ZA (talk) 18:33, 6 May 2018 (UTC)

As far as I know, Croatian coalitions are always joint lists (albeit with open lists), while in Turkey, the lists in an alliance are separate. If we use the Croatian one, I fear we could give the impression that the Turkish alliances are much more compact than they really are. After all, the one and only purpose of Turkish alliances under this system, is to eliminate the threshold for smaller alliance partners. Having columns showing the alliances' totals illustrates that well, without making it look like the constituent parties are actually running jointly. And I don't think we have to remove the AKP numbers that are AKP+MHP, because, technically, they do not include BBP and are as such not strictly the final Cumhur total. The "alliance" columns show the totals for each coalition regardless of minor changes over time like that. Μαρκος Δ 18:42, 6 May 2018 (UTC)

If you go with the Danish model, which I prefer for the reasons you explained above, I think it's best to start the Cumhur numbers on the day the alliance was created (February 20, like it is in the article right now), same with Millet. The way your template is right now, it is most obvious with the May 1 polls that the "original research" doesn't really make sense. --Gbuvn (talk) 19:51, 6 May 2018 (UTC)

One question: why should the seat projections be listed with a "lead" column or use {{composition bar compact}}? I don't think that it's particularly useful to show it (especially in the case that there's a party with a plurality but short of a majority), as the individual differences in seat numbers aren't notable, and it's redundant to the content that's already there. I also don't really see the necessity in using the bar templates – yes, they look nice visually, but they don't provide any information that can't just be relayed using text or formatted like the table above it. Mélencron (talk) 21:54, 6 May 2018 (UTC)

The seat projections only show the number of seats above majority, elsewise it says "hung" (in case of plurality only). So I'm not sure I get your point. --Gbuvn (talk) 22:12, 6 May 2018 (UTC)

My point is that I don't think it's a useful column. Mélencron (talk) 22:14, 6 May 2018 (UTC)

I wouldn't say that having a column showing whether a party has a majority or not is "not useful" in a table meant to show the parliamentary situation. It contains the composition bars because 1) that is how it was done for the 2015 election, and 2) the number of total seats is changing this year, which I'd say is pretty important to illustrate. There's no way to show that in the table other than either using composition bars as we do now, or by typing it out as a note next to the table, which strikes me as redundant when it can be done easier like this. And if it, as you say yourself, also look nice visually, I think it's pretty clear we should keep it that way. Μαρκος Δ 23:36, 6 May 2018 (UTC)

Surely the HDP, as a major party which could get a substantial amount of seats, should be given a seperate table in the Alliances section? Masterpha (talk) 10:15, 7 May 2018 (UTC)

I think there are two ways to look at this. You could either see the alliances table as a separate table where all blocs should be represented, or you could see it as an addition to the main table which sums up the numbers of certain parties. Considering that there should be a section from February to May where only Cumhur exists as an alliance, the latter makes more sense to me. --Gbuvn (talk) 10:51, 7 May 2018 (UTC)

Besides, HDP is not an alliance, even if the current circumstances has left it as its own sort of "bloc". I suggest the following: in addition to Cumhur and Millet columns, we add an "others" column to the "alliance" section. There we can place HDP – as well as the Millet alliance before it was founded. This could be a solution to the issue of the two coalitions having been founded on separate dates. It is also the solution I have used for the seat projection table. Μαρκος Δ 11:12, 7 May 2018 (UTC)

I would suggest that in your example the "lead" column should still have yellow numbers (party or alliance with plurality) as "others" isn't actually a thing that would have a plurality or majority. --Gbuvn (talk) 11:36, 7 May 2018 (UTC)

Yes, that's a good point, and I'll change it. Anyway, it seems like most here would be fine with a switch to the new layout overall. So, if there are no objections or further input, I'll place it into the article later today or tomorrow. I'll also do some minor tweaks to the rest of the article so that it's consistent with the new layout (specifically, reduce the text size and move years into the date columns in the other tables as well). Μαρκος Δ 12:51, 7 May 2018 (UTC)

Final draft?

Okay, so now I have reached the point where I think we have a good balance of many of the suggestions that have come up. What I now hope is the final draft can be found here. After originally following the Danish model by placing alliance totals to the right of the party vote, it quickly became clear that it wouldn't work; the table became far too wide, and with the "comment" rows required to mark alliance founding dates, also far too messy and complicated. If it looks complicated to us, I can only imagine what a casual, non-political reader would think. That's why I've made the move to instead go for the Italian model suggested by Gbuvn, by moving the alliance table just below the party table, which also makes it more consistent with the seat projection section. This leaves only one issue: shading of alliance totals.

It was mentioned that we should shade AK Parti's seat shares even if they were smaller than the "others" category, since the category "others" technically isn't a group that can be in the lead. However, in the table for alliance vote shares, some polls conducted before the formal creation of the Millet alliance give scenarios with opposition alliances a lead over AKP. Which means that even though AKP are the only alliance, they don't have the lead, and shading them would go against the actual data in our sources. Therefore, I've decided to open up for the shading of the "others" category whenever it's in the lead in that table. This has been done in the case of Japan and likely a few others, so I suppose there's a certain precedence for it. With that, I think we have found the only solution that could work. Like I said, I'll insert this into the main article within the next day or two. Μαρκος Δ 15:10, 7 May 2018 (UTC)

I allowed myself to change the link from National Alliance (Turkey) to Nation Alliance (Turkey) which is not the same. --Gbuvn (talk) 15:43, 7 May 2018 (UTC)

Good catch, thank you. :) Μαρκος Δ 16:01, 7 May 2018 (UTC)

Great work! Although I do wonder why the size of the font was reduced to 90%? Masterpha (talk) 20:46, 7 May 2018 (UTC)

Thank you! I reduced it so that it's easier to get a quick overview of the tables; previously, they spanned more or less the entire screen from left to right, which could make that harder. The polling articles for Spain, Italy, and Germany also use somewhat reduced text sizes, for what I can only assume is the same reason. I hope it won't be a problem for nearsighted readers. Μαρκος Δ 04:28, 8 May 2018 (UTC)

I think HDP and others can be added seperately to the Alliance section and we should amend the part as "with Alliances", so that the results can be seen completely. We can note that HDP is not part of an alliance. Can be easier to understand. Just a suggestion. Berkaysnklf (Berkaysnklf) 23:29, 8 May 2018 (UTC)

You might be right, but personally I think it is already implicit that HDP, as a non-alliance party, is just listed as part of the "others" column. Especially since we already clarify in the table itself which parties belong to which coalitions. Μαρκος Δ 11:37, 9 May 2018 (UTC)

IYI Party abbreviation

Whilst I know that ‘IYI’ may be seen as a suitable abbreviation for the party due to the fact that it is only 3 letter, I do think that it would be better for us to refer to it as ‘IP’. IYI is the first word of the party, and it’s full name is actually ‘Iyi Parti’. Same with AKP- ‘Adalet ve Kalkınma Partisi’, CHP-‘Cumhuriyet Halk Partisi’ or SP: ‘Saadet Partisi’ but we don’t abbreviate them as ‘Adalet ve Kalkınma’, ‘Cumhuriyet Halk’ or ‘Saadet’ because that would literally mean ‘Justice and Development’, ‘Republic Peoples’ and ‘Felicity’. If we put ‘IYI’ we’re literally just putting ‘Good’ as an option for a party. ‘IYI Parti’ is it’s full name, and therefore ‘IP’ should be it’s abbreviation. Masterpha (talk) 13:14, 13 May 2018 (UTC)

If IP is actually an abbreviation that is used in media, then I guess it could be changed. But if it's just about the logic that you're using, then Labour can't be an abbreviation for Labour Party, same for Liberal or Green. --Gbuvn (talk) 16:28, 13 May 2018 (UTC)

Here, here and here. Also in turkey they don’t leave out the ‘parti’/‘partisi’ out of the acronym of the parties, so it’s different to the UK/NZ Masterpha (talk) 17:44, 13 May 2018 (UTC)

Fair enough then. However, I'd like to point out that in the UK/Canada/NZ there are absolutely parties that use the P for party in their acronym, like UKIP, NDP, TOP, SNP... My point is that one way to abbreviate a party's name doesn't necessarily stop the other way from existing in the same country as well. Sometimes multiple abbreviations are used for the same party. Like Lega or LN. --Gbuvn (talk) 19:09, 13 May 2018 (UTC)

Well, believe it or not, some western media actually does use simply 'AK' to refer to the AK Parti, while in Turkey, it is extremely common to use simply 'Saadet' for Saadet Partisi. Meanwhile, the use of İP is close to non-existent. Yes, it is occasionally used, but not even remotely close to as much as İYİ, not by a long-shot. Considering there is no doubt that the full 'İYİ' designation is (1) more commonly used, and (2) the party's own preference, not to mention it is more reflective of the party's identity, in that it is meant to symbolize the Kayı banner. That banner is essentially the party's only reason for picking the name and color. If parties choose to embrace a specific, deliberate branding in the way İYİ does, then we have no right to impose our own obscure, "artificial" abbreviation on it. The only statement that comes close to being an argument for a change, namely "this is how some other countries do it", does not count for anything here. Finally, let me just ask you to go on Google or, better yet, Google Images, and search for the word "iyi". That should make it completely clear to anyone in doubt that İYİ is the abbreviation of this party. Meanwhile, search results related to "İP" in the context of Turkish politics will yield results for the now-rebranded Workers' Party (İşçi Parti), which could cause confusion. Μαρκος Δ 23:12, 14 May 2018 (UTC)

İP is absolutely definitely not the abbreviation for this party... three (well, two) media sources incorrectly abbreviating it does not supersede the 1000s of other, more credible media articles that use İYİ. Nub Cake (talk) 00:55, 26 May 2018 (UTC)

Ordering of parliamentary section

@Μαρκος Δ: As I stated when making that edit, I think the alliance vote table should come first as some polls are not even asking about parties separately but only alliances. We also talked about how polls where parties are asked separately are still summed and added to this table anyways - so the alliance table is more comprehensive anyways. Also you are saying that people vote for parties but even if you stamp your ballot smack in the middle of two allied parties it will still be a valid vote for the alliance (even though it normally would not be valid for a single party). The graph is also not very useful as I said since so many pollsters are biased or straight up propaganda machines, and this inaccuracy or misleading picture can be seen with a 10% change within a month in the AKP vote there. Junk2711 (talk) 21:14, 8 June 2018 (UTC)

Well, your personal evaluation of what is more important does not really matter here. It is convention for polling articles which include alliance summaries to place those at the bottom of the article. It is also nearly universal that polling graphs are placed above the main table. I recommend that you take a look at virtually any other polling article to see what I mean. Here's an example. We are not breaking that convention because of your personal view, which I frankly don't even understand. For example, your argument against the polling graph makes no sense. You say it's not reliable because of biased pollsters. Uh, yeah, but the same applies for the polling tables as well. So, are we to omit the tables, too, then? The graph illustrates polling averages, which should logically be more reliable than individual polls. That should not be hard to understand. And for the record, attacking a user for ethnicity, language or religion, as you did in your edit summary, is naturally unacceptable. I am sure you are aware of this, and will take a more civil tone in the future. Μαρκος Δ 10:47, 9 June 2018 (UTC)

@Μαρκος Δ:You cannot accuse me of attacking you "for ethnicity, language or religion" for this, get a grip. That's a serious offense and I won't take it. Typical immature behavior on here to accuse someone of a serious violation in even the smallest dispute - don't do this. I don't know anything about you and I stated your low level of Turkish (as stated publicly on your user page) could be why you are not understanding the situation - where is there an "attack" on you as a person?

Second, I don't mean the graph is useless because the pollsters are biased. I am saying that there are crazy swings on there due to biased polls - therefore making the graph pointless as it doesn't mean anything when someone says the AKP is at 40% and the next poll says they are at 55% or whatever. This just causes spikes indicating a trend that isn't there. A table gives you all the data whereas a graph here is supposed to indicate trends - this is impossible for the reasons I have given. This is the issue and why I'm saying it doesn't help. You are saying it should be more reliable as it represents averages and I agree, but only if the pollsters were all honest.

Third, unless there is some written rule I am missing, "convention" is not at all binding here. I don't know the situation with those other elections but they are different elections, with different systems, in different places. The only point of mine that you have refuted is regarding the graph. OK, let's leave that aside for a moment. What about my other points? I have given valid reasons which you haven't disputed, I don't see what the issue is besides you not wanting to change because of "convention" and how the article was originally. You said to seek a consensus for the change so let's look for one then?

I am also stating this again: all polls now ask an alliance vote while I have seen numerous polls without separate results now. We have also previously summed up results from the party table and added them to the alliance table. The alliance table is therefore inherently more comprehensive. -Junk2711 (talk) 23:08, 9 June 2018 (UTC)

People vote for parties, not alliances [...]
— User:Μαρκος Δ

They can vote for alliances. betseg|g 17:15, 12 June 2018 (UTC)

@Betseg, yet political parties remain the basic electoral unit, as opposed to e.g. joint lists or the loose type of coalition that may be formed now. That's my point. The alliance system is a new addition to the Turkish electoral system, and for all we know it might be abolished again by 2023. It should not be hard to see that alliances are less central to the functioning of the electoral system than individual parties are. @Junk2711, I suggest you calm down; you have to be able to discuss these issues without resorting to that tone.

Firstly, it is clear your targeting of my level of understanding of the Turkish language was meant as nothing other than a snarky insult. Any well-informed contributor understands that there is just as much information on the Turkish electoral system in English as there is in Turkish. Yours was not a factual argument, but a blatant ad hominem, reminiscent of the "you don't even live here so butt out" that so often found in online comment sections. That tone does not belong here.

As for your second argument, it is still completely nonsensical, and I therefore don't even know how to respond to it.

To respond to your third point, yes, there is very much a convention with regards to the structure of polling articles. The placement of graphs above tables is virtually universal, and as has been pointed out in discussions above, the alliance system in Turkey serves no other purpose than allowing for smaller parties to circumvent the threshold. The Italian coalition system is a far more important determinant of elections there, yet they still place alliance totals below party totals. Considering the Turkish electoral system is not nearly as special or unique as you apparently seem to think, and that it is in fact very much comparable to the systems found in a variety of other countries — there is thus no doubt that this article should mirror those of countries with similar systems as much as possible. If you want to change the structure that has been accepted and implemented in a large number of articles based on nothing but your personal, objective views, then I suggest you take it up with this project and community at large. Though preferably in a more mature tone than the one you have regrettably resorted to here. Μαρκος Δ 17:52, 12 June 2018 (UTC)

@Μαρκος Δ: It seems you must have the article remain as such so have your way. I have no personal reason for this change, I am just trying to correct it. I think this makes 3 of us changing the order or saying you can vote for alliances but you seem to be the only one arguing otherwise. It does not matter what the situation is with other elections because there is no set rule and every situation is different. I never claimed this system is unique or special so I'm not sure where you're getting that from. I am simply saying every country or election is a different case and should be judged independently rather than by what other past articles have.

I also do not appreciate you still claiming "ad hominem" and it is certainly not "clearly nothing other than a snarky insult". This is typical immature behavior to just nitpick and claim a personal attack in the slightest dispute. I also have no idea where you live which is why it seems bizarre that you think I am here to just target you for that. Lastly, you may not agree with a point but calling it nonsensical is not very constructive, it is still explaining something. — Preceding unsigned comment added by Junk2711 (talk • contribs) 21:23, 13 June 2018 (UTC)

About the graph: If the consensus is that it doesn't help at all it should be removed rather than placed below the tables which doesn't make sense at all. I'd like to point out though that heavily differing polls by different institutes are not unique to Turkey (see here (VVD, FvD)). It is definitely possible to get rid of false trends by changing the method to draw the graph.

About the order of the tables: From what I can see, the parties are mainly campaigning under the party's brand rather than an alliance logo. That leads me to thinking that the parties table is more important. --Gbuvn (talk) 20:08, 13 June 2018 (UTC)

I agree with @Gbuvn on all points, though I must again stress that graphs are in fact our strongest tool to display overall trends in countries with strongly different or inconsistent polls. So, again, I have to say that @Junk2711's argument simply does not make any sense. There is also another important point I think the latter user could benefit from learning: section order/structure is not necessarily indicative of relative importance; specifically, the sections at the top article are not necessarily more important than those further down. As such, the entire argument that we should restructure the article based on what is more important, falls completely and utterly flat. Other than that, the only pseudo-argument you've conjured up is that there is supposedly no convention that blocks a change. But come on, look at other polling articles – they all follow the same pattern, and have extremely similar structures. Please show me one article where the graph is below the table, for example. I mean, at this point it really looks like you just want to shake up the article out of spite. Why insist so vigorously on fixing something that clearly isn't broken? Μαρκος Δ 21:52, 14 June 2018 (UTC)

Biased selection of polls

The polls here seem to have been selected to push an anti-AKP/MHP/Cumhur ittifaki/Erdogan agenda and show them as less popular as possible.

There is no Andy AR or Metropoll which are considered big polling companies. Yet there is PIAR which is biased and so far out even compared with opposition friendly polling stations like Sonar and Gezici.

My suspicion is that this biased selection will be used to fuel accusaions of "rigging" - you know what the opposition uses to explain the fact that the Turkish public rejected them again. — Preceding unsigned comment added by Ozan454 (talk • contribs) 12:59, 9 June 2018 (UTC)

We include all polls we can find here. We are not going to remove any of the ones you are complaining about, but you can add additional polls from reliable agencies, if you feel that there are any agencies that are under-represented. A good start would be adding those Andy AR and Metropoll surveys. That being said, it seems very clear that you have an agenda, so be careful to remain neutral if or when you make edits here. Μαρκος Δ 14:08, 9 June 2018 (UTC)

The people who purposely ignored major polling A&G and Metropoll. Piar is so ridiculous that it is about 10-15% from even polling organizations that are positively biased towards the opposition. Piar's twitter account show's it own political leanings very openly.

Haha you have even included CHP which is not even a polling organization, but a opposition political party. There is not even a sample or methodology given. I guess that when the election results come in and don't match their biased so called "polls", they will be crying about "rigging" and "cheating" as always :)) — Preceding unsigned comment added by Ozan454 (talk • contribs) 11:30, 13 June 2018 (UTC)

Instead of explaining your political view, which no one here cares about, you could just add the polls that are missing instead of lamenting. --Gbuvn (talk) 12:19, 13 June 2018 (UTC)

Lead in 1.1.4 Abroad?

In 1.1.2, Lead #1 is the difference between the highest and second-highest score, and Lead #2 is the difference between the 2nd and 3rd highest. In 1.1.4, Lead #2 is the difference between highest and second-highest, and I just can't figure out what Lead #1 is. No possible difference amounts to the 34.8 percent given in the table. Is that an error? --Thorwyn (talk) 16:18, 15 June 2018 (UTC)

That's not the only one. Sometimes the lead #2 is wrong because the 4th placed is used accidentally. --Gbuvn (talk) 21:43, 15 June 2018 (UTC)

As I've said before, I think we should simply remove the 'lead #2' column altogether, since it seems to be a recurring point of confusion, and since it is not done in any other artice. Μαρκος Δ 11:35, 16 June 2018 (UTC)

Parliamentary simulation for "plus kayak" poll doesn't have the right calculation in alliances section. Basically it won't make 600/600 Guest

It has the right calculation, but the missing 26 seats are listed as "uncertain" in the source. Unfortunately, it seems to be a recurring problem for us that many Turkish pollsters display their findings in unorthodox ways. The numbers for each alliance in this case are their minimum seat shares. I'm not sure what to do about this particular case. Μαρκος Δ 11:33, 16 June 2018 (UTC)

Summary of the last 9 polls

https://indigodergisi.com/2018/06/en-son-secim-anketleri/

A very good collection of the last nine opinion polls in Turkey.

Abroad election poll

There is a significant percentage (9.1%) missing, in special if you take in account that there is a share for 'others'. So that does not seem to make it a reliable poll then. Maybe removing it?Nurkartal (talk) 16:08, 21 June 2018 (UTC)