Talk:Opinion polling for the 2019 Greek legislative election

PAMAK opinion poll
The latest PAMAK poll was added, removed, added and removed once more. The last remover said the poll only concerned the Thessaloniki A constituency. However, I see no mentioned of that in the PDF. What I see is this: "Κάλυψη - Σύνολο της Επικράτειας". Which means, "covers all the country". Shouldn't the poll be added again? There are already 2 other PAMAK polls. I have also read about about a Thessaloniki A opinion poll by the same people but the results were different so IMO the last remover is mistaken and the poll should be added once more (see http://greece.greekreporter.com/2016/05/15/syriza-20-points-down-in-thessaloniki-poll). Orgyn (talk) 20:30, 26 May 2016 (UTC)
 * The PDF added for the last edit was from PAMAK's March poll. It is, indeed, a poll for the Thessaloniki A constituency, so this is not its place. Impru20 (talk) 20:36, 26 May 2016 (UTC)
 * Ah yes, I see. The PDF link for the poll was wrong, it was actually the same as the previous PAMAK poll and the results in the corresponding row were for the Thess A constituency... Orgyn (talk) 11:26, 27 May 2016 (UTC)

New opinion polls
Here are some new opinion polls that have not yet been added to the article, feel free to add them: Orgyn (talk) 11:36, 20 January 2017 (UTC)
 * 17-19 January http://www.parapolitika.gr/article/anigi-ki-allo-psalida-iper-nd-provadisma-9-monadon-enanti-tou-siriza
 * 15-17 January http://www.skai.gr/files/1/000000/NewFolder/Final_SkaiTaseis_15-17Jan17%20(1).pdf
 * 21-22 December http://kaparesearch.com/index.php?option=com_k2&view=item&task=download&id=82_96d7e3b0299c56819b1c41bfc3202d0a&Itemid=137&lang=el

New Updates
hi can somebody update the Graphical summary it hasn't been updated since December 2016. thanks Torygreen84 (talk) 10:23, 8 October 2017 (UTC)

ALCO/Kontra channel opinion poll
The latest ALCO/Kontra channel opinion poll of the 22 May is not weighted like the other surveys, because they consider undecided people (19,6%) as part of the sum. What should we do? --Broncoviz (talk) 16:07, 4 June 2018 (UTC)

Misleading presentation of the actual data of the opinion polls
The data provided in almost all the opinion polls presented are completely wrong. This is extremely easy to check by crosschecking the actual data published by the media and the opinion polling companies with the data presented in this wiki page. Following I indicatively examine the last 5 opinion polls:

Rass/in.gr, 21–24 Jan 2019

Percentages presented in wiki page: ND 36.6/ SYRIZA 25.2/ X.A 8.6/ KKE 8.4/ KINAL 10.3// ND leading by 11.4

Percentages as published by in.gr:  ND 31.4/ SYRIZA 21.6/ X.A 7.2/ KKE 7.4/ KINAL 8.8//  ND leading by  9.8

source (that is the SAME presented at the wiki page): https://www.in.gr/2019/01/24/politics/kyvernisi/megali-dimoskopisi-sto-gr-stis-10-monades-diafora-nd-syriza-posa-kommata-mpainoun-sti-vouli/

"Pulse RC/Skai, 20–22 Jan 2019"

Percentages presented in wiki page: ND 38/ SYRIZA 26/ X.A 9/ KKE 6.5/ KINAL 7.5// ND leading by 12

'''Percentages as published by skai.gr:  ND 33/ SYRIZA 22.5/ X.A 8/ KKE 5.5/ KINAL 6.5//  ND leading by  10.5 '''

source (that is the SAME presented at the wiki page):http://www.skai.gr/news/politics/article/395230/dimoskopisi-skai-to-62-kata-ton-prespon-dipsifio-provadisma-nd/

Opinion Poll/parapolitika.gr

Percentages presented in wiki page: ND 42.5/ SYRIZA 22.2/ X.A 5.8/ KKE 5.5/ KINAL 5.1// ND leading by 20.3

'''Percentages as published by skai.gr:  ND 31.5/ SYRIZA 16.5/ X.A 4.3/ KKE 4.1/ KINAL 3.8//  ND leading by 15 '''

source (that is the SAME presented at the wiki page):https://www.parapolitika.gr/article/dimoskopisi-meta-tin-apochorisi-kammenou-sto-15-diafora-nd-siriza

Pulse RC/Skai, 16–18 Dec 2018

Percentages presented in wiki page: ND 38/ SYRIZA 26/ X.A 8/ KKE 7/ KINAL 8/ ND leading by 12

'''Percentages as published by skai.gr:  ND 33.5/ SYRIZA 23/ X.A 7/ KKE 6/ KINAL 7//  ND leading by 10.5 '''

source (that is the SAME presented at the wiki page):http://www.skai.gr/news/politics/article/392732/epestrepse-se-dipsifio-provadisma-i-nd-pos-pane-ta-alla-kommata/#ixzz5aGdKEKqo8

Alco/Open, 5–10 Dec 2018

Percentages presented in wiki page: ND 32.8/ SYRIZA 25.4/ X.A 8.6/ KKE 7.4/ KINAL 6.4/ ND leading by 7.4

'''Percentages as published by skai.gr:  ND 27.6/ SYRIZA 21.4/ X.A 7.2/ KKE 6.2/ KINAL 5.4//  ND leading by 6.2 '''

source (that is the SAME presented at the wiki page):http://www.alcopolls.gr/portfolio/%CF%80%CE%B1%CE%BD%CE%B5%CE%BB%CE%BB%CE%B7%CE%BD%CE%B9%CE%B1-%CE%B5%CF%81%CE%B5%CF%85%CE%BD%CE%B1-open-%CE%B4%CE%B5%CE%BA2018

As it can be seen, ALL polls are mispresented and should be changed as being misleading and manipulated. — Preceding unsigned comment added by Ka79 (talk • contribs) 13:11, 31 January 2019 (UTC)
 * As it is explained in the note shown above the table, results shown are merely those shown in opinion polls but excluding refusals, don't knows and abstentions. It is a simple WP:MATH operation which is frequently conducted by many of the pollsters as well, and brings no difference proportionally. Impru 20  talk 13:24, 31 January 2019 (UTC)

The actual numbers of the polls are not the same as the ones in the table. Someone has put intentionally wrong infrmation on almost all of them. As a result, the graph is also wrong.Gomoloko (talk) 10:41, 15 May 2019 (UTC)
 * It is not wrong. Results show are poll results excluding undecided and abstention rates. This is perfectly valid under WP:CALC (note that in the edit summary I wrongly referenced WP:MATH), and is needed as there is no other way to make all polls comparable to each other as well as to election results (since some pollsters do not publish results including abstentions and/or undecided, and election results are typically not shown including abstentionists). This is the way many pollsters do calculate their results, so it is also a reported and valid system for Greek opinion polls. Impru 20  talk 16:01, 15 May 2019 (UTC)

Yes, but one cannot assume that the votes of the undecided will be distributed equally, actually this does not happen. Making the assumption you said, the numbers become extraordinary. Also, almost all of the polls i corrected are of the same type (they contain undecided, white vote, invalid, and abstention). I insist that we should use the real numbers, all polls contain the initial information, the rest are arbitrary assumptions Gomoloko (talk) 16:13, 15 May 2019 (UTC)
 * Again, this is what Greek pollsters do to their own results: substracting undecided and abstentionists from their own results, then give out the results as the headline vote intention. Metron Analysis, Alco or Marc, among others, have been shown to do this; Pulse RC also frequently does this but rounding results to the closes .0 or .5. Indeed, nearly all Greek pollsters have done this at some point or the other for the past two decades, the only notable exception of a pollster not doing that are Palmos Analysis and Public Issue (and as a result, this method is not applied to their polls. Public Issue hardly ever gives out raw data, however).
 * This is not only customary practice in Greece (your assumption of numbers becoming "extraordinary" a result of your own opinion), but this is the only way for Greek polls being comparable to each other. Impru 20  talk 16:20, 15 May 2019 (UTC)
 * Further, once you get reverted, you don't revert again before discussion ends, as per WP:BRD, lest you risk falling into edit warring. This might have took "a lot" work of you, yet it also took a lot of work for many other users to add the very poll results you now seek to unilaterally change. Unless consensus goes your way, you shouldn't try to impose your view. Impru 20  talk 16:22, 15 May 2019 (UTC)
 * Ok, we have the true numbers, somebody(who?) changes them using methods that even some pollsters doubt, but it is me the one who unilaterally!?!? changes and misleads!?!?! The person that puts in the real numbers! Also most of the polls using these methods prove wrong after the events they are involved in, most of the times. Furthermore, the difference between the fisrt 2 parties changing a lot, is not my opinion, anyone can do the math! I don't try to impose anything as you say, I am just talking about what is right! We have the numbers but we put different in the table. Great! Gomoloko (talk) 16:36, 15 May 2019 (UTC)
 * Yes, it is you who mislead. The absurdity of the complain may be evidenced evidenced just by looking at this: take the Metron Analysis's latest poll, for example. You repeteadly changed the numbers from 27.3% Syriza and 37.5% ND to 21.0% and 28.8%, respectively. Then, you check the source and... suprise! You see the source itself making the calculation into 27.3% and 37.5%.
 * You came here assuming bad faith from everyone in here (please, consider how WP:AGF works): assuming that people intentionally put "wrong numbers" in order to mislead people. Indeed, there's a note just above the table clearly stating that Refusals are generally excluded from the party vote percentages, while question wording and the treatment of "don't know" responses and those not intending to vote may vary between polling organisations. This can be made clearer if required, but no one is seeking to mislead anyone here. Indeed, the presentation of raw data not comparable between pollsters due to varying undecided/abstention methods is what can lead to misleading readers: you couldn't have a chart with raw data, and you couldn't have a table in the first place due to the impossibility of comparing polls and results. We have more than enough evidence that Greek pollsters do use methods to translate raw results into vote estimations comparable to each other, so what is done here is a basic rule of calculation that is supported by sources themselves. Impru 20  talk 16:47, 15 May 2019 (UTC)
 * You should take it back, I am not trying to mislead anyone, and you accuse me this is a personal attack. I am not a pollster, I just saw differnt numbers in the table than the actuall on the polls, so I started correcting them. Now that we 're talking I am trying to see why this happened. And at the same time I explain to you some things. First ywho changed the numbers and with what method(if this is not clarified how can we check them?).Second, do the differences between the fisrt two parties and between the others change(creating wrong impressions) or do they not? Third, I suggested that we can use the same type of poll numbers (excluding those who will not vote, white vote and invalid) but Fourth not excluding undecided because pollsters in tv admitt, and also logic says that the votes will not be distributed equallyGomoloko (talk) 17:01, 15 May 2019 (UTC)
 * Someone has put intentionally wrong infrmation on almost all of them This is what you said back in your first comment in this discussion. This may be seen as a clear breach of WP:AGF, and unless you are accusing yourself of personally attacking others, you should maybe put the personal language down in the first place. You're the one who came here accusing everyone else of intentionally misleading people, so don't get surprised when your own behaviour comes under scrutiny precisely because of that. Indeed, you are not a pollster, nor a reliable source under Wikipedia standards, so we may better bring this part of the discussion to an easy end.
 * On the calculation method, it's fairly easy: undecideds and abstentions are excluded (blank ballots as well, when the pollster does give the number, which is not always), then the numbers re-calculated to 100%. An easy rule of three. I wouldn't apply it for other countries, but for Greek polls it does not only come as a necessity for easy comparison and functionability, but it's also a method frequently used by pollsters themselves and backed up by sources, so that basically supports its use. On the lead/party comparison and so on they are fully proportional to raw data numbers, so it's mathematically impossible they can give out misleading information to the reader: it's just the raw data amplified. So, if ND is high in the raw data, the re-calculation gives an high result; if ND and Syriza are nearly tied in the raw data, they'll be tied after the calculation as well.
 * Finally, on the possibility of also showing results without excluding the undecided and so, it was tried in the past but it is a nightmare to handle and unworkable to make it practical. Because pollsters do not give out similar raw data information, you can't really make a table of opinion polls with the full raw data. Some pollsters do not show blank ballots, some others already show results re-calculated by excluding abstentionists and/or undecided, some others combine their undecided with a "not sure" column and others may show up to six different "not sure/undecided/abstention/not voting/blank/etc" columns. This without discussing the logistical and navigation issues resulting from trying to present different tables (possibly more than just two) with such information.
 * As you see, there is no will to mislead or distort results in any way. Much to the contrary, the current scheme is designed so as to present information as pure as possible while making them visually operational and functional for readers, maintaining some coherence throughout the use of the only method most pollsters do agree with (and pollsters using different methods do publish their estimations anyway, so they can be presented as well). Impru 20  talk 17:30, 15 May 2019 (UTC)
 * I did not accuse everyone as you say, just the person/s that changed the numbers. That was my first thought when i saw the table, i am now not sure someone did it on purpsose. But this is not the important. So I am not a reliable source, the person who put the initial numbers, nor the citations and the data of the companies polls, but someone else who changes the numbers in his opinion is. Anyway. The only reason i continue this is because when the original data change, the differnces between the parties also changes and this is something both pollsters admit and logic says (undecided votes don't distribute equally). But in th end I have no gain through all these, so I don't care. This is the way you like it? Ok. Polls never predicts what will happen, they only try to manipulate the voters. Keep going, but without me. Last quiz, since you mentioned it above: if we meke the calculation 37.5-27.3 do we get the same result as with 28.8-21.0. I also invite everyone to see the numbers in the table after the elections.. Gomoloko (talk) 18:13, 15 May 2019 (UTC)
 * Considering that many people did change/add the numbers, you were accusing many people by default. This, and the fact that you yourself acknowledge you were accusing someone. You should have never started a discussion launching any such accusation against anyone. Read WP:NPA on this.
 * if we meke the calculation 37.5-27.3 do we get the same result as with 28.8-21.0. Proportionally, yes. But in comparative terms, you can't work with raw data to directly compare it to other polls and to election results over valid votes. Further, the source itself gives the 37.5-27.3 calculation, so there's little way you can discuss that.
 * I also invite everyone to see the numbers in the table after the elections. Because we now report opinion polls based on how accurate they end up? Hahaha let's delete 99% of opinion polling articles in Wikipedia then! I invite you to bring some justice to Opinion polling for the 2015 United Kingdom general election. Numbers were clearly flawed when compared to actual election results, so there must be some manipulation and something must be done! (irony)
 * Now seriously: as provided, sources and pollsters do use this calculation method so it's not as you say. However, if you keep going on your accusation that this is somehow done in purpose to manipulate numbers despite it having been disproven, then that's outright bad-faithing from yourself, and I will leave you alone in that. Cheers. Impru 20  talk 18:37, 15 May 2019 (UTC)
 * Why do I have to say things twice? After the conversation we made I don't believe you did it on purpose. As for the rest, I am long gone from this conversation, so guess who will be alone. :) Gomoloko (talk) 18:56, 15 May 2019 (UTC)