Talk:Opinion polling for the May 2023 Greek legislative election

Proposal that three parties are added

Course of Freedom, Ean and Niki have appeared in the latest polls after the ban on National Party – Greeks. Their percentages are around 1,5%, but I think they could be added on the list as the sum of these figures show consistency with the banned party's. Please tell me what you think. Argybz (Argybz) 16:34, 11 May 2023 (UTC)

@Argybz: No, as per what has been explained in both my edit summaries (here and here). I had not even seen this comment here, but you yourself acknowledge the polling numbers of these parties are not so much. This comes after you yourself acknowledged the ban of EKE is not a big deal (surely not even close to requiring a table split between before and after the ban). Also, parties run independently; what their combined numbers may add up is irrelevant in terms of reaching the 3% threshold. Do this really require a major table split? Really? Surely not. Also, consider that there is a previous consensus to remove parties of little relevance from the table.

Plus, don't use edit summaries as a way to reinsert contested edits: you made your edit, you were contested and reverted, so follow WP:BRD instead of engaging in an unnecessary edit war, please. Impru20^talk 20:53, 14 May 2023 (UTC)

@Impru20: The exit polls are out and Course of Freedom seems to reach the 3.0 threshold. What is your opinion on the matter now? Argybz (Argybz) 16:04, 21 May 2023 (UTC)

One exit polls give them seats, not the others. If after the count they win seats, then yes, we should add them. If not, then they shouldn't. Impru20^talk 17:43, 21 May 2023 (UTC)

Proposal that two parties are removed

The list of parties polled is very large. I think some are unnecessary. Neither the Union of Centrists (listed as EK in the table), nor Recreate Greece (listed as DIXA in the table) have any members of Parliament and both haven't been included in any opinion polls for almost a year. I think if more than a year elapses and a party hasn't been included in any polls since and has no members of Parliament its reasonable that we remove them from the table. So, in summary if neither of these two parties are included in any polls by 20 November, I advocated we remove these two parties. Please leave your thoughts below. Helper201 (talk) 18:20, 9 November 2020 (UTC)

​​​​​​ Please see my post above about possibly removing two of the listed parties. Your feedback would be much appreciated. Helper201 (talk) 20:44, 19 November 2020 (UTC)

Yes, I don't really understand those parties' presence in the tables. Their current polling data is almost non-existant, and it's fairly clear at this point that they won't get polled anytime soon. They're merely occupying space. Impru20^talk 15:48, 20 November 2020 (UTC)

Impru20 is there any chance you could remove these two parties from the table please? I have tried to do so myself but upon viewing my edits in preview I seem to be making some errors. Helper201 (talk) 21:16, 20 November 2020 (UTC)

Proposal to include a separate section where the percentages are not re-calculated

Although it certainly is helpful to see what the results would be by re-calculating the numbers by excluding those who will abstain or haven't decided yet. I think it would be equally helpful to have a section where it is just the voting estimate and the percentage of people who plan to abstain or have not yet decided is shown. Politics in Greece are not that stable so I think that these categories should be included so that people can see the whole picture. There is no guarantee that the people that have not decided who they are going to vote for will go for in the end. By not including the information that shows that a substantial amount of people are undecided, it is quite possible that people will make false assumptions about the popularity of the parties. Lastly, I think is important to have a category for the "other" parties because it is an important factor that shows how pleased the people are with the parties in parliament and their percentage will determine how many seats each party will get after the elections are over. Please consider adding another section that includes those categories. C0rsp1m (talk) 12:58, 17 April 2021 (UTC)

I'd agree, since there already is one for the January 2015 election, if im not mistaken. Thepottato (talk) 11:02, 29 July 2021 (UTC)

I think that the formula that converts the results should be provided beforehand. So that it is clear for the reader to understand how the poll numbers were calculated (simple-rule-of-three), and why they do not match their corresponding referenced article. On the same grounds, I think it is far more useful and self-explanatory to have the original poll numbers in a table, as they provide an intuitive way to compare results, either between polls or elections. I think the calculated results are still useful, as a second table, having their formula explained. Even giving a single row example. NordMarios (talk) 15:55, 27 September 2021 (UTC)

It's literally explained just above the table: Polls that show their results without disregarding those respondents who were undecided or said they would abstain from voting (either physically or by voting blank) have been re-calculated by disregarding these numbers from the totals offered through a simple rule of three in order to obtain results comparable to other polls and the official election results.

Also no, precisely the calculation is done to allow for a proper comparison between polls and elections. Elections results do not include a field for "undecided", and pollsters often do not match each others' when deciding how to present their results: some polls offer results for "undecided" only, others for "undecided/abstaining", others for "undecided/abstaining/invalid", others for "undecided/abstaining/invalid/not answering"... and as many combinations as you can think of. Without the recalculation, polls are not comparable between each other. Impru20^talk 16:06, 27 September 2021 (UTC)

Thank you for the answer Impru20. What I mean about providing the formula, is actually providing the formula. Not the explanation of the formula. And also an example e.g take a row of the opinion poll table and show how it calculates a row in the "voting intention" table. I understand that the "voting intention" table's purpose is to provide a comparison between poll results and election results. But it does not provide what the title suggests. The opinion poll results. I would recommend that in this page there should definitely be the opinion poll results. And a different page with the calculated opinion poll results with a different title, maybe : "voting intention estimates from opinion polls". Or in the same page, but as a second table, as it is a derived result. NordMarios (talk) 10:45, 29 September 2021 (UTC)

The formula is already explained to consist on a simple rule of three disregarding undecided, abstainers and any other figure not actually voting for a party. It is not directly provided because the link to the rule of three article is already given, as well as the link to WP:CALC; the actual mechanism of how a rule of three works is surely much more nicely explained in that article that it would ever be in this one. We do not need to explain basic concepts such as what a political party is either, right?

Independent polling aggregators (such as EuropeElects or Electograph, to name two, but you could find more) do present polling results in a similar fashion. Considering this, as well as WP:CALC, it is somewhat daring to argue that the article does not present "opinion poll results". It does. Cheers. Impru20^talk 07:50, 29 September 2021 (UTC)

Thanks again for your prompt response Impru20. And thank you for all of your detailed explanations. But I feel I may have offended you somehow. Forgive me if I have. My intention in this discussion is to maybe provide some insight into some improvements. And make this article that contains such valuable information, even more friendlier to newcomers, like I was when I first read it. But, I see that there is no intention of discussing any changes, so I can leave you be. NordMarios (talk) 11:35, 30 September 2021 (UTC)

I think this is a good place to replay to revert and find consensus. I'd like to start with reason why I even bothered with expanding the table with polling results. I found Dimiourgia started to appear in latest polls, but it was absent in the table, so I just wanted to check if it was missing in other places (turns out - it was) and also add other missing entries if there were any. Now replies:

"mere inclusion in a poll is not enough" - I'd like to argue here. I see opinion polling pages on wiki as an archive, a place where every poll is saved in case original source is deleted, but also aggregator for polls, a place where you can easily check the support for political parties and changing trends.

"it leads to whole empty columns" - This is a valid point. I have two solutions for that, both equally good - a) use footnotes (like AGROunia in Polish page; it appears once in a while but not frequently enough to have its own column), b) divide the table into parts by years or months based on when party starts to appear more/less frequently in polls (like in Ukrainian page; not perfect, but it's a good example).

"The scarcity of results is such for these that they cannot be properly represented in the chart with R, since their results are negligible. The "Others" column is also worthless since it each poll includes a different range of parties within that category (cannot be properly represented in the chart, either)" - Minor and negligible parties as well as "others" may be skipped in the chart (like Russian page - only 5 largest parties are shown, no minors, no "others", chart is clear). These columns are mainly informative. What one does with such information is another story.

Regarding "Others" meaning many things - I went through literally every poll on the page and everywhere it's either "other parties" or there are listed results for "undecided/abstain/invalid/not answering" leaving no other option than "other parties", or results already exclude people abstaining and giving blank votes. Besides, when I saw "others" in the polls, it always meant "I choose an option, but it's not listed here". I see no reason to treat Greek polls differently in that matter. Leo0502 (talk) 17:40, 24 May 2022 (UTC)

Replying to these points:

• "I see opinion polling pages on wiki as an archive, a place where every poll is saved in case original source is deleted, but also aggregator for polls" First error. Wikipedia is not a database, an archive or an indiscriminate collection of information. Information collected in Wikipedia articles must be noteworthy. You'll notice that for almost all countries with opinion polling entries in Wikipedia, not all parties get a mention; basically because there are not notable.
• Use of footnotes should be limited. You cannot have an endless string of footnotes (something that would be a consequence of your proposal here) to mention party results that are not noteworthy. Instead, if a party gets enough coverage in polls and sources, we can consider adding a column for that party and listing all of their results. ED is a possible candidate for that, but we still need to see whether that trend consolidates or not: it wouldn't be the first time that a party seems to get coverage in a large number of polls with minor support (<3%) only for such coverage to disappear in a matter of weeks/months. We are not in a rush to add parties here.
• On your third point, that is contradictory: what is the point of listing parties with negligible support that cannot even be represented in charts due to lack of data? Again, the "archive" reason is not enough in itself as depicted in the first point. On the Russian page, it is a really big clutter (listing parties with 0% support in polls... lol) in which you seem to be the main contributor from the beginning. Basically, the page's current design is your work, so I don't think that counts as an objective example since you seem to have implemented your own vision of which parties should be in the tables.
• The point with "Others" is that it does not include the same parties, since from time to time some pollsters give results for some of these separately (i.e. Antarsya, EP, etc.). It is not homogenous and gives none info in terms of trends (since it basically depicts "the remaining % to add up to 100%").

I think this addresses all points. Cheers. Impru20^talk 19:28, 24 May 2022 (UTC)

Your points are valid.

I know that every case, at the end of the day, is dealt with separatelly, but it may be a good idea to have a guideline when party is worth adding or worth being considered to be added - how many times it should be mentioned or what % support it should receive, e.g. if party appears after election, gets 2-3% support and disappears (dissolved, merged, reason is not important) before the next election or if party appears only three times but (for some reason) had enough support to pass threshold - is it enough to be noteworthy, to be seen as a serious candidate, at least at the time, with noticeable (potential) support?

Regarding Russian page - yeah, that wasn't the best example as I'm the main contributor right now, Slovenian one would be better to show that graph and table may differ.

Looking on a bright side - all past results are written down in one place are ready to use from my revision.

That's all from my side. Cheers. Leo0502 (talk) 09:18, 30 May 2022 (UTC)

It seems like almost all of us agree that this article could be improved.

Almost all of us agree that an "others" category would be helpful. One person disagrees. The fact is that this a pretty common practice in many other polling articles on Wikipedia. To deny that is to deny reality. Here are some examples: https://en.m.wikipedia.org/wiki/Opinion_polling_for_the_next_Italian_general_election https://en.m.wikipedia.org/wiki/Opinion_polling_for_the_next_German_federal_election https://en.m.wikipedia.org/wiki/Opinion_polling_for_the_next_Finnish_parliamentary_election https://en.m.wikipedia.org/wiki/Opinion_polling_for_the_2022_Swedish_general_election https://en.m.wikipedia.org/wiki/Opinion_polling_for_the_next_United_Kingdom_general_election https://en.m.wikipedia.org/wiki/Next_Irish_general_election

The "others" category is very valuable in understanding the current political climate. For instance, in the 2012 elections we had over 15% of the vote going to smaller parties showing the disappointment of the general public with the traditional ruling parties. By arbitrarily deleting that information from the polls we're fabricating and inaccurate and biased image about the Greek political climate.

That can also be said about using how polls are generally presented on this article. It's true that a lot of the companies include an estimate of the vote as long as unweighted voting preferencs in the results of their surveys,but in this article only the former is allowed to be depicted.

That again has resulted in an inaccurate depiction of the political climate in Greece often showing a bias towards the governing party by showing them performing better than they are only based on estimates.

This issue has been brought up time and time again but nothing has been done about it. One person disagrees with making changes and is not really up for discussing making any as seen here. I was under the impression that Wikipedia works by reaching a consensus, not by following what one person thinks its right.

So I'm once again proposing to add an "others" category and also presenting the polls in two lists: one showing what the raw data of what the people responded and one where the estimate of the polling company is depicted. C0rsp1m (talk) 19:31, 13 November 2022 (UTC)

Proposal for addition 2 more parties

I think Free People and Creation, which includes New Right and Recreate Greece, should be listed in the opinion polling section, because they have been icluded in an opinion poll. Μοναχικός Λέων (talk) 13:02, 13 July 2021 (UTC)

I saw some polls with National Popular Consciousness, they should be include as well Braganza (talk) 14:46, 30 August 2021 (UTC)

@Impru20, Αθλητικά, KnightPower17, Thepottato, PatHRK, and Σπύρος Αあるふぁ.: your thoughts? Braganza (talk) 14:47, 30 August 2021 (UTC)

We should probably wait somewhat to see if the trend of including these parties in opinion polling is maintained into the future or is just occasional. Creating a whole column for parties that get mentioned in only one or two polls with negligible support does not seem appropiate. Impru20^talk 14:53, 30 August 2021 (UTC)

Well, it's been a while since I've written something. The opinion polls are too important for some people to ignore. The following are the results of my search on Google News. the results, however, do not appear on the webpage. KnightPower17 (talk) 15:30, 30 August 2021 (UTC)

i personally think we should make notes for parties which only appears less than 10 times or so (like some other poll pages do) Braganza (talk) 15:44, 30 August 2021 (UTC)

Party logos

User:Derzki has twice reverted my addition of party logos. My attempt to obtain an explanation for this has thus far gone unanswered. It should go without saying that we cannot have logos for some parties and not others, as this would violate WP:NPOV. •ΘΕΟΔΩΡΟΣ• 13:45, 30 October 2021 (UTC)

@ΘΕΟΔΩΡΟΣ: idk why he made but Bots usually undo addition of non-free party logos Braganza (talk) 09:31, 20 November 2021 (UTC)

Party logos that are non-free will be automatically removed by bots anyway. You have to check the party logo image page to see if these are copyrighted or not, because many logos only can be used in the party's main page, not outside of it. Impru20^talk 09:54, 20 November 2021 (UTC)

Is there no way to go around the non-free images thing? I.e. adding an "appropriate copyright tag" and a "fair use rational" as described in this page: https://en.wikipedia.org/wiki/Wikipedia:Non-free_use_rationale_guideline.

Also, regarding the PASOK logo, it needs to be reflected that this is not the party PASOK, but the coalition PASOK-Movement for Change, which used to be called just Movement for Change but recently changed its name. The English Wikipedia still uses the old name but the new name and logo can be seen in the Greek Wikipedia. I think the new logo should be added and it should link to the Movement for Change page (which should also be updated asap), rather than the PASOK page. Am34114 (talk) 10:49, 30 May 2022 (UTC)

Not for this case; use of non-free images is limited only to where they are absolutely necessary to provide context (i.e. the parties' own articles is an example). They're not needed to provide context in this article, but are rather only a minor aesthetical thing. Thus, there is no way to circumvent the non-free prohibition. Impru20^talk 11:10, 30 May 2022 (UTC)

Ah ok, I see, thank you! But what about my comment on the PASOK logo and link? Is there any possibility to fix that? Am34114 (talk) 11:17, 30 May 2022 (UTC)

An AFD should be opened

This article should be deleted, IMHO. GoodDay (talk) 19:37, 30 October 2021 (UTC)

What sort of reasoning would you use for your AfD? The article (like most opinion polling articles) seems to be sourced to reliable content and would pass GNG in that regard. Bkissin (talk) 03:17, 31 October 2021 (UTC)

Palmos Analysis/Zougla.gr Poll

I think this poll should be removed. The vote changes, especially for New Democracy and EgtP, are wildly different to current trends. Additionally, the poll was performed to a rather unconventional method, as the source explains. Anyone else in agreement? 81.110.178.252 (talk) 19:04, 15 January 2022 (UTC)

Zougla polls are not reliable. These have never been added before and will not be now. They are essentially pro-far-right propaganda. Impru20^talk 19:24, 15 January 2022 (UTC)

The polls presented by Zougla.gr are not included in the polls as they are fake which give fake percentages to the parties that are more right-wing in New Democracy.--Derzki (talk) 11:41, 16 January 2022 (UTC)

Poll of abacus/alpha

This poll has receive very criticism and probably is biased 2A02:587:A609:100:39AB:680:19BC:4A4D (talk) 20:58, 28 January 2022 (UTC)

The poll is backed by reliable sources and its comissioner one of the biggests TV channels in Greece (Alpha TV). Personal opinions (a.k.a original research) matter little here, and I dunno about the "criticism" on the poll, but if Wikipedia starts removing verified information only because it comes under criticism, there would be no Wikipedia at all. Cheers. Impru20^talk 21:27, 28 January 2022 (UTC)

That is certainly an interesting approach. So because it was broadcasted by a national TV channel that makes the information reliable. So if Alpha TV reported that according to a research, a flying monkey shooting laser beams out of its eyes is the frontrunner, that would automatically make it true. Obviously what matters is who is reporting it and not the actual research or the company that did it. All polling companies receive criticism but unlike most of the other companies, this one was nonexistent prior to this poll and its registered information was outdated. Those facts should have mattered more than that Alpha TV commissioned this unknown company to do a poll. Especially given the fact that as has been reported by many reputable sources, like RSF, media freedom in Greece has been declining and the media are heavily influenced by the government. C0rsp1m (talk) 18:29, 13 November 2022 (UTC)

How are voting estimates calculated

I know this is supposedly explained on the main page, but I've been trying to do the math myself on some polls (the 2 most recent ones, being one by Pulse and the one by Real Polls) and the numbers I get aren't the same so evidently there's something I've misunderstood. I'd greatly appreciate it if someone could explain the process to me. — Preceding unsigned comment added by Oh Tassos (talk • contribs) 18:51, 2 March 2022 (UTC)

@Oh Tassos Μέθοδος τたうωおめがνにゅー τριών , αποκλείονται αναποφάσιστοι, δでるたξくしー δでるたαあるふぁ αποχή άκυρο λευκό 2A02:85F:F886:5B00:69F5:2A4C:49E7:FFED (talk) 04:51, 27 April 2022 (UTC)

Realpolls/Dimokratia

There are new researches that are not included, such as https://www.dimokratia.gr/politiki/544477/dimoskopisi-realpolls-katholiki-apodokimasia-tis-kyvernisis/. It seems that sometimes this research is included and others is not. Please enlighten us about the guidelines of the published researches. As we know the institution that is resposible for the eligibility of the researches is the ESR (National Council of Radiotelevision) and this research is in full compliance with this institution.

The researchers have collaborated with the institution in order to publish the research. — Preceding unsigned comment added by 79.107.183.97 (talk) 13:31, 9 April 2022 (UTC)

Speaking of Realpolls/Dimokratia, I've noticed those polls have been consistently giving lower percentages to KINAL than pretty much all other polls we have on this page, any idea why that could be? (I can't notice any particular party benefiting from that though, which is also slightly odd) — Preceding unsigned comment added by Oh Tassos (talk • contribs) 12:34, 14 May 2022 (UTC)

Vandalism and Removal of polls

Recently the article has been vandalized by people that remove the most recent polls.

There are two polls, one by ProRata and one by Data Consultants, both showing the difference between the two main parties decreasing. I'd understand it if the latter is not considered a proper source since it hasn't been included before. But ProRata is an established polling company and all of the other polls conducted by them are there!

So now the article has been locked and can't be edited. Not by correcting the bias though but by embracing it! You have prevented people from adding a valid poll that happens to show that the difference between ND and SYRIZA has decreased significantly.

For how long do you think you can keep up such ridiculous tactics that attack Wikipedia's credibility? There's a new poll out now further proving ND's decline. Will you censor every poll that is not favourable to the government from now on? C0rsp1m (talk) 09:12, 4 May 2022 (UTC)

Polls are once again being constantly vandalised. We have had trolls presenting months old polls as new, others completely making up numbers. This vandalism is getting out of hand, especially during August when we get no new polls and as such we have less activity. Could someone ask to protect this page? — Preceding unsigned comment added by Oh Tassos (talk • contribs) 15:21, 25 August 2022 (UTC)

Add National Creation party to the polls

The newly formed Coalition between Recreate Greece, New right amd National Agreement is polling at around 1.5%-2% , hence it should be added to the polling graph and catalog. Alexispapp (talk) 22:39, 22 May 2022 (UTC)

That's not true - the party was literally founded and presented officially three days ago, whilst there is not enough data to back your claims. 2% is a total exaggeration. There was just one poll showing a modest 1% so far. NikolaosFanaris (talk) 23:13, 22 May 2022 (UTC)

The party used to exist by the Dimioyrgia name alone and now the National Agreement party was added to the coalition. Dimioyrgia was polling steadily at 1.5% in the past months and I really cannot see how it is not there when you have Plefsi Eleftherias and Antarsya. I don't know whether it's a form of bias against conservative parties or not but it's unacceptable for not including us Alexispapp (talk) 12:44, 23 May 2022 (UTC)

It's a brand new party without any recognision yet in recent polls. If something changes, we could consider adding it. Dimiourgia Xana has been completely absent from polls with very low numbers in recent years. Your above statement is inaccurate. NikolaosFanaris (talk) 13:27, 23 May 2022 (UTC)

It not at all inaccurate and what you're doing is completely biased, almost all polls include Dimiourgia (the coalition between DIXA and New right) at a consistent 1.5% and at the same time the party is seeking the 3% threshold Alexispapp (talk) 13:49, 23 May 2022 (UTC)

https://twitter.com/europeelects/status/1528615837258719233?s=21&t=BEsmfWduWDL9Cd13LcfQQQ

https://twitter.com/europeelects/status/1528437802739896322?s=21&t=BEsmfWduWDL9Cd13LcfQQQ

Here you have the two most recent polls and it's obvious that what I am saying is 100% accurate, please add Ethniki Dimiourgia to the polls as soon as possible Alexispapp (talk) 14:03, 23 May 2022 (UTC)

"Consistent" lol - two polls from the last couple of days because of hype?! Nah, that won't be possible under the current circumstances. NikolaosFanaris (talk) 14:29, 23 May 2022 (UTC)

The party will most likely get the 3% threshold in the 1st election. You once again prove your left wing bias through declining simple facts. I'll add it and i'll keep adding ED to the polls from different accounts even if you ban me until you realize that you eventually have to. Alexispapp (talk) 15:35, 23 May 2022 (UTC)

I think it's now time to add the party, it has appeared in 4 consecutive polls, polling from 1.4%-1.7%. I also created a page on the party so everything is set, just add the party. Alexispapp (talk) 21:29, 26 May 2022 (UTC)

WP:TOOSOON NikolaosFanaris (talk) 21:40, 26 May 2022 (UTC)

Distortion of percentages

Hi everyone - I am starting to think that the majority of numbers are made up in the latest polls. Can someone verify that the last two polls (MARC and Pulse) are correct? None of the two attached sources show those numbers in any of the poll results. NikolaosFanaris (talk) 12:01, 5 July 2022 (UTC)

If you mean the weird numbers being randomly changed by IP users, yup, that's outright trolling. I have repeteadly asked for a permanent protection of this article due to the permanent nature of this kind of disruption, to no avail.

If you refer to the provided figures in the table not considering the abstention/undecided rate, yup, that's intended and explained in the text just above the table:

The table below lists nationwide voting intention estimates. 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. Polls that show their results without disregarding those respondents who were undecided or said they would abstain from voting (either physically or by voting blank) have been re-calculated by disregarding these numbers from the totals offered through a simple rule of three in order to obtain results comparable to other polls and the official election results (...).

Cheers! Impru20^talk 14:22, 5 July 2022 (UTC)

Remove Golden Dawn from polling table

Golden Dawn is consistently not appearing in polling anymore. Its polling numbers, when they appear, are very low, at the same level as parties that are not included on the table for that very reason. It is also highly unlikely that this party gets any seats in the next election; hence, there is no reason for it to be included anymore. (Arguably it makes more sense to include National Creation instead, but that's a different matter.) Am34114 (talk) 13:21, 7 July 2022 (UTC)

Golden Dawn had significant polling numbers at the beginning of the parliamentary term (as well as a significant election result, only 0.1% away of winning a seat), so that's the reason they're not being removed. Impru20^talk 13:24, 7 July 2022 (UTC)

Recalculation of percentages

I observed this following the 3 Pulse RC polls published after ΔでるたΕいぷしろんΘしーた was concluded.

We've always recalculated percentages to exclude non-voters/blank votes/undecided/etc. Sometimes, however, articles also publish recalculated results themselves, typically only excluding blank votes (in other words, leaving the undecided votes untouched). This makes us have to do a 2nd recalculation (the 1st was done in the original article) which can sometimes distort percentages.

Take the Pulse RC poll conducted on 12-15 September for example (between Mr Mitsotakis and Mr Tsipras speeches). Doing only a single, full, recalculation based on the live TV footage, MeRA25 gets 3.0% of the vote. Basing the numbers on the linked article, however (with a 2nd recalculation), it's 2.5%. That difference is huge because in the latter case it means the party would stay out of parliament.

I propose we do not count sources that use recalculated results and only link the original data, but I don't know how realistic that is. Anyone have any ideas? Oh Tassos (talk) 13:04, 22 September 2022 (UTC)

The issue of recalculating the various polls based on a "simple rule of three" so as to make them comparable has been raised again in the past in this talk page, and in my view has not been persuasively addressed. The rationale may be explained above the list (to make the polls comparable) but the fact that a reason has been provided does not mean that the reason is a good one. The main problem is that employing the rule of three actually rests on the assumption (and it is a big assumption) that somehow we can exclude undecided voters from the calculus by assuming that they will probably split along the lines of decided voters. This assumption is unsound as it does not take into account the fact that different parties have different percentages of their original voters in the last election that will certainly vote for this party again, and some pollsters actually measure this (συσπείρωση). So a party which keeps 80% of its original voters will have fewer former voters in the undecided pool than one which kept, say 50% of their original voters. It is not safe to assume that those undecided people will just split similarly to those that have made up their mind already. Most pollsters are not confident disregarding those from their recalculated results, which tend to only exclude invalid votes and abstentions. The results one gains from doing a recalculation based on a simple rule of three may make the polls comparable but are also quite crude and methodologically suspect. If companies make recalculations of their own (αναγωγή επί τたうωおめがνにゅー εγκύρων) then I think we should be including those in a separate table and just report what the companies themselves are reporting, including undecided voters and "others" which are oddly omitted here despite being included in every single poll. The fact that some of the smaller parties are counted separately in some polls and under "others" in others is not really a big problem. When Wikipedia readers see a dash under a party name they can easily surmise that in this particular poll it was counted under "others". Kkostagiannis (talk) 11:43, 24 September 2022 (UTC)

Well, the rule of three is used because pollsters use it, both at the example raised by Tassos (which follows such a rule of three to detract blank votes), but also in many cases when undecideds are discounted as well to show a "final" figure over actual voters. It is not a randomly-picked criterion, but one that pollsters use.

Since polls calculate their undecided/blank/abstainer results in a different way (and, some time, omitting some of these categories altogether), it will be nigh to impossible to design a table that puts all of these polls together, since their raw figures are not comparable across different companies. The current may not be a perfect approach, but it is the better one, given the circumstances of Greek polls (unless of course you defend to scrap any kind of comparability between polls, including the polling chart or the chronological table). Impru20^talk 13:27, 24 September 2022 (UTC)

Actually the cases where pollsters discount undecided voters, as you suggested, are not really that many. From a quick sample of the most recent polls I could check, among those pollsters that present recalculated figures, only one (Metron analysis) discounted them. Everybody else (pulse, MRB, marc, OP) did not. They just discounted blank/invalid/abstention and included figures for undecided in their recalculated polls. When it comes to the recalculated results, pollsters follow a quite uniform approach: they include don't know/won't answer (which they often cluster under "unclear vote" or "αδιευκρίνιστη ψήφος") and exclude invalid/blank/abstention. From the admittedly small sample I checked, this seems to be the rule, not the exception. So, I think a separate table including those pollsters that make the recalculation themselves is possible (the few exceptions can be signaled as such in notes).

Now, as regards the remainder of polls (which are the majority) and the other things that they measure differently. The same few pollsters I indicatively checked do indeed vary in the categories they use, but those categories overlap quite significantly and could possibly be merged under more generic ones with notes added to clarify where individual pollster diverge.

Almost all pollsters count "abstentions" and "blank/invalid" votes. Most will count those separately, but some fuse them into a single category (Pulse). This is the sort of metric that pollsters disregard when it comes to recalculating results and I think it could be turned into one extra field in the table. It will include the total of abstention/blank/invalid.

All pollsters also count "indeterminate vote" in one way or another: some have a "don't know/won't answer" category, others count the "don't know" and "won't answer separately" (which means they can be merged in a more generic category as they actually measure the same thing, just in more detail). Others just have one "undecided" or "don't know" category which is a bit more problematic.

Adding three extra table fields ("others," "invalid/blank/abstention," and "indeterminate vote") of course involves clustering together things that some pollsters consider worth having as different categories, but those fields are already entirely disregarded in the current version of the article. I think merging them into broader categories would sacrifice less in terms of accurately representing the polls than the current version. Kkostagiannis (talk) 17:58, 24 September 2022 (UTC)

I am not saying they are that many. I say they use it as a method, so it's a valid method backed by sources. And no, it's not just Metron Analysis (though you would have to check Greek polls for many years and elections to get the whole picture, something you clearly acknowledge you did not). Even so, when you say "they just discounted blank/invalid/abstention and included figures for undecided in their recalculated polls", that's already a rule of three (discounting these figures, then recalculating the remaining figures to add up to 100%). It's simple and easy.

You are really not even answering what I stated on the main issue of the raw data of the different pollsters not being comparable among themselves, instead going into different thinkings of how data works. As stated, the current version is not perfect, but it offers an accurate picture in which you can compare polls' data among themselves and you can measure trends without difficulty and in an accurate way (and all abiding to Wikipedia's guidelines). What you propose would basically break any usefulness for any table, as it would require possibly many tables with the many different methods used by the various pollsters, which would make this very same article (and the trend chart) pointless. Impru20^talk 20:04, 24 September 2022 (UTC)

I'd like to remind you two of the original purpose of this section: that percentages are distorted regardless. As I said, because many pollsters discount certain votes anyway in their attempt to give us a clearer result, the percentages get quite distorted when we discount even more (like in the example I gave with Pulse RC, recalculating the raw numbers shown on TV gave different results to recalculating the already-recalculated results from in.gr and kathimerini). I don't know how accessible the raw data is most of the time, but it'd really help create a more accurate image. So I would suggest using that from now on (or, since it's too much work to fix everything retroactively, we could implement that starting after the next election). Oh Tassos (talk) 11:36, 25 September 2022 (UTC)

To clarify because perhaps I wasn't being clear enough. I'm not saying we shouldn't do what we're already doing and recalculate the results based on a simple rule of three. I'm just saying that we should apply the rule of three to the raw data, not the recalculated results shown on many articles. Oh Tassos (talk) 11:39, 25 September 2022 (UTC)

@Oh Tassos: Oh yes, yes, I got your point. You basically raise an issue which happens with those polls whose results are rounded up or down to the nearest .5 o .0 figure (which as of now applies mostly to ProRata and Pulse RC). From my experience: A) if these polls show only one set of results (either with or without the partial recalculation) then there is no issue, since you can just do the math with the numbers that are provided. B) If these polls show both, then I typically go with the following scheme: 1) Make the recalculation with both results and compare them; 2) Where both calculations would give the same rounding, then that's it; 3) Where the rounding is not the same, then common sense applies: the recalculation that is closer to the variation between the party's current poll and the previous poll should be used. To be illustrative, let's say that one party goes from 1.5 to 2.0 in the raw data, but with the recalculations it then translates into both a 3.23 result (which would be rounded down to 3.0) and a 3.26 result (which would be rounded up to 3.5), compared with, let's say, a 2.5 in the previous poll's recalculation. In this case, the 3.5 figure could be considered a mathematical outlier, so it would be more correct to adjust it to the 3.23 figure (3.0, that's it). Not sure if this helps! Impru20^talk 13:08, 25 September 2022 (UTC)

That does make sense, thanks a lot! I was mostly raising this point that way future editors have some written precedent to act on, as the situation is honestly quite a bit hectic until you get a bit more familiar with the page (you can find another section by me regarding how the numbers are calculated, that matter's been long resolved luckily). Obviously a deviation of 0.5 on just one specific poll isn't that important when there are so many other polls to "cover" any potential inaccuracy, I just wanted to clarify the situation for future reference. Oh Tassos (talk) 14:32, 25 September 2022 (UTC)

Semi-protected edit request on 18 November 2022

There are two surveys that are not included: https://www.in.gr/2022/11/12/politics/kommata/dimoskopisi-gpo-meta-tis-ypoklopes-anatropi-sti-diafora-nd-syriza/ https://realpolls.gr/surveys/realpolls-panelladike-ereuna-gnomes-04-11-2022-07-11-2022 https://1voice.gr/dei-to-62-5-ton-ellinon-thelei-na-epistrepsei-sto-dimosio/ 46.190.26.254 (talk) 08:07, 18 November 2022 (UTC)

Not done: When creating edit requests you are required to present what you want a reviewing editor to do specifically. That is to say, you should phrase your request to sound something like "Please change X to Y" or "Please remove/add Y" with X and Y being quoted prose. If adding content, you must also specify where specifically in the article you wish to insert it. —Sirdog (talk) 09:13, 18 November 2022 (UTC)

@Oh Tassos, Thepottato, Impru 20, Αθλητικά, Sirdog Copy and paste the URL below which includes the corrected poll and 2 new ones.--79.107.220.117 (talk) 11:07, 23 November 2022 (UTC)

|- | Interview/Politic | 10–12 Nov 2022 | 1,000 | style="background:#BFD5F7"|| 35.7 | 28.0 | 11.3 | 6.7 | 5.3 | 3.8 | – | 1.7 | 1.4 | style="background:#1b5cc7; color:white;" | 7.7 |- | Alco/Alpha TV | 14–19 Nov 2022 | 1,000 | style="background:#BFD5F7"| 35.3 | 26.8 | 13.3 | 6.1 | 5.7 | 3.2 | – | 2.5 | – | style="background:#1b5cc7; color:white;" | 8.5 |- | GPO/Parapolitika | 10–12 Nov 2022 | 1,000 | style="background:#BFD5F7"| 38.1 | 31.1 | 12.7 | 7.1 | 4.3 | 2.8 | – | 2.1 | 1.8 | style="background:#1b5cc7; color:white;" | 7.0

References

Error at last GPO poll

ND 31,6 -> 36,6 SYRIZA 25,8 -> 29,9 PASOK 10,5 -> 12,2 KKE 5,9 -> 6,8 EL.L. 3,6 -> 4,2 MERA25 2,3 -> 2,7 E.P. 1,7 -> 2,0 E.D. 1,5 -> 1,7 Other parties 3,4

Someone mistook the numbers because he did not take into consideration other parties which are mentioned at the poll — Preceding unsigned comment added by 2A02:587:7E38:C300:754E:F2A:30BA:18A5 (talk) 15:28, 24 November 2022 (UTC)

Fixed! My bad Oh Tassos (talk) 20:10, 24 November 2022 (UTC)

The last Poll of PRORATA is not included

HI, the last PRORATA's poll is not included in the list: https://www.efsyn.gr/politiki/kybernisi/368553_ligo-prin-apo-tin-eytheia-gia-tis-ethnikes-kalpes

Please add the latest poll (citation above) of this company that was published for Efimerida twn Syntaktvn newspaper. ProRata/Efimerida ton Sintakton 19-22 Sep 2022 1,905 participants ND 35.5 SYRIZA-PS 30.0 PASOK-KINAL 12.5 KKE 5.5 ELLINIKI LYSI 4.5 MeRA25 3.0 EP 2 (OTHER PARTY) 7 LEAD ND-SYRIZA 5.5 Hamnsick (talk) 10:59, 2 December 2022 (UTC)

Hi! I just added the poll! Oh Tassos (talk) 13:09, 4 December 2022 (UTC)

Why you removed National Party-ELLHNES from the graphical summary?

This party hasn't suspended from the elections and has more than 3% in most of the polls. 62.1.222.220 (talk) 12:23, 12 March 2023 (UTC)

Real polls

I think that there's another poll from Realpolls. Astronaftis (talk) 12:52, 2 April 2023 (UTC)

σしぐまτたうηいーたνにゅー εφημερίδα voice Astronaftis (talk) 10:31, 3 April 2023 (UTC)

A brief study on the numbers of the polling companies' biases

Hello!

Having been contacted recently by RealPolls on Viber to participate on a new poll-a polling company said to be affiliated with members of SYRIZA as covered on newspapers such as Parapolitika and Ethnos and whose credibility I had expressed concerns about two weeks before that-the topic of how reliable each polling company is has been on my mind again. So I spent a night analysing in an excel sheet on my computer how the reported numbers from each company compare to the mean (average) from that time period. The order chosen to comment on each company is alphabetical.

Period covered: All of 2023 so far (from Alco 9-11 Jan 2023 to ProRata 27-30 Apr 2023)

Number of polls: 38 (details on each company below)

Mean taken from: 2 weeks before and after the semisum of the start and end dates of each poll

Alco (3 polls): nothing noteworthy from the polls conducted in 2023. All of them fall <1pp of the mean for their respective time period which is very well within statistical error. In fact, only 1 party in each poll is >0.9pp of the mean. So again, nothing noteworthy.

GPO (4 polls): nothing noteworthy from the polls conducted in 2023. Unlike Alco, the larger parties are consistently 1-2pp above the mean, though that applies to all of them so I don't think we can deduce much of anything from that. Nothing noteworthy overall

Interview (5 polls): nothing noteworthy from the polls conducted in 2023. Similar case to the previous two. A few instances of the numbers deviating 1-2pp from the mean, otherwise all <1pp. All in all nothing noteworthy. Interview is also the company with the most polls published in 2023 so far.

Marc (4 polls): nothing noteworthy from the polls conducted in 2023. The case is identical to Interview's, a few deviations in the range of 1-2pp, otherwise all <1pp. Nothing noteworthy.

Metron Analysis (4 polls): this is where some potential issues start to appear. Out of the 4 polls conducted, SYRIZA is ~2.4pp below average on one of them, and ~3.6pp on another, in the latter the difference between ND and SYRIZA is also ~3.1pp larger than average. These numbers could be within the statistical error, but there could also be semi-consistent deviations at the expense of a particular party. Metron Analysis has also had this tendency in past polls outside of 2023.

MRB (4 polls): nothing noteworthy from the polls conducted in 2023. Same case as Interview and Marc, I don't think I need to comment on much else: nothing noteworthy.

Opinion Poll (3 polls): in the poll in January Nea Dimokratia is given ~2.3pp more than the average for that period. Looking back at past polls too, it appears that they consistently give the party a higher percentage than other companies, though in the cases I checked its technically within what you'd call a statistical error.

Palmos Analysis (1 poll): this one is weird. Both Nea Dimokratia and SYRIZA are below average by ~3.2pp and ~3.1pp respectively. Overall since 2019 this company has conducted 3 polls, and while I didn't check the numbers in detail it appears like this pattern might apply to those as well, or maybe on the other two its just Nea Dimokratia's numbers that are lower. Further investigation is probably needed to comprehend what's up, but it appears to be at the expense of the top 2 parties and the rest have virtually no deviation from average so I really do not know how to judge this one.

ProRata (4 polls): there is an instance of SYRIZA being above the mean by ~2.4pp lowering the difference with Nea Dimokratia by ~2.5pp compared to other polls. In some polls before 2023 this pattern appears to exist too, which should not be a surprise considering most are conducted for Efsyn. Other than that though, the rest of the numbers seem to be fine.

Pulse RC (4 polls): nothing noteworthy from the polls conducted in 2023. In the category of Interview, Marc, and MRB.

Rass (2 polls): consistently higher numbers for PASOK, ~2.5pp in the most recent poll and ~4pp in January. Outside 2023 this pattern continues. In the most recent poll SYRIZA is also ~2.1pp lower than the mean, which may or may not carry any meaning. Keep in mind these polls are also conducted for a smaller news network, Action24, if that means anything.

Conclusion: in no case can we speak of forged numbers or anything like that since the sample size is small and in most cases the deviations from average are not that significant. In addition, there's the risk the average itself is biased in some way, we have no way to account for that as far as I'm aware. Assuming the above analysis is correct though: Rass might be giving higher numbers to PASOK, Metron Analysis lower to SYRIZA, Opinion Poll higher to Nea Dimokratia, ProRata higher to SYRIZA, GPO higher to both Nea Dimokratia and SYRIZA, and Palmos Analysis is doing... whatever, I'm not certain. Interview, Marc, MRB, Pulse RC, and especially Alco seem to be completely in line with the average of all polls (though they do account for ~53% of all polls, so again the average might be biased).

Thank you for your time! Oh Tassos (talk) 00:38, 5 May 2023 (UTC)

Source needed for the new GPO poll

Sone days ago a new poll was released by GPO on the newspaper Ta Nea. https://www.protagon.gr/epikairotita/dimoskopisi-gpo-sto-65-to-provadisma-tis-nd-pws-allakse-o-xartis-tis-voulis-44342718132

Unlike the other polls, there is no source about the data, i.e. the number of participants and the time range. Can someone find that information? 2A02:1388:14E:BD0B:6C8D:3654:4236:D948 (talk) 09:01, 8 May 2023 (UTC)

New parties

After the ban of National Party - Greeks new parties emerged in polls. Like Course of Freedom (PE), EAN and Victory (Niki). What about adding them? 2A02:1388:14A:ED14:CB2C:6210:B50B:AE93 (talk) 12:42, 11 May 2023 (UTC)

Euractiv's "poll"

Euractiv's mention of a secret poll conducted on behalf of an undisclosed multinational by an undisclosed polling agency has no place in the page. 2A02:2149:8B0A:6200:15FD:52CB:BA43:D1FA (talk) 19:00, 17 May 2023 (UTC)

Reappearence of Euractiv "poll"

Euractiv is neither a polling firm nor a commissioner of a poll so in order for the article to have some sempblence of integrity and since-for some uknown reason-it is still on the list, it should have a warning label. 2A02:2149:8B0A:6200:897F:4FC5:924C:7E3 (talk) 14:02, 19 May 2023 (UTC)