Talk:Opinion polling for the 2023 Turkish presidential election
![]() | This article is rated List-class on Wikipedia's content assessment scale. It is of interest to the following WikiProjects: | |||||||||||||||||
|
Outdated info
[edit]There have been new polls.
Data compiled from multiple sources at Europe Elects and their twitter.com/search?q=T%C3%BCrkiye%20(from%3AEuropeElects)&src=typed_query&f=live
Twitter Account Axadem (talk) 22:16, 10 January 2023 (UTC)
- The ALF one is for parliamentary election and not presidential. The Metropoll one is only between people who claim themselves as religious. I could not even find the Optimar one from the link you provided. Yanekyuk (talk) 10:49, 5 March 2023 (UTC)
- There are 4 candidates but Metropoll has %11 in others section, this is wrong. That %11 should be distributed candidates proportionally. Oozkan48 (talk) 17:47, 8 April 2023 (UTC)
What surveys should be included should be reconsidered
[edit]Checking Areda survey's website and Twitter account, it can be seen that it was never published officially. I wanted check other sources as well and seen that these mostly consist of news outlets that either close to the opposition or the ruling party. The main problem is that these surveys do not give any information other than their sample size and the main results. There are no official reports on them. I believe it is crucial only to include survey results that have been published by the polling firm. We might even be breaking some copyright laws here. For instance, Optimar is a paid-only polling firm, meaning that we can only see the outline when we enter their polling results, and not the actual results. But some news outlets just claim what the results are with no verifiable sources.
To keep it short: I think it is important to include only the polls that are published by the polling firms and not third-party news outlets. Yanekyuk (talk) 15:08, 6 March 2023 (UTC)
- To understand what I mean, you can check (https://en.wikipedia.org/wiki/Nationwide_opinion_polling_for_the_2020_United_States_presidential_election) to see difference in source quality. If there is a poll conducted by X firm and published by Y firm and if X firm does not present the poll's existence in any form whatsoever, does this poll belong to X firm or Y firm? Yanekyuk (talk) 15:21, 6 March 2023 (UTC)
It is better to include the publisher than to omit a result altogether.
[edit]I did try to delete Yeni Safak's poll because the polling firm did not publish it. Then I realized that almost half of these results are not published by the polling firms, but by the news outlets that are dominantly either close to the ruling party or to the opposition party.
Again, ff we decide to delete secondary-source polls, the page will be half of what it is right now. I think it is quite enough to include "published by" column and let people know that it was not published by the polling firm but by the news firm. Yanekyuk (talk) 09:39, 7 March 2023 (UTC)
Placement of Polls
[edit]Shouldn't polls with only Erdogan and Kilicdaroglu be in the second round like they are on Turkish wikipedia? https://tr.wikipedia.org/wiki/2023_T%C3%BCrkiye_cumhurba%C5%9Fkanl%C4%B1%C4%9F%C4%B1_se%C3%A7imi_i%C3%A7in_yap%C4%B1lan_anketler 173.62.118.190 (talk) 16:18, 19 March 2023 (UTC)
- That would make more sense to me. Some of the polls (for example ORC March 6, ALF March 7) ask specifically about a two-person race between them.Lilactree201 (talk) 07:55, 24 March 2023 (UTC)
- Just noting that this change has now been undertaken by a third user. Lilactree201 (talk) 21:17, 2 April 2023 (UTC)
Ser-ar
[edit]There is no ser-ar polls show in here ser-ar is as valid as areda and other it should be also included 5.176.67.10 (talk) 12:19, 21 April 2023 (UTC)
Polls
[edit]Most of the survey companies on this page are actually fake. In terms of reliability, it should be edited at the same rate as Turkish. KreiosII (talk) 20:17, 23 April 2023 (UTC)
- How are the editing standards different? Both Turkish and English sources are from Twitter, and I don't see much difference in terms of reliability. If anything, on a purely mathematical level the English seems more representative. Uness232 (talk) 11:02, 30 April 2023 (UTC)
Kilicdaroglu seems to be ahead but not in the chart
[edit]I found here a source for it by Gercek, which says Kilicdaroglu is the first on eight polls while Erdogan in two. I do not know how to adapt the chart though.Paradise Chronicle (talk) 23:30, 24 April 2023 (UTC)
- Erdogan appears to have polled about 11 times above 45%, while three of those polls where from a pro Erdogan pollster and Kilicdaroglu was 16 times above 45% and only one was from a pro-Kilicdaroglu pollster, specially the last ones with several just below 50%. and the last one above 50%. Yet Erdogan appears with over 45% almost side by side with Kilicdaroglu in a chart updated on the 29 April. How is that possible? Paradise Chronicle (talk) 22:35, 29 April 2023 (UTC)
- Two things. The graph is from the Turkish Wikipedia page, since you objected to the old one which was slightly outdated. This means it has a different poll set to get data from. Secondly, LOESS has a tendency to not work as well soon after when new, sparser data comes out. Since the data is organized time-wise, LOESS reacts to the lower amount (less representative) of data in the last few days (Areda just had a poll, for example). The chart has to reflect that, unfortunately. Anyway, charts may not be necessary for this, we can add monthly means like the Turkish page if necessary. Uness232 (talk) 10:49, 30 April 2023 (UTC)
- Charts are important. Aintabli (talk) 21:03, 1 May 2023 (UTC)
- I don't quite see why. Sure, they'd be nice, but the current setup also works? I can get all the information I need (Kılıçdaroğlu will likely get the most votes, but there'll have to be a second round). We don't need charts to get that from this article. Of course, if somebody can maintain them on a regular basis, we should include them, but that doesn't seem to be the case. Uness232 (talk) 15:37, 2 May 2023 (UTC)
- Charts are only as valuable as the date. they give a false precision when in fact most of these companies were not around for the last election, there are no reliable pollster ratings, and most appear to be associated with one side or the other. It's GIGO. 2601:2C6:4300:B8C0:B576:416F:2518:BDC6 (talk) 18:36, 6 May 2023 (UTC)
- I wouldn't go that far. There are pollster ratings at 600vekil, for example, and while there are some apparent biases, we can only be so confident about what we're accusing them of. Generally, it seems, more AKP leaning pollsters are new, and we can confidently say that Areda for. ex. is doing something wrong, but what about ORC or Optimar, two seemingly accurate pollsters (considering the last elections anyway) that have consistently produced different results? We can't just say that they're all lying manipulators, especially on Wikipedia. Uness232 (talk) 13:02, 10 May 2023 (UTC)
- Charts are important. Aintabli (talk) 21:03, 1 May 2023 (UTC)
- Two things. The graph is from the Turkish Wikipedia page, since you objected to the old one which was slightly outdated. This means it has a different poll set to get data from. Secondly, LOESS has a tendency to not work as well soon after when new, sparser data comes out. Since the data is organized time-wise, LOESS reacts to the lower amount (less representative) of data in the last few days (Areda just had a poll, for example). The chart has to reflect that, unfortunately. Anyway, charts may not be necessary for this, we can add monthly means like the Turkish page if necessary. Uness232 (talk) 10:49, 30 April 2023 (UTC)
- The readdition of the graph is not according to the polls that follow. Opinion polls on elections are to show who is leading in the polls, and if Kilicdaroglu is up most of the times in the polls and several times even over 50%, he should also in the graph on the polls.Paradise Chronicle (talk) 17:58, 9 May 2023 (UTC)
- The lead is 1) not that large, 2) this is about LOESS, and how we think about polls, not about the polls not agreeing with the graph. Uness232 (talk) 12:56, 10 May 2023 (UTC)
- @Paradise Chronicle, I've added the chart back. Now that people are objecting to averages, we need a simple way to read this information. The charts are now up to date, and use this poll set. Uness232 (talk) 11:09, 11 May 2023 (UTC)
- It's ok now, thank you for your updates. And sorry I didn't adapt it myself but I just don't know how to create or adapt graphs like this.Paradise Chronicle (talk) 15:14, 11 May 2023 (UTC)
- Here is a gitlab repository with some instructions in case you're interested in creating LOESS graphs yourself. Gbuvn (talk) 15:26, 11 May 2023 (UTC)
- The updates are not mine, they are by @CoaxAndBotany. I just saw that they were updated, and that unlike the Turkish version, they seemed to be using these polls. Uness232 (talk) 16:14, 11 May 2023 (UTC)
- It's ok now, thank you for your updates. And sorry I didn't adapt it myself but I just don't know how to create or adapt graphs like this.Paradise Chronicle (talk) 15:14, 11 May 2023 (UTC)
- @Paradise Chronicle, I've added the chart back. Now that people are objecting to averages, we need a simple way to read this information. The charts are now up to date, and use this poll set. Uness232 (talk) 11:09, 11 May 2023 (UTC)
- The lead is 1) not that large, 2) this is about LOESS, and how we think about polls, not about the polls not agreeing with the graph. Uness232 (talk) 12:56, 10 May 2023 (UTC)
Pre-election polling ban not enforced?
[edit]Why are there still polls being conducted and published past the 10-days-before polling ban? Dingers5Days (talk) 17:12, 9 May 2023 (UTC)
- It has a fine and it is enforced, pollsters just pay the fine. Betseg (talk) 17:56, 9 May 2023 (UTC)
- Thank you, was very confused by that. Dingers5Days (talk) 19:58, 9 May 2023 (UTC)
- How do you know that?
- https://www.bbc.com/turkce/articles/cj70250n4gko
- "During the publication of public opinion research and surveys, it will be necessary to disclose which organization conducted the research, the number of subjects, and who financed the research."
- There is no violation of electoral law. Mysticmauler (talk) 02:01, 11 May 2023 (UTC)
- The first paragraph,
Oy verme gününden önceki 10 gün içinde yazılı, sözlü ve görsel basın ve yayın araçları ile kamuoyu araştırmaları, anketler, tahminler, bilgi ve iletişim telefonları yoluyla mini referandum gibi adlarla, adayın lehinde veya aleyhinde veya seçmenlerin oyunu etkileyecek biçimde yayın ve herhangi bir surette dağıtım yapılamayacak.
translatedIn the 10 days prior to the voting day, there shall be no publication with names such as "polls, forecasts" in written, oral or visual media in favor of or against any candidate that could affect the voters' decisions.
BBC is also a secondary source, if you want the primary source, here is legislation, relevant article is 55/B paragraph 3. Betseg (talk) 05:36, 11 May 2023 (UTC)
- The first paragraph,
Mostly to @Betseg: I've realized that you've deleted the averages, calling them WP:OR. Considering the fact that graphs made by editors are allowed, I would argue that a simple average is more of a routine calculation than a graph, which involves a data interpolation technique that many struggle to fully understand, and the absence of both graphs and averages make the polling section hard to read. WP:OR does state that "A primary source may be used on Wikipedia only to make straightforward, descriptive statements of facts that can be verified by any educated person with access to the primary source but without further, specialized knowledge." I would argue, however, that "this is the average of all the polls we have" is a descriptive statement of fact, not a subjective synthesis. Indeed, anyone can check 1)the calculation, 2) the sources of these averages. With graphs, that was not possible. Uness232 (talk) 07:16, 11 May 2023 (UTC)
Misleading accuracy
[edit]The lead is given to two decimal places, even though polls are generally accurate to no more than a few percentage points. The poll results and lead should be changed to the nearest whole number. 92.30.11.38 (talk) 10:03, 11 May 2023 (UTC)
- I disagree. Normally the lead is given with one decimal place (and two decimal places are generally only given in situations with 49.9), but it's not false accuracy to convey what the result of the poll was. Poll results, as you can see from this page, can actually be up to 10-15 points off, based on methodology; but we're not making a prediction here, the polls are simply saying, "of the people we surveyed, X percent said this." It's not misleading to say that directly. Uness232 (talk) 11:00, 11 May 2023 (UTC)
- Plus one decimal place is standard across WP election pages. Uness232 (talk) 11:00, 11 May 2023 (UTC)