Talk:Opinion polling for the next Polish parliamentary election

What to do with Research Partner polls
Research partner is known for polls that are inaccurate. The company is linked to PiS and overshoots it, the state media under Poland's administration used to order their polls

https://sprawdzamysondaze.pl/parlamentarne-2019/wybory/sondaz/#ariadna I don't think we should list something that clearly misinforms and has stayed under the 3% margin of error for one party in 2 consecutive elections.

I want to see if there's a consensus of doing something with this. For now I added small refs but perhaps we should move them from the main 2024 page with polls into "Pollsters known for bias" on the bottom of the page. Opinions? Octilllion (talk) 23:04, 15 February 2024 (UTC)


 * I do not think that this is warranted. If we chose to label pollsters because of their bias, it would be way more warranted to label CBOS for example, because it is a publicly funded institution and has consistently overestimated the support for whatever party is in power, for at least the last 15 years (so Civic Platform between 2007 and 2015, Law and Justice from 2015 to 2023, and since the last election it's back to overestimating PO and underestimating PiS again).
 * Furthermore, labelling pollsters like that is like opening a can of worms, since there are few pollsters who are truly independent and objective. For example, IBRiS has often been accused by some of overestimating Confederacy in the past, IBSP has allegedly overestimates Poland2050, Social Changes was implied to be biased in favor of PiS, CBOS is a mess as described above, and so on and so forth. So in order to do this page justice and not cherry-pick pollsters and their bias, we would have to label basically almost every poll published, which would unnecessarily clutter the page.
 * I would also argue that the labels are currently not phrased in an appropriate way - Wikipedia is an encyclopedia of knowledge, not a publishing site, so there should be no subjective statements like "take the results with the grain of salt" - encyclopedias do not advise readers on how to interpret the information presented, they only provide the facts. That is why opinion polling pages display all polls conducted - so that the readers of Wikipedia could see all polls and compare them against each other, and come to their own conclusions based on that.
 * Which is also why something like putting some polls in a table with a "Pollsters known for bias" label would be completely unacceptable, since it introduces a subjective opinion of writers as to which pollsters can be considered biased and which can't. This goes deeply against Wikipedia's principles, which states that everything should be written from a neutral point of view. It would also be an extension of the problem mentioned above, where to be just, most pollsters would have to end up in such a section, rendering it pointless, and making the article itself very biased and opinionated.
 * I personally would advise against any labels, to let the readers analyze the presented information by themselves. However, one solution that could inform of potential bias is something similar to what has been done on the page of Hungarian 2026 election polling, where above the table with the polls there is a separate section listing the polling companies and their claimed bias, with the caption "The following table displays the pollsters which have alleged funding and links to the political parties:". Note however, that it is incredibly important to use words like "claimed", or "allegedly", in order not to introduce opinions into the article. Marcin8612 (talk) 22:44, 17 February 2024 (UTC)
 * I too was thinking about doing something like on the Hungarian election page. I think I'll get to it now Octilllion (talk) 15:27, 18 February 2024 (UTC)

Last IBSP "poll"
I think we should delete last 'poll' by IBSP for Stan Polityki. It is not a poll, thats their forecast based on other polls and their algorithm. By the way, where did someone get the fact that it was based on a sample of n=1000, if it is clearly written that it is a forecast and not a poll? HerrmanNotGerman (talk) 18:17, 12 March 2024 (UTC)


 * IBSP has always been shown on this page, and I don't see any reason why we should change that. You'd also have to delete all past IBSPS on this page and from the 2023 election opinion polling.
 * Please also refrain from putting SOVPOL into ZP when the scenario in the poll is sovereign poland being seperate Octilllion (talk) 17:00, 13 March 2024 (UTC)
 * No, it is clearly written here that this is a forecast, not a poll. "Według najnowszej prognozy OGB Pro..." (via X).
 * IBSP also creates poll and publishes them from time to time, and they should be included here because they are an opinion polls on a n=? sample. But the election forecast is based only on some algorithm, it is not the type of data this page is about. 213.134.169.155 (talk) 13:22, 15 March 2024 (UTC)

Do not add United Surveys polls after the 18th of March
https://www.inforegister.ee/en/14804038-UNITED-SURVEYS-OU

The company has been deleted from the register on that day and de facto doesn't exist Octilllion (talk) 11:37, 26 March 2024 (UTC)


 * Apparently, United Surveys polls are now "by IBRIS" therefore scrap that Octilllion (talk) 14:27, 26 March 2024 (UTC)