Talk:Opinion polling for the 2022 Italian general election/Archive 3

Government approval
User:Ritchie92 inspired me to share this draft with you, in the event that government trust numbers are added, either to this page or to the Conte I and II Cabinet pages.

As you can see I cannot do a work that is as thorough and complete as his. Kahlores (talk) 01:59, 7 October 2019 (UTC)

Talk
Thank you for the tables. I think this does not really belong to this page, since in the Italian elections the Parliament is elected, not the government. This might be a table in another article about "Opinion polling for the Italian government approval", or something similar (see Leadership approval opinion polling for the next United Kingdom general election), however I don't think it's very relevant (Italy is a parliamentary democracy, so government or leadership approval does not matter as much as party voting intentions). Of course if we decide to start such a page I would surely contribute to it. --Ritchie92 (talk) 10:14, 7 October 2019 (UTC)
 * I agree. Separate articles would be great, provided there are users willing to keep them up to date. --Checco (talk) 06:35, 8 October 2019 (UTC)

Cambiamo!
Maybe it's time to remove Cambiamo! --89.96.185.33 (talk) 12:46, 9 October 2019 (UTC)
 * I oppose that. Why? It will surely fire up. I would also add We Are Europeans. --Checco (talk) 06:20, 10 October 2019 (UTC)
 * Instead of giving random opinions, User:Broncoviz and myself are maintaining the table in this page. The purpose is indeed having a consistent criterion (instead of a random one) to decide which parties to include in our polls table. So I suggest you contribute to that discussion with the formation of the criterion itself. The only consistent one that was proposed (by Broncoviz) was to add a new party only when it was polled by 50%+1 of the "high-quality" pollsters. In this moment neither C! nor SE satisfy this criterion. However the parallel discussion is about which are the high-quality polls themselves... It's a difficult issue at the moment, but I am sure that once settled with a consensus it will save us a lot of time in the future decisions. --Ritchie92 (talk) 06:39, 10 October 2019 (UTC)
 * For clarity, I am responding here, even though the subject is discussed above. Please excuse me for my bluntness. I am not convinced both by the 50%+1 and, above all, the idea of "high-quality" pollsters. It is OK to have a table classifying pollsters, but that table should have no effect on the opinion polls listed, let alone the qualification of parties to the list. I would make things easier and simpler. --Checco (talk) 06:50, 10 October 2019 (UTC)
 * You have not given any reason for your opinions apart from "I don't like" or "I like". Also, you "would make things easier and simpler". Ok that's great, but how? Please explain what is the criterion you suggest to insert parties in our table. --Ritchie92 (talk) 06:59, 10 October 2019 (UTC)
 * Consensus, through this talk page. --Checco (talk) 07:18, 10 October 2019 (UTC)
 * Ok, so discussing consensus every time. That's perfect (also far from "easier and simpler" but whatever), but then the table of the pollsters is useless (which is OK). And by the way there is no consensus about the addition of C! to the table, it was added by an IP user on a whim, and it stayed there. --Ritchie92 (talk) 07:57, 10 October 2019 (UTC)
 * No good reason to include Cambiamo in the table, and neither is it a good reason to have columns for others. Onetwothreeip (talk) 00:04, 19 October 2019 (UTC)

RfC on "Government/Other" columns
As per the discussion above, requesting comments on the issue of the Government-support columns. Two questions: (Q1) Should the Government-support columns in the opinion polling tables be kept or removed? (Q2) Open-ended question: Notwithstanding your answer to Q1, in the event the columns were to be preserved, what should their formatting be? 19:01, 30 August 2019 (UTC)

Q1 discussion
*Keep – Finding a consistent wat to pick the government is simple: the government of the time! There is so nothing misleading: the result in the column changes consistently with the government of the time. --Checco (talk) 06:25, 2 September 2019 (UTC)
 * Delete - For several reasons: (1) Possible WP:SYNTH violation as this is presented as if the polls actually asked on the government's approval, when this refers to the aggregated totals of the registered voting numbers for two separate parties (A + B does not equal to C if sources do not establish that it equals to C). As a side note, some polls may actually ask on the government's approval in a separate question, which may add to the confusion; (2) Failing to meet WP:VERIFY, as a majority of sources do not report these aggregated numbers; (3) WP:NPOV and WP:UNDUE issues raised as a result of the choosing of this particular combination of parties over any other, maybe more likely electoral arrangements. Government coalitions shouldn't be mixed with electoral alliances, and in the particular case of M5S and Lega (as well as M5S and PD) no such electoral alliances exist. This would require for us to report on all of the other unofficial but possible electoral combinations or none of them at all, in order to avoid giving undue prominence to any specific POV. All in all, it's not for us to choose on any one of this; (4) At large, potential conflict with what to do with parties not in government but supporting it externally (something which at times becomes difficult to properly measure, due to the varying nature of external support); (5) Redundancy issues in situations where all government parties are an electoral alliance (and thus, would be reported anyway in the Coalition vote section). Similarly, redundacy issues in situations of single-party governments (not common in Italy, I concede, but not an entirely disposable idea considering that some opinion polls in the last months have hinted at some chances for a Lega absolute majority). Impru 20  talk 19:01, 30 August 2019 (UTC)
 * Delete - Per User:Impru20. Also, finding a consistent way to pick the government majority at any moment in time is really tricky (see my comment to Q2). Complete removal is the best option. --Ritchie92 (talk) 09:10, 1 September 2019 (UTC)
 * Delete - I totally agree with Impru20, government's support columns are quite misleading, especially in a country like Italy, where governments change so frequently. -- Nick.mon (talk) 13:55, 1 September 2019 (UTC)
 * Delete – Although nice but irrelevant Braganza (talk) 15:18, 2 September 2019 (UTC)
 * Keep - I do totally agree with Checco. All that is need is put an orizontal line to indicate the important event of the government change and then start the calculation with the new government. --Thorin III (talk) 17:41, 5 September 2019 (UTC)
 * Comment - Just as a supplement to my previous explanation: we've just had a situation where the government column has proven to be absolutely impractical, as evidenced by this 's edit. (6) What happens when there is a minor party within the government, but an opinion poll does not return a figure for it? Impru 20  talk 13:56, 10 September 2019 (UTC)
 * That is a good point, enough to make me change my vote. Now I am for delete. --Checco (talk) 14:40, 10 September 2019 (UTC)
 * Delete I'm not a usual contributor of this page, but I think it's misleading, it's becoming difficult to keep track of the sum of the parties that approved the new government and it occupies a lot of space that could be used in the future for new parties (the new two parties of this summer, Cambiamo and We Are Europeans, are being polled by some pollsters).--Broncoviz (talk) 14:15, 11 September 2019 (UTC)
 * Keep Just like Opinion polling on the Emmanuel Macron presidency is worth an article, the sum of the support in opinion polls for governement parties is worth a column. --PJ Geest (talk) 07:15, 14 September 2019 (UTC)


 * The sum of the support for the parties in the coalition government does not represent necessarily an estimate of the support for the government. This constitutes a big WP:OR. The Macron presidency article in this sense is a completely different example, because in France (Presidential Republic) the polls actually directly ask and measure the support for the President and the government (like in the US too). --Ritchie92 (talk) 08:40, 14 September 2019 (UTC)
 * The use of the sum of governement parties is used by political commentators as measure for the support of the government, if not in any case (and maybe a more neutral formulation), they use it just to know how much of the electorate the governement parties reach (see for example the director of an italian polling agency which uses it: Governo, il patto giallorosso non piace agli italiani and Sondaggi politici. All wikipedia does is represent this measure, which is also used by political commentators. So this is not original research. --PJ Geest (talk) 09:58, 15 September 2019 (UTC)
 * What you are doing here is a clear exercise of WP:SYNTH. Firstly, the sources you provided do not present the combined data for a PD/M5S alliance as "support for the government", but rather as aggregations of the data for both parties as opposed to the centre-right coalition. This is, it is presented as an hypothesized electoral alliance of both parties (btw, excluding LS/LeU, which is also in government), rather than as some sort of measure for government support (which, if anything, should belong to the electoral coalition section if such an alliance becomes formalized). This is coupled with the fact that these sources do show separate results for actual questions on the government's support/trust among the population.
 * Secondly, the aggregation made by the source refers to the Noto poll of 4 September; this is, with the Conte I Cabinet still in place. Thus, an additional argument that breaks your thesis that "the sum of governement parties is used by political commentators as measure for the support of the government". This is a conclusion you reached on your own, not what the sources you provide actually state. Nonetheless, this is a perfect example of the SYNTH problems the use of such aggregations as "Government support" do represent. Impru 20  talk 10:11, 15 September 2019 (UTC)
 * I agree you should not say - a measure for support of governement (if you say this, I agree this can be interpreted as original research), sorry for the confusion. But the current version of the article is just a representation of how much of the electorate the governement parties reach. This is just a neutral fact. For me you are using a quite autistic definition of original research. It is just a plus sign you use. The graphical summary also uses math to make a trend line which is not given in the original source, will you remove this also? This article also gives an overview of all different sources of polling companies, so if you go over the results you can discover a different trend then if you consult each individual polling institute. So should we split the list for each individual polling institute? Because now it is a synthesis of material which is not in an original source. You can go quit far if you use an autistic definition of original research. --PJ Geest (talk) 11:03, 15 September 2019 (UTC)
 * It has been already explained above why it is problematic for us to try to infer this ourselves, so I'll not repeat myself again. However, this is vastly different from the chart itself or the mere presentation of opinion polls within a table, because those both are a mere presentation the direct data provided by RS, without us re-touching or re-interpreting such data. What you are defending here is to show a specific aggregation of results, over many other possible ones, that is not backed by sources. Indeed, the calculus itself is allowed under CALC, but the issue here is not the math itself, but the selective presentation of such data as some sort of "government support" figure, when this is absolute false (and, btw, this causes a large number of issues, as pointed out above). SYNTH is defined as "combining material from multiple sources to reach or imply a conclusion not explicitly stated by any of the sources"; this is exactly what is being done when combining the separate support from various parties into one single bloc in order to convey something that is not suggested by sources (i.e. "how much of the electorate the governement parties reach", which btw is exactly the same as "support for the government", just using more words). I can't think how this is an "autistic" definition of SYNTH or OR. You can define this exercise in many ways, but neutral is not one of them. Impru 20  talk 12:11, 15 September 2019 (UTC)
 * I don't know if you are native English speaker or not, but why would you use the word "autistic" to define another editor's reasoning? This is pretty bad. --Ritchie92 (talk) 12:21, 15 September 2019 (UTC)
 * No I am not a native English speaker. Sorry, did not meant to hurt people. If this is the case my apologies. At, I understand your reasoning. Still find it a pity the column would not be included. But maybe this is indeed more conform to the rules. Sorry, I maybe I don't know the rules as well as you do. I will pull back from this discussion. Thanks anyway to all the editors (including and ) which do such an effort for keeping this article up to date. I did not want to start a conflict. --PJ Geest (talk) 12:37, 15 September 2019 (UTC)


 * Delete or split. Might be worth its own article but not really useful here. Onetwothreeip (talk) 08:37, 14 September 2019 (UTC)
 * Keep (at least the government column). Unless there is a one-party government, we should always have, for all countries, a government/opposition tally, for this is a fundamental aspect of political life in a (liberal) democracy, and especially in a multipartite democracy where coalitions are the norm. Whether the coalition changes is unimportant. The reader of today, or the reader of the year 2026, needs to know what kind of electoral support the then-coalition had on DD/MM/YYYY. Kahlores (talk) 20:55, 19 September 2019 (UTC)
 * P.S.: As for the overall support (also known as "governement trust"), I should mention that last May, I suggested on this talk page to add a section about the Italian polls that are specifically surveying government trust, but then I received negative responses. If the government/opposition columns are finally deleted, I think it should be discussed again. So far I've gathered numbers from one pollster. Kahlores (talk) 20:55, 19 September 2019 (UTC)
 * This is basically going around in circles on the same specific issue: a self-made aggregation of government parties does not equal to "government trust" or "popular support for the government"; rather, it is a NPOV violation, as it is being presented as such when it isn't. The discussion on whether to include specific tables for "government trust" figures is entirely independent of the fate of the self-made "government column" here, because they do not refer to the same thing. Impru 20  talk 21:10, 19 September 2019 (UTC)

Decision
So, what have we decided? I see 6 users for "Delete" and 3 for "Keep". -- Nick.mon (talk) 13:46, 30 September 2019 (UTC)
 * I think it's not up to "us". A closer who was not involved must WP:CLOSE the discussion. --Ritchie92 (talk) 16:32, 30 September 2019 (UTC)
 * Ok thanks, I'm not so skilled in these practices :) -- Nick.mon (talk) 19:09, 30 September 2019 (UTC)
 * I've now requested for the discussion closure by an uninvolved editor, so that we may have a decision soon already. Thank you all for your participation. :) Impru 20  talk 15:31, 12 October 2019 (UTC)
 * Thanks to you, User:Impru20, for crafting this RfC. --Checco (talk) 16:21, 12 October 2019 (UTC)

Q2 discussion

 * Comment - I don't have a strong opinion on this, but this would inevitably require for the "Other" column to be removed, as it provides absolutely no information in addition to the "Government" colum, is redundant ("Other" = 100.0 – "Government" figure. Doesn't add anything the "Government" column does not provide already) and rather arbitrary (because it includes support for parties outside parliament that are not even formally reported in opinion polls). Impru 20  talk 19:01, 30 August 2019 (UTC)
 * Comment: I agree with User:Impru20. The "Others" column should definitely be removed in any case. Regarding the "Government" column, in the case it is kept, there is an issue with consistency about which government majority to consider at a certain date. Of course, during the normal course of a government there is no ambiguity in picking which government majority is the one. However, in periods such as during a government formation (e.g. from 4 March 2018 to 1 June 2018), formally another government is in charge, even though the PM has presented their resignation (in the example, the acting government was the Gentiloni Cabinet, but at the moment the table lists the "future" Conte Cabinet majority). I see only two possible options here, but both of them are imperfect and might incur in what Impru20 said in their comment on Q1. Here are the options:
 * 1) At each given date in time, report the government majority of the acting government, no matter if it is during a future government formation (i.e. under resignation) or not. This means modifying the current table to show the Gentiloni cabinet majority in the period going from 4 March to 1 June 2018.
 * 2) If a certain date corresponds to a period of government formation, that is the former PM has officially resigned and the President has started the party talks, the "Government" column is kept blank. The resignation usually occurs during government crises, or in correspondence of a new election, right after the Presidents of the Houses of Parliament are elected. This means modifying the current table, and blank the entries from 23 March to 1 June 2018 (Gentiloni resigned on 23 March), and from 20 August 2019 to today. I don't think in this case we should blank also 4 March to 23 March 2018 because in that time nothing formally happened to the Gentiloni cabinet.
 * In the moment I cannot think of another consistent and rational option for that column. In any case there is no reason to show the Conte majority in the period before Conte has been appointed. Finally, I don't have a strong preference for any of these two options, but in case it comes to a decision I would pick the second one. --Ritchie92 (talk) 09:09, 1 September 2019 (UTC)

In my opinion, you should leave the columns empty in the time before June 1st and after August 9th/20th Braganza (talk) 15:29, 2 September 2019 (UTC)
 * Comment – "At each given date in time, report the government majority of the acting government, no matter if it is during a future government formation (i.e. under resignation) or not", that's what I would like. --Checco (talk) 06:25, 2 September 2019 (UTC)
 * Comment – What does the reader find out when the polls of the parties are presented to him by a resigning government
 * Comment – Did anyone notice that More Europe (+Eu), which voted for the Conte bis investiture but has no portfolio, has been "forgotten" in the government tally? Either that or the footnote wrongly says "sum of parties supporting the cabinet". Kahlores (talk) 16:20, 20 September 2019 (UTC)
 * The party More Europe officially does not support the cabinet. Only the +Eu group in the Chamber of Deputies voted in favour, but against the party's decision. --Ritchie92 (talk) 09:59, 21 September 2019 (UTC)
 * My mistake. I had only looked at the confidence vote. Kahlores (talk) 13:28, 21 September 2019 (UTC)

Opinion polling graph
I noticed a weird thing with the curves plotted in the graph File:OpinionPollingItalyGeneralElectionNext.png (version 21:52, 2 January 2020) that you are maintaining: why do the lines already go beyond the line of January 2020 if there are still no new polls in January that have been included in the updated version of the graph? I think there might be some unwanted shift in the plotting routine. --Ritchie92 (talk) 13:21, 8 January 2020 (UTC)
 * Hi there! Yes, that's relatively intended and comes as a consequence of the averaging at the final end of the trendline. Polls obviously stop abruptly with the last published one; if this was depicted literally in the chart it would look heavily distorted as the final average would rely mostly on the latest poll's results (a situation further aggravated in case such a latest poll was an outlier for some or all parties). In order to avoid this, the average is manually extended by a small number of days (10 for this chart) so that the end point of the trendlines come up more softly and stylized. It's actually technical routine intended precisely to avoid unwanted shifts in plotting.
 * Take for instance the latest Ixè poll showing Lega at 29.5% (not yet included in the plotting, btw). Ixè tends to give Lega a couple points below what other pollsters typically record; if shown as such, you would have a drastic plotting shift for Lega from around 32% to below 30%, which wouldn't be representative of actual polling trends since Lega only loses 0.5 points from the previous Ixè poll, not >2 points. Hope I've been able to solve your doubts. Impru 20  talk 13:38, 8 January 2020 (UTC)
 * I think one simple solution to this would be not plotting the average for the very first points at the beginning of the graph, and for each date plotting the average of the previous polls. This way we would have a plot that ends at the real current date, doesn't seem to "make up" future data, and we lose the curve on the first few days after the latest general election, which is not a great loss. --Ritchie92 (talk) 14:17, 8 January 2020 (UTC)

do you think you could apply my suggestion and solve the problem of the current graph, looking like it "predicts the future"? This is along the lines of WP:CRYSTAL I think, or at least it looks like it. --Ritchie92 (talk) 23:49, 19 January 2020 (UTC)
 * Considering that I updated the chart earlier today and fixed the issue by setting the last day at the date of the last opinion poll, I am seriously surprised that you think this "predicts the future". Could you please elaborate? Impru 20  talk 01:52, 20 January 2020 (UTC)
 * Right, I had to refresh my cache to update the picture. Sorry for the inconvenience, Ritchie92 (talk) 08:12, 20 January 2020 (UTC)

SuperAverage
Does it make sense to include numbers (YouTrend) which are base on a super-average of polls? That is not a poll as such and hopefully all the polls used to calculate the average are in the table already --FantinoFalco (talk) 21:29, 22 January 2021 (UTC)

New opinion polling graph
I created a new opinion polling graph and I think it should replace the "old" graph beacuse it can be updated by everyone every month. What do you think about it?--Facquis (talk) 12:34, 11 May 2021 (UTC)

@User:Broncoviz, User:Civiltalatina, User:Altorio, User:Eddy world, User:Holapaco77, User:Impru20, User:Nick.mon, User:Ritchie92, User:Scia Della Cometa, User:Checco


 * I see that this is not compliant with how opinion polling charts are designed everywhere else in Wikipedia. Because of one simple fact: trend lines are not subject to monthly time periods. And by doing monthly averages, you'd be basically: 1) Conducting WP:SYNTH by taking a specific set of opinion polls with which to conduct calculations, and not others; 2) Raising a number of questions regarding the time period picked (why monthly? Why not weekly? Or every two weeks? Etc.) and how people calculate it. The current chart, by using a LOESS average, avoids all those issues and poses a much bigger improvement.
 * Also, your proposal takes a lot of article byte space in an already massive article. We should look toward solutions that minimize the amount of space used, not enlarge it.  Impru 20 talk 13:06, 11 May 2021 (UTC)
 * I think the current situation is problmetic as the graph is managed by a single user and is uneditable, so I think using this template is the best solution to adopt. To solve the problems you have listed we could use the "Youtrend supermedia" instead of the pure monthly average of the polls listed in the table and this is in my opinion a good solution already used in the pages of american politics (for example here).--Facquis (talk) 13:19, 11 May 2021 (UTC)
 * Much to the contrary; anyone can edit it! Just by using R and collecting the data from the tables, you can easily update the chart. It's tricky to learn at first if you are not experienced at R, but I wasn't before the 2021 Catalan regional election and got quickly accostumed to it. Opinion polling for the 2021 German federal election or Opinion polling for the next United Kingdom general election, to name a few, use this system, and the trend throughout Wikipedia is towards using LOESS as the preferred method of showing opinion polling averages, since it's the one that conflicts the less with SYNTH and NPOV. About US elections, those are not good examples to take (and that's why almost no one in Wikipedia uses US opinion polling charts as example of anything. They also tend to vary every election cycle, which is problematic)  Impru 20 talk 13:25, 11 May 2021 (UTC)
 * The Graph:chart template doesn't really state the method that it uses for interpolation. But you can tell there is something seriously wrong with it (look at the PSRM trend, it doesn't make any sense). I would generally avoid it, especially when better solutions are available. --Gbuvn (talk) 13:38, 11 May 2021 (UTC)


 * I am not an expert, but I like this version more than the current one!
 * Also the one at Opinion polling for the 2021 German federal election would be better. --Checco (talk) 17:12, 11 May 2021 (UTC)
 * The current one is the one at Opinion polling for the 2021 German federal election.  Impru 20 talk 18:11, 11 May 2021 (UTC)


 * I agree that monthly averages should not be taken as a synthesis, while the regression is the best solution (much better than the former 7-point average). So I support on this. Anyway, Impru20 keeps the graph up-to-date every 1-2 months, so it's not that bad. The frequency of the update could certainly be improved, but as they said, anyone can make the same R script that produces that plot ( do you have a template for the script?) --Ritchie92 (talk) 10:01, 13 May 2021 (UTC)
 * It is based on this, although, if other users wish to update this graph, it could be helpful, if the exact script for Italy was made available as well. --Gbuvn (talk) 11:39, 13 May 2021 (UTC)

Free and Equal
I'd like to discuss about splitting up Italian Left and Article One in the table since half of the pollsters are doing it so.

Right now, Demos & Pi, Termometro Politico, Euromedia, Piepoli, Noto and Demopolis keeps polling the leftist parties as a whole, while Bidimedia, Index, Tecnè (only Italian Left, they stopped polling Article One a month ago), SWG, EMG and Ipsos are polling both parties. Lab2101 is messy since they poll either LeU AND The Left (since both lists were led by SI, are they counting the latter twice...?).

The reason of this change of mind is that, since the Draghi Cabinet, most pollsters started to split the two parties as Article One supported the new cabinet while Italian Left ultimately joined the opposition. As nobody knows how the leftist party will stand in the upcoming general election, I think we should consider rather to split them up or not.

@User:Civiltalatina, User:Altorio, User:Eddy world, User:Holapaco77, User:Impru20, User:Nick.mon, User:Ritchie92, User:Scia Della Cometa, User:Checco, User:Facquis --Broncoviz (talk) 16:38, 15 May 2021 (UTC)


 * I surely agree with you and, if you are available for the job, please let me thank you in advance.
 * Several pollsters, including Euromedia and Piepoli, track also Us with Italy, which has become a full-fledged party. --Checco (talk) 16:48, 15 May 2021 (UTC)


 * I wouldn't include NcI right now since Euromedia changed NcI to a generic "Others CDX" while only Noto (Piepoli doesn't include it) and EMG keeps polling it, if in the future other pollsters will follow them we should include them as well.--Broncoviz (talk) 19:09, 15 May 2021 (UTC)


 * On NcI, you are right (I meant Noto, not Piepoli) and I see your point on not including it.
 * On the main issue, I confirm that I would split Art.1 from SI. --Checco (talk) 19:54, 15 May 2021 (UTC)


 * I'd like to address how we should split the parties and how they should be considered for the graph: right now we polled LeU as a whole party and some pollsters are still doing it so, but how should we handle in the graph a poll that contains LeU if we're going to split the parties? The table can be easily arranged like this in order to put only one new column instead of two, but it's not the same thing for the graph. @User:Impru20
 * {| class="wikitable collapsible" style="text-align:center;font-size:85%;line-height:13px"

! rowspan="2" style="width:100px;"| Date ! rowspan="2" style="width:170px;"| Polling firm ! rowspan="2" style="width:60px;"| Sample size ! style="width:35px;"|M5S ! style="width:35px;"|PD ! style="width:35px;"|Lega ! style="width:35px;"|FI ! style="width:35px;"|FdI ! style="width:35px;"|SI ! style="width:35px;"|Art.1 ! style="width:35px;"|+Eu ! style="width:35px;"|EV ! style="width:35px;"|C! ! style="width:35px;"|A ! style="width:35px;"|IV ! rowspan="2" style="width:35px;"|Other ! rowspan="2" style="width:25px;"|Lead ! style="background:;"| ! style="background:;"| ! style="background:;"| ! style="background:;"| ! style="background:;"| ! style="background:;"| ! style="background:;"| ! style="background:;"| ! style="background:;"| ! style="background:;"| ! style="background:;"| ! style="background:;"|
 * - style="height:30px; background-color:#E9E9E9"
 * }


 * I like it! I would have Art.1 first because it obtained more elects in 2018. --Checco (talk) 18:19, 22 May 2021 (UTC)
 * I think we should continue to keep the LeU label as most pollsters do.--Facquis (talk) 14:28, 30 May 2021 (UTC)
 * I am not sure most pollsters keep LeU. However, that is factually incorrect. Art.1 and SI are separate parties and, at this time, no longer support the government together. They even split in the Senate. --Checco (talk) 19:33, 31 May 2021 (UTC)
 * More and more pollsters track Art.1 and SI separately. Additionally, CI should supplant C!. --Checco (talk) 18:18, 5 June 2021 (UTC)
 * I remember spending quite some time producing this table exactly for this purpose. I would suggest interested editors to please update it. Thanks, Ritchie92 (talk) 21:28, 6 June 2021 (UTC)

I do not know who finally got SI and Art.1 separated, but thanks to him/her and to all the other editors for not challenging that. That, as well as the creation of a new column for CI, was a great improvement. If someone is willing to do that and has time for that, SI and Art.1 should be separated also before May 2021, at least since Draghi's government. Unfortunately, I have no time for that. --Checco (talk) 14:05, 26 June 2021 (UTC)

update polling map!
the polling map should be updated every month… 5.168.17.74 (talk) 13:10, 19 May 2022 (UTC)

refresh the graphical summary!
even more so now that the elections in Italy will take place in two months, the map should be updated often impru20 do something! 151.71.84.89 (talk) 14:34, 23 July 2022 (UTC)

More Europe and Action single column
@Broncoviz Why don't we make a single column +E-A for the whole 2022? In this moment it looks weird that a lot of rows have +E and A together, and starting from some random point in time we merge them. We can have a note saying that in the few cases where the parties were polled separately we show the sum of the two. Yakme (talk) 08:55, 8 July 2022 (UTC)


 * Sure, I think it's pretty reasonable Broncoviz (talk) 17:59, 8 July 2022 (UTC)


 * I think that from now on Azione and +Europa will be polled separately again, as the latter will likely be part of the centre-left coalition, while Azione will not. CorradoMor (talk) 07:36, 9 August 2022 (UTC)

Weird polls
The last two polls added seem a little bit dubious. The first, by one prof.Enzo Rizzo, links only to the screenshot of an infographic from somewhere; the second is by G.D.C, Giornale di Chieti, the local newspaper of a small town. Both polls have been added by the same IP. 151.75.90.44 (talk) 12:53, 16 August 2022 (UTC)


 * G.D.C is published also on the governo's website http://www.sondaggipoliticoelettorali.it/ListaSondaggi.aspx?st=SONDAGGI 93.34.149.131 (talk) 15:18, 16 August 2022 (UTC)

G.D.C. Poll
Yakme (talk) removed the G.D.C poll with no real reason. He cannot assume that this poll is a fake poll considering that it is published on the site of Presidenza del Consiglio, it respects what the law requires and contains all element requested such as the date, the sample size and features and the method of interview. It should be considered as a real survey until proven otherwise. EspertoPolitico (talk) 09:41, 17 August 2022 (UTC)

Copyright of opinion polls
Is it legal to copy all the results of opinion polls conducted by private companies (published on their copyrighted sites or on the copyrighted site of the commissioner), and then paste all these copyrighted data into the table of the Wikipedia entry? All regular? Just an example: in the table I see the results of the 8-11 August opinion poll made by Tecnè srl (whose site says "Copyright @ 2022 Tecnè Italia. All rights reserved") on behalf of the commissioner RTI spa-Mediaset (whose site used as the source of the poll expressly states: "Copyright © 1999-2022 RTI S.p.A. - All rights reserved"). How "copyrighted data" can be compatible with the free licence used by wikipedia? Holapaco77 (talk) 06:03, 21 August 2022 (UTC)

update polling map!
please impru10 update polling map! 158.148.112.179 (talk) 15:57, 25 August 2022 (UTC)

Politiche2022
Many seat projections by Politiche2022 have been added, taken from their website. This website updates the projection almost daily, and editors are adding entries for each day, linking to the web.archive links archived on the day of the projection. However there is a problem: at the moment each of the Politiche2022 entries in the table looks like an independent new projection, while it is certainly not, because each projection is based on an average of other polls in the previous days. So for example the projection relative to 27 August is based on an average of polls that in part are the same polls used for the 26 August estimate, and so on. Note that the website does not say which polls it used for each new projection. So I do not think we are doing a good job in terms of clarity if we keep adding the updated projection of Politiche2022 as a new entry each day on top of the former ones. As far as I can see (this is the first time I learn of Politiche2022) this website in principle does something similar to what FiveThirtyEight does for the US elections; the difference is that the 538 projections are not added in opinion polling articles of US elections day by day. So I would suggest that – in order to avoid having a table that is over-representing Politiche2022 – we remove these kinds of projections made by poll aggregators from our table, and keep only the ones made by the pollsters themselves. Another option could be to create a "Polling aggregation" subsection to handle such type of data – in that section we only have to keep the latest up-to-date estimate (see the US case for example). Yakme (talk) 20:42, 29 August 2022 (UTC)


 * The "Polling aggregation" option is a good idea to fix the issues your highlighted while keeping data that may be useful. Davide King (talk) 18:49, 30 August 2022 (UTC)