Talk:Opinion polling for the 2017 French presidential election/Archive 1

Disruptive editing detected
Someone has obscured the second round voting, if anyone could help undo that it would be very helpful.

le pen vs hollande
does anyone know a poll with hollande vs le pen?81.58.144.30 (talk) 10:52, 2 May 2013 (UTC)

Le pen will win the presidency anyway. Who cares? — Preceding unsigned comment added by 79.71.170.209 (talk) 15:07, 26 March 2017 (UTC)

Table needs reversing
The table seems to be set out far-right...centre...far-left. But is displayed with the far-right at the left hand side and vice versa. Surely it makes some sense to reverse that and have the left displayed on the left and the right to the right, as it were.... ?? 86.156.183.71 (talk) 19:54, 21 February 2015 (UTC)
 * More important, should be reversed to put the latest polls on top.--MiguelMadeira (talk) 02:18, 20 November 2015 (UTC)


 * Agree with OP. The table needs reversing. So that the far left are on the far left... and the far right are on the far right. If no-one objects I will change that around in a week or so. 87.115.88.94 (talk) 20:21, 12 December 2015 (UTC)
 * Viewing the History, it seems someone has intentionally made the polling inconsistent. --Weegeeislyfe (talk) 00:47, 4 January 2017 (UTC)

Table change
I suggest some major changes for the table covering opinion polling for the first round of elections. Rather than listing all the candidates -- which takes up far too much space, and makes it difficult to get a quick overview of the polling numbers -- we should merge the candidates from the same party into a single column. I realize that this is a presidential election, and thus is personality-focused, but it makes the table far neater, and as long as we include the name of the candidate in each cell, there shouldn't be any problems.

This has been done in at least one other article, though I can't find it at the moment; the point is that this system has been used before. In the below proposal, I have removed the micro-candidates (I think we can all agree that Cheminade is not going to be president), but if anyone gains support to the point where it is significant, we can naturally just add them in later.

If there are no objections in about a week, I might implement this in the article, with some modifications, like I said. :)

Μαρκος Δ (talk) 19:14, 8 February 2016 (UTC)
 * The non-free logos are not allowed here on Talk (WP:NFCC) and they won't be allowed in the article (WP:NFCC). I have commented them out. – Finnusertop (talk ⋅ contribs) 09:29, 9 February 2016 (UTC)

A *Lot* of Missing Polls
I've been comparing this table to the one in the French version of this article, and only a small fraction of the polls listed there seem to be here (polls from Ifop and Elabe in particular seem to be ignored here). Is there some (perhaps methodological) reason for ignoring this polls? If there are no objections, I'd like to copy those polls over here to make this article more complete. Thanks, Chuborno (talk) 21:46, 11 October 2016 (UTC)

Article is inconsistent and needs changing
The first round polls are shown with most recent poll first. The second round polls are shown with most recent poll last... List them one way or the other, but not both. 86.215.104.82 (talk) 19:25, 7 December 2016 (UTC)

Chart
Any reason why there isn't a fancy/pretty chart?

James Tamim (talk) 20:10, 15 January 2017 (UTC)


 * Complexity of implementation (multiple candidates for PS/LR, Bayrou in/out, etc.) and lack of enough polls to actually do so; going by 2012's level of polling, I might be able to throw one together by the end of January or early February. Mélencron (talk) 20:44, 15 January 2017 (UTC)
 * Now that pollsters are only surveying two scenarios, I anticipate that we'll probably have enough data points by mid-February. We'll see then. Mélencron (talk) 04:45, 5 February 2017 (UTC)
 * That's when Bayrou says he will decide whether to stand or not. Greengreen2 (talk) 19:23, 7 February 2017 (UTC)
 * Yes, and there have been many polls conducted with and without Bayrou's candidacy already. Mélencron (talk) 21:30, 7 February 2017 (UTC)

Why there are several rows with different results for every single poll?
2804:7F7:DC80:E242:0:0:0:1 (talk) 21:49, 31 January 2017 (UTC)
 * These are all different possibilities based on the inclusion of certain candidates, whether a candidate wins a primary, or simply scenarios surveyed by the pollster. Mélencron (talk) 23:20, 31 January 2017 (UTC)
 * In particular, many pollsters asked respondents how they would vote in different cases according to who the Socialist Party chose as its candidate and whether or not François Bayrou decides to stand. The SP has now nominated Benoît Hamon, but Bayrou has not yet announced his intention. Greengreen2 (talk) 13:40, 4 February 2017 (UTC)

Hamon vs Le Pen
It would be interesting to see a poll about this second-round scenario if one exists. Now that Hamon is the official PS candidate he seems to be catching up with the front-runners. MFlet1 (talk) 09:46, 2 February 2017 (UTC)


 * I'm sure we can expect one in the next few weeks. My guess is that it'll be less decisive than Macron v Le Pen but more decisive than Hollande v Le Pen was. Melenchon v Le Pen was also a compelling scenario but probably not one we'll ever see tested. Maswimelleu (talk) 09:57, 2 February 2017 (UTC)
 * I think we should let our reliable sources produce what they deem fit. I can not see any evidence of a bounce in the polls that would give Hamon a credible chance of getting through to the 2nd round, which does make me wonder whether the reliable sources will bother to poll this question. However, it is early days yet. Just a reminder; Wikipedia is not a forum. 2407:7000:875B:C370:A5C8:43FE:B396:A942 (talk) 08:26, 3 February 2017 (UTC)
 * Hamon's poll results have improved rapidly and Mélenchon's have dropped, as have Fillon's. I won't be at all surprised if the second round is between Hamon and Le Pen, whether or not Mélenchon withdraws. In the latest poll, the second to fourth candidates scored 20.5%, 18.5%, and 16.5%. The 18.5% candidate's scores are dropping and he is beset with scandal, while the 20.5% candidate's use of ministerial money for entertainment has been questioned. Remember what happened in the first round in 2002. It is unfortunate that we have no polling figures regarding a run-off between Hamon and Le Pen, but when we get some we can put them up. Figures for Mélenchon-Le Pen would also be interesting. That contest is less likely but it is by no means outlandishly improbable. I could offer further reasoning, but this is not the place. Greengreen2 (talk) 13:47, 4 February 2017 (UTC)
 * Just a friendly reminder that this is not a forum. The talk page is there to discuss what should and should not be included in an article and how we should present it; it's a means of seeking agreement. We should not be using talk pages to fantasise or speculate. 118.93.49.168 (talk) 20:57, 7 February 2017 (UTC)

Are those three February 2017 polls by Idop-Fiducial independent of each other?
We now have three polls by Idop-Fiducial, with overlapping periods stated for fieldwork, namely 29 Jan - 1 Feb, 30 Jan - 2 Feb, and 31 Jan to 3 Feb. Each poll had about 1400 respondents. Are we sure that on 30 Jan and 31 Jan the respondents for each poll were all different people from the respondents for each of the other polls held by the same company on the same day? Greengreen2 (talk) 11:27, 4 February 2017 (UTC)


 * Most likely the respondents of the 30 and the 31 are the same across the polls, but of course we can't know.  Rami  R  12:15, 4 February 2017 (UTC)


 * We can know. The first word in each of the documents we link to, in which they publish their methodology and results, is "Rolling". So our suspicions are right and they are conducting a rolling poll. The responses from the fourth most recent day drop out when the latest day's responses are added in. But since each time they are quoting results from three days of fieldwork, two-thirds of the responses considered in each day's announcement were also considered in the announcement the day before. We shouldn't be treating every poll they publish as if it has the same significance as a discrete poll. Otherwise Idop-Fiducial is going to dominate this article, not because they are doing huge polls but just because of how they aggregate their data. Perhaps a note is needed to say that their reported poll results aren't independent of each other? Greengreen2 (talk) 13:24, 4 February 2017 (UTC)


 * I don't mind a footnote saying "rolling poll" or something, but per WP:NOR we really should treat it like any other poll.  Rami  R  16:01, 4 February 2017 (UTC)


 * Company 1 gathers data five days running and publishes it on the sixth. Company 2 gathers the same amount of data, similarly on five days running, but publishes three-day aggregations on the fourth, fifth and sixth days. Result: Company 2's results gets presented on the foundation's website as though taken together they have triple the noteworthiness of Company 1's. But if deciding not to present them here would be against the foundation's policy, then that's what it would be. Greengreen2 (talk) 17:23, 4 February 2017 (UTC)


 * I don't think the issue here would be undue prominence given to the Ifop-Paris Match rolling poll, but it might be better, say, to merge rows so as to indicate in some way that it's a rolling poll. I've provided an example below:

Blocked out images in case of licensing issues, as above. Of course, it'd also be necessary to indicate separate files by date – but I think it might suffice to simply link to Paris Match's site in this case. Mélencron (talk) 22:58, 4 February 2017 (UTC)


 * I think this suggestion is fine, combined with Rami R's suggested footnote saying "rolling poll", or "tracking poll" as it also called, linked to the section of the Opinion Polls entry about tracking polls, which I will edit to make clearer. Greengreen2 (talk) 11:52, 6 February 2017 (UTC)

Presentation of first round table

 * I made the above changes to the table to make it more reader friendly. Previously the leading candidate was highlighted and the leading 2 candidates had their figures in bold. I have swapped it so that the leading 2 candidates are highlighted and just the leading figure is in bold. This makes it more reader friendly for those of us who struggle to distinguish between the bold and standard print on the white background. It also better reflects the article and electoral system as the top 2 candidates go through to the next round, so there is not actually 1 winner of the first round, there are 2; it makes no difference whether you come first or 2nd in the first round.
 * However, Mélencron objects to this change for some reason, saying that it is inconsistent, I am not sure what it is suppose to be inconsistent with but there we go, perhaps he/she can explain. Though I am not hopeful as if you look on the Talk:French presidential election, 2017 they fail to engage properly with the substance and do not give a final response in order to end the discussion. I am really curious as to what possible objection anyone could have to making a table more reader friendly and reflective of the electoral system.
 * As a side note I am also curious as to what others think about introducing a lead column? 2407:7000:875B:C370:6C29:16B0:BB81:1B4E (talk) 21:59, 7 February 2017 (UTC)
 * I agree with Mélencron here. Shading is commonly used in opinion polling to depict most voted parties/candidates, and, specially for French opinion polling in previous elections, it has been common usage to just bold the two runner ups and shade the one topping the polls. Having several candidates' numbers shaded may actually mislead readers for this reason, as shading is usually reserved for the most voted party/candidate only. Besides, bold is designed so that it's really hard to actually "struggle" to distinguish between bolded and unbolded text, so if you're actually having issues with it, it's not a Wikipedia one. Also, you put it as if your design is "more reader friendly and reflective of the electoral system", but that is your opinion, not a fact.
 * On the lead column issue, it may cause issues for smaller screens in the wider tables with many candidates for previous years, since there's no more space for it. And if we can't have it in some of the tables, it's just non-sense to put in in the others. Impru20 (talk) 22:07, 7 February 2017 (UTC)

I agree with the change - the important thing for the first round is who is in the top two, not the winner overall. Absolutelypuremilk (talk) 23:36, 7 February 2017 (UTC)
 * I still believe there ought to be a distinction drawn between who's farther ahead within the first round, as it gives an idea of the gap that separates the two candidates. Frankly, I find that the background-color change makes it harder to tell who's first/second and doesn't make it any clearer than the current method does which candidates are the top two. Personally prefer the two bold/one highlight to two highlight/one bold; as the background colors are soft on the eyes, it frankly doesn't draw visual attention as easily as simply bolding does. Second method is far clearer/more distinct, imho. Mélencron (talk) 00:20, 8 February 2017 (UTC)
 * Mélencron, the distinction of who is farther ahead is easily discernible by reading the numbers. In two round systems it's more important who the two leading candidates are then their individual percentages.  Nevermore27  (talk) 23:03, 8 February 2017 (UTC)
 * Look, Nevermore27, I can read numbers, and I'm fully aware that it's a two-round system. However, it's also a matter of where one's eyes are drawn to – and the second option is far clearer in this distinction. The background-color highlight is soft and draws little contrast with the table itself. The bold text does – as Impru20 points out, it's explicitly designed to be clear. I don't see the value in making the table less clear to readers. Mélencron (talk) 23:15, 8 February 2017 (UTC)
 * This might be a function of my browser (Google Chrome), but the bold text without shading for the second place candidate is not clear at all to me.  Nevermore27  (talk) 14:50, 9 February 2017 (UTC)
 * Then, under that rule, absolutely everything is discernible by reading the numbers and then no bold nor shading is needed at all. Yet, customary practice all throughout Wikipedia and for past French elections has been to shade the most-voted party. French and English media everywhere have been highlighting for the past years how Le Pen has been topping the polls, and this coupled with this customary practice must lead us to believe that the most-voted candidate does indeed get some special attention and interest just for coming in first. As well as the fact that, while not always, topping the polls in the first round usually makes it easier for that candidate to win the second round (not always the case, but it's very frequent). Also, highlighting the candidate topping the polls allows for easier comparisons of the other candidates' relative electoral strength.
 * It is not customary practice "all throughout Wikipedia", look at Opinion polling for the Austrian presidential election, 2016 or United States Senate election in Georgia, 2014, both of which shade AND bold any candidates who are making the second round.  Nevermore27  (talk) 14:50, 9 February 2017 (UTC)
 * I don't see your point. I didn't say it is used in every election in Wikipedia, but that it's used widely, and specially for French elections: Opinion polling for the French presidential election, 2012, The Republicans (France) presidential primary, 2016, French Socialist Party presidential primary, 2017... even in Portuguese presidential elections, using a two-round system, the second-placed candidate is not highlighted at all: Portuguese presidential election, 2006, Portuguese presidential election, 2011, Portuguese presidential election, 2016 (and all of this not counting legislative elections throughout the world). Impru20 (talk) 15:09, 9 February 2017 (UTC)
 * Your Portuguese examples are simply invalid, because in all but two instances the leading candidate is taking more than 50% in polls, which means there wouldn't be a second round at all. And the French examples aren't proof positive of anything except that your preferred status quo holds sway.  Nevermore27  (talk) 15:15, 9 February 2017 (UTC)
 * Also, the Austrian presidential election you just posted does indeed give a special shading for the most voted candidate, so even one of your examples does indeed prove the fact that the most-voted candidate gets a special attention. As for the US Georgia Senate election, aside from the fact that US elections have different Wikipedia mechanics than the rest of the world, the article you just put is the perfect example of how chaotic a table may become if you just choose to equally highlight everyone (how many candidates at once are you willing to highlight through shading without no distinction to other polls in the event of a tie?). Also, you fail to acknowledge where's the issue with the current system in place. Impru20 (talk) 15:09, 9 February 2017 (UTC)
 * I have repeatedly explained my issue with the current system, you just disagree that there's an issue. I would be fine with implementing the Austrian system here if that would be amenable to everyone. And the Georgian poll table is indeed a unique case because they're all the same party, but that isn't an issue when there are different party colors to use.  Nevermore27  (talk) 15:20, 9 February 2017 (UTC)
 * You explained it just below. I don't recognize it as an issue because I'm using Google Chrome and I definitely don't have such an issue nor did I have it for years. So I don't know which is the issue you may be having with that browser. Bold text is designed for being discernible and recognizable. If you're having issues with it, then it may be another issue you have, but definitely not Wikipedia's or your Chrome browser. The Austrian system would be to maintain the current bold text, shade the second placed candidate and change the shading to a random color (and if you propose the Austrian system, that wouldn't be using the different party colours but some kind of "neutral" color. Not the same). It's a major change and would break with consistency with other French election articles, so it will surely need a firm consensus. Impru20 (talk) 18:36, 9 February 2017 (UTC)
 * My preference would be to shade the top two candidates in their respective party colors, which I think is perfectly reasonable and not disruptive at all. Anyone glancing at the table would be able to quickly see who the top two candidates are, and if they cared to look further and find out who the top candidate was, numbers aren't hard to read. You seem to be hung up on readers being able to tell who the #1 candidate is, but in a top-two system like France, that's not as important as knowing who the two candidates that are going to move on to the next round!


 * This is so easy to read that I'm honestly baffled that there are arguments against it.  Nevermore27  (talk) 21:26, 9 February 2017 (UTC)
 * But this choice doesn't highlight the most-voted party, which was one of the points above. You say that it's not important, but in two-round presidential systems nearly everywhere and, indeed, for past French elections, the #1 candidate is highlighted. So, obviously, past and current practice by a large variety of users throughout Wikipedia has been to, indeed, find this relevant. As a result, you can't just dismiss it just because you feel it as not important. You must acknowledge that others may find it important, and that this has been frequently the case when making opinion polling tables. This is a fact.
 * I honestly don't mind if you find yourself baffled or whatever if someone doesn't like your proposal, but I kindly ask you to respect that others disagree. It's obvious you have your own tastes on this, but you give contradictory reasonings to defend your point. You did put the Austrian case as an "example", yet your solution does not follow the Austrian example and you refuse to highlight the top candidate at all, in contrast to what is done in Austria. You also keep failing to acknowledge that the current system has been in use for years in both recent French presidential elections and primaries without causing any issue with the bolding, and suddently it has to be changed because you (somehow) have readability issues with bold text with your browser (which, btw, looks like the same than I use, and I don't have those readability issues at all). Yet you also say that numbers aren't hard to read to defend your point. So, what gives? I can also just reduce it to the same argument that you: bold isn't hard to read. And then, if numbers aren't hard to read, I don't get why you want to make those changes at all. Just apply your very same arguments to yourself and read the tables instead of forcing a change that, under your very own premise, are not necessary and could potentially be harmful to others, given that you can just distinguish candidates by reading the numbers. Impru20 (talk) 21:47, 9 February 2017 (UTC)
 * "Others disagree" with your preferred method too, including in this very discussion. So there isn't consensus that the status quo is best. I only brought up Austria because you claimed it was "customary all throughout Wikipedia" to only highlight the top candidate in polls. I have proved that that is simply not the case. And there is not a single winner in the first round of French presidential elections. There are two (assuming no one gets a majority). So unless a candidate is getting a majority in a poll, the top candidate in an opinion poll is not relevant. The top two are. The top candidate doesn't win anything except a spot in the runoff with the other candidate. Both candidates in the top two should be shaded and bolded. We seem to be repeating ourselves so unless we can call a vote of some kind to achieve consensus, I'd rather not continue this discussion.  Nevermore27  (talk) 22:12, 9 February 2017 (UTC)
 * Thing is, my preferred method is "just" the one that has been in use for years throughout the latest French elections, and one that is also in use for several other countries, either in full or in essence. Consensus is also achieved through editing, and the fact that the status quo has endured these many years without changes and in several articles is something that you, again, should acknowledge, because that also reflects consensus. It means that the current version works and has worked for much time. It's not "your opinion" against "my opinion". I support the status quo version not only because I find it the most appropiate, but because it has been the most widely accepted one during all this time.
 * Sorry, I thought that you said I would be fine with implementing the Austrian system here, and then you came up with something very different from the Austrian version. Maybe I'm mistaken.
 * Agreed. Given our point have been already exposed, we should better not repeat ourselves over and over again. Impru20 (talk) 22:29, 9 February 2017 (UTC)
 * Just to clarify, I said I would be fine with the Austrian method if that's what everyone agreed to do, but my preferred solution is as I demonstrated.  Nevermore27  (talk) 22:31, 9 February 2017 (UTC)
 * The two top-candidates are still highlighted in bold, which is clearly discernible, so I don't know where the issue is. User 2407:7000:875B:C370:6C29:16B0:BB81:1B4E is just seeing an issue where there's none. IMHO, this proposal makes no improvements on the distinction of the two most-voted candidates from the rest (it just intends to change bold text by shading) yet it hampers those readers interested in checking who's the one coming first, as well as the readability of the table as a whole. Impru20 (talk) 23:34, 8 February 2017 (UTC)
 * The issue User:Impru20 is that bold text without shading is not easily discernible at all to people using browsers like Google Chrome. Maybe you use a different browser where it's not a problem, but it's a problem for me to see who the second place candidate is. It is not at all a universal practice to shade only the top candidate in a two-round system, and I not only think that it's not a heavy lift to look at two shaded cells and find out which number is bigger, but I think it would be more immediately helpful to readers to clearly demonstrate (via shading) that there are two "winners" in the first round.  Nevermore27  (talk) 15:09, 9 February 2017 (UTC)
 * I've been using Google Chrome for years and I can assure you I can clearly distinguish bold text from unbolded text. I myself do frequently use bold text (without any shading) to highlight some relevant aspects within articles I edit and am still able to perfectly discern between the different types of text. I can't see where's the issue, except that it is one of personal tastes maybe. Impru20 (talk) 15:12, 9 February 2017 (UTC)
 * I think the top two candidates for the first round should be highlighted, since just bolding them makes it harder to easily see who's in the top two. Highlighting them will be much more convenient for users. Mydabo (talk) 22:29, 9 February 2017 (UTC)
 * As things stand it seems we have 4 editors in favour of highlighting 1st and 2nd place with the leader in bold and on the other hand we have 2 in favour of highlight 1st place with 1st and 2nd in bold. It seems that several people have highlighted that the bold text in the columns against the white background is not very distinguishable on some devices/screens, or due to the readers vision; this highlights the issue of reader friendliness. One thing that is clear, is that it is easier to see bold text against a coloured background.
 * It seems there is a consensus to highlight 1st and 2nd with 1st place also in bold. This discussion has been inactive for a few days and we do seem to have a settled majority. 2407:7000:875B:C370:11E9:8FB2:AB08:ED90 (talk) 20:57, 10 February 2017 (UTC)
 * Yeah, I don't have a strong opinion either way, but shading the top two does seem to fit better with the format of the election. Jdcooper (talk) 21:29, 10 February 2017 (UTC)
 * If we're doing that, can we at least, well, bold the second-place candidate in addition to shading them? Same reason as cited earlier. Mélencron (talk) 21:32, 10 February 2017 (UTC)


 * This brings us to 5-2, Mélencron please can you stop edit warring. 2407:7000:875B:C370:11E9:8FB2:AB08:ED90 (talk) 21:36, 10 February 2017 (UTC)
 * If bolding makes it clearer on other users' interfaces (personally mine shows the shading much more clearly) then yeah, I would support that too. Jdcooper (talk) 21:37, 10 February 2017 (UTC)


 * I'd like to iterate a point that never got brought up in this discussion, and it's that despite the French presidential election being a two-round system, almost all media outlets report in the headline the candidate who's in first place during the first round: e.g., "Le Pen en tête d'un sondage pour la présidentielle", "Présidentielle: Le Pen en tête, Fillon recule, Macron progresse", "Présidentielle : Le Pen en tête au premier tour, battue par Fillon...", "Présidentielle 2017 : Le Pen se maintient en tête des sondages...", etc. and much more infrequently the second round. I think that's reflective of the view of who's ahead in the first round still does matter to an extent – enough so that it's the main way that they report polls in the first place. Mélencron (talk) 21:44, 10 February 2017 (UTC)
 * I'm sorry but this point did get brought up and was actually addressed in the consensus that was reached, ie for the 1st place candidate to have their figure both highlighted and in bold and second place merely to be highlighted. I can see the change you have made since, to put 2nd place in bold too.
 * Personally, I do not have a problem with that change you have made and would have been my preferred option; in terms of the outcome there is no difference between 1st and 2nd place in this round. However, we did try to take account of the point you made above. For this reason, I am not going to take issue with you on it. If you are happier for both to be in bold, then I'm inclined to go along with you there.
 * I am often at pains to point out that Wikipedia is not a news outlet, we are an encyclopedia; we should not be swayed by news headlines; we should focus on the substance. The solution that has been reached better reflects the reality of the situation. If we were to merely look at headlines, we would all be putting big bets on Marine Le Pen to be the next President but if we look at the reality of the polls and electoral system, this appears to be most unlikely. Again, I must stress that is not a prediction or anything like that; anything can happen, I am merely talking about the evidence we have and working with substance as opposed to headlines and media narratives.
 * I hope feelings aren't riding as high now and that we can all move on. All the best 118.93.49.168 (talk) 21:58, 10 February 2017 (UTC)

Macron's highlighter colour

 * I am amazed that despite consensus being sought and reached that further changes have been made, without taking to the talk page. Where has this new colour come from for Macron? The grey is so pale that it is almost white and you have virtually put things back as they were!
 * I can see why you would want to change Macron's colour from being different to the socialists and I do see an argument for it being grey with him effectively standing as an independent (new self founded party with no other candidates in selection process) but this is just too light and violates the consensus that was reached above.
 * Please alter to a slightly darker shade; not a solid colour but a darker highlighter colour, similar depth of colour to the others. Thanks 2407:7000:875B:C370:ADC7:AE4B:9C37:7821 (talk) 00:57, 12 February 2017 (UTC)
 * Looks fine to me! Jdcooper (talk) 01:00, 12 February 2017 (UTC)
 * Yeah, seems fine. Significantly clearer than the highlight color being used on fr.wikipedia, at least. Mélencron (talk) 01:22, 12 February 2017 (UTC)


 * See above for comparison.
 * Does this not highlight that there is a better option for a similar depth of colour. Some of us do not have great eyesight, so subtle colour differences do make a difference. When I first saw the new colour I thought someone had switched it back to white before I had a closer look.
 * Please see what you think. Please bare in mind that the consensus established was for both 1st and 2nd to be clearly highlighted. 2407:7000:875B:C370:FC42:96FF:242E:1827 (talk) 04:48, 12 February 2017 (UTC)


 * Seriously, the colour is not that lighter. And specially with the new font, if you aren't able to discern that Macron figures are highlighted, then you actually have visual problems.
 * I obviously changed the pinkish shades because it was too similar to PS color, and because pink was (actually) not EM official colour (it actually doesn't have an official colour, but its website uses different shades of grey and white)
 * I tried to use lighter shades for the #2nd placed candidate so both the 1st and the 2nd could be shaded without completely ignoring who the top candidate is. I may add a slightly darker shade for Macron, but the one you propose is just too dark. And sincerely, as I said, with the new font highlighting the bold in such a way, there's no way you can get lost in guessing who the second placed candidate is. Cheers. Impru20 (talk) 09:38, 12 February 2017 (UTC)
 * I agree that getting rid of the pink is sensible, I think grey is the right colour because of the reasons you have said but also because to all intents and purposes Macron is an independent candidate and traditionally independent candidates adopt the colour gray. So yes, I am completely with you on the idea of changing to grey.
 * I still think the grey highlight is a bit to weak. Fair point the colour I picked is probably too bold and I was probably a bit lazy in searching for a colour (I just picked silver). I think somewhere half where between the 2 colours above would be ideal. That is quite literally the faintest shade of grey i have ever seen.
 * Thank you for your constructive approach to finding the best solution. 2407:7000:875B:C370:D803:59D2:1A7D:3D7E (talk) 10:22, 12 February 2017 (UTC)
 * I think the shading can be worked out. The issues with grey, however, is that darker shades tend to look too dark, while lighter shades look nearly white. So one has to find a correct proportion between both darker and lighter shades that fits all uses. I also wanted to avoid using pure grey for Macron (it has a small pinkish hue), since his candidacy is not actually that of an "independent", but is built around his "En Marche!" platform which, arguably, also intends to contest the legislative elections.
 * In any case, I believe the changes I made should be satisfying to all by trying to integrate all views. Bold is now greatly highlighted, and the two-candidate shading is maintained, but darker shades are now used for the top candidate so it is highlighted too. Obviously this is in a testing stage, but if this goes uncontested and is widely accepted I think this may become a satisfying solution for everyone, respecting both the consensus reached here and the minor views. Impru20 (talk) 10:35, 12 February 2017 (UTC)
 * I see the reason that pink isn't an ideal color due to En Marche! seeming to prefer blacks and grays, but I think pink is better than the pinkish-gray that is currently settled on. If gray is going to be used as a shading, it needs to be dark enough to be clearly seen against the white of the other cells. I think we should either use a darker gray or go back to pink. Yes it's not ideal, but since Macron is a former PS minister and for all intents and purposes represents the Left in the current three-way race, I think it's appropriate.  Nevermore27  (talk) 20:53, 15 February 2017 (UTC)
 * The problem with pink is that it's the color traditionally referring to the left – and in France, to the miscellaneous left (i.e., left-wing candidates unaffiliated with a major party or the PS). Mélencron (talk) 21:12, 15 February 2017 (UTC)
 * Pink would still give issues due to the comparison with PS. Given that you voted for a system that allows for two candidates to be shaded at once, it wouldn't be impossible for both Macron and Hamon to get to the top. In that case, both would be shaded pink (yeah, doesn't seem likely right now, but such an eventuality must not be ruled out given how many "unlikely" things have happened as of recently to other candidates). Macron was a PS member, but so was Melenchon at some point and he doesn't uses pink. Light gray is the closest to one of En Marche's website "colours" (it also uses light blue at times, but that would then enter into conflict with Fillon's colours). And dark gray is just horrible: both for contrast with other candidates' shadings, as well as with text itself (which is black).
 * Also, I think that the issue was that in the initial version of the table, the second most voted candidate wasn't visible enough for some people, and shading was chosen as a way to reinforce the second position. Then, bolded was also maintained. Then I introduced an alternative font which greatly reinforces bold text. And finally, I darkened Macron's shading so that it's both kept light as second candidate, but it's also visible enough to clearly distinguish him from others. The consensus to shade the top two candidates was on the basis of the issue that the numbers were "hard to read". Where's the difficulty now? Impru20 (talk) 21:23, 15 February 2017 (UTC)
 * I agree, I think it looks fine now, and I don't really see how it matters that much what colour it is. Jdcooper (talk) 21:34, 15 February 2017 (UTC)
 * If your response to everything I say is going to be sneering, then fine, I don't want to argue with you.  Nevermore27  (talk) 21:41, 15 February 2017 (UTC)

Dates
Can we have the dates in a standard anglosphere format, such as 9 February 2017?--Metallurgist (talk) 01:02, 9 February 2017 (UTC)

Jacques Cheminade
Is there any use including someone who's been polling at 0% continuously this month? Only one pollster is actually tracking his support. Maswimelleu (talk) 19:20, 15 February 2017 (UTC)
 * He's still included in opinion polls, and his presence in the table doesn't cause any problem as far as I know. What's the issue? Impru20 (talk) 21:26, 15 February 2017 (UTC)
 * Just seems irrelevant. More clutter in the table that people have to sift through when they're scanning it for useful information. Maswimelleu (talk) 09:51, 20 February 2017 (UTC)

Also makes the table very hard to read in on a mobile device B A Thuriaux (talk) 11:20, 27 March 2017 (UTC)

Second Round Table Ordering
The fact that the tables for the second round are ordered from oldest to newest is incredibly unintuitive and confusing, especially considering the first round is ordered newest to oldest. I'm not sure if there's an easy way to reverse the lists, but if I don't find a good reason for why the tables are that way I'll probably just do it all manually to make the lists more intuitive. Net User (talk) 23:18, 19 February 2017 (UTC)
 * Reverse it manually. I agree that the tables should be in the same order. Maswimelleu (talk) 09:52, 20 February 2017 (UTC)

Could someone fix this, please? Boscaswell  talk  10:02, 20 February 2017 (UTC)
 * I'll do it later if someone else hasn't sorted it. Maswimelleu (talk) 10:20, 20 February 2017 (UTC)

Should we add a graph of opinion polls?
I've made a graph of the 7-day rolling average of opinion polls since the start of January. Link to the graph and spreadsheet. Should I upload this graph to Wikipedia and include it in the article? If so, I'm not sure what to do about Bayrou, as he's only included in some polls. Should he be shown on the graph or not? Chessrat ( talk, contributions ) 05:20, 21 February 2017 (UTC)
 * I've got a sheet and graph myself, but I've chosen to weight polls by pollsters only once within any given period, so as to ensure that the rolling polls flood the rest out. Mélencron (talk) 06:13, 21 February 2017 (UTC)
 * Regarding Bayrou (and Cheminade) I suppose there needs to be a fainter or dashed line indicating their support. Or just make a distinct graph for them. If they're combined then you'll need to have some kind of disclaimer to indicate why the polling figures add up to more than 100%. Maswimelleu (talk) 10:05, 21 February 2017 (UTC)
 * The way I deal with Bayrou is simply to create two separate graphs; one for polls with him and one for polls without. I just include Cheminade in an "other" column. Mélencron (talk) 15:21, 21 February 2017 (UTC)
 * I'll get two versions uploaded to Commons a few hours from now, one with Bayrou and one without. Usual method I use for rolling averages, with two-week dropoffs for pollsters and only using the most recent poll by any pollster so as to ensure that the rolling polls don't drown out other data. Mélencron (talk) 22:04, 22 February 2017 (UTC)
 * I've added a few (current scenario, scenario with Bayrou, two second-round graphs); I'll see if I can get some polling graphs for Juppé/Sarko done as well. I'll likely just use a gallery to include the three outdated scenarios below the current one. Mélencron (talk) 16:27, 23 February 2017 (UTC)

Thanks for doing that - it makes it much clearer what is going on! B A Thuriaux (talk) 11:21, 27 March 2017 (UTC)

Major events in tables
Would there be any objection to including major occurrences in tables, e.g. primaries and official announcements of non-candidacies in the cases of Hulot, Hollande, and Bayrou? I don't see a serious problem in doing so; the French Wikipedia article already does so, but only in the main tables. Mélencron (talk) 21:21, 22 February 2017 (UTC)


 * Seems useful to readers. Support.  Rami  R  21:36, 22 February 2017 (UTC)
 * The English Wikipedia doesn't tend to use events in opinion polling tables; only UK opinion polls do it, and with strong limitations (nearly everything is left out). The issue would come because: 1. when it comes to specify what "major events" are, 2. what's the purpose of their inclusion in the tables (there must be a strong reasoning for including some events. Either ALL events are included (which is obviously impossible) or you'll have issues of WP:UNDUE because you'd be including some events and leaving others out). For the UK, only elections and changes in national leaderships are included. Impru20 (talk) 21:51, 22 February 2017 (UTC)
 * I think the instances I named (primaries and notable non-candidatures) would be fine, mainly as it explains why candidates who were included in some election polls at one point no longer do so. It'd also make sense to remove the bands from the second-round tables – with the exception of Bayrou's non-candidature/proposed alliance with Macron – if that's the case. On fr.wiki it seems that they also decided to remove the Penelopegate notice from the table because they wished to keep it quite strictly related to events within the campaign. When I was working on the French Socialist Party presidential primary, 2017 article the way I did it was to note 1. Hollande's non-candidacy, 2. the beginning of the official campaign, and 3. the televised debates for first-round polling. In addition to that, I also created a separate table upon the release of the candidate list/certification of nominations – which in the case of the presidential election will occur in March. It shouldn't be hard to figure out what to include, that is, essentially nothing – 2 notes each (top two advancing to the second round/primary winner) for the 3 primaries, so 6 there; add the three non-candidacies and you have 9 "major events" in all at this point. Just use it sparingly and it'll be fine. Mélencron (talk) 22:03, 22 February 2017 (UTC)
 * Ah, that's a further issue: second round polls. You'd have to add events to every one of the possibilities, thus repeating information and increasing the article's size needlessly. Also, when you point out "primaries" and "notable non-candidatures", you forget that the tables are already split so that they do coincide with one of such relevant events (mainly because, as a result of those, pollsters stop asking for specific candidates, who no longer appear in opinion polls). So, you'd end up having events just at the beginning and end of the tables... Impru20 (talk) 22:10, 22 February 2017 (UTC)
 * And a further one would be on the point on listing Fillon and/or Hamon primary wins in run-off tables where one or none of these candidates are listed. Impru20 (talk) 22:28, 22 February 2017 (UTC)
 * I don't think we should do it because a lot of the polls are rolling and have 3 day fieldwork, so the actual effect of any given event is delayed. Maswimelleu (talk) 10:47, 23 February 2017 (UTC)
 * I think we can assume our readers are capable of understanding this.  Rami  R  13:43, 23 February 2017 (UTC)

Color change?
The addition of the rolling average of polls is nice, but the colors make it a bit difficult to read. Perhaps fewer shades of red? 23:49, 23 Feb 2017 (UTC) — Preceding unsigned comment added by M4ry73 (talk • contribs)
 * That's just what happens when the French left is still split four ways and between two Trotskyists, I guess. I could maybe give JLM a darker red and make Hamon pinker, but I'm not sure what else I could do. Mélencron (talk) 00:39, 24 February 2017 (UTC)
 * I would suggest removing the minor candidates and replacing them with an "Other" line, as well as maybe putting the candidate names in the graph itself. Here's a rough image showing what I mean: http://i.imgur.com/6ZiayL3.png Chessrat  ( talk, contributions ) 08:29, 24 February 2017 (UTC)
 * Too nonstandard in my view – we've minor parties and candidates have always been included in polling graphs so long as they're regularly included in polls. I think I've figured out a clearer palette to distinguish between the left-wing candidates. Arthaud very dark red, Poutou less dark, JLM reddish, Hamon pink. Looks much better distinguished on the graph now, imho. I'll update the graphs after today's Ifop-Fiducial rolling poll is published. Mélencron (talk) 15:36, 24 February 2017 (UTC)
 * Looks fairly good. The candidates could still be distinguished a bit better though, in my opinion. I'd suggest using this palette; it has a lighter shade of pink for Hamon to more easily distinguish him from Mélenchon; I also replaced Macron's grey with a yellow. (Grey is being used for Cheminade in the table, and even if Cheminade isn't being included in the graphs I don't think there should be two candidates with slightly different shades of grey). Chessrat  ( talk, contributions ) 03:54, 25 February 2017 (UTC)
 * This looks much better, thanks for addressing this issue. m4ry73 18:59, 27 February 2017 (UTC)