Talk:Statewide opinion polling for the 2020 Democratic Party presidential primaries/Archive 1

Archive 1

Transclusion

Why not transclude the sections from this page onto the individual primary election articles? Onetwothreeip (talk) 22:22, 24 April 2019 (UTC)

@Onetwothreeip:, from Talk:Statewide opinion polling for the 2020 United States presidential election:

The last series of edits was made to implement a workaround for named notes across different states; as the instances of the named notes cannot be properly transcluded on state articles without explicitly providing the associated text with each note, the polls have instead been moved to each respective article with the explicit notes and re-transcluded here, such that the original errors have been eliminated and only one note is associated with each name on this article, as long as both the name and content of each note is identical. (This is a maintenance note for future reference in case other editors encounter issues editing.)

This is a technical fix for a slightly abstruse issue which should have no visible difference on either page other than displaying the transclusion template on this page rather than each state's article (where they already exist). In the case of the other article, it resolved two different problems: before that, it was either a choice between (1) unresolvable note errors or (2) notes with duplicate content in order to solve (1). Mélencron (talk) 00:33, 25 April 2019 (UTC)

It's not about visible differences, surely it makes it harder to edit these pages when they are strewn across many articles, rather than being able to edit them all in the same article. Onetwothreeip (talk) 00:50, 25 April 2019 (UTC)

I don't agree – polls for Senate/gubernatorial/House/statewide row officer elections are all strewn across numerous states and I've made thousands of edits to add polls on those articles over the last few years and never found it an issue. As you're not actively adding polls on any articles, I don't see how this really affects you or readers at all. (The "edit" button is right there in the transclusion headers.) Mélencron (talk) 00:56, 25 April 2019 (UTC)

It does make it harder to be aware of when new polls are added and altered. Onetwothreeip (talk) 03:20, 25 April 2019 (UTC)

That's not my problem; this solves a technical issue with no other workaround in order to avoid littering articles with citation errors over time. In any case, if you just want to follow U.S. polls, everything I add is added here or here, and their interactive interfaces are easier to use than Wikipedia's static one to parse trends, so I recommend you check these out if that's what you're interested in. If you want to be alerted on Wikipedia, then go to Special:EditWatchlist/raw and add this and hit save. (You can just manually watchlist the four current primary/caucus articles; no others have been created yet.) Mélencron (talk) 04:28, 25 April 2019 (UTC)

It's not about following polls, it's about the edits on polling articles, of which additions of polls are included. Onetwothreeip (talk) 05:26, 25 April 2019 (UTC)

what is the point

What is the point of posting that New York poll,it is inaccurate,and doesn't ask about the other candidates in the race for president.Alhanuty (talk) 14:47, 10 June 2019 (UTC)

They're both from New York, which is why Siena (a New York-based pollster which circulates its results in NYS media) specifically asked about them. Mélencron (talk) 15:44, 10 June 2019 (UTC)

Transclusion order

Per my revert, we should really be transcluding sections out from here, rather than transcluding them from 50+ places into here. Onetwothreeip (talk) 01:24, 23 June 2019 (UTC)

As I've already explained, this is to resolve a technical issue and it's not something you have to worry about. Mélencron (talk) 01:33, 23 June 2019 (UTC)

What's the technical issue? Onetwothreeip (talk) 01:39, 23 June 2019 (UTC)

When giving a note the same name, it is impossible to properly transclude on any state article if it's used on multiple states other than the state under which the full note was originally presented. Because of that, the only way to work around it and ensure both the proper transclusion of notes on both the master list and state articles is to explicitly provide both the name and full text for each note on individual state articles such that the content of each note under a single name is identical, and transclude them here. It's not something that you should really care about; it's just that there is no workaround to this other than this. (FYI, if you just want to monitor new polls, Twitter, FiveThirtyEight, and RealClearPolitics are all better resources for this than your Wikipedia watchlist.) I'd also note that the need to transclude state articles into this list is inevitable; the total byte size in 2016 would have been around 1 million bytes on a single page each for both the Democratic and Republican primaries in 2016, which is simply uneditable, which is why the polls were simply transcluded instead (reducing the overall byte size of the master lists to a much more manageable 10,000 bytes of text). Mélencron (talk) 01:47, 23 June 2019 (UTC)

I've told you this before, I'm not using Wikipedia to monitor polling. Transcluding the lists would not decrease the downloading size of the main article, which is a more pressing concern. This supposed technical issue seems to be easily fixed by simply giving the notes different names, especially if the notes are named when they are created on the respective state articles, which would prevent the need for all this. Onetwothreeip (talk) 02:00, 23 June 2019 (UTC)

I don't think you understand the technical issue behind this if you're saying that it can be fixed by "giving the notes different names" – do you know what reference naming does? Notes are functionally identical in that respect; when you define a reference, you are able to assign a single name to identical content multiple times, so you can refer to it elsewhere within the article with the same name and don't create multiple copies of the same reference. The same is true of notes. If you attempt to transclude a note with the same name, it transcludes only as much of the note as is defined – i.e., if you refer to a note in multiple sections but transclude a section in which it is invoked but not the one where it was originally defined, then it produces a reference error. The only way to avoid this is to provide the definitions of the notes the other way around.

Forward transclusion

#	Name	Definition	Initial page	Other pages
1	X defined on initial page	Some text here	X	X
2			X	Reference error
3			X	Reference error

Reverse transclusion

#	Name	Definition	Initial page	Other pages
1	X defined on other pages	Some text here	X	X
2			X	X
3			X	X

It isn't clear to me that you understand the underlying why the underlying technical issue is problematic to doing anything other than this, so I'm going to reiterate that this is not something that is going to be changed since this is the only way this can be achieved, and really not something you should actually care about. Mélencron (talk) 02:14, 23 June 2019 (UTC)

It seems like you're just trying to cut down the size of the note list then, which doesn't seem like a good enough reason to transclude 50 articles into one, rather than one article into 50. However, if we're going to prevent the article getting too long (either editing size or download size), then we shouldn't have one article that has every single state poll for the Democratic primary. The article would naturally be split anyway, regardless of transclusion. Onetwothreeip (talk) 02:22, 23 June 2019 (UTC)

You're right about that, of course – I don't actually agree with the principle of a list consolidating all statewide polls as it's redundant to list them twice (both on a full list and on individual state articles), in addition to the fact that it isn't clear what a consolidated list would do better than listing them on states' respective articles – this isn't done for anything other than presidential election articles in the United States, and pretty much only for one reason (to create a polling map... the 2016 article even uses the phrase "statistical tie" with a wikilink to margin of error, which is... eyeroll-inducing), the principle of which I also disagree with – and which isn't even used here. As this list has now been rendered obsolete, I'll mark this with a G7 speedy deletion request. Mélencron (talk) 21:34, 23 June 2019 (UTC)

It would make more sense to simply split off polling for states as we go along, and more likely to their own articles rather than the state primary articles. That's another reason to transclude sections from this article onto the state primary articles, but honestly I think we should just provide a link to the relevant section of this article on those state primary articles. Onetwothreeip (talk) 02:28, 25 June 2019 (UTC)

Furthermore, why have polling data on the state primary articles at all? They can much more easily be transcluded from this article if necessary, but really we can suffice with just linking to this article on those articles. If we were to remove all the polling data from this article and merely transclude information into it, this would really just be a shell and would not be of much use at all. There are clearly a lot of people who want this article to exist and to contain the relevant information. Onetwothreeip (talk) 23:37, 25 June 2019 (UTC)

You seem really not to understand what's going on – literally nothing is different except for where the wikicode is stored and that the transclusion fixes a bug. You don't ever actually update anything here; this doesn't affect you. This is literally zero difference in the appearance when content is transcluded; it's just that the transclusion of named notes the other way around produces a bug and this way does not. There is zero visual difference, and as you never actually update any of these lists with new polls or anything of the sort, you are not in any way affected by this. Please just drop this: it's a waste of time. I cannot emphasize this enough: there is no visual difference at all whether the lists are transcluded from the articles or the other way around – the one visual difference there is that reference errors result from named references if transcluded the other way around. The only thing this affects is editors who actually update these lists, which is a list, I'm going to point out, that basically consists of only one user. Mélencron (talk) 23:56, 25 June 2019 (UTC)

You haven't shown the bug at all and I've never seen such a bug happen. I'm fully aware that this regards the editing code and not the appearance of the article. I would suggest you stop trying to persuade me that this doesn't affect me since that clearly isn't working, and instead we can work through this constructively. Let's look at this from the start, what notes exactly aren't working properly? It seems like you just really want all the notes at the bottom of the page, rather than after each table. Onetwothreeip (talk) 00:07, 26 June 2019 (UTC)

This is what the error looks like. It's what's produced when you invoke a reference (or note) that isn't defined on a page. If you're wondering why this might be an issue, it's because if you give a reference a name (<ref name="refname">content of reference</ref>), and then invoke it in a separate section which you transclude onto another article (<ref name="refname"/>), then while no error will be produced on the original page, there will be a reference error on the page that it's transcluded on. Do you understand that part?

Now, the reason why it's pertinent here and why I've structured the transclusions on this article this way is to avoid producing such an error. If you include the same content from each section on another article and transclude those back here, then there won't be an error as they're explicitly defined on each page and share the same name.

There is really no workaround to this (which is why I've reverted and repeatedly asked that this not be changed from this way). While it's not a major issue yet, there is already one such case which would produce an error if the transclusion were done the other way around. Ultimately, it won't be so much an issue for the candidates but for partisan/internal polls which are few in number right now but will begin to multiply over time (and it's likely to be an issue with other pollsters testing limited candidate lists across multiple states).

Now, again, there isn't a workaround other than what I've done, but I'll again try my best to emphasize that this isn't something that affects readers (as the display is identical... minus any reference errors) or editors (who simply edit a different page) instead. Mélencron (talk) 00:17, 26 June 2019 (UTC)

We could just as easily define the notes each time. Obviously we just shouldn't name the notes. It doesn't make sense why you would use the same note more than once anyway. This is all more reason why note lists should be for each table rather than all combined into one. Let's be sensible here, what you're doing is clearly not "the only way" to avoid reference errors. It might just be the only way when you restrict yourself from making any other changes to how we write notes. Onetwothreeip (talk) 00:23, 26 June 2019 (UTC)

That's an intriguing idea which I think might actually work? I don't think it would look great, but it would be better for state-by-state navigability, I think. Give me a sec to set it up. Mélencron (talk) 00:26, 26 June 2019 (UTC)

And done. From a navigability perspective I think it's a big plus over the old chonker of a notelist that there used to be, though obviously the state-by-state ones will grow over time (though some might not grow much at all – there were barely any Nevada caucus polls in 2016 either). I think this should also resolve the note naming issues, though they can still be named individually to avoid duplication within each state if needed. Mélencron (talk) 00:30, 26 June 2019 (UTC)

The incidence of notes being exactly repeated is sure to be a low number, and we would probably be splitting the tables off from this article anyway if the same notes were appearing more than a few times, given that it means the exact same results for the same low-polling candidates in the same state. Onetwothreeip (talk) 00:37, 26 June 2019 (UTC)

Strictly speaking, that's pretty mathematically unlikely! Try flipping a coin 18 times repeatedly and seeing if you'd ever get the same result. My concern was with notes for internal/partisan polls, but that's fixed by simply keeping the notes under every state. Mélencron (talk) 00:39, 26 June 2019 (UTC)

Maine and Mississippi

why where their polls removedAlhanuty (talk) 19:57, 14 July 2019 (UTC)

How should we be organizing states

Currently it appears we are organizing them by the date of the primary. If we keep this, might I suggest adding the scheduled date of the primaries in parenthesis after the name of each state in the section headers. Example: Alabama primary (March 3). However, I would think it would make the most sense to instead order states alphabetically. People do not have the order of the primaries memorized, and it makes it difficult to scroll through the states to find the one they are looking for. However, we can assume all readers are going to be familiar with alphabetical order, which would make it easier to navigate. Any thoughts??? SecretName101 (talk) 21:45, 18 July 2019 (UTC)

New chart for polling in early primary states

I added a chart depicting the polling numbers of the candidates in the four early primary states as discussed there. Xenagoras (talk) 23:11, 21 September 2019 (UTC)

Hawaii Poll

Should we add the Hawaii Poll from Hawaii 2nd Congressional District,considering that it cover nearly all of Hawaii.Alhanuty (talk) 23:16, 1 October 2019 (UTC)

The 2nd Congressional District covers about 51% of Hawaii's population. It doesn't make sense to add it. WittyRecluse (talk) 02:47, 2 October 2019 (UTC)

Change Research Sep 29-Oct 2 poll

Change Research deleted all traces of the Sep 29-Oct 2 poll from their website[1]. Adding it to Wikipedia therefore causes a WP:FAILEDVERIFICATION. Xenagoras (talk) 20:09, 9 October 2019 (UTC)

Missing polls

some of the polls that have been put in the article are missing,what is happeningAlhanuty (talk) 19:22, 2 November 2019 (UTC)

Removal of Bloomberg from Chart

Just have a question about Bloomberg related to his removal from the chart. Basically I saw JFG remove Bloomberg from the chart for the listed reason of Bloomberg isn't in any early state, which I understand. I did have a thought though, if Bloomberg was polling high enough, perhaps he could earn a significant amount of write in votes, which would allow for a case to be made for his addition to the chart. Obviously for this to occur, Bloomberg would have to be eligible for write in votes, but I can't figure out who is eligible for write in votes, much less if Bloomberg is eligible. So I guess the question I have is, how is eligibility for write in votes determined, and do we know if Bloomberg made this qualification? As a quide aside, I am aware this the polling page, so if neccessary I will move this to the main 2020 Democratic Party presidential primaries page. WittyRecluse (talk) 10:35, 30 December 2019 (UTC)

Normally, I'd say that write-ins won't make a large enough difference for a write-in candidate to be included on the graph. However, in this case, the first two primaries' aggregate polls are within the margin of error, so I think he could — perhaps — be included. --Comment by Selfie City (talk about my contributions) 14:22, 30 December 2019 (UTC)

Sounds speculative to me. It's clearer to keep him out. Perhaps add a footnote to explain why he isn't there? — JFG ^talk 08:33, 1 January 2020 (UTC)

Updating the New Hampshire section

I don't really know how to make the tables, but I think it's about time that the New Hampshire table gets split into two separate sections, one for 2019 and one for 2020 - if someone would be able to do that, I would greatly appreciate it!Fjantelov (talk) 20:41, 21 January 2020 (UTC)

That section is transcluded from 2020 New Hampshire Democratic primary, so any changes made to the table will have to be made there. WittyRecluse (talk) 06:41, 22 January 2020 (UTC)

Alright great, I've just made the appropriate changes.Fjantelov (talk) 12:03, 22 January 2020 (UTC)

Replace chart?

The polling chart for the first 4 states is non-standard and rather illegible. Could a volunteer replace it with a similar chart to what we use for the nationwide polling? — JFG ^talk 06:23, 16 December 2019 (UTC)

I thought we were trying to update the nationwide graph to look more like the one fom 2016? Shouldn't we wait until we garner a consensus on the graph there before we start updating the graph here? WittyRecluse (talk) 14:38, 16 December 2019 (UTC)

I updated the data per December 15 and restored the previous style. The newly styled chart had error in its data that I could not fix. The line chart also makes recognizing a candidates' line very difficult. Xenagoras (talk) 02:04, 17 December 2019 (UTC)

@JFG, WittyRecluse, and Xenagoras: I agree with JFG that the nationwide polling format is much easier to read and user's can quickly glance at it to understand how candidates are ranked along with how they changed week to week. Right now I have to hold my cursor over each candidate to even tell if they have gone up or down in an individual week and trying to figure out how they've changed over time takes a long time. Comparing them to other candidates is that whole process over again. I don't see how the current version is better at all, but I'm interested in other opinions. Xenagoras, what's the error in the data? I might be able to fix it. - Leftist Commentary (talk) 02:39, 17 December 2019 (UTC)

In light of comments by Xenagoras, I have made several improvements to the line graph format, and used the same colors as in the nationwide polling. Also fixed data errors and formatted the markup for ease of editing. Let me know what you think. — JFG ^talk 05:10, 17 December 2019 (UTC)

I have also transcluded the early states polling graph into the main primary page. Easier for readers to compare with national polling there. — JFG ^talk 15:39, 17 December 2019 (UTC)

The new graph is wrong, however — Steyer is not doing nearly as well as implied by the graph. Polling data from the first four primaries shows Steyer with an average of 3-4% across all four, while the polling graph shows him closer to 10%. --Comment by Selfie City (talk about my contributions) 00:06, 22 December 2019 (UTC)

@JFG: What is wrong with Steyer's line on the graph? Has someone made a mistake, or is he actually at 10%? --Comment by Selfie City (talk about my contributions) 14:17, 22 December 2019 (UTC)

@SelfieCity: There is nothing wrong now. The data is from Morning Consult's poll. I corrected data for December 16 where some candidates had wrong data. You can look up the data for yourself, it is linked in the text below the header. Xenagoras (talk) 23:00, 28 December 2019 (UTC)

@Xenegoras: Tom Steyer has been above 5% since November according to our graph, but the archive posted in the link does not have Steyer above 3% ever. WittyRecluse (talk) 08:54, 29 December 2019 (UTC)

@WittyRecluse:Try again.[2] PDF Xenagoras (talk) 12:31, 29 December 2019 (UTC)

((ping|Xenegoras}} That is very strange. The internet archive doesn't have it at 10% but the pdf certainly does. WittyRecluse (talk) 23:36, 29 December 2019 (UTC)

Yes, how does this graph make sense? Aggregators suggest that Buttigieg is in first or near it in Iowa and New Hampshire, and still ahead of Steyer in South Carolina and Nevada, yet Steyer somehow edges him out in the average. Looking for the data on the linked website (Morning Consult), the data is there; however, that section has been taken offline on the live website. Since the data is inconsistent with what we present on the rest of the page, perhaps the graph should be taken down and replaced with individual graphs for each early primary state in the same way that the nationwide polling was done? Cookieo131 (talk) 23:48, 23 December 2019 (UTC)

@Cookieo131: The data on the source website is updated weekly. For June 30 and December 23 there was no early state data but the weeks after Dec 23 will have that data again. Old data is in the web archive which is also linked. If we wanted to replace the Morning Consult data, it should be by the average of the 270ToWin aggregates from the four states because they use the biggest sample of polls (although they update their data slower than RealClearPolitics). It would create a problem though: There has been a severe "drought" of polls in early primary states especially since November while Morning Consult has been polling weekly since February 2019. (And the DNC has not endorsed any early primary state poll since November 17.) Xenagoras (talk) 23:00, 28 December 2019 (UTC)

@JFG: Thank you for fixing errors and transcluding the chart. The line chart is much more difficult to read because the lines are so close to each other, which was my reason to chose a stacked bar chart in the first place. The data for a certain candidate on a certain week cannot be determined by the reader anymore. My previous stacked bar chart allowed for that (by hovering the mouse over any bar). Xenagoras (talk) 23:00, 28 December 2019 (UTC)

@Leftist Commentary:, June 30 and December 23 must not have data in the graph. Ignoring this caused the time axis to be wrong. Additionally, the "interpolation = basis" graph property caused a strong smoothing of the data and displayed lines a full percent away from the data points. I changed this property to "interpolation = cardinal" to remove smoothing. Now all lines go exactly through the data points. Xenagoras (talk) 23:00, 28 December 2019 (UTC)

Thanks for fixing the data. I have restored the "basis" interpolation method because it creates adequate smoothing of the lines. Going exactly through all data points is in my opinion less important than showing the trends clearly. Our source for this graph only publishes data rounded to the nearest percentage point anyway, and error margins in polls are several points wide. — JFG ^talk 12:26, 29 December 2019 (UTC)

Steyer's value in the graph is now somewhere between 15 and 17 %, in front of Buttigieg and Warren while in all other statewide polls and in the Morning Consult national poll, he is somewhere near 3-5. How can that be? Stefanbw (talk) 21:22, 29 January 2020 (UTC)

Steyer line on graph (again)

I know this was brought up before, but it continues to bother me. Steyer's line is too high and last time I was told that it was becuase that the Morning Consult source lists it as such. However, I have pulled up a bunch of the archived pages, two I've listed in February and in January and neither has Steyer near the line on the graph. Can anyone verify this or tell me what im still missing? WittyRecluse (talk) 08:01, 12 February 2020 (UTC)

RfC participation invitation

? : Please participate in the RfC about change proposal for infobox for caucus results. Xenagoras (talk) 21:29, 13 February 2020 (UTC)

What do you guys think about including the actual popular vote percentages recorded in each primary just below the average from the poll aggregators, instead of grouped with all the polls?

Also, maybe include the deviation in percentage points of the actual result from the polls? It would make reading this article after the primaries more interesting to see how close the polls were compared to the actual result. Vandergay (talk) 20:30, 15 February 2020 (UTC)