Out of the polling in the article, there still aren't any polls posted that polled respondants over RFK Jr. or Asa Hutchison, maybe Vivek and Elder but I'll have to double check.
If there are polls with these candidates out there already and can be added, can they?
ALSO - PLS CAN HEAD-TO-HEAD OF BIDEN vs. HALEY BE UPDATED? THERE'S PLENTY OF POLLS TO ADD BUT FOR SOME REASON ONLY BIDEN vs. TRUMP HAS BEEN UPDATED RECENTLY. THIS MAKES IT HARD FOR ANYONE TO VERIFY OR OTHERWISE, THE CLAIMS OF A CANDIDATE TO BE MORE LIKELY TO BEAT JOE BIDEN IN NOVEMBER AS THEY CAN'T COMPARE THE TWO FOR THE SAME POLLING PERIOD. THANK YOU!
RCP has not updated their 4-way race polls since November. Their 3-way race lists only one March poll, where The Hill has 20.[1] In addition, the 2-way and 5-way charts (which are being updated) link to one another but not to the 3 and 4 way races. This implies RCP has abandoned maintenance on these pages and we should not be using them. GreatCaesarsGhost13:43, 1 April 2024 (UTC)
Polling by Demographic
I was just thinking if they have polling based on Demographics we should add them to this page, just like they had polling based on Demographics in the 2016 election.Muaza Husni (talk) 10:31, 13 January 2024 (UTC)
???? Can somebody start posting those polling numbers on the page.Muaza Husni (talk) 08:01, 7 April 2024 (UTC)
So if you go back to the 2020 pages, the graphs are all currently broken because of a bug: https://phabricator.wikimedia.org/T334940. Apparently there is a fix being worked on right now but there seems to be a stopgap measure on the republican primary page of occasionally updating an image. I imagine that once the situation gets clearer we will know how to progress with polls but my understanding is that temporarily graphs are in limbo right now. Przemysl15 (talk) 21:58, 29 November 2023 (UTC)
Would it make sense to make a time series figure that don't rely on any sensitive machinery? The article is quite useless without the figure IMHO. 82.147.226.185 (talk) 08:57, 7 March 2024 (UTC)
@Niokog as others have said, the inbuilt wiki graphs are broken. However, a lot of other election pages use static local regression graphs - do you think that would be good? Quinby (talk) 09:04, 20 March 2024 (UTC)
@Quinnnnnby I think if the inbuilt graphs are broken, then a static local regression graph sounds like the next best thing, as it would complement the table just as well. If you can make one, go ahead. :) Niokog (talk) 09:57, 20 March 2024 (UTC)
How's this? Local regression of polling conducted up to the 2024 United States Presidential election. Excludes 'Others' and 'Undecided'Quinby (talk) 11:33, 20 March 2024 (UTC)
It looks good. Perhaps it would be even better to add others/undecided as an additional line (e.g. gray). Niokog (talk) 10:39, 21 March 2024 (UTC)
@Niokog I considered it, but the problem is that undecided and other are merged, while in most countries they are not. I would be happy to put other on a line, but when some pollsters remove the undecideds and others dont, it throws off the graph and the lines. It would mean some pollsters could throw off the average and nothing would be standardised. Unless there's a way to separate out the others and undecideds, I don't see how it would be possible. Quinby (talk) 10:54, 21 March 2024 (UTC)
Sounds great - and really glad to see a chart, it would be very useful for everyone.
I'd suggest you put this at the top - common in polling for other countries' elections - and put the polling aggregator sections below the main list of national polls. BarrySpinno (talk) 15:14, 31 March 2024 (UTC)
Aren't there any hypothetical electoral college results based on opinion polls yet? I don't see where we have these covered in Wikipedia.
The overall percentages of support only have a loose statistical relation to the likely electoral college results. For example, if Californian support for Biden jumped up by 20% (and nothing else changed), then the US-wide support for Biden would jump up by about 2%, but the electoral college result would be unaffected. Boud (talk) 17:10, 1 June 2024 (UTC)
Thanks, that's exactly what I was looking for. I clarified the wording to make it clearer and easier to find. Boud (talk) 19:27, 2 June 2024 (UTC)
Michelle Obama
Hello.
I found the following highly relevant information, and would greatly appreciate if it can be incorporated into the article in an appropriately structured manner. Thanks in advance for any help.
According to a Reuters/Ipsos public opinion poll among 892 registered voters released on July 2, 2024, Michelle Obama was the only listed Democrat option who would defeat Trump in a confrontation, with 50% of the votes for Obama versus 39% for Trump. 55% of the voters also had a favourable view of Obama versus 42% toward Trump.[1][2]
I think the polls including all the candidates should be listed before those which give a false impression of the race by failing to include some of those who will be on voters' ballots. Starchild (talk) 02:09, 22 July 2024 (UTC)
Who Funds Polls?
Should this article identify who commissioned and paid for each poll? Sometimes this information is related to the results (Democratic-funded polls are statistically more likely to show favorable results for Democrats and Republican-funded polls more likely to show favorable reports for Republicans), as well as to which candidates get included in the polls and which candidates do not. — Preceding unsigned comment added by StarchildSF (talk • contribs) 02:35, 22 July 2024 (UTC)
I'm not sure why we ever had one of the many hypothetical matchups listed among the actual contest. But Trump and Biden are now the presumptive nominees and Harris is not running. GreatCaesarsGhost20:22, 24 April 2024 (UTC)
I vote that we remove the Harris-Trump polling table. It's purely hypothetical at this point and the sources used don't appear to be actively updating those match-ups anymore.
While we're on the subject, is there any value to keeping the Biden-Trump-Kennedy-West table separate from the B-T-K-W-Stein table? It looks like Race to the WH uses the same numbers for both match-ups now under "National + 3rd Party" with Stein included in the line graph, but not the bar graph. Woko Sapien (talk) 16:18, 25 April 2024 (UTC)
I concur, with the obvious caveat that we can revisit this subject as aggregators change their behavior, or new aggregators emerge. GreatCaesarsGhost17:31, 25 April 2024 (UTC)
The polls with Biden are a sizable chunk of the article, and will end up needing to in some way be separated from those of his ultimate replacement for the Dem party nominee.
I propose spinning off the polls for matchups with Biden into its own article. Or perhaps spin-off all polls conducted prior to July 22 into a separate article, as the dynamic of the race has shifted so drastically now. SecretName101 (talk) 19:58, 22 July 2024 (UTC)
In the margin column on the overall average row for each matchup table, are we averaging the margins in the above rows, or taking the margin between the averages of the leading candidates? 103.8.18.128 (talk) 22:40, 5 August 2024 (UTC)
Adding VP selection note on polling chart
Should we add the date/selection of Vance/Walz as their party's respective VP nominees to the polling chart? A note like that one that currently informs readers of the date/when Biden dropped out and Harris announced her campaign? Sometimes VP pick announcements have an impact on polls (which we could see in the fallout of Vance's cat lady comments). I think it would be an interesting note to add so we could see the change in polling trends post-VP announcement. TDKR Chicago 101 (talk) 21:29, 6 August 2024 (UTC)
I really dislike this change [2], but want to solicit input rather than reverting. The aggregators names used to go directly to the source; now they go to their WP articles (or often redlinks). I understand why this may be preferable to some and more in line with other WP articles, but given the purpose and nature of this article, this is a big downgrade to me. GreatCaesarsGhost12:01, 8 August 2024 (UTC)
I dislike it too, though I get what they were trying to do (follow the WP:EL guideline). I began creating redirects to deal with some redlinks without the need to edit each article (there are so many edits just like this one). I quickly found that this would still take too much of my time. I haven't completely gone through my watchlist, but from what I have seen, it looks like this editor may have refined their method at some point to reduce the amount of redlinks. I don't know what the best solution is. --Spiffy sperry (talk) 14:39, 8 August 2024 (UTC)
Revert the edit. I think the intention was good, but the results are an absolute mess – just a sea of redlinks and disambiguation page links. Things were perfectly fine the way they were; this was a solution in search of problem. Woko Sapien (talk) 16:35, 8 August 2024 (UTC)
Yes, it's jarring, and especially inconsistent when some still link directly to the poll. I think there's a case to be made for an exception when a piece of information refers to a specific poll result. What has been done here might be better achieved by including a reference to a list of pollsters. 103.8.18.128 (talk) 21:08, 8 August 2024 (UTC)
My apologies, but I did not get notified that it was reverted until this message. As noted, I was trying to get pages to follow WP:EL, in part to prevent future problems of links dying and then being repaired and ballooning the columns when {{webarchive}} gets added, as I had seen on some pages (particularly with many of the older zogby.com pages now being dead links). Recent polling tables appear to have their own format of using inline external links instead of references that are sometimes in their own column; I perhaps wrongly assumed it was an oversight compounded by editors then copying the most recent examples (Nationwide opinion polling for the 2008 United States presidential election, for example, was tagged for not following policy and I realized the issue extended well beyond that page). The external links don't always go to the polls, and instead a news story that mentions the polls, so they're not transparent, especially when the news story doesn't explain who the polling source beyond an abbreviation; I feel that internal links make the information more useful for people who don't know these polling institutions. I tried to dab or use the correct links where possible, but WP:AWB does not clearly show which are dabs or redlinks (which I don't personally find jarring and instead indicative of potential pages that should be created or unlinked if they truly are not notable), and I've built out the replacements individually as I realized the problem. I originally did not fully grasp the scope of the issues and I'm sorry for the extra cleanup that others performed before I addressed them; I also had missed links with preceding spaces and the first pass only caught the first slashes between names, which led to the noted inconsistencies. I was trying to go back and resolve these, which is why there were additional edits. I can build some more find/replace to remove specific redlinks if that is desired. —Ost (talk) 15:08, 14 August 2024 (UTC)
Including polling since 2020... why?
Why do we include polling as far back as 2021? No one new that it would be Kamala v. Trump until like a month ago. I'm not sure 2021 polls really tells us anything about the 2024 election. Particularly given there appears to be a large span of time before Kamala became the presumptive nominee where we have little to no polls. (Courtesy ping: @Quinnnnnby). Endwise (talk) 12:41, 15 August 2024 (UTC)
The article isn't about Kamala vs. Trump, but rather it's a list of all opinion polls for the 2024 election. It focuses on the presumptive/declared/confirmed nominees as they are most relevant to the topic, but you'll see that it also includes matchups such as Biden vs. DeSantis, or Buttigieg vs. Trump. 103.8.18.128 (talk) 22:34, 15 August 2024 (UTC)
Weird outlier poll in first diagram (but not in tables)?
According to the first diagram, one poll of ~August 15-20 has Harris at ~48% and Trump at ~52%. I cannot find anything like this in the table yet, and it is a major outlier: two contemporary polls - probably ActiVote and Outward - have the reverse, and Trump is consistently 50% or less even among the most GOP-biased pollsters. Quinnnnnby, can you clarify what happened here? 2A02:8071:5BD0:D4C0:0:0:0:A265 (talk) 02:36, 18 August 2024 (UTC)
Hello, I'm aware it is an outlier but I cannot decide myself whether to exclude them just for being an outlier. All polls are normalised (with others/undecideds removed), which is why you cant find the exact numbers. Regards, Quinby (talk) 16:48, 18 August 2024 (UTC)
It normalises all polls, so a 49/45 becomes roughly a 52/48. In all polls, 1 candidate will be above 50% because others and undecideds are removed. There is another graph for polls with other candidates. Quinby (talk) 08:59, 19 August 2024 (UTC)
Candidate Missing
Why isn't Libertarian candidate Chase Oliver listed anywhere in this article? Are all the polls really failing to include him, or is the article missing the polls that do? Perhaps there should be an introductory paragraph talking about who's actually running, versus who gets included in the polls, and why (e.g. campaigns with lots of money paying for the polls). — Preceding unsigned comment added by StarchildSF (talk • contribs) 01:06, 22 July 2024 (UTC)
Here is one poll I found without too much trouble which has a question that does include Chase Oliver, along with fellow independent candidates RFK Jr., Jill Stein, and Cornel West:
I agree that Oliver should be included in the table. Polls have started including him lately, and he'll be on the ballot in 40 states so I think nationwide polling on him is relevant. Benintogo (talk) 22:31, 22 August 2024 (UTC)
Improvements needed to second loess graph
The time axis range should be shortened to make data more visible:start from mid 2021 (there's no datapoints before that) and end within 2024 (no need to extend to 2025). 176.200.207.225 (talk) 09:05, 27 August 2024 (UTC)
It is useful context to see the entire election to election timeline, and also the 'extension' to 2025 is automatic because the 2024 election is so late in the year. The lines represent the elections. All other graphs follow these precedents. Quinby (talk) 14:46, 27 August 2024 (UTC)
What events do we include?
Major events noted in the polls are ones that explain odd shifts in the polls, like when Joe Biden's withdrawal led to a surge in Harris's polling. In that case, why are we including the DNC, RNC, and Kennedy's suspension? None of these events have changed polling numbers between Harris and Trump, and we didn't include the DNC/RNC in past elections. Realsmartcraft (talk) 01:34, 27 August 2024 (UTC)
I'm not personally a fan of including events on this page, as it can veer too far into providing commentary on polling trends. There's better pages for that information such as 2024 United States presidential election. I feel notable events on this page should maybe be limited to events that impacted which candidates are represented in the polling. 103.8.18.128 (talk) 02:39, 27 August 2024 (UTC)
For some reason, the page doesn't show the bar for when RFK Jr. dropped out, even if I look at changes made to the article prior to when this presumed change would've been made. ThrowawayEpic1000 (talk) 21:50, 2 September 2024 (UTC)
With RFK Jr. remaining on the ballot in many states (including Michigan and Wisconsin), should aggregates that include him be moved out of "hypothetical polls"? There are still voters that are going to vote for him, and his presence on the ballot can still affect the election by acting as spoiler. Additionally, many aggregates still include him in their 3rd party aggregates. CountyCountry (talk) 16:33, 14 September 2024 (UTC)
In the list of poll results in there are references to the sample size with RV/LV/other qualifiers. I wonder if something could be added to explain what those mean? Flibble (talk) 08:32, 18 September 2024 (UTC)
Intro Vandalism?
Surely the latest revision to the intro ("According to data from...") is vandalism or, at best, unsourced. (I have no idea how to refer to the edit more precisely; WP is a mystery to me.) — Preceding unsigned comment added by 202.128.114.90 (talk) 03:42, 19 September 2024 (UTC)
Polling aggregate averages
For most of this page's editing history, there seems to have been general consistency in terms of taking "Average" in the polling aggregation tables to mean the arithmetic mean, since, that is already the most typical meaning for the word "average" in a mathematical context. This also holds consistent on all of he other nationwide and statewide US election polling pages that include averages of aggregate polls. Despite the consistency, I do not believe that the matter of how averages ought to be being calculated has been directly discussed anywhere. I am bringing this up because recently (starting from the the 13th) there has been a user, 159.250.17.61, who has been editing the page using mid-range for the average values, rather than arithmetic mean. I therefore think it worthwhile to bring this topic to the talk page, to gauge thoughts in terms of how the averages should be being calculated. – Jamie Eilat (talk) 13:32, 16 August 2024 (UTC)
I don't think averages should be displayed at all. The aggregators value is already an average (sometimes weighted). Displaying multiple aggregators gives the reader the understanding that there is different ways to interpret the data. A simple mean of those values adds nothing, and implies the achievement of some greater truth. Further, the fact that we are averaging such a small set of data (often two values) decreases value added. GreatCaesarsGhost01:59, 18 August 2024 (UTC)
You just said two contradictory things. If it is not implying a greater truth, how is it educating readers? If I just tell you that RCP's average is 45. and The Hill's average is 49, you might be curious about the machinations that caused two observers to come to different conclusions using the same data. You might even click through and learn that RCP used a RV model when an LV was available. But an average obscures both the variance in the results and the source of those variances (i.e. the aggregators house style). GreatCaesarsGhost14:57, 18 August 2024 (UTC)
You can average averages, but generally you only want to do this if your data sets are relatively consistent. With these aggregators, however, they don't all draw their numbers from the same set of polls, so you may wind up with a table average that includes one poll 5 times, and another poll only twice. So as the user above said, using table averages draws a conclusion that isn't necessarily up to us to decide.
But my main argument against including column averages is the consistent inability for users to calculate it correctly. So I'd be in favour of removing them. 121.98.231.4 (talk) 22:45, 25 August 2024 (UTC)
That user is still reverting edits several times a day to change the numbers to use mid-range. They're actually doing it on a multitude of articles related to US election polling. 103.8.18.128 (talk) 21:20, 19 September 2024 (UTC)
Extended confirmed protection until election day (5 november 00:00 UTC)
This is because people can easily edit the article to show inflated support for either candidate, or even an independent one, Sale1925 (talk) 07:12, 18 September 2024 (UTC)
The first one is mentioned on the chart, but not the second one. Those things can widely affect public opinion. I think it should be restored. Bkatcher (talk) 12:18, 23 September 2024 (UTC)
It's not really our place to speculate on which events may have influenced public opinion. This page used to have various notable political events, but they were removed as the convention across many such pages is to only include events that affect which candidates are competing (campaign launches, suspensions, deaths, etc). I don't think assassination attempts should be included, nor should debates. There are better places to draw attention to such topics, such as the 2024 United States presidential election article. 103.8.18.128 (talk) 01:21, 26 September 2024 (UTC)
Hi all, I was just updating the aggregation graph and had not noticed the addition of the PoSSUM polls. These seem to be samples of Twitter users (see here for crosstabs), and then AI is used to guess their demographics for weighting. I don't see why this should be included, especially as I cannot find that any polling aggregator publishes their polls. I would welcome some other opinions before deletion. Regards, 11:46, 9 October 2024 (UTC) Quinby (talk) 11:46, 9 October 2024 (UTC)
RCP included or not?
I see Real Clear Politics removed in the polling aggregation. Then added again, then removed, now added.
But if RCP's numbers are included, Harris is not 2.8% ahead as indicated, but 2.5%.
I understand that RCP can be considered untrustworthy, but if it's included, get the numbers right! 178.197.223.108 (talk) 15:55, 15 October 2024 (UTC)
I personally support exclusion, because RCP includes Republican-leaning pollsters like Rasmussen and Trafalgar; the same applies if aggregates include Democratic-leaning pollsters. The other aggregates have better criteria for the polls they include--i.e. nonpartisan pollsters like Emerson college, Quinnipiac, Yougov, NYT/Siena, ActiVote, etc. JohnAdams1800 (talk) 00:23, 16 October 2024 (UTC)
538, Decision Desk HQ, Silver Bulletin, and 270toWin also include right wing pollsters. Aggregates including them is a common occurrence. CountyCountry (talk) 00:34, 16 October 2024 (UTC)
See their Georgia aggregate [5]. I see Insider Advantage, Fabrizio, and Trafalgar. Pollsters typically known as right-wing. See Arizona as well. [6] You can take a look at other states. CountyCountry (talk) 04:11, 17 October 2024 (UTC)
The removal of RCP causes a loss of credibility for Wikipedia
Despite the fact that some editors here seem to think they have the right to hide a major polling aggregator from the general public, this significantly takes away credibility from Wikipedia. Whether you personally think RCP is reliable or not, is completely irrelevant to the content of this article. RCP is a major polling aggregator the same as all the others. Each polling aggregator uses its own formula and has its own bias. The idea of an average and the idea of showing all the aggregators is to allow the user to see the overall picture of the election and make his own conclusions. This is pure censorship by editors with personal biases who believe themselves to have the right to hide information and present their own personal point of view as well as to influence the United States election.
It is situations like this that take away all credibility from this website and the reason why Wikipedia cannot be used as any serious or reliable source. 207.162.26.51 (talk) 18:25, 19 October 2024 (UTC)
Wikipedia articles are not required to give equal weight and consideration to all sources, and they are not designed to present all viewpoints and encourage the reader to form their own opinion, but rather to try and present an objective picture of reality.
Without knowing the results of the election yet, the closest we can get is to exclude sources that appear to be trying to push a particular narrative rather than a good-faith effort at predicting the election. If, after the election, RCP turns out to have been very accurate I would support their retroactive inclusion.
It doesn't? Last edit was the 21st. The file's description even says: This graph will be updated at least weekly, but I hope to update every couple days. --Super Goku V (talk) 10:03, 27 October 2024 (UTC)
Are you conflating a trend in polling with bias? Regardless, the issue is with staleness, not bias. Race to the WH appears to have stopped updating its section on head-to-head polling, in favor polling with 3rd party candidates, which could explain why it was so far from the other aggregates (a Dem. lead nearly double that of all the others). You will see that they are still included in the table here with 3rd party candidates. --Spiffy sperry (talk) 20:54, 28 October 2024 (UTC)
The figures are correct. What is happening involves this text, †The average includes all polls within seven days of the date of the most recent poll, limited to one poll from any source. If there are fewer than five, the window is expanded to 30 days or five polls, whichever comes first. [...]
By itself, it isn't a problem as averages help to smooth out the outliers. But, when you have multiple polls that don't all provide the same list of candidates to choose from... then you cannot smooth out the outliers. In this case, the main issue is that Marquette Law School was the only one to have an option for Kennedy while the others didn't, presumably because he has withdrawn. Since Kennedy got 9% in that one poll, his average is high. Indeed, if we were to instead calculate it as if no polling equals zero percent, then Kennedy would be 9/5 → 1.8% and Oliver would be 3/5 → 0.6%; this added together would be 46.4% + 1.2% (47.6%) + 1.8% (49.4%) + 0.8% (50.2%) + 0.6% (50.8%) + 45.6% (96.4%). This would leave us with figuring the average for the other column which is 18/5 → 3.6%. Add that to 96.4% to get 100%.
So, why would 270ToWin calculate it the way they do? Because they are trying to average out the candidates to get a better idea of what percentage of the vote they will get. But, because not all polls are asking the same list of candidates, they are forced to figure out a compromise, which is to keep the average they get from the polls that did let a specific candidate be selected.
Some but not all of the electoral votes predicted.
Not really. All the information is there (thank you) but it's presented state by state and not together as a national result of electoral seats. Unless you count the map at the top, but this has only some of the states filled in with red or blue while others are brown, so you don't get the whole picture. GotoGothenburg (talk) 23:10, 22 October 2024 (UTC)
Yes, that is the whole current picture. The total electoral votes listed at the bottom of the table are for all states, including "safe" states not listed in the table. To the extent that it is up-to-date, the states that are not shaded red or blue on the map are tossups, and won't be known until closer to the election (as polling uncertainty decreases) or possibly after vote counts are certified. --Spiffy sperry (talk) 01:16, 23 October 2024 (UTC)
I think the whole point here is that the polls do not give the total picture. They are lazy and say some states are 'toss-ups' in stead of making a prediction. Are they afraid of being wrong? They shouldn't be. They show the poll as it is now, not on election day. — Preceding unsigned comment added by 37.247.31.205 (talk) 18:50, 29 October 2024 (UTC)
Fixed. --Spiffy sperry (talk) 15:46, 10 May 2023 (UTC)
_polling.svg)
