Talk:Nationwide opinion polling for the 2016 United States presidential election/Archive 1
![]() | This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page. |
Archive 1 |
Suggestion: Thicker or Darker Lines Between Polls
On the polling pages I would like to see a thicker/darker line to separate polls from different polling firms. It would be simpler to discern which polls are from when and from whom.
Currently it is a little confusing (or at least has the potential to to so) when I see the same line between different candidates within poll X when compared to the separation line between poll X and poll Y.
This is more so on my mobile phone when I have to zoom in to see it and can't automatically see the part of the graph indicating what firm commissioned the poll and when.
Thanks! 98.253.175.243 (talk) 06:01, 19 February 2015 (UTC)
General Elections Table
A bug I noticed in the table is if you sort by date, instead of sorting oldest/newest or newest oldest (i.e. chronologically) poll, it sorts alphabetically. [User retrograde62@yahoo.com 15:04 PM PST 31 May 2015] — Preceding unsigned comment added by 70.170.11.65 (talk) 22:08, 31 May 2015 (UTC)
Ya this is still a problem.
--98.190.133.83 (talk) 07:36, 2 June 2015 (UTC)
- Fixed by adding sort keys in the yyyy-mm-dd format where string sorting is the same as date sorting.[1] PrimeHunter (talk) 13:09, 2 June 2015 (UTC)
New polls?
Are there any new polls we can add?173.66.197.57 (talk) 22:52, 20 June 2015 (UTC)
External links in Body of Article
Hello, Editors! You may have seen me removing External Links from this article. I've also been reverting new additions to the Polling articles that contain new External Links. This is all to meet WP:EL and is not a judgement on the subject or quality of your contributions here. Please convert those External Links into cites when adding new content so your hard work isn't reverted. For the most part, editors have been using cites instead and that is great!
If I or another editor revert your edit containing ELs, please feel free to revert back *and then immediately* change the ELs into cites/references so you don't have to redo any of the rest of your edit.
We can get rid of the EL issues tag at the top of the page once all of the external links are removed (and converted to cites where appropriate). I've started the process and hopefully others can help out with a few here and there until they're all gone. Thanks, Stesmo (talk) 20:26, 1 August 2015 (UTC)
- Don't remove useful content just because the source is given as an external link in the body. Formatting guidelines don't say you can remove material which isn't formatted as recommended. PrimeHunter (talk) 01:22, 3 August 2015 (UTC)
Done I have changed all the external links in the article except for the two aggregate poll links. I did not change those because they are not static references and will continue to change over time. --Stabila711 (talk) 07:19, 19 August 2015 (UTC)
Aggregate polls
What do mean the aggregate polls? Is it like an avergage? — Preceding unsigned comment added by 181.91.125.217 (talk) 01:32, 1 September 2015 (UTC)
- @181.91.125.217: Yes. Aggregate polls are averages. They differ depending on the poll model that is used and how each poll is weighted. --Stabila711 (talk) 01:39, 1 September 2015 (UTC)
CNN/ORC poll
It should be registered voters only, not all Americans94.211.104.84 (talk) 10:08, 12 September 2015 (UTC)
Date column
The aggregate polls really need a date column. Prcc27 (talk) 16:42, 6 October 2015 (UTC)
- @Prcc27: So I tried it in preview mode and it just looks terrible. Since all the different polls have different dates associated with them it makes the table harder to read. When I originally put in the aggregate table I had a line in the header that said when it was last updated. Perhaps "last checked" would be a more accurate description but putting in a date column actually makes the table harder to read. Thoughts? --Stabila711 (talk) 18:06, 6 October 2015 (UTC)
- I guess that works... Prcc27 (talk) 21:25, 6 October 2015 (UTC)
Gravis Marketing/One America News
http://www.oann.com/pollnational/94.211.104.84 (talk) 10:17, 15 October 2015 (UTC)
three-candidate matchup
Trump vs Clinon vs Bloomberg(and others)http://morningconsult.com/2016/01/new-poll-could-bloomberg-win/83.86.208.191 (talk) 20:37, 21 January 2016 (UTC)
Rasmussen
Keep an eye on Rasmussen: [2], once the gears start moving faster it is a good reliable source for polling. - Knowledgekid87 (talk) 14:27, 5 February 2016 (UTC)
External links modified
Hello fellow Wikipedians,
I have just added archive links to one external link on Nationwide opinion polling for the United States presidential election, 2016. Please take a moment to review my edit. If necessary, add {{cbignore}}
after the link to keep me from modifying it. Alternatively, you can add {{nobots|deny=InternetArchiveBot}}
to keep me off the page altogether. I made the following changes:
- Added archive http://web.archive.org/web/20160204023120/http://www.mcclatchydc.com/news/politics-government/election/article43727049.ece/BINARY/Poll%20details to http://www.mcclatchydc.com/news/politics-government/election/article43727049.ece/BINARY/Poll%20details
When you have finished reviewing my changes, please set the checked parameter below to true to let others know.
An editor has reviewed this edit and fixed any errors that were found.
- If you have discovered URLs which were erroneously considered dead by the bot, you can report them with this tool.
- If you found an error with any archives or the URLs themselves, you can fix them with this tool.
Cheers.—cyberbot IITalk to my owner:Online 12:59, 26 February 2016 (UTC)
New NBC poll
I'll add this later if no one else does. —Torchiest talkedits 15:19, 12 April 2016 (UTC)
New polls
http://www.foxnews.com/politics/interactive/2016/02/18/fox-news-poll-national-presidential-race-february-18-2016/ https://www.scribd.com/doc/299591277/NBC-WSJ-February-Poll-GOP?secret_password=JnP82QeNxD9kHWxwZuHr http://www.cbsnews.com/news/cbs-news-poll-gop-race-trump-remains-on-top-hell-get-things-done/ http://www.quinnipiac.edu/images/polling/us/us02182016_Urpfd42.pdfGhostmen2 (talk) 11:07, 22 February 2016 (UTC)
Done 1st and 4th were added. 2nd/3rd are Rep. primary polls and are at the appropriate page. --Spiffy sperry (talk) 19:45, 12 April 2016 (UTC)
Aggregate polls
The Aggregate polls don't have sense because simple polls measure what would be the result of an elections at one point of time. So creating polls that mix the results of different moments of time don't mean anything and can't predict what will the result of the election. — Preceding unsigned comment added by 181.91.131.85 (talk) 16:23, 18 June 2016 (UTC)
- Uh...no. Exact opposite actually. Single polls are subject to bias and can change drastically depending on who exactly is polled. Aggregate polls are averages and provide a much much better look at what the "real" numbers are. Just like how any average takes into account outliers on either side. In any case, are you suggesting that we remove the aggregate polls section? That can always be put towards the community but if so, I oppose that for my reasoning above. --Majora (talk) 16:28, 18 June 2016 (UTC)
Graph
Has not been updated in 20+ days. Seems like it would make more sense to drop Sanders going forward, and do either a 4 way graph? (perhaps a 2 way and 4 way, unless one of the 3rd parties gets into the debates) Gaijin42 (talk) 15:26, 27 June 2016 (UTC)
Two way race graph
Please could someone add a key explaining what the four different coloured lines represent? Thanks. --Dweller (talk) Become old fashioned! 09:03, 29 June 2016 (UTC)
Polls
The third party candidates need to be included in the data sets. — Preceding unsigned comment added by 73.184.59.163 (talk) 06:55, 13 July 2016 (UTC)
I have a new one, but i have no time to add it now http://apps.washingtonpost.com/g/page/politics/washington-post-abc-news-poll/1562/83.80.208.22 (talk) 16:29, 22 January 2015 (UTC)
- Done, thanks. Tiller54 (talk) 20:00, 22 January 2015 (UTC)
again new one :) http://www.foxnews.com/politics/interactive/2015/01/29/fox-news-poll-voters-believe-romney-clinton-remain-top-picks-for-2016-believe/83.80.208.22 (talk) 10:56, 30 January 2015 (UTC)
- again http://www.rasmussenreports.com/public_content/politics/elections/election_2016/clinton_vs_walker_it_s_very_close83.80.208.22 (talk) 14:54, 4 March 2015 (UTC)
- http://www.mcclatchydc.com/2015/03/06/258941/clinton-loses-ground-against-gop.html83.80.208.22 (talk) 11:30, 7 March 2015 (UTC)
- new one http://www.rasmussenreports.com/public_content/politics/elections/election_2016/clinton_vs_cruz_clinton_vs_paul83.80.208.22 (talk) 13:36, 14 April 2015 (UTC)
- The Economist /YouGov Poll https://d25d2506sfb94s.cloudfront.net/cumulus_uploads/document/olylop8wkh/econTabReport.pdf83.80.208.22 (talk) 08:48, 1 June 2015 (UTC)
- ABC http://www.washingtonpost.com/page/2010-2019/WashingtonPost/2015/06/02/National-Politics/Polling/release_396.xml83.80.208.22 (talk) 13:10, 2 June 2015 (UTC)
- CNN http://i2.cdn.turner.com/cnn/2015/images/06/01/2016.poll.pdf83.80.208.22 (talk) 13:10, 2 June 2015 (UTC)
- NBC http://msnbcmedia.msn.com/i/MSNBC/Sections/A_Politics/6_22_PollPDF.pdf83.80.208.22 (talk) 09:36, 23 June 2015 (UTC)
Dubious Pollsters
There appears to be many clearly biased pollsters that objectively shouldn't be included here. I'm especially looking at "Rasmussen Report" and "One America News Network/Gravis Marketing". The credibility of the entire page and the community created poll chart that include these biased pollsters are in danger. — Preceding unsigned comment added by 2001:999:11:4A35:2D01:9C92:EC2D:ABA9 (talk) 13:19, 25 July 2016 (UTC)
Article size
Is anyone opposed to me splitting the older polls (2013, 2014, 2015) into templates and using those as transclusions on this page? That would drastically reduce the page size while still maintaining the information. The only downside to this would be that the page information would be spread out across four different places. Three templates and this main page. --Majora (talk) 17:46, 30 July 2016 (UTC)
Actually, I am probably going to do it with one template and labelled section transclusion. So two pages. Still would like to know if there is any opposition to this before I go ahead and make the changes. --Majora (talk) 18:09, 30 July 2016 (UTC)
Done I just went ahead and boldly changed it. Cut the article size down substantially. You can also now edit the whole section without freezing your computer. --Majora (talk) 02:37, 31 July 2016 (UTC)
Graph trend
The graph currently only shows the trend over a very ling period of time. It would be interesting to make the graph more accurate so that we can see smaller trend changes in the graph aswell. — Preceding unsigned comment added by 62.195.197.219 (talk) 15:36, 1 August 2016 (UTC)
Penn Schoen Berland
According to the Huffington Post, the 3 candidate Penn Schoen Berland poll which here shows Johnson at 15%, is actually Johnson + Would Not Vote + Don't Know. Huffington Post breaks it down to 5% Johnson, 1% Would Not Vote, 9% Don't Know/Undecided. But, I don't see a direct source from Penn Schoen Berland which shows that. http://elections.huffingtonpost.com/pollster/polls/penn-schoen-berland-25048 — Preceding unsigned comment added by AJPEG (talk • contribs) 00:28, 6 August 2016 (UTC)
Graph shows sizable Trump lead?
Ignorant limey here, but the 2 horse race tabular figures all show Clinton leading, other than one poll, yet the graph indicates a sizable Trump lead. How does that work? --Dweller (talk) Become old fashioned! 10:44, 3 August 2016 (UTC)
- Due to the complexities of generating the graph (it's updated by a volunteer who knows how to run an R script), there is an unavoidable lag between it and the polls in the table. --Spiffy sperry (talk) 14:08, 3 August 2016 (UTC)
- I'd get that if the graph's line joined all poll results, but it's supposed to be a trend line. How can it possibly be showing a trend of Trump in the lead, when only one poll amongst many recent ones shows him in the lead? --Dweller (talk) Become old fashioned! 14:31, 3 August 2016 (UTC)
- I couldn't tell you why the trend looks like it does, but in any case I think it bothered enough people that I updated it again with the newer polls. Abjiklɐm (tɐlk) 15:12, 3 August 2016 (UTC)
- That looks much more reliable. Nice work, thank you. --Dweller (talk) Become old fashioned! 15:36, 3 August 2016 (UTC)
- In truth, nothing is more reliable about the updated curves. Merely by chance do they "look" better than the previous iteration. It's not unlikely that the next iteration will look just as ridiculous as the previous one. It's the same exact formula, which is effectively an amateur election forecast. Since that amounts to original research, the trend lines should be removed from the graph unless their specific application in this context can be justified by a reliable source. (To see that the trends shown are arbitrary, let's get into the weeds a bit. Based on the code (here, using the relevant command 'geom_smooth' documented poorly here), it uses LOESS with span = 0.8. Essentially, each point in the LOESS curve is determined by first calculating a trend line (part of the arbitrariness here concerns the use of a trend line as opposed to a quadratic, etc.) over a weighted average of surrounding points (the weights being determined by that span = 0.8 embody a large part of the arbitrariness) and then using that trend line to determine the point in question. To answer the original question, the problematic part of graph in question (preserved here) is the result of unjustifiably extrapolating forward the trend upward resulting from Trump's convention bounce. Set the span higher or lower and the end of the graph would probably look more reasonable.) At the very least, someone who thinks we should keep the current trend curves should explain why this particular method and parameter value should be used. I should say that the graph looks nice and could, in principle, be helpful to have. But precisely because the trend lines are so effective at boiling down all the numbers on this page to a simple visual, it's really unacceptable to use such an arbitrary method to determine them. -hugeTim (talk) 20:39, 5 August 2016 (UTC)
- p.s. See also WP:SYNTHESIS and a related (archived) discussion on poll aggregation/averaging here in which consensus was reached to only use poll aggregates/averages published in reliable sources.
- I'm someone who thinks we should keep the graph, and the reason I used local regression is because I didn't want to assume the curve takes a specific shape. The span is arbitrary. Do you have an alternative method to suggest or do you want to do away with the graph? Abjiklɐm (tɐlk) 00:42, 6 August 2016 (UTC) edit: Ah I just saw you did remove it and someone replaced it with their own graph. That didn't take long. Abjiklɐm (tɐlk) 00:46, 6 August 2016 (UTC)
- That looks much more reliable. Nice work, thank you. --Dweller (talk) Become old fashioned! 15:36, 3 August 2016 (UTC)
- I couldn't tell you why the trend looks like it does, but in any case I think it bothered enough people that I updated it again with the newer polls. Abjiklɐm (tɐlk) 15:12, 3 August 2016 (UTC)
- I'd get that if the graph's line joined all poll results, but it's supposed to be a trend line. How can it possibly be showing a trend of Trump in the lead, when only one poll amongst many recent ones shows him in the lead? --Dweller (talk) Become old fashioned! 14:31, 3 August 2016 (UTC)
Could we rescale the two-way race graph's y-axis?
The current graph runs from 0% to 100%. This makes the data hard to read, since the race is always fairly close to 50-50. Some previous edits (e.g. [ here]) used a graph that started at 30%. I realise that graph was removed because its trend lines were OR, but I think a y-axis running from (for instance) 30% to 70% will be better than one from 0% to 100%, at least until the race becomes more lopsided than that (which is unlikely). I realise also that the current graph updates slowly due to a dependence on someone's R skills; this is just a suggestion for long-term future edits to it. 86.162.159.218 (User ) 13:58, 7 August 2016 (UTC)
- Agreed. The current y axis is decidedly unhelpful, making even a huge swing for either candidate appear negligible. Neither of them is or will poll in the range 0-30 or 60-100, so including them is just obscurism. --Dweller (talk) Become old fashioned! 20:12, 7 August 2016 (UTC)
- Change of y-axis is under way (30-70) if other scale is wished for please tell! --Avopeas (talk) 23:13, 7 August 2016 (UTC)
LA Times/USC
Question from an outside observer. Is this poll not mostly the same data, only slightly changed day by day? - Galloglass 19:19, 8 August 2016 (UTC)
- There has been some concern raised by the NYT considering that the LA Times/USC poll does what no other polling source does. - Knowledgekid87 (talk) 22:24, 8 August 2016 (UTC)
Trend line
The current graph also doesn't mention what method is used for the trend line. Is it a weighted or unweighted average, is it day-by-day average or over a span of a few days? Add to that the fact that only its author can update it, as no way is given for other users to update it. I fail to see how this is an improvement. Abjiklɐm (tɐlk) 18:43, 7 August 2016 (UTC)
- The trend line is a useful thing, but I agree explaining it would be good. As I pointed out above, there have been times when it's shown a trend that doesn't appear to be a trend at all. --Dweller (talk) Become old fashioned! 20:12, 7 August 2016 (UTC)
- The trend line is an average of the ten latest published polls. Meaning if a new poll is published it goes into the counting, and the poll that was published ten before it goes out of the average count. This means that the lines could go from, for example. 45% to 46% on the same date, since several polls are published on the same date. --Avopeas (talk) 23:23, 7 August 2016 (UTC) edit: I am open to change out the graph with an other if someone has the skills to make it editable for all and still maintain a more ingoing trendline. --Avopeas (talk) 23:27, 7 August 2016 (UTC)
- How was the 10-poll average determined to be suitable for a trend line? (The old graph was better, imo, but my objection applies to both.) I appreciate the boldness of providing a new graph, but I'm going to remove it because any trend line using a method arbitrarily decided by a random Wikipedian is WP:SYNTHESIS. I'm going to continue being in favor of removing any graph with a trend line until someone at least addresses this concern. -hugeTim (talk) 17:57, 8 August 2016 (UTC)
- My suggestion would be to plot a simple average of the most reliable poll aggregators (perhaps NYT, HuffPost, and RealClear for the two-way) as the trend line rather than doing our own analysis of poll trends based on the raw polling data presented as a scatterplot. Alternatively, a single external poll aggregator or a simple average of all aggregators displayed in the table would also be reasonable ways to avoid WP:SYNTHESIS. -hugeTim (talk) 18:06, 8 August 2016 (UTC)
What about a daily weighted average? That's hardly arbitrary is it? Not more arbitrary than averaging a select group of poll aggregators that use different methods for their numbers. Abjiklɐm (tɐlk) 22:12, 9 August 2016 (UTC)I tried it and it looks terrible. I welcome suggestions. Abjiklɐm (tɐlk) 15:47, 10 August 2016 (UTC)- Thank you for trying out a new idea and also for providing the two-way scatterplot without a trend line for now. I agree with you that averaging poll aggregators seems arbitrary. How about my alternative suggestion of choosing one poll aggregator to plot? I wonder if we can find consensus about which aggregator to use. Based on my current assessment of reputation for reliability, I would suggest New York Times for the two-way, FiveThirtyEight for the three-way, and RealClearPolitics for the four-way. Somewhat comfortingly, these sources roughly agree when two or more of them have averages in the same category, unlike TPM's Polltracker at the moment. Another reasonable option would be to use RealClearPolitics for all three (two-way, three-way, and four-way), for consistency's sake. -hugeTim (talk) 20:21, 10 August 2016 (UTC)
- The trend line is an average of the ten latest published polls. Meaning if a new poll is published it goes into the counting, and the poll that was published ten before it goes out of the average count. This means that the lines could go from, for example. 45% to 46% on the same date, since several polls are published on the same date. --Avopeas (talk) 23:23, 7 August 2016 (UTC) edit: I am open to change out the graph with an other if someone has the skills to make it editable for all and still maintain a more ingoing trendline. --Avopeas (talk) 23:27, 7 August 2016 (UTC)
Three-way race
There have been only two opinion polls released in August for Three-way race compared with 14 for Four-way race (and 25+ for two-way race). Should we shift the place for the three-ways polls to after the Four-ways?Yger (talk) 15:33, 10 August 2016 (UTC)
- No. It should stay the same until the time being.DrFargi (talk) 23:01, 16 August 2016 (UTC)
- I think 4 should be moved up. The reality of the race is 4+, although the chances of any but Trump or Clinton winning is low. ResultingConstant (talk) 23:53, 16 August 2016 (UTC)
Will there be polling for Evan McMullin?
https://www.evanmcmullin.com/ Abe 92.192.59.12 (talk) 13:42, 11 August 2016 (UTC)
Unlikely, since he is not expected to be on most state ballots. --Eliyak T·C 22:15, 11 August 2016 (UTC)
The McMullin campaign expects to anyway: https://www.evanmcmullin.com/how_evan_mcmullin_will_compete_nationally Abe 92.192.107.90 (talk) 10:37, 12 August 2016 (UTC)
- They don't even have polling for Darrell Castle who as of right now has ballot access in 21 states. - Knowledgekid87 (talk) 11:47, 12 August 2016 (UTC)
Well, the difference is McMullin is actually in the news/on national TV. ABC News, MSNBC, BBC, CNN, Foxnews ... He is also featured at PredictIt (https://www.predictit.org/Market/1234/Who-will-win-the-2016-US-presidential-election), where he is performing better than Jill Stein at this time. He barely started his campaign and has 40.000 followers on Twitter. On fundraising he is already doing better than Stein. Has a clear chance of winning Utah. Abe 92.192.42.26 (talk) 12:34, 13 August 2016 (UTC)
- In this case, none of that matters if he isn't in any poll conducted by a reliable source. - Knowledgekid87 (talk) 23:57, 16 August 2016 (UTC)
He actually already is, but only on state level afaik. PPP is about to do a Utah poll. Abe 92.192.97.144 (talk) 13:59, 18 August 2016 (UTC)
References for 2015, 2014, 2013 polls
Is there a way to access these polls so I might be able to fix the references by adding proper citations and dates? DrFargi (talk) 05:07, 20 August 2016 (UTC)
USC/Los Angeles Times
This is surely he most weird poll. It consist of a predefined set of 3000 voters who are asked over and over again of their preferences. And the main curve is very much different from any other pollresults, and it swings unproportional day from day. Should we really take this in, and if so should it not only be one per week, not every day?Yger (talk) 13:52, 20 August 2016 (UTC)
- Considering the way it is conducted, I don't see the polling as reliable. - Knowledgekid87 (talk) 14:35, 20 August 2016 (UTC)
- Here is an article by the NYT regarding the matter: [3]. As for polling sources, some use the LA Times polls, some don't. The Huffington Post, Pure Polling, New York Times, and 270 to Win for example do not include the polling. - Knowledgekid87 (talk) 14:44, 20 August 2016 (UTC)
Even RCP and 538 include LA Times poll. I don't see any problem. — Preceding unsigned comment added by Gladkyandrey (talk • contribs) 14:46, 20 August 2016 (UTC)
- The problem is how the polling is conducted, some sources include the polls while others don't. - Knowledgekid87 (talk) 14:47, 20 August 2016 (UTC)
- Note that 538 does, indeed, include the USC/Los Angeles poll in their figures, but as they explain here, their model "weights each poll by its sample size, how recently it was conducted, and the historical accuracy and methodology of the polling firm. The model then adjusts each poll based on other factors." I think it's important to note that they currently show the results of the USC/LA Times poll of August 13-19 as "Leader: Trump +1", "Adjusted Leader: Clinton +3". And whereas most of the polls they cite are given letter grades, A+ to C-, the USC/LA Times poll's grade rating is blank. I can see a rationale for excluding this poll here, because we can't perform the OR necessary to correct for systemic errors in the polling methodology, and it may be misleading to our readers to pass numbers like the USC/LA Times poll results through without filtering them, but to a greater or lesser degree that can be said of all the polls. NameIsRon (talk) 17:41, 20 August 2016 (UTC)
- Removing it would also be OR based on a personal idea that it is not reliable. The poll is included in aggregates, as already mentioned. Leave it. There could be a lot of factors that cause poll numbers to fluctuate. That is why individual polls are separate from aggregate polls and why aggregate polls come first on the article. Wikipedia should not be deciding what polls to add and what polls to exclude. A poll happened, results were given, the poll gets included. Leave it be. --Majora (talk) 17:49, 20 August 2016 (UTC)
- Like it, or hate it the fact is that the polling is controversial. - Knowledgekid87 (talk) 18:22, 20 August 2016 (UTC)
- This whole election is controversial. Content should not be removed because it is controversial. Unless it can be shown that they committed some sort of polling fraud the poll should just be included. We shouldn't be deciding what gets included and what doesn't. That isn't for us to do. We report what has already been reported. Nothing more. We are an impartial source, and we need to stay that way. It is for the readers to decide what polls they want to give more weight to, or if they want to give any weight at all to the individual polls. --Majora (talk) 18:33, 20 August 2016 (UTC)
- Seeing other sources are posting the polling that is a reason to include it, but seeing some aren't it needs to be mentioned somehow per WP:NPOV. - Knowledgekid87 (talk) 18:42, 20 August 2016 (UTC)
- The USC/LA Times poll does not fulfill basic criteria for what constitutes a poll. Wikipedia is an encyclopedia which puts some basics demands on what we publish. A playaround among a set of 3000 persons does not qualify to be presented here, and to publish its weird figers every day as they choose to publish it makes the whole table unserious.Yger (talk) 18:47, 20 August 2016 (UTC)
- I am fine with removing the USC/LA Times polls if there is a problem with the way they are conducted. If only 2 polling average sites out of 7 are mentioning them, then that gives more reason to remove them. However that still doesn't change the fact that the polling averages for RCP and 538 will keep changing irregardless of whether we include the USC/LA Times polls or not.DrFargi (talk) 23:41, 20 August 2016 (UTC)
- Don't know if this is TMI, but this morning RCP updated their summary, and the only poll in their list that changed is USC/LA Times, which changed from a Tie to a net Trump +2. This resulted in a change in the RCP Average from Clinton +5.7 (for 8/1-8/19) to Clinton +5.3 (for 8/1-8/20). That's a larger change than might have been expected for a two-point change in a single poll that has an unknown MoE. I think the size of its effect is probably due to its timeliness and large sample size. I think the point DrFargi was making was that if we think the USC/LA Times poll is unreliable, we need to be aware that if we include RCP and 538 in our tables, we're allowing some suspect numbers ("noise") to be passed through. I don't think that's a problem per se; I trust RCP and 538 to use their judgment in what polls to track and how to process their results. NameIsRon (talk) 16:10, 21 August 2016 (UTC)
- Seeing other sources are posting the polling that is a reason to include it, but seeing some aren't it needs to be mentioned somehow per WP:NPOV. - Knowledgekid87 (talk) 18:42, 20 August 2016 (UTC)
- This whole election is controversial. Content should not be removed because it is controversial. Unless it can be shown that they committed some sort of polling fraud the poll should just be included. We shouldn't be deciding what gets included and what doesn't. That isn't for us to do. We report what has already been reported. Nothing more. We are an impartial source, and we need to stay that way. It is for the readers to decide what polls they want to give more weight to, or if they want to give any weight at all to the individual polls. --Majora (talk) 18:33, 20 August 2016 (UTC)
- Like it, or hate it the fact is that the polling is controversial. - Knowledgekid87 (talk) 18:22, 20 August 2016 (UTC)
- Removing it would also be OR based on a personal idea that it is not reliable. The poll is included in aggregates, as already mentioned. Leave it. There could be a lot of factors that cause poll numbers to fluctuate. That is why individual polls are separate from aggregate polls and why aggregate polls come first on the article. Wikipedia should not be deciding what polls to add and what polls to exclude. A poll happened, results were given, the poll gets included. Leave it be. --Majora (talk) 17:49, 20 August 2016 (UTC)
- Note that 538 does, indeed, include the USC/Los Angeles poll in their figures, but as they explain here, their model "weights each poll by its sample size, how recently it was conducted, and the historical accuracy and methodology of the polling firm. The model then adjusts each poll based on other factors." I think it's important to note that they currently show the results of the USC/LA Times poll of August 13-19 as "Leader: Trump +1", "Adjusted Leader: Clinton +3". And whereas most of the polls they cite are given letter grades, A+ to C-, the USC/LA Times poll's grade rating is blank. I can see a rationale for excluding this poll here, because we can't perform the OR necessary to correct for systemic errors in the polling methodology, and it may be misleading to our readers to pass numbers like the USC/LA Times poll results through without filtering them, but to a greater or lesser degree that can be said of all the polls. NameIsRon (talk) 17:41, 20 August 2016 (UTC)
As my propsal to only publish these figures on a weekly base has not been accepted, I then think we should pubclish them every day to be consistent.Yger (talk) 13:11, 22 August 2016 (UTC)
- Just FYI the Washington Post has also just raised its eyebrows at how the polling is conducted. [4] - Knowledgekid87 (talk) 14:03, 22 August 2016 (UTC)
- Just so you know that there will always be differences in terms of how New York Times and Washington Posts conduct their own polls compared to how the LA Times conducts theirs. They are competitors and they do raise issues with each others polls from time to time to fit their narrative. The title beginning with 'Donald Trump's New Favourite Poll' tells you a bit about W. Post DrFargi (talk) 01:06, 24 August 2016 (UTC)
- Readers can choose to read into it or not, it is a viewpoint that is presented though. - Knowledgekid87 (talk) 01:10, 24 August 2016 (UTC)
20 August Gladkyandrey enters a poll for LA/USC. Today the same user deletes this and instead includes one from two days later, being even closer to the latest one? Which rule apllies to thse polls, I beleived once a week as he norm, but now it seems it is to pick whatever you "like"?? I noticed at least one other uses all the time the latest, and I thought this was good and then include on a full week older, in order not to have duplicate polldates Yger (talk) 17:29, 24 August 2016 (UTC)
USC/Los Angeles Times August 15–21, 2016 & August 16–22, 2016
Are the person(s) who added these aware that these two polls share a 90% common data set? - Galloglass 18:02, 22 August 2016 (UTC)
- I don't know, at this point I already added a note per WP:NPOV that explains concern raised by reliable sources. I don't care if Clinton or Trump lead by as much as 10 pts, the way we include these polls has to be with caution. - Knowledgekid87 (talk) 18:22, 22 August 2016 (UTC)
- This is one of the reasons I have recommended only one poll per week from them, but the deletion of the too frequent poll results was reverted. I try once more to remove too frequent poll results from them.Yger (talk) 10:07, 23 August 2016 (UTC)
- Which was reverted so back to daily poll figures for LA.Yger (talk) 11:05, 23 August 2016 (UTC)
- Since this is updated daily, would it fall under being a aggregate poll? - Knowledgekid87 (talk) 12:03, 23 August 2016 (UTC)
- Which was reverted so back to daily poll figures for LA.Yger (talk) 11:05, 23 August 2016 (UTC)
- This is one of the reasons I have recommended only one poll per week from them, but the deletion of the too frequent poll results was reverted. I try once more to remove too frequent poll results from them.Yger (talk) 10:07, 23 August 2016 (UTC)
- Why is it that we have to put out the polls for USC/LA Times once a week? IF we are to to only include the polls once a week then which set of polls in that week do you include? Do you see where I am coming to in regards to this discussion? In my opinion it is not to take it on ourselves to remove the daily polls but to include all polls. Only polls that are suspect in terms of missing sample numbers, margin of errors and polls that are not quoted by one of the big pollsters like Huffington, Real Clear Politics, etc.DrFargi (talk) 00:54, 24 August 2016 (UTC)
- It doesn't matter what the polling source is, if you notice RCP doesn't show back to back daily polls. [5] Per WP:UNDUE we cant give preference to one polling source. - Knowledgekid87 (talk) 00:58, 24 August 2016 (UTC)
- Granted that RCP doesn't show back to back daily polls, but is have we don't all our checks and balances? Have you check the other poll sources FiveThirtyEight, 270 to Win, etc to see whether they have included the some of the polls that RCP has not included? DrFargi (talk) 01:11, 24 August 2016 (UTC)
- Fair enough, I still feel that we should add the LAT as an aggregate poll though as it is conducted daily. - Knowledgekid87 (talk) 01:16, 24 August 2016 (UTC)
- No, an aggregate poll combines results from other polls. See Poll aggregator. PrimeHunter (talk) 22:04, 24 August 2016 (UTC)
- Fair enough, I still feel that we should add the LAT as an aggregate poll though as it is conducted daily. - Knowledgekid87 (talk) 01:16, 24 August 2016 (UTC)
Removed the graph
I have no idea here why the lines for Trump/Clinton are drawn to the points of which it shows. Are these polls more accurate, if so then why, or why not? - Knowledgekid87 (talk) 01:38, 20 August 2016 (UTC)
- I think it would be contact the creator of the graph and ask him.DrFargi (talk) 05:07, 20 August 2016 (UTC)
- Creator here. The lines show daily averages. When the line goes through a point, it only means that there was only one poll published on that day. Otherwise, you'll notice that the line stops between points. Abjiklɐm (tɐlk) 15:38, 20 August 2016 (UTC)
- Adding daily lines to the graph rather distorts the whole thing. It would be much better left as it was. - Galloglass 19:05, 20 August 2016 (UTC)
- Could you explain what you mean by distortion? Abjiklɐm (tɐlk) 13:31, 21 August 2016 (UTC)
- I can offer an observation. The version up to August 10 ( [6] ) included a trend line, and I think you were looking for comment here on the talk page as to what people thought about it ... there is an element of original research about your providing a "value-added" enhancement to the data presented in the article. The line is a trend line, so it changes gradually as polls change. In the newer version ( [7] ), because different polls show up each day, the line is jagged because it connects points that only represent an average of the polls sharing a given date. My opinion is that the trend line was helpful; the jagged line exaggerates the accidental differences that occur between various polls whose reports are not synchronized. NameIsRon (talk) 14:10, 21 August 2016 (UTC)
- Could you explain what you mean by distortion? Abjiklɐm (tɐlk) 13:31, 21 August 2016 (UTC)
- Adding daily lines to the graph rather distorts the whole thing. It would be much better left as it was. - Galloglass 19:05, 20 August 2016 (UTC)
- Abjiklɐm, are you willing to remove the line in the graph? The previous design was much better.DrFargi (talk) 23:32, 20 August 2016 (UTC)
- Of course I'm willing :) Will do so soon. Abjiklɐm (tɐlk) 13:31, 21 August 2016 (UTC)
- Creator here. The lines show daily averages. When the line goes through a point, it only means that there was only one poll published on that day. Otherwise, you'll notice that the line stops between points. Abjiklɐm (tɐlk) 15:38, 20 August 2016 (UTC)
Has this graph now been orphaned by its editor as its not been updated since the 17th? - Galloglass 21:16, 27 August 2016 (UTC)
True Polls
Question: Why is there nothing here reflecting the TRUE, UNBIASED, and ACCURATE polls that we all know show Trump clearly winning by a landslide, even in New York and California?
http://www.allenbwest.com/michellejesse/two-new-polls-show-trump-florida-thats-not-really-big-news
https://www.youtube.com/watch?v=zaskdfi7iyQ
http://fortune.com/2016/05/25/election-polls-hillary-clinton-donald-trump/
https://www.youtube.com/watch?v=hIUEeru1L84
https://www.youtube.com/watch?v=w2784OBlwjU
https://www.youtube.com/watch?v=eAua8JovI7Y
www.infowarscom/trump-defeats-hillary-in-poll-monmouth-amends-it-to-give-hillary-victory/ [unreliable fringe source?]
https://www.youtube.com/watch?v=NHyf80lekeA
https://www.youtube.com/watch?v=acENoDuAFyo
https://www.youtube.com/watch?v=WHlFYmzhbHg
https://www.youtube.com/watch?v=d3snvTSEEXg
www.infowarscom/app-maker-trump-will-win-election/ [unreliable fringe source?]
www.infowarscom/trump-obliterates-clintons-lead-in-latest-reuters-poll/ [unreliable fringe source?]
--Concerned Ex-Pat — Preceding unsigned comment added by 74.216.73.120 (talk) 16:28, 28 August 2016 (UTC)
- Because, shockingly, we require reliable sources. Not fairy tale websites that attempt to "unskew" polls that aren't actually skewed. Infowars and youtube don't meet that bar. --Majora (talk) 18:07, 28 August 2016 (UTC)
Very funny, but obviously I didn't mean to claim those websites (Infowars, Youtube, etc.) as sources in and of themselves, but rather as links to the actual polls that they cite. Please do your research before criticizing. --Concerned Ex-Pat — Preceding unsigned comment added by 74.216.73.120 (talk) 13:18, 29 August 2016 (UTC)
- I stopped reading at "Unbiased, Independent Poll" for one of the links. You might want to read WP:SOAPBOX. - Knowledgekid87 (talk) 14:04, 29 August 2016 (UTC)
- Wrong article, Ex-Pat. The Florida polls are already covered, over in the "Statewide" article. I didn't see any specific polls regarding NY or CA in what I checked of what you posted here. NameIsRon (talk) 14:09, 29 August 2016 (UTC)
- Even if there were some, all polls are inaccurate in some way shape or form. As for "bias" this is in the eye of the beholder, we don't want to go down this route. - Knowledgekid87 (talk) 14:15, 29 August 2016 (UTC)
- What Ex-Pat doesn't appreciate is that in the articles on on-going polling, Wikipedia is more willing to pass along polls' raw reports of their results than poll aggregators such as fivethirtyeight.com. The recent (Aug. 19-22) poll by Florida Atlantic University, for example, shows Trump leading by 2%; Nate Silver's site adjusts that based on their analysis of the inherent slant of this poll to indicate Clinton leading by 1%. No such thing as a "true poll", of course, but Wikipedia passes along results from that Florida Atlantic University poll, rated by Nate Silver as a "C+" poll, just the same as better-respected polls such as Ipsos (A-) and YouGov (B). NameIsRon (talk) 14:50, 29 August 2016 (UTC)
Thanks, I understand now. Ex-Pat — Preceding unsigned comment added by 74.216.89.78 (talk) 08:24, 3 September 2016 (UTC)
Likely voters to supersede Registered Voters
It is really unnecessary to have both registered and likely voters shown shown for each individual poll. If there is only registered voters are only shown then that should take precedence here. If likely voters are shown instead then they should take precedence. If both are shown then likely voters should take precedence. Remember that the aggregate pollsters are taking into account likely voters into their calculations first, not registered voters. Please shown me any of the aggregators such as Huffington Post, Real Clear Politics, etc who are including registered voters over likely voters.DrFargi (talk) 12:51, 7 September 2016 (UTC)
Why old polls
Why years 15 14 13 etc in an article titled 2016? It makes it too busy SaintAviator lets talk 07:31, 17 September 2016 (UTC)
- 2016 is the election year for United States presidential election, 2016. The article has said 2016 since it was created. Many countries don't have fixed election schedules so we also have articles like Opinion polling for the next United Kingdom general election. They are renamed when the year is decided. PrimeHunter (talk) 11:05, 17 September 2016 (UTC)
new type of poll
Is it appropriate to add info about a new polling app? Info here. It measures response to speeches/debates in real-time so provides a different type of data than typical polls. Sunshine Kim (talk) 19:32, 23 September 2016 (UTC)
Remove the polling map
OK this new polling map doesn't do much more of a visual favor as the past ones. The truth is the map will be outdated pretty much everyday at this point. And showing the sample size doesn't mean the poll is more accurate. I propose it should wait until after the election to create a graph like the one for the 2012 page. The truth is that it is too visually unpleasing and really doesn't help the article. Manful0103 (talk) 04:24, 28 September 2016 (UTC)
Five-Way race
We need a Five-Way race, ppp and Echelon Add Evan McMullin. http://echeloninsights.com/wp-content/uploads/2016/09/post_debate_survey_toplines.pdf http://www.publicpolicypolling.com/pdf/2015/PPP_Release_National_83016.pdf http://www.publicpolicypolling.com/pdf/2015/PPP_Release_National_92916.pdfGhostmen2 (talk) 10:00, 29 September 2016 (UTC)
Evan McMullin - Aggregate polls
Evan is now polling higher than Jill Stein: http://www.publicpolicypolling.com/main/2016/09/clinton-leads-by-4-nationally-trump-hurt-himself-in-debate.html
You can vote for him in 34 states: https://www.evanmcmullin.com/34_states_and_counting
Come election day the campaign expects you can vote for Evan in at least 40 states. Official PPP polling results: www.publicpolicypolling.com/pdf/2015/PPP_Release_National_92916.pdf
How many polls are required to be featured in Aggregate polls? Abe92.192.103.43 (talk) 11:09, 30 September 2016 (UTC)
- Editors here did not decide who is featured on aggregate polls, nor are the aggregate polls calculated by wikipedia editors. We're simply taking existing aggregate polls and displaying that information. Abjiklɐm (tɐlk) 10:54, 1 October 2016 (UTC)
Sorting
I have tried sorting by polling source so I can see the trend for each particular source, however the polling dates do not line up in chronological order. The problem might be the 'single digit' dates that might need a zero in front of the digit, and I don't know how the dates will be sorted once we have July August and September dates and beyond. I might be asking the impossible but could be possible to have exact date order when you sort by polling source? Juve2000 (talk) 02:44, 31 August 2016 (UTC)
- If you sort by Poll source in Nationwide opinion polling for the United States presidential election, 2016#Since convention nominations then it uses the whole cell for sorting, i.e. both the letters and the reference number including the square brackets. The character "]" sorts after digits so "[1]" sorts after "[19]". You could use Help:Sorting#Specifying a sort key for a cell on all the cells and make sort keys like "Ipsos/Reuters 2016-08-29", but that's a lot of work. It will sort as one long string so I gave a date format where alphabetical sorting and date sorting gives the same result. PrimeHunter (talk) 11:22, 31 August 2016 (UTC)
- Thank you for your response. I consider myself an amature, and some of this stuff goes over my head. I apologise in advance if my next suggestion is laughable, but I was thinking of adding a new column #2 in between the column 1 (name of polling company) and old column 2 (date range) which would number the polling results FOR EACH company starting at 1 with the oldest polling date. Would that give you the required result regardless of the date format? Juve2000 (talk) 19:37, 28 September 2016 (UTC)
- @Juve2000: That wouldn't work. Sorting by the Poll source column would still have used the whole cell including the reference number for sorting. The reference could have been moved to another column but I have done something else: I created {{sad}} and used it in each Poll source cell.[8] Some of the collapsed tables are transcluded from {{Nationwide opinion polling for the United States presidential election, 2016 (older polls)}}. {{sad}} isn't used there yet but I expect to do it Sunday with some regex. PrimeHunter (talk) 02:24, 2 October 2016 (UTC)
- {{Nationwide opinion polling for the United States presidential election, 2016 (older polls)}} is also done now.[9] PrimeHunter (talk) 15:16, 2 October 2016 (UTC)
- Thank you for your response. I consider myself an amature, and some of this stuff goes over my head. I apologise in advance if my next suggestion is laughable, but I was thinking of adding a new column #2 in between the column 1 (name of polling company) and old column 2 (date range) which would number the polling results FOR EACH company starting at 1 with the oldest polling date. Would that give you the required result regardless of the date format? Juve2000 (talk) 19:37, 28 September 2016 (UTC)
USC/LA reliable source?
Can USC/LA really be considered a "reliable source"? Its results are apparently heavily biased because of one single individual. [10] Mlewan (talk) 04:46, 13 October 2016 (UTC)
Seeking advice
Editors, thank you for maintaining a very good article.
As an outsider it's a mystery as to why it's so difficult to find information on en.WP about the polls for house and senate elections—specifically the interpretation of those polls in relation to the likely numbers for the parties (and even the right, centre, and left of each of the two parties). Or perhaps I'm not searching properly for this information. Tony (talk) 12:01, 19 October 2016 (UTC)
- This article is about the presidential election as the name says. The box at the top right links to other races in a blue box at the bottom. You can find some information by state at United States House of Representatives elections, 2016#Competitive districts and United States Senate elections, 2016#Latest predictions of competitive seats, and the individual state articles linked in those articles. The place to discuss the house and senate election articles is Talk:United States House of Representatives elections, 2016 and Talk:United States Senate elections, 2016. We don't appear to have statements of form "Poll x predicts y seats to party z". PrimeHunter (talk) 13:14, 19 October 2016 (UTC)
- Thanks, I do realise that those articles exist, and I would say the same there (I'm assuming that many WP editors work on more than one of these articles); but it's also a mystery that there's nowhere on WP that readers can go to find a summary of the whole thing. It's as though there are three disconnected elections that just happen to be held on the same day; but the result in all three elected organs of state is critical in its potential for interplay in the new electoral cycle. It's just very hard for outsiders to make sense of, with the focus in the media so much on the presidential race alone. It's a pity that WP doesn't provide a summary (review-type) article embracing the whole thing in terms of likely outcomes—opinion polling is the best prediction we have, but I can't get a grip on the three interacting parts in that respect. Tony (talk) 13:57, 19 October 2016 (UTC)
- There is a rather short article on United States elections, 2016 which should cover all the elections this year. Perhaps you could suggest some improvements/additions over at that article's talk page? Abjiklɐm (tɐlk) 18:24, 19 October 2016 (UTC)
- Thanks, I do realise that those articles exist, and I would say the same there (I'm assuming that many WP editors work on more than one of these articles); but it's also a mystery that there's nowhere on WP that readers can go to find a summary of the whole thing. It's as though there are three disconnected elections that just happen to be held on the same day; but the result in all three elected organs of state is critical in its potential for interplay in the new electoral cycle. It's just very hard for outsiders to make sense of, with the focus in the media so much on the presidential race alone. It's a pity that WP doesn't provide a summary (review-type) article embracing the whole thing in terms of likely outcomes—opinion polling is the best prediction we have, but I can't get a grip on the three interacting parts in that respect. Tony (talk) 13:57, 19 October 2016 (UTC)
rasmussen polling 20 ok 2016?
rasmussen polling 20 ok 2016? — Preceding unsigned comment added by 62.101.0.30 (talk) 09:48, 21 October 2016 (UTC)
IBD/TIPP - Tracking Polls
Do people think it's right to include every (almost daily) update from some tracking polls, especially ones like IBD/TIPP which's results seem a little out of line at the moment?
Maybe there should be another section for tracking polls, or updates from them should only be published every few days or week or something...
Tom W (talk) 14:58, 26 October 2016 (UTC)
- Well to be fair we are also using ABC News which has had Clinton up by double digits. The polling averages are the ones that should have more weight. Right now it looks like she has a 5 to 6 point lead which is in the middle of those two. - Knowledgekid87 (talk) 15:09, 26 October 2016 (UTC)
- Should we have ABC News' tracking there as well then? Would a new section for tracking polls be better? Tom W (talk) 17:54, 26 October 2016 (UTC)
- I don't think they make a difference here, as for tracking polls you mean the Aggregate polls? - Knowledgekid87 (talk) 00:42, 27 October 2016 (UTC)
- Naa like IBD/TIPP and now ABC News where they've got a group of respondents and they're asking a portion of them questions each day, adding it to the results from the previous few days or week and combining them and publishing results everyday...
- It's alright to include the ABC and IBD polls as they always include a different sample size each day. I don't think they are important enough to warrant a new section in this article.DrFargi (talk) 12:26, 30 October 2016 (UTC)
External links modified
Hello fellow Wikipedians,
I have just modified one external link on Nationwide opinion polling for the United States presidential election, 2016. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
- Added archive https://web.archive.org/web/20160930234942/http://www.foxnews.com/politics/interactive/2016/09/30/full-fox-news-poll-results-30/ to http://www.foxnews.com/politics/interactive/2016/09/30/full-fox-news-poll-results-30/
When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.
This message was posted before February 2018. After February 2018, "External links modified" talk page sections are no longer generated or monitored by InternetArchiveBot. No special action is required regarding these talk page notices, other than regular verification using the archive tool instructions below. Editors have permission to delete these "External links modified" talk page sections if they want to de-clutter talk pages, but see the RfC before doing mass systematic removals. This message is updated dynamically through the template {{source check}}
(last update: 5 June 2024).
- If you have discovered URLs which were erroneously considered dead by the bot, you can report them with this tool.
- If you found an error with any archives or the URLs themselves, you can fix them with this tool.
Cheers.—InternetArchiveBot (Report bug) 12:06, 20 November 2017 (UTC)