Talk:Statewide opinion polling for the 2016 Republican Party presidential primaries/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 |
Note by the 29 Jan Iowa Poll?
As Mike Huckabee (who was leading in the other two polls) and Jeb Bush (Who had around 10% support) as well as Condoleezza Rice (Who got 9% support in the last poll) were not included in the "Harper Polling" poll should a note be attached as it looks odd that the candidate who was leading in both the other polls is suddenly no where to be seen on this Harper Polling poll? Guyb123321 (talk) 11:56, 2 February 2013 (UTC)
New New Hampshire Poll
http://www.unh.edu/survey-center/news/pdf/gsp2013_spring_2016primary042513.pdf Guyb123321 (talk) 12:21, 28 April 2013 (UTC)
Margins of Error
Ted Cruz is technically within the margin of error vs. Chris Christie in the only Illinois poll that's posted here. Shouldn't this be reflected in the color-coding? Meanwhile, without a margin of error reported, it's rather arbitrary that the color coding suggests that Rubio is competitive with Paul in Arkansas, even though they're 4 points apart. TBSchemer (talk) 07:25, 8 February 2014 (UTC)
Also, in the most recent Oklahoma poll listed, Rand Paul could be within the margin of error of Ted Cruz, but this possibility is obscured only by a discrepancy in the number of significant figures used. TBSchemer (talk) 07:30, 8 February 2014 (UTC)
Confusing and Not Up to Date
Some of the polls listed on the map are over a year old. They don't really apply to recent polling and can make the map more confusing to read. And if you look at the map totals, you would think that Chris Christie is the frontrunner. He used to, but that data isn't very relevant. And if you really want to show the complexities, you can turn this map into a timelapse video of different polling at different times. States' opinions change over time and maps should represent this. If anyone has the time for this task, it would make the map more educational and representational and time conscious. And hopefully less confusing. If that is too difficult, a simpler solution would involve only recent polling and not all polls. 198.228.228.24 (talk) 00:49, 2 June 2014 (UTC)
- In 2012, we ended up eventually splitting the article into multiple articles as more and more polls were taken. First we split off older polls into new articles, but later they were reorganized based on when the primaries were held in each state. I'm sure we'll end up doing something similar here. But I only see two polls that are extremely out of date. One is New Jersey, where Christie is probably still leading, and the other is Wyoming, which honestly isn't too crucial. Every other state has been polled within the last six months, which isn't too bad. Poll frequency will only increase, so things will continue to get more up-to-date as time passes. —Torchiest talkedits 12:38, 2 June 2014 (UTC)
- New York, Montana, Maine, and Illinois are also more than 6 months out of date. These states are quite significant, and except for Montana, were all polled within a two week period of each other- never before and never again. I think it's fair to say that the situation in these four states in November 2013 misrepresents the current state of the race. TBSchemer (talk) 20:05, 13 June 2014 (UTC)
Map Template
I was wondering if the original creator of the map could post the template used for striping the states, so that the community can update the maps more completely. The most similar blank, fully-striped version I can find in the Wikimedia archives is here: https://commons.wikimedia.org/wiki/File:Blank_US_Map,_striped2.svg
But of course, the stripes go in the opposite direction for that one. Not sure if anyone here cares that much about the aesthetics or not. TBSchemer (talk) 21:40, 21 July 2014 (UTC)
- I have added the map on wikimedia under the name "Blank map usa states.png". I hope that at this point it is still of some use. You can find it here: https://commons.wikimedia.org/wiki/File:Blank_map_usa_states.png — Preceding unsigned comment added by Leo Chapin III (talk • contribs) 03:41, 13 August 2015 (UTC)
Mitt Romney's map color designation
I think Rand Paul's and Mitt Romney's map color chosen to represent their poll numbers are too similar. They are both orange-ish. What if we try gray instead? (For Mitt Romney) Btw, I'm not an editor, so I can't change it. — Preceding unsigned comment added by 98.253.175.243 (talk) 01:24, 22 August 2014 (UTC)
Mitt Romney's States: Suffolk Poll Methodology Problem
Currently, two states on the map are labeled as Mitt Romney's, because of his victory in the second part of two-part Suffolk University polls. However, these Suffolk polls have a severe methodological problem, in that Part 1 gives the candidate list without Mitt Romney, and then Part 2 prompts the participants to consider switching their vote to Mitt Romney. It is no surprise, then, that Mitt Romney always wins in Part 2 of these Suffolk primary polls.
I thought it was for this reason that the Suffolk Part 2 polls were previously ignored when coloring the map and tallying the states. But it seems a few users have broken this convention and given North Carolina and Massachusetts to Romney. I do not believe it is scientifically justifiable to give him either state. Shouldn't we change it back? TBSchemer (talk) 22:31, 27 August 2014 (UTC)
- I agree they should be ignored, but I don't think they should be removed. Tiller54 (talk) 23:15, 4 September 2014 (UTC)
Unaccessible map
The map is unaccessible and should be removed. Way too difficult to read. Prcc27 (talk) 09:25, 10 November 2014 (UTC)
- I disagree. I think it's legible. TBSchemer (talk) 11:49, 15 November 2014 (UTC)
- I hold the same sentiment as TBSchemer, the map is easy to read and I also think it's easy to edit and update. Bullshark44 (talk) 13:40, 15 November 2014 (UTC)
- I agree. I feel like the striped state format isn't well suited to ties of more than three candidates. It becomes difficult to distinguish who and how many candidates are leading. Furthermore, there are so many candidates that several of the colors look to similar. I think the ultimate problem is that it's just too early for a map to be informative. Most of the polls on the page are many months old, making them basically useless. Even current polls are largely meaningless, because of how far away the primary is, and how undefined the field is. As candidates start declaring or declining this will become more meaningful, but that's still a ways off. ThrawnRocks (talk) 17:48, 24 November 2014 (UTC)
- I would also agree with removing the map for now, and adding it again later when we can agree that it's providing more useful information. It is pretty messy right now, and yes many polls are months old as well. —Torchiest talkedits 20:46, 24 November 2014 (UTC)
- Even if it's frustrating that the races are too divided to have any idea who might take the lead, I think that, in itself, is information worth of inclusion and visualization. This situation is rather distinct from what the Democratic Party has in the works for this round. The map shows that some states are much more divided than others, and some candidates are at much higher levels of support than others. Besides, we're already at the point that the races will consolidate fairly soon, now that the 2014 elections are over. TBSchemer (talk) 05:10, 25 November 2014 (UTC)
- When opened up, the map is much more accessible and very easy to read. TBSchemer's points are also very well made: it highlights the fractured Republican field but will settle down more once candidates actually make up their minds and more frequent polls are taken. Tiller54 (talk) 16:34, 25 November 2014 (UTC)
Mitt Romney Map
Add Mitt Romney on the map again?83.80.208.22 (talk) 12:54, 24 December 2014 (UTC)
- I think that would be appropriate. He's no longer being polled as the second question, "now who would you vote for if Romney were running as well" option. Although, this would only presently affect New Hampshire. Tiller54 (talk) 12:56, 24 December 2014 (UTC)
- I agree. The methodology problems I brought up earlier have been resolved in some recent polls. TBSchemer (talk) 01:38, 27 December 2014 (UTC)
- At present this would only result in one change - New Hampshire being marked for Romney whereas previously it was tied Christie/Paul/Bush. Tiller54 (talk) 03:14, 27 December 2014 (UTC)
- I agree. The methodology problems I brought up earlier have been resolved in some recent polls. TBSchemer (talk) 01:38, 27 December 2014 (UTC)
- I think we should add him now, because he is now with Publicly expressed interest83.80.208.22 (talk) 20:24, 6 January 2015 (UTC)
- Then is he the winner of: Arkansas, Colorado, Kansas, Massachusetts, Michigan and New Hampshire83.80.208.22 (talk) 11:54, 8 January 2015 (UTC)
- Source: http://www.reuters.com/article/2015/01/09/us-usa-politics-romney-idUSKBN0KI26Q2015010983.80.208.22 (talk) 12:09, 10 January 2015 (UTC)
- The question of whether to include him on the map has never been whether or not he's an "official" candidate. Most of the candidates included are not yet "officially running" or "exploring" or whatever they want to call it. The determining factor in whether or not Romney polls should be included in the map is whether Romney was included in the initial lineup of candidates suggested to the poll respondents, or if the respondents were given a no-Romney lineup, and then prompted to change their vote to Romney. The Suffolk polls listed for Arkansas, Colorado, Kansas, Massachusetts, and Michigan use the bad-methodology "prompting" tactic to include Romney after the fact, so Romney cannot be considered the winner of those states until we get better polls. On the other hand, the latest New Hampshire poll includes Romney in the initial lineup of candidates, and then prompts Romney voters to choose someone else. So in this case, the first lineup, that includes Romney and gives him the victory, is the correct poll to use. We can include both the initial candidate lineups and the prompted lineups in the poll listings (as we have already done), but the map should only reflect the results from the initial candidate lineups presented to poll respondents.
- Hence, the map should be adjusted to show Romney as the sole winner in New Hampshire, but at this time we should not change Arkansas, Colorado, Kansas, Massachusetts, or Michigan. TBSchemer (talk) 23:26, 10 January 2015 (UTC)
- Then we only change New Hampshire?83.80.208.22 (talk) 19:30, 11 January 2015 (UTC)
- Can anyone change it?83.80.208.22 (talk) 12:54, 14 January 2015 (UTC)
- What TBSchemer said. Tiller54 (talk) 19:55, 14 January 2015 (UTC)
- Although now Romney leads in Iowa too. Tiller54 (talk) 23:40, 16 January 2015 (UTC)
- What TBSchemer said. Tiller54 (talk) 19:55, 14 January 2015 (UTC)
polls
2 new Iowa. http://loras.edu/LorasCollege/files/d6/d69775e6-870f-465d-98fa-370cba8097b6.pdf http://www.desmoinesregister.com/story/news/politics/iowa-poll/2015/01/31/iowa-poll-walker-leads-tight-pack/22659477/83.80.208.22 (talk) 11:43, 1 February 2015 (UTC)
- Done. Tiller54 (talk) 12:05, 1 February 2015 (UTC)
- Iowa, New ham and South Carolina:
http://newscms.nbcnews.com/sites/newscms/files/iowa_february_2015_annotated_questionnaire_nbc_news-marist_poll.pdf http://newscms.nbcnews.com/sites/newscms/files/new_hampshire_february_2015_annotated_questionnaire_nbc_news-marist_poll.pdf http://newscms.nbcnews.com/sites/newscms/files/south_carolina_february_2015_annotated_questionnaire_nbc_news-marist_poll.pdf83.80.208.22 (talk) 17:25, 19 February 2015 (UTC)
- again
California Republican Presidential Primary http://www.field.com/fieldpollonline/subscribers/Rls2496.pdf
- Iowa Republican Presidential Caucus http://gravismarketing.com/uncategorized/iowa-poll-walker-garners-24-of-gop-support-paul-clinton-beats-walker-others-head-to-head/83.80.208.22 (talk) 17:21, 19 February 2015 (UTC)
- All done. Tiller54 (talk) 13:14, 21 February 2015 (UTC)
- Nevada http://gravismarketing.com/polling-and-market-research/current-nevada-polling-scott-walker-leads-bush-does-best-against-clinton/83.80.208.22 (talk) 11:39, 28 February 2015 (UTC)
- again http://gravismarketing.com/polling-and-market-research/florida-poll-jeb-walker-lead-in-gop-sunshine-state-poll/ and http://gravismarketing.com/polling-and-market-research/current-south-carolina-polling-republican-primary-polling-2016/83.80.208.22 (talk) 14:54, 4 March 2015 (UTC)
- http://gravismarketing.com/polling-and-market-research/current-new-hampshire-primary-political-poll/83.80.208.22 (talk) 15:22, 26 March 2015 (UTC)
- http://gravismarketing.com/polling-and-market-research/current-montana-polling-2/83.80.208.22 (talk) 11:21, 29 March 2015 (UTC)
- new NH http://www.bostonherald.com/sites/default/files/blog_posts/Franklin_Pierce_Herald_GOP_Poll.pdf83.80.208.22 (talk) 09:50, 31 March 2015 (UTC)
- SC http://gravismarketing.com/polling-and-market-research/south-carolina-political-polling-ted-cruz-bounce/145.52.142.104 (talk) 13:59, 31 March 2015 (UTC)
- NC http://www.nccivitas.org/wp-content/uploads/2015/03/GOP-poll-PR-xtabs-3-31-15.pdf83.80.208.22 (talk) 13:50, 1 April 2015 (UTC)
- Ohio, Florida and Pennsylvania http://www.quinnipiac.edu/news-and-events/quinnipiac-university-poll/2016-presidential-swing-state-polls/release-detail?ReleaseID=2182145.52.140.192 (talk) 11:16, 2 April 2015 (UTC)
- NJ http://www.quinnipiac.edu/news-and-events/quinnipiac-university-poll/new-jersey/release-detail?ReleaseID=221983.80.208.22 (talk) 13:18, 21 April 2015 (UTC)
- WV http://harperpolling.com/docs/default-source/default-document-library/hp-15-04-wv-gov-gop-toplines.pdf83.80.208.22 (talk) 19:42, 21 April 2015 (UTC)
- NH http://gravismarketing.com/polling-and-market-research/new-hampshire-poll-ayotte-opens-up-6-point-lead-walker-others-lead-clinton/83.80.208.22 (talk) 11:40, 27 April 2015 (UTC)
- Iowa http://www.quinnipiac.edu/news-and-events/quinnipiac-university-poll/iowa/release-detail?ReleaseID=222383.80.208.22 (talk) 11:36, 6 May 2015 (UTC)
- NH https://cola.unh.edu/sites/cola.unh.edu/files/research_publications/gsp2015_spring_gopprim050615.pdf83.80.208.22 (talk) 07:11, 7 May 2015 (UTC)
- Georgia http://landmarkcommunications.net/wp-content/uploads/2015/05/Georgia-Presidential-Primary-Poll.pdf83.80.208.22 (talk) 10:05, 20 May 2015 (UTC)
- CALIFORNIA http://www.field.com/fieldpollonline/subscribers/Rls2506.pdf83.80.208.22 (talk) 16:28, 25 May 2015 (UTC)
- NH http://gravismarketing.com/blogs/current-new-hampshire-republican-polling/ and NY http://www.quinnipiac.edu/images/polling/ny/ny06082015_n7p28rr.pdf83.80.208.22 (talk) 09:45, 11 June 2015 (UTC)
- SC http://morningconsult.com/wp-content/uploads/2015/06/150502_crosstabs_mc_SC_v1_AD.pdf83.80.208.22 (talk) 09:40, 17 June 2015 (UTC)
- NH http://morningconsult.com/wp-content/uploads/2015/06/150502_crosstabs_mc_NH_v1_AD.pdf83.80.208.22 (talk) 09:40, 17 June 2015 (UTC)
- Iowa http://morningconsult.com/wp-content/uploads/2015/06/150502_crosstabs_mc_IA_v1_AD.pdf83.80.208.22 (talk) 09:40, 17 June 2015 (UTC)
- Ohio, FL, PA http://www.quinnipiac.edu/images/polling/ps/ps06182015_Sk32gth.pdf83.80.208.22 (talk) 09:35, 23 June 2015 (UTC)
- NH http://www.realclearpolitics.com/docs/2015/6_23_for_posting_FINAL_NH_GOP_Marginals.pdf/ https://www.suffolk.edu/news/60149.php83.80.208.22 (talk) 20:01, 23 June 2015 (UTC)
Map
The map does not match with the polls, I think? — Preceding unsigned comment added by 83.80.208.22 (talk) 17:25, 19 February 2015 (UTC)
- Is there a particular state that you think needs updating? There's a consensus to not include those who have officially dropped out, so that accounts for some of the mismatch. Also, polls are added and maps are updated by different editors, so it takes time for one to catch up with the other. --Spiffy sperry (talk) 06:49, 20 February 2015 (UTC)
- Sorry my a fault it was a bug in my screen, stupid mobile :)83.80.208.22 (talk) 11:32, 20 February 2015 (UTC)
Illinois/Old Data
So Illinois's data is from November 22–25, 2013. Should we keep Illinois's colors on the map? I note this is the only piolling data for IL yet on this page. Perhaps a better question could be: How long do we keep state polling data on the map to keep it timely?98.253.175.243 (talk) 04:15, 22 April 2015 (UTC)
- I have no particular problem with keeping old data, but if there is a consensus for removal, I wouldn't object either. If I recall, previous election polling articles used lighter shades for old data, but with so many colors at this point, that method wouldn't work well. --Spiffy sperry (talk) 04:03, 2 May 2015 (UTC)
- Perhaps we could cross-hatch states with polls more than, say six months old with diagonal black lines, in the opposite direction to the colored stripes? Polls from before the beginning of 2015 are pretty useless at this point. I mean, come on, the map shows Sarah Palin sharing a state, and she's not even remotely going to run. It's worse to have misleading data than none at all. Antony–22 (talk⁄contribs) 22:23, 11 July 2015 (UTC)
- FYI, there are currently 24 states where the newest poll is from 2015, 12 from 2014, and 3 from 2013. That leaves 11 states, all 5 territories, and DC unaccounted for. Antony–22 (talk⁄contribs) 22:29, 11 July 2015 (UTC)
Utah Poll
The numbers for Utah's single poll have been changed numerous times now. The only relevant line from the linked article is "Utah Republicans mostly favor Jeb Bush (21%), followed by Scott Walker (14%) and Rand Paul (10%)." That leaves us with no information on other candidates, on the exact sample size, or the margin of error. Admittedly, this is a poor amount of data, but it is all we have. The other numbers reported in the linked article are the combined percentages for polled Republicans and Democrats, that is, the general population. The point of this Wikipedia article is Republican primaries, thus we include polling results of those who will vote in Republican primaries, not the general population. --Spiffy sperry (talk) 13:27, 12 May 2015 (UTC)
Sarah Palin is not a candidate
Sarah Palin is not running for President, so I don't think she should be on here. When Mitt Romney announced he was not running, his map color was removed. Tenor12 (talk) 05:20, 30 July 2015 (UTC)
Hideous map and other clarifications needed
![](http://upload.wikimedia.org/wikipedia/commons/thumb/7/77/Statewide_opinion_polling_for_the_Republican_Party_presidential_primaries%2C_2016.svg/400px-Statewide_opinion_polling_for_the_Republican_Party_presidential_primaries%2C_2016.svg.png)
First the lesser of the evils - the table needs some clarification on what is going on. It is not clear what the pink/red shading of the names indicates. At a glance it seems to indicate that the frontrunners are within a certain margin of error in terms of their respective leads, but if this is the case, and what the margin for cutoff is, is not mentioned anywhere in the article and thus renders the shading useless. With no context, the shading could mean virtually anything.
Secondly .. what gives with the map being used here? It is quite possibly the most hideous and useless infographic I have stumbled upon recently. No offense meant to those who are spending the time and effort maintaining it, but .. what on earth is going on? I presume that the intent is to show the current frontrunner in any given state, and that the striping is intended to communicate when frontrunner status is split among more than one candidate. If that is the case, it suffers from the same lack of context as the table - what is the margin required before one candidate is considered a frontrunner? What are the circumstances in which two or three candidates "share" frontrunner status but not the others? None of this information is offered anywhere and thus it makes the distinction useless and practically nonexistent. Also, at what point do you yield the point and say that there simply is no frontrunner rather than trying to show everyone who is "close"? I mean, dear god, California is a rainbow of 8 or 9 different colors. I cannot tell what is going on there. Even as someone with good color vision, I have a hard time distinguishing that there are 2 color stripes in Connecticut and Massachusetts. This map has serious accessibility issues for colorblind folks.
I must strongly suggest and propose that the map be remedied and propose the following:
- The map be replaced with a SVG version so as to be editable without the need of graphics software and for superior display. If help is needed creating an initial SVG file I would be happy to oblige, and they are easy to modify via text editors.
- An attempt should be made to select colors that are more accessible for colorblindness (though this may be a challenge due to the number of candidates)
- States with no recent polling data should be omitted (Maine and Wyoming are both from 2013 and thus useless)
- The map be modified to show only the frontrunner and no "shared" statuses:
- To qualify as a frontrunner a candidate should have a lead of at least some minimum percentage, say 5%
- A darker shade could be employed to show when a candidate has at least a lead of at least 10%
- If no candidate has a lead of at least 5%, instead of displaying a tie, a neutral "no clear frontrunner" color could be employed
Please consider the above or some variation on the above to make the situation clearer and more useful. Shereth 21:48, 5 August 2015 (UTC)
- SUPPORT: I am in support of a neutral color to indicate lack of a clear front runner, and also removing states whose polling is more than three months old. The sheer number of candidates makes striping inefficient as a manner of indicating shared frontrunner status, especially when it's four or five different candidates. At most, striping should be used for two candidates at most. Otherwise, the infographic becomes too cumbersome and fails to provide clear, useful information. --Vrivasfl (talk) 22:32, 5 August 2015 (UTC)
- I've thrown together a simple SVG map based on the latest polls as of this morning. I used the 5% cutoff, and didn't bother with shading to represent stronger leads (there's just too many colors). The map should now be visible above/to the right, along with a simple legend that could probably use some formatting. I've only included candidates that RealClearPolitics is currently tracking in their polling average and using their color scheme (with the exception of Fiorina, as they use light gray) as they are a fairly well known and neutral source. I've employed striping for 2 way ties but fall back to a generic color for 3 or more. I've also arbitrarily set a cutoff for polls for anytime in 2015 but that is of course easily modified. The source SVG file is also easily modified, and if we employ it we can create a set of simple instructions that anyone can follow to make easy changes to the map. It's still not perfect - Bush and Trump's colors are hard to tell apart in Florida, for example - but there's only so much you can do when dealing with over a dozen colors on one map. Still, I believe this map communicates the information much more clearly. Unless someone voices some reasonable opposition to using this in lieu of the rainbow-stripe disaster of a map currently in place I intend to replace it with this. Shereth 15:07, 7 August 2015 (UTC)
- I generally like this idea. It's similar to something I've been trying, except I use more striping for statistical ties. I particularly like the idea of using a single color for 3+ ties. However, I have two thoughts/concerns. First, I'm not sure it's a good idea to use an arbitrary 5% cutoff rather than the actual margin-of-error for each individual poll. It seems to skirt the Original Research line. Second, while I like the idea of using a color palette that another site uses, the Real Clear Politics colors are simply too close together. For instance, the Fiorina/Perry/Walker and Kasich/Trump combinations are difficult for me. For what it's worth, the colors that are currently used on the "hideous" map are close to the palette found here, with a few expections. --Spiffy sperry (talk) 20:36, 7 August 2015 (UTC)
- I originally thought about staring with the old web specification, but I remembered why it's been more or less abandoned - the high contrast is just hard on the eyes. Granted that's a subjective sort of thing, and of course the actual colors that we use are quite mutable. Perhaps the existing selections could be tweaked for better clarity? In any case we're kind of stuck between a rock and a hard place given the number of candidates, so we'll never actually find a "perfect" solution. Any discussion on the color palette to use would be welcome. I also tend to agree about the cutoff, as long as whoever is doing the maintenance on the map is willing to do the legwork and select shades based upon the actual margin of error in the polls (rather than a set amount) then that is probably better. Shereth 21:11, 7 August 2015 (UTC)
- I agree that the suggested map would be an improvement on the current map, and that the margin of error should determine statistical ties. As far as colors are concerned, I have no problem with the current color scheme in place and would rather not change it, but the colors is a conversation that we could continue to have separately. The map itself would be a huge improvement. After all, if someone really wanted to know which eight candidates were statistically tied in California, all they would have to do is scroll down. Rainbow map is entirely unhelpful.--Vrivasfl (talk) 21:49, 7 August 2015 (UTC)
- I don't mind switching to the current color scheme, looking at it it should be fine for the most part. The current map, however, has colors for candidates who are not running (ie. Palin) and lacks colors for some who are (ie. Fiorina). There are 17 declared candidates (and indeed 17 candidates who participated in yesterday's debates). Should we pick a color for each of them, even if they are not now, nor likely will be, on the map? Or shall we have some other minimum bar for inclusion? Shereth 23:19, 7 August 2015 (UTC)
- I think colors reserved for cadidates not running (like Palin) should be taken and reused for candidates currently unassigned a color (like Fiorina), but we can cross that bridge when we come to it. There is no need to assign Jim Gilmore a color right now, for example, and there's a pretty good change there never will be. Go forward with as it is now. --Vrivasfl (talk) 00:28, 8 August 2015 (UTC)
All right, I've finally gotten around to updating the map to use (where possible) the colors already in use in the extant PNG map. I attempted to make a judgement call on candidates who were not previously in the list. If anyone has issues with the specific colors in use (at any time) it would be relatively trivial to update them. I also switched to using the published margin of error for each poll, and where that is not available defaulted to a 5% margin to determine frontrunner status.
I created a tool that makes updating this map vastly simpler. I may look into getting it hosted on the toolserver since it is dedicated to use here, but in the meantime anyone can access the tool to create an updated version of the map. I am creating a short list of instructions that will be included at the top of this page on how to use the tool and how to update the map. Please provide any feedback on how I can make this process easier. Shereth 20:08, 12 August 2015 (UTC)
- Suggestion: Could we please switch the gray colors? I think it would make more sense for the states with no polling data to be light gray because dark gray grabs the reader's attention more and the "3 or more candidates statistically tied for the lead" states actually have polling data and should get more attention than then states without data. Prcc27 (talk) 03:54, 13 August 2015 (UTC)
- "Grabs the attention" is subjective. In my own subjective opinion a darker shade is indicative of something missing like polling data. However if there's a consensus for swapping the colors it'll be pretty easy to do. Shereth 04:02, 13 August 2015 (UTC)
- While you do have a point about my opinion being subjective, it would be more consistent to use light gray for "no data" since light gray is used for that purpose on several maps: the Democratic polling map (2016), the General election polling map (2012), the GOP polling map (2012) (example), and the public opinion for same-sex marriage in the United States map Prcc27 (talk) 07:35, 13 August 2015 (UTC)
- Good points. I've swapped out the grays. Shereth 13:58, 13 August 2015 (UTC)
- I'd like to suggest using stripes to denote three or maybe even four way ties between candidates. Grey makes sense for the ten-way ties in California or Colorado, but the map seems to be less helpful, at least in my opinion, with so many grey states. Also, as a side note, the map needs to be updated for the Arkansas poll, which was left off when the greys were switched. Leo Chapin III (talk) 15:18, 13 August 2015 (UTC)
- Arkansas has been updated, if it's still showing gray you might have to refresh. While I'm not strongly opposed to striping for 3 way ties, I don't think it's necessary and I feel that it's preferable to have the cutoff at two. Triple striping becomes visually cluttered, troublesome in smaller states, and by the time a user has sorted out which 3 stripes they are looking at and which 3 candidates it corresponds to, it's not really any more effort to see the gray and click down to the table to get the info. Shereth 15:32, 13 August 2015 (UTC)
- If that's the consensus, then it is what it is. My concern is that should Donald Trump decline in the polls, for example, the map might end up completely grey, and not terribly informative. Especially if we start using the statistical margin error as it is supposed to be used, as you brought up in the section below. (talk) 17:02, 13 August 2015 (UTC)
Glad to see someone finally took the time to update the map...I agree that the rainbow explosion was entirely unhelpful. If five or six candidates are that close together in the polls, then the real message of the poll is "nobody has a substantial lead". That said, the colors need changed. The pastels are making it look like an Easter egg. I understand the desire to assign every candidate a color and stick with that color through the entire primary season, but lets face it: some colors are more desirable than others. And the purple and brown aren't distinguishable enough from each other in a small state like MASS. Stick to primary colors where possible: red, blue, green, yellow. Then we can add in other colors as needed. Also, I recommend striping dark/light gray for states with more than two leaders. 2602:304:CE63:B160:8019:4A54:FBA0:1CA4 (talk) 19:45, 15 August 2015 (UTC)
Good proposal! The rainbow candy striping was insane and confusing. This map looks much better, altho the pink annoys me, but that doesnt matter.--Metallurgist (talk) 14:05, 23 August 2015 (UTC)
Margin of error
After reading a discussion on the margin of error for the Democratic primaries page [1] I realized that we are handling our math for margin of error incorrectly. For those who don't want to read the linked discussion, here is an example that gets the gist of it :
- A poll is released with a margin of error of 4%
- Candidate A polls at 37%
- Candidate B polls at 31%
The naive application of margin of error (which both our table and our map currently employ) say this is not a statistical tie, as the difference between candidates A and B is 6%, greater than the margin of error of 4%. However, the 4% means any candidate could be 4% higher or lower. So candidate A could actually be anywhere from 33%-41%, and B anywhere from 28%-34%. Thus there exists a scenario in the above hypothetical poll where candidate B is leading candidate A, 34% vs 33%.
In short: for a candidate's lead to be truly outside of the margin of error, they must be leading by at least twice the margin of error.
The way I see it we have two choices. One, we can conform to the proper definition of "margin of error" and "statistical tie", which will result in more of our map going gray (Kentucky, Massachusetts, Michigan, New Jersey and Connecticut become 3+ ties) or striped (Mississippi and Utah become 2 way ties) and alterations to the shading in the table. Alternatively we can leave the coloring and shading as-is and find some alternative wording as we are no longer displaying true statistical ties.
The data purist in me says that we should conform to the true definition and modify our table/the map to show true statistical ties. If there is a consensus to make that switch I can alter the logic in the map generator tool accordingly. Shereth 14:32, 13 August 2015 (UTC)
- I support using the true definition of statistical ties. So I would consider candidates A and B in a statistical tie in your scenario! Prcc27 (talk) 17:28, 13 August 2015 (UTC)
- Technically speaking, there is no true definition of statistical tie. It is a term that is not accepted amongst statisticians. See Margin of error. Nitroxium (talk) 19:10, 13 August 2015 (UTC)
- According to the following article (http://blogs.wsj.com/numbers/whats-a-statistical-tie-anyway-234/) statistical tie is used to describe when two numbers are within that number of margin of error or it isn't used at all. This article (http://www.motherjones.com/kevin-drum/2012/10/one-last-encore-great-statistical-tie-fallacy) also claims that the statistical tie is a myth, as does this article (http://www.washingtonmonthly.com/political-animal-a/2012_10/no_such_thing_as_a_statistical040271.php). I propose we get rid of statistical ties altogether. Nitroxium (talk) 19:22, 13 August 2015 (UTC)
- In case there is further doubt, Nate Silver (Founder of FiveThirtyEight) also has written against the use of the term "statistical tie". https://twitter.com/natesilver538/status/628950531172638721 Nitroxium (talk) 19:30, 13 August 2015 (UTC)
- @Nitroxium: Let me get this straight. You're proposing that we ignore the margin of error? So if the margin of error is 4% and Candidate A has 50% support and Candidate B has 49% support you think we should just color a state in Candidate A's favor? That makes no sense! Prcc27 (talk) 19:50, 13 August 2015 (UTC)
- Did you read the articles? I have provided actual sources showing the myth of "statistical ties". Nitroxium (talk) 20:07, 13 August 2015 (UTC)
- @Nitroxium: Okay... so when do the margin of errors come in to play? Are we every going to use them at all or from here on out are we going to ignore them? Also, the opinion of the people in your sources might be a different point of view than that of other sources. We shouldn't give undue bias to either. Prcc27 (talk) 20:27, 13 August 2015 (UTC)
- If there are statisticians who agree with the use of the "statistical tie", then by all means post them. Otherwise, there seems to be a consensus on it being a myth. Nitroxium (talk) 20:29, 13 August 2015 (UTC)
- You still aren't answering my first question. When do the margin of errors come in to play; are we going to use them at all? Prcc27 (talk) 20:45, 13 August 2015 (UTC)
- If there are statisticians who agree with the use of the "statistical tie", then by all means post them. Otherwise, there seems to be a consensus on it being a myth. Nitroxium (talk) 20:29, 13 August 2015 (UTC)
- @Nitroxium: Okay... so when do the margin of errors come in to play? Are we every going to use them at all or from here on out are we going to ignore them? Also, the opinion of the people in your sources might be a different point of view than that of other sources. We shouldn't give undue bias to either. Prcc27 (talk) 20:27, 13 August 2015 (UTC)
- Did you read the articles? I have provided actual sources showing the myth of "statistical ties". Nitroxium (talk) 20:07, 13 August 2015 (UTC)
- @Nitroxium: Let me get this straight. You're proposing that we ignore the margin of error? So if the margin of error is 4% and Candidate A has 50% support and Candidate B has 49% support you think we should just color a state in Candidate A's favor? That makes no sense! Prcc27 (talk) 19:50, 13 August 2015 (UTC)
- In case there is further doubt, Nate Silver (Founder of FiveThirtyEight) also has written against the use of the term "statistical tie". https://twitter.com/natesilver538/status/628950531172638721 Nitroxium (talk) 19:30, 13 August 2015 (UTC)
- According to the following article (http://blogs.wsj.com/numbers/whats-a-statistical-tie-anyway-234/) statistical tie is used to describe when two numbers are within that number of margin of error or it isn't used at all. This article (http://www.motherjones.com/kevin-drum/2012/10/one-last-encore-great-statistical-tie-fallacy) also claims that the statistical tie is a myth, as does this article (http://www.washingtonmonthly.com/political-animal-a/2012_10/no_such_thing_as_a_statistical040271.php). I propose we get rid of statistical ties altogether. Nitroxium (talk) 19:22, 13 August 2015 (UTC)
- Technically speaking, there is no true definition of statistical tie. It is a term that is not accepted amongst statisticians. See Margin of error. Nitroxium (talk) 19:10, 13 August 2015 (UTC)
- Having read the linked articles, I don't think the argument is so much that the concept of the statistical tie is a myth, but referring to it as such is misleading since it implies something that is not quite the truth. So we shouldn't call it a statistical tie. I get that. But I think Prcc27's question still needs to be answered : if we are going to concede that a "statistical tie" does not exist, how do we intend to handle situations when two or more candidates are close enough in polling that there exists a statistically significant probability that any one of those candidates might actually be in the lead? Shereth 21:44, 13 August 2015 (UTC)
- The problem is that with statistics, there is not only a margin of error but also level of confidence. According to the Mother Jones article, it is actually improbable for the person in second place to actually be in the lead. On that basis, there should be no assumption on our part that there is a "statistical tie" or that the second place candidate could be in first place, due to the less than 50% probability of it. If we were to base ourselves on less than 50% probabilities, then due to levels of confidence, polls could be completely off and in one way or another, every candidate has a probability of actually being in first place. See Confidence interval. Nitroxium (talk) 12:14, 14 August 2015 (UTC)
- By definition it is always improbable that a second place polling will be in the lead. I don't think anyone has any misconceptions about that; if there were a poll where it was probable, that would be an inherently useless poll. The whole point of margins of error is not to show that something is probable but that there exists a statistically significant chance that the result is wrong. And, yes, technically there exists some probability that any given candidate might actually be in first place - there is never a 100% probability of a sample being correct - but all we are talking about is statistical significance. What we end up with are 2 possible outcomes to this scenario.
- We can continue to show striping/gray states for polls where two or more candidates polling results fall within the poll's margin of error. Rather than use loaded language like "statistical tie", we word this as simply and neutrally as possible. Something to the effect of "Striping indicates two results are within the poll's margin of error" and "3 or more results are within the poll's margin of error" for the gray color. It is then left to the reader to draw any conclusions.
- We can eliminate the concept of stripes and gray states, except in the case of true ties, and concede the probability is great that the leader in the poll is, in fact, the current frontrunner in a given state.
- Technically there are other options, such as eliminating the map altogether or switching to an "average of polls" method to get around the margin of error issue.
- Personally, I fall on the side of keeping the indicator for the margin of error. I believe it is useful information, and so long as it is put in a way that is neutral and factually correct, I think it should stay. Shereth 14:26, 14 August 2015 (UTC)
- Having read the linked articles, I don't think the argument is so much that the concept of the statistical tie is a myth, but referring to it as such is misleading since it implies something that is not quite the truth. So we shouldn't call it a statistical tie. I get that. But I think Prcc27's question still needs to be answered : if we are going to concede that a "statistical tie" does not exist, how do we intend to handle situations when two or more candidates are close enough in polling that there exists a statistically significant probability that any one of those candidates might actually be in the lead? Shereth 21:44, 13 August 2015 (UTC)
- What if we just give the highest percentage the state and just word it in a way that shows the highest percentage in that state is candidate X, irrespective of margin of errors or probabilities. Bullshark44 (talk) 14:02, 14 August 2015 (UTC)
- that new map is horrible94.211.104.84 (talk) 16:35, 14 August 2015 (UTC)
- I'd like to point out that the 2012 General Election polling map used margin of error. And they used margin of error the right way (a candidate must lead by twice the margin of error to actually be leading). I think we should do the same so we don't ignore the margin of errors. Prcc27 (talk) 17:37, 14 August 2015 (UTC)
- We either need to fully implement the margin of errors or not use them at all. The status quo is inconsistent! Prcc27 (talk) 05:38, 15 August 2015 (UTC)
- There's no hurry. Let's give people ample time to discuss. Shereth 15:27, 15 August 2015 (UTC)
- Actually, the statistical tie term refers to when they are in a lead that is less than the number of the margin of error, not what Prcc is saying. The Wall Street Journal article makes this clear. While not an actual statistical term, it is a term that has been used by journalists in this sense. So either we don't have the margins of error or we have them only when the lead is within the margin of error, as shown in the WSJ article. Another example of this is what Shereth has shown, the past article for 2012 republican primaries. Nitroxium (talk) 21:11, 15 August 2015 (UTC)
- In the 2012 GOP polling pages, the general consensus was to count the margin of error as the maximum difference to be considered a tie, not double the MoE. That seems to match up with what sources presented here have defined as a statistical tie, so I'm fine with doing it that way again this go round. —Torchiest talkedits 17:53, 20 August 2015 (UTC)
- I'm also fine with Shereth's idea about mentioned being within the MoE and not referring to the controversial term statistical tie at all. —Torchiest talkedits 17:58, 20 August 2015 (UTC)
- @Torchiest: From one of the sources: "The sampling margin of error applies to each measurement of the poll. That means that a difference between two candidates that is less than twice the margin of error is considered statistically insignificant." Prcc27 (talk) 05:36, 21 August 2015 (UTC)
- I'm also fine with Shereth's idea about mentioned being within the MoE and not referring to the controversial term statistical tie at all. —Torchiest talkedits 17:58, 20 August 2015 (UTC)
- In the 2012 GOP polling pages, the general consensus was to count the margin of error as the maximum difference to be considered a tie, not double the MoE. That seems to match up with what sources presented here have defined as a statistical tie, so I'm fine with doing it that way again this go round. —Torchiest talkedits 17:53, 20 August 2015 (UTC)
- Actually, the statistical tie term refers to when they are in a lead that is less than the number of the margin of error, not what Prcc is saying. The Wall Street Journal article makes this clear. While not an actual statistical term, it is a term that has been used by journalists in this sense. So either we don't have the margins of error or we have them only when the lead is within the margin of error, as shown in the WSJ article. Another example of this is what Shereth has shown, the past article for 2012 republican primaries. Nitroxium (talk) 21:11, 15 August 2015 (UTC)
- There's no hurry. Let's give people ample time to discuss. Shereth 15:27, 15 August 2015 (UTC)
Vermont
The Vermont poll has Democrats in it too so should we remove it..? Prcc27 (talk) 03:44, 19 September 2015 (UTC)
- The problem is that the poll asked "Who do you hope to see elected President in 2016?" (broken down by party affiliation) rather than "Who will you vote for in the (Democratic/Republican) primary?" Since Vermont has an open primary, afaik the Republican supporters would vote in the Democratic primary. So I'm not sure how we should include the poll either here or on the Democratic article. SPQRobin (talk) 18:18, 21 September 2015 (UTC)
California
California should be shaded in pink because Trump is leading in the moest recent polls. --77.176.43.141 (talk) 22:20, 23 October 2015 (UTC)
- By 2 percentage points, in a poll with a margin of error of 7 percentage points. The map is correctly shaded gray. --Spiffy sperry (talk) 15:55, 26 October 2015 (UTC)
New Hampshire
I changed New Hampshire to statistical tie between Trump and Carson before realizing there are two polls that were conducted on the exact same days. One of them says Trump is ahead, the other says there's a statistical tie. How should we color the map..? Prcc27 (talk) 03:58, 5 November 2015 (UTC)
- I'd say leave it for Trump, the one with Trump winning over the MoE has slightly more weight because it has a somewhat larger sample size and most previous ones have Trump doing the same. If the one with him in a tie with Carson had the larger sample size then it would need to be striped.ShadowDragon343 (talk) 04:36, 5 November 2015 (UTC)
- I guess, but it seems ridiculous because the larger poll is only larger by ten people. We need to come up with a better system especially since this could happen again. If the larger poll found that Trump and Carson were tied and the poll with 10 less people found that Trump was ahead- I'd be reluctant to color it as a tie. We know from these polls that Trump is either tied or ahead while Carson is either tied or in 2nd place. But I still think we should find a way to reflect both polls.. Prcc27 (talk) 18:05, 5 November 2015 (UTC)
- Iowa also needs to switched back to Trump; according to the last poll. --77.178.100.200 (talk) 19:54, 5 November 2015 (UTC)
- I guess, but it seems ridiculous because the larger poll is only larger by ten people. We need to come up with a better system especially since this could happen again. If the larger poll found that Trump and Carson were tied and the poll with 10 less people found that Trump was ahead- I'd be reluctant to color it as a tie. We know from these polls that Trump is either tied or ahead while Carson is either tied or in 2nd place. But I still think we should find a way to reflect both polls.. Prcc27 (talk) 18:05, 5 November 2015 (UTC)
Order of candidates in the map key
The order of the candidates in the map key has, until this week, always been in order of the most states ahead (even last election cycle). Why did this change, and was there a discussion about it? Light-jet pilot (talk) 17:24, 23 December 2015 (UTC)
- It was alphabetical, with grays at the top, for three months from Aug. 12 (when the new map was implemented) to Nov. 11. I have no strong preference for either alphabetical or numerical, as long as the grays are together at the top or bottom of the list. --Spiffy sperry (talk) 18:20, 23 December 2015 (UTC)
- Oh, ok. I like the grays at the bottom. Personal preference: I do think the candidates look better in order of most states. But I was just wondering why, that's all. Light-jet pilot (talk) 18:45, 23 December 2015 (UTC)
Referencing
Is it just me or do some editors refuse to add the title, dates and proper links when referencing polls in this article? DrFargi (talk) 23:52, 3 January 2016 (UTC)
- Take a minute to scroll down the whole page. Then go back to the Statewide opinion polling for the Republican Party presidential primaries, 2012 page. They are done that way intentionally, and have been for years. I don't actually know if that is consensus or not, but it is the way ALL of them are done. I am not against doing it your way, but the page needs uniformity, so all or none in my opinion. Light-jet pilot (talk) 06:47, 4 January 2016 (UTC)
- Do it the correct way. If there is a consensus, it would have been discussed in this talk page. If you need help with referencing, look up this page: WP:REF.DrFargi (talk) 07:38, 4 January 2016 (UTC)
- I know how to ref properly. But in most cases, these polls do not fall under the category of a proper reference. They are direct links to PDF files. And just because it's not on this talk page doesn't mean there is not consensus. The only reason I referenced "consensus" is because you seem to be the only one objecting. You are also, as far as I can tell, the only editor to do it that way in the last 5 years. Hundreds of editors vs. one is pretty good consensus. Did you look at the page I linked? This page when it was created and four months ago, and the equivalent pages from last election cycle (2012). My point being, out of the thousands of edits since, you are the only one doing it your way. Therefore, we are already doing it the 'correct' way. Light-jet pilot (talk) 19:38, 4 January 2016 (UTC)
- Well if you look at hundreds of articles in Wikipedia, we should be doing it the right way. By the way did you look at the Democratic state polls - Statewide opinion polling for the Democratic Party presidential primaries, 2016, which is properly referenced. Using PDF is not to be used as an excuse for the way referencing is done on this page.DrFargi (talk) 05:27, 5 January 2016 (UTC)
- Do it the correct way. If there is a consensus, it would have been discussed in this talk page. If you need help with referencing, look up this page: WP:REF.DrFargi (talk) 07:38, 4 January 2016 (UTC)
Reformat tables
@173.52.18.146: Made a good point at the talk page for the Democratic polling page about reformatting the tables. I think it would be a great way to get rid of all of the unnecessary state sections on both articles. Btw, the maps to the left have nothing to do with this section. Here is what they said/proposed (obviously we would be using GOP candidates for this page instead of the Democratic ones):
“ | I don't understand why the tables are split up by state and don't have the following format allowing the user to either sort by state, date, or polling source. Am I missing something? | ” |
State | Date(s) Conducted | Date Published | Polling Source | MOE | Sample Size | First | Second | Third | Other |
---|---|---|---|---|---|---|---|---|---|
Alabama | 2016-02-02 | Overtime Politics | 5.1 | 385 | Hillary Clinton 54% | Bernie Sanders 43% | Martin O'Malley 1% | undecided 2% | |
Alabama | 2015-08-11 | News-5/Strategy Research | 2 | 3,500 | Hillary Clinton 78% | Bernie Sanders 10% | |||
Alaska | 2016-01-23 | Alaska Dispatch News/Ivan Moore Research | 3.8 | 651 | Bernie Sanders 48% | Hillary Clinton 34% | Martin O'Malley 6% | undecided 14% | |
Arizona | 2015-11-05 | Behavior Research Center | 7.3 | 186 | Hillary Clinton 47% | Bernie Sanders 19% | Martin O'Malley 2% | uncommitted 32% | |
Arkansas | 2015-12-23 | Overtime Politics | 5.2 | 209 | Hillary Clinton 47% | Bernie Sanders 42% | Martin O'Malley 4% | undecided 7% | |
Colorado | 2015-12-17 | Overtime Politics | 319 | Hillary Clinton 49% | Bernie Sanders 36% | Martin O'Malley 4% | undecided 11% |
Prcc27 (talk) 07:03, 12 February 2016 (UTC)
- The only issue I have with a combined table with all the states is that the it will be very unwieldy with polls from 2013, 2014, 2015 included. The tables as they are should remained with the different states allocated, as and when each caucus or primary results appear too.DrFargi (talk) 13:54, 13 February 2016 (UTC)
- I don't know what you mean. The table is sortable by date which means you can sort it so then the one's from 2013 would be at the very bottom. There is already a page for results so I really wouldn't mind getting rid of the results altogether anyways. Prcc27 (talk) 20:34, 13 February 2016 (UTC)
I think its great to have the easy to access state sections. This is about statewide polling, which means people come looking for specific states. A sortable table is great, but if I want Missouri, theres no easy way to sort it for that except as it is. Your proposal is a good idea, however, it would make things more confusing and complicated here.--Metallurgist (talk) 19:16, 23 February 2016 (UTC)
Six months?
I understand having six months of polling accepted before the primaries begin, but now we are in full swing and a lot has changed since many of these polls. I think it would be better to limit the polling window for the map to three months at most.--Metallurgist (talk) 19:14, 23 February 2016 (UTC)
- Support: polls that are 6 months old are too old. Prcc27💋 (talk) 01:43, 24 February 2016 (UTC)
- Support. I was thinking this myself the other day. Six-month-old polls are useless at this time. 75.164.96.208 (talk) 05:10, 24 February 2016 (UTC)
- Support Many polls showing Carson in the lead are from back when he was peaking, these states have since not been polled and he is far lower nationwide, it is very likely Ted Cruz or Trump is leading them nowShadowDragon343 (talk) 07:03, 24 February 2016 (UTC)
- Three months: In the last six months, Carson has crashed, twelve candidates have withdrawn, and four states have held primaries. pbp 14:17, 25 February 2016 (UTC)
Looks like its been done. Thanks to whoever!--Metallurgist (talk) 17:55, 2 March 2016 (UTC)
Primary dates and ultimate winners
I think it would be useful to note the primary date after the subheaders. I would do it, but I dont have the time to devote to it, so if someone is bored, heres a suggestion. Also, it might be worth noting the ultimate vote count at the top of every table. This has been traditional on Wikipedia with opinion polling.--Metallurgist (talk) 17:57, 2 March 2016 (UTC)
2 Big problems with the main map
I have 2 big problems with the main map.
1. Not clearly differentiating between those contests that are already finished versus those that are upcoming or worse ongoing. It also leads to the uncomfortable question of what to do when someone wins who the polls said wouldn't.
2. The map was showing several states colored Trump's color when there was a statistical tie. I'll need someone who's actually done polls to confirm but I believe to actually say someone is winning you need double the margin of error, which for most of these polls would mean winning by about 10 points.
Ultimately Wikimedia is a media organization. And while TV and most other forms of media have veered far into fringe territory, we can at least do better. Polloromantico (talk) 23:54, 26 March 2016 (UTC)
- I'm not quite sure what you mean by the second point, but I certainly agree with you on the first point. I remember after Iowa that the state was shaded mixed between Trump and Cruz. Someone tried to change it to show all Cruz, and it was then reverted. I concur that there should be something to indicate what states have already voted. (Actually, Polloromantico, this has already been discussed. See above in the "Map to be Updated?" section.) Display name 99 (talk) 03:56, 6 April 2016 (UTC)
Shading and margins of error
If the second place poll-getter is within, twice the margin of error, then they need to be shaded as well. That means a lot of the shading on this article needs to be fixed. For example, in the latest Wisconsin poll, Trump is 42% and Cruz is 32%, but the margin of error is 5% - this means Trump is not ahead of Cruz. I have fixed that one, and tagged the article for the others. StAnselm (talk) 19:04, 5 April 2016 (UTC)
- It's never been like that, always been within one MoE along with the national and democratic pages. ShadowDragon343 (talk) 19:18, 5 April 2016 (UTC)
- Well, it doesn't matter what the practice on these pages has been, if it is incorrect usage, then it should be changed. The margin of error articles says "The terms "statistical tie" and "statistical dead heat" are sometimes used to describe reported percentages that differ by less than a margin of error, but these terms can be misleading. For one thing, the margin of error as generally calculated is applicable to an individual percentage and not the difference between percentages, so the difference between two percentage estimates may not be statistically significant even when they differ by more than the reported margin of error." StAnselm (talk) 21:51, 5 April 2016 (UTC)
- Unfortunately, your reasoning of twice the margin of error is incorrect. We shouldn't be fixated on the difference of percentages and reinterpret what may be statistically significantly or not. So as in the past practice, this is the correct usage. The only issue I have are that of the referencing.DrFargi (talk) 05:50, 6 April 2016 (UTC)
- Well, certainly referencing is a big part of it. We can only add shading if there is a reliable source saying that someone as "won". StAnselm (talk) 07:52, 6 April 2016 (UTC)
- The bit I mentioned about referencing is that about inserting the author, website and date. Currently all polls on this table are from reliable sources as they are derived from website polling aggregators Real Clear Politics and Huffington Post. We have removed one pollster Overtime politics because of its unreliability. So its not our call on saying who has won or who hasn't. The current system is as it should be and shouldn't be change just because someone has reinterpreted the margin of error for a particular pollster.DrFargi (talk) 08:15, 6 April 2016 (UTC)
- So who's call is it? When we do have shading, on what basis do we have it? StAnselm (talk) 18:42, 6 April 2016 (UTC)
- Like I said before we use the margin of error that is provided by the particular poll. If any second placed poll getter's figures falls within that margin of error, then we shade that figure. We don't inflate or reinterpret that margin of error because that's original research.DrFargi (talk) 16:57, 7 April 2016 (UTC)
- There is nothing inflated. As it says on the margin of error Wikipedia page, "the margin of error as generally calculated is applicable to an individual percentage and not the difference between percentages". In a hypothetical scenario, if Trump had 41% and Cruz had 34%, with a margin of error is 5%, Trump's score ranges from 46% to 36%, while Cruz's score ranges from 39% to 29%. While Trump leads Cruz by 7% in the poll, when the margin of error is applied, Cruz could actually be leading by 3%. Therefore, should they not both be shaded? — Preceding unsigned comment added by 184.146.48.95 (talk) 16:12, 9 April 2016 (UTC)
- Again that is an assumption that the 2nd placed poll-getter's range will increase his or her percentages because of the margin of error. If going by that you also have to increase the first place poll-getter's numbers too. Then it doesn't fall into the margin of error as you say. You also have to take into account that there percentages for 3rd placed poll-getter and undecideds and others. The shading system should stay as it is and remain so.DrFargi (talk) 03:33, 10 April 2016 (UTC)
- Unfortunately, your reasoning of twice the margin of error is incorrect. We shouldn't be fixated on the difference of percentages and reinterpret what may be statistically significantly or not. So as in the past practice, this is the correct usage. The only issue I have are that of the referencing.DrFargi (talk) 05:50, 6 April 2016 (UTC)
- Well, it doesn't matter what the practice on these pages has been, if it is incorrect usage, then it should be changed. The margin of error articles says "The terms "statistical tie" and "statistical dead heat" are sometimes used to describe reported percentages that differ by less than a margin of error, but these terms can be misleading. For one thing, the margin of error as generally calculated is applicable to an individual percentage and not the difference between percentages, so the difference between two percentage estimates may not be statistically significant even when they differ by more than the reported margin of error." StAnselm (talk) 21:51, 5 April 2016 (UTC)
Where is Delaware? and the Dakotas?
Where is Delaware? and the Dakotas? — Preceding unsigned comment added by 79.19.144.33 (talk) 13:57, 11 April 2016 (UTC)
- No major polling outfit has polled those states, as far as I'm aware. Delaware's primary is on April 26, at which time its results will be added. South Dakota's primary is on June 7, at which time its results will be added. If either state is polled before their primary, those results will be added. North Dakota will not have a Republican candidate preference vote, so it is likely that that section will not be created (although someone can create a section to explain that if they wish). --Spiffy sperry (talk) 14:41, 11 April 2016 (UTC)
- Thank you for the reply — Preceding unsigned comment added by 79.19.144.33 (talk) 19:36, 11 April 2016 (UTC)
Map to be updated?
In Statewide opinion polling for the Republican Party presidential primaries, 2012, as each state completed its caucus or primary, the map would be colored in for that candidate with a star placed on the state to indicate that it was an actual result and not a polling projection. Is this going to be done again this year? (I don't know how to edit the map myself.) --Metropolitan90 (talk) 13:26, 2 February 2016 (UTC)
- I added a note on Commons that says "The Iowa caucuses have already been completed." Prcc27 (talk) 03:54, 3 February 2016 (UTC)
- I don't think that's accurate if you are trying to differentiate the actual results and the last poll.
![](http://upload.wikimedia.org/wikipedia/commons/thumb/7/77/Statewide_opinion_polling_for_the_Republican_Party_presidential_primaries%2C_2016.svg/400px-Statewide_opinion_polling_for_the_Republican_Party_presidential_primaries%2C_2016.svg.png)
![](http://upload.wikimedia.org/wikipedia/commons/thumb/a/a6/Republican_Party_presidential_primaries_results%2C_2016.svg/400px-Republican_Party_presidential_primaries_results%2C_2016.svg.png)
The state opinion polling map should have Iowa should reflect the last poll, not the results of the caucus. DrFargi (talk) 14:48, 5 February 2016 (UTC)
- I was one of the editors who came up with the star system last cycle, and it is one of the worst things that I've ever done. I think it should just reflect the most recent poll before the actual election. 168.16.182.210 (talk) 21:01, 10 February 2016 (UTC)
- Really? I think that the star system is actually a good idea. Once the election has taken place, the final polling is of much less interest than it had been before the election. I hope that we can go back to the star system, rather than leaving Iowa in a stasis between Trump and Cruz. --Metropolitan90 (talk) 19:57, 14 February 2016 (UTC)
- Could the stars be added without interfering with us using a text editable map? Prcc27💋 (talk) 20:08, 14 February 2016 (UTC)
- That would be great because then we can add a "States won" number with Trump having 3 and Cruz having 1ShadowDragon343 (talk) 11:54, 25 February 2016 (UTC)
- First, kudos to all for creating a terrific site for this information. I very much like the map, but I agree that I would like to see the states colored in with the result of the final caucus or vote results. Even if I accept the premise that the map should show only the most recent polling results, well then can't we consider the actual vote a polling of 100% of definite voters with a 0% margin of error? Rrm151 (talk) 17:18, 30 March 2016 (UTC)
- That would be great because then we can add a "States won" number with Trump having 3 and Cruz having 1ShadowDragon343 (talk) 11:54, 25 February 2016 (UTC)
- Could the stars be added without interfering with us using a text editable map? Prcc27💋 (talk) 20:08, 14 February 2016 (UTC)
- Really? I think that the star system is actually a good idea. Once the election has taken place, the final polling is of much less interest than it had been before the election. I hope that we can go back to the star system, rather than leaving Iowa in a stasis between Trump and Cruz. --Metropolitan90 (talk) 19:57, 14 February 2016 (UTC)
- I was one of the editors who came up with the star system last cycle, and it is one of the worst things that I've ever done. I think it should just reflect the most recent poll before the actual election. 168.16.182.210 (talk) 21:01, 10 February 2016 (UTC)
- I agree with Metropolitan90. The ultimate opinion poll is the actual primary or caucus result. The other polls only matter as a predictor for the actual result. For example, if the day before the New York primary some poll comes out - and turns out to be the last one before the primary - that happens to show Kasich leading, it would be silly to recolor the map green if Trump actually wins the primary. Once the primary/caucus has occurred the map should reflect the winner. Rlendog (talk) 15:17, 15 April 2016 (UTC)
Map changes as contests are completed
On the polling map, Prcc27 has removed the coloring for Minnesota and Alaska, as well as the footnote I wrote for the light gray color. Prcc27 feels that these polls are old and should no longer be reflected on the map. I respectfully disagree. At this rate, the entire polling map will be a useless gray by October 2016, if we continue to remove old polls after the primaries are over. For the 2012 contest, we didn't have separate polling and results maps, so we didn't have this problem. Instead, the polling and results were reflected in a single map (polling for states that hadn't voted, results for states that had voted). I think the 2016 polling map would be more useful if it reflected the most recent polling prior to each state's contest. So, if a poll is older than 3 months (the current cutoff), but was within 3 months of that state's contest, then the candidate(s) coloring should remain.
The other objection to this method was that it doesn't match the Democrat polling page. I agree, and propose that both pages should use this method instead. --Spiffy sperry (talk) 14:30, 26 April 2016 (UTC)
- The current consensus is to color states light gray when the polls are older than 3 months old. Anything older doesn't matter at this point. Prcc27🌍 (talk) 17:00, 26 April 2016 (UTC)
- That was the consensus before voting started. We are in a different phase now. Your position only makes sense if we plan to remove the polling map when the primary is completed. My position is that we have a polling map and a results map, and we should be able to quickly compare the two to see how the polls stacked up against the results. This is a page about polling, and it does matter to have a helpful image related to polling, not an image of all gray states (which is where we are headed with your method). And based on the edit by Jvikings1 after you first changed Minnesota, I'm not alone in this opinion. The Minnesota and Alaska polls are not recent, but they are not out of date. --Spiffy sperry (talk) 18:42, 26 April 2016 (UTC)
- Actually, this consensus was reached after the first three states had already voted. You can see the discussion here. Prcc27🌍 (talk) 23:15, 26 April 2016 (UTC)
- These polls are not outdated because the state has already passed, so there will be no recent polling. Maybe the wording should be no polling from 3 months before the election. Because, these polls are accurate for the time that they voted.Jvikings1 (talk) 00:05, 27 April 2016 (UTC)
- This article isn't meant to "predict" the winner in a contest. The article's purpose is to show what people's opinion of the candidates are in each state based on recent polls irregardless of the election outcomes. On the Public opinion of same-sex marriage in the United States article we colored states in based on recent opinions in those states irregardless of whether same-sex marriage was already legal there or not. Since the polls in MN and AK are older than 3 months they don't meet the "recent" criteria which maps on public opinion articles use. And if a new poll was to come out from IA we would update the map to reflect that even though IA already voted. FYI, please try not to vote stack as it could unfairly skew this discussion. Prcc27🌍 (talk) 04:42, 27 April 2016 (UTC)
- These polls are not outdated because the state has already passed, so there will be no recent polling. Maybe the wording should be no polling from 3 months before the election. Because, these polls are accurate for the time that they voted.Jvikings1 (talk) 00:05, 27 April 2016 (UTC)
- Actually, this consensus was reached after the first three states had already voted. You can see the discussion here. Prcc27🌍 (talk) 23:15, 26 April 2016 (UTC)
- That was the consensus before voting started. We are in a different phase now. Your position only makes sense if we plan to remove the polling map when the primary is completed. My position is that we have a polling map and a results map, and we should be able to quickly compare the two to see how the polls stacked up against the results. This is a page about polling, and it does matter to have a helpful image related to polling, not an image of all gray states (which is where we are headed with your method). And based on the edit by Jvikings1 after you first changed Minnesota, I'm not alone in this opinion. The Minnesota and Alaska polls are not recent, but they are not out of date. --Spiffy sperry (talk) 18:42, 26 April 2016 (UTC)
- The purpose of this article is to show opinion polling of each state prior to the state's primary or caucus. The "recent" criteria should be determined in terms of the date of the state's primary or caucus. There will be no reason to poll a state on this issue after the primary or caucus, so no polling firm will perform one. Once again, I do not think it is useful to display a fully gray map, which is what will happen if we continue to remove colors from states with completed contests. --Spiffy sperry (talk) 17:19, 27 April 2016 (UTC)
- Determining "recency" based on the primary or caucus itself makes no sense. This article deals with opinion polling, not primaries or caucuses. We have a separate map for that. Prcc27🌍 (talk) 06:27, 28 April 2016 (UTC)
- It deals with opinion polls of primaries and caucuses. The other map shows the winner of the states after the election is completed. These are showing what the polls are like coming up to that state's election. So, if there is a poll a week before the election, then that polls is not outdated. And, there will be no recent polls in the states that passed because that would be useless and would just waste money.Jvikings1 (talk) 12:33, 28 April 2016 (UTC)
- It might be WP:CRYSTAL to assume that. Prcc27🌍 (talk) 22:56, 28 April 2016 (UTC)
- What am I assuming that is predicting the future? It is common sense that a state would not have a poll if the primary/caucus has passed. Why? Because it does not help anyone. The way that you are prosposing will just lead to a blank map that is completely usless.Jvikings1 (talk) 01:48, 29 April 2016 (UTC)
- Polls are not always conducted for remaining primaries/caucuses i.e. national polls include people from states that have already voted. Prcc27🌍 (talk) 02:17, 29 April 2016 (UTC)
- This article and map is for statewide polling. Pollsters will not do individual states(which this is about) after the election passes.Jvikings1 (talk) 21:56, 29 April 2016 (UTC)
- The purpose of this article is to show opinion polling of each state prior to the state's primary or caucus. The "recent" criteria should be determined in terms of the date of the state's primary or caucus. There will be no reason to poll a state on this issue after the primary or caucus, so no polling firm will perform one. Once again, I do not think it is useful to display a fully gray map, which is what will happen if we continue to remove colors from states with completed contests. --Spiffy sperry (talk) 17:19, 27 April 2016 (UTC)