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 | Archive 2 |
The map does not match with the polls, I think? 83.80.208.22 ( talk) 17:24, 19 February 2015 (UTC)
Is Vermont with 46% Clinton en 42% Sanders not shared? 83.80.208.22 ( talk) 12:00, 24 February 2015 (UTC)
Clinton has a weak lead in Washington according to the map, but the poll we're using to justify this was taken a day after another poll indicated Clinton has a strong lead. I think we should stripe states that have conflicting percentages within a 3 month time. We would probably need to use an svg text editable map which isn't actually a bad idea. Also, IMHO there only needs to be three shades of green at the most! I think we should consolidate the two darkest shades of green. Prcc27 ( talk) 16:03, 28 July 2015 (UTC) @ Nitroxium: Sorry to ping you and revert you again.. But per above the other map was problematic and inaccurate. the Public opinion of same-sex marriage USA map uses the striping system for conflicting polls just like the map I made. And the GOP map is way more cluttered than the Democrat map and I'm pretty sure the Democrat map won't get as bad especially since there are a lot of states that haven't even been polled/aren't polled that often. It's usually the early voting states that get polled the most. The reason why I picked the 3 month time period is because public opinion isn't likely to change too much in that amount of time and just because the most recent poll says something doesn't mean it's necessarily true. I might be willing to decrease it to 2 months- 1 month would be pushing it IMO. But the current system will not work because it's absolutely bogus to think that Clinton had a 33% lead in Washington one day, and then the very next day all of the sudden she only has a 9% lead. Prcc27 ( talk) 08:33, 29 July 2015 (UTC)
You already explained to me that both polls are accurate with each other within 1%, but that's if and only if Clinton has a 21% lead which the map does have Washington colored as. And it's not the polls that are conflicting, it's that the map doesn't take into account the margin of error and quite frankly a footnote isn't enough. If readers see Washington striped with two colors and they see the margin of error footnote- they will be more likely to scroll down to find out why Washington is striped and to find out what those margin of errors for Washington are. Then they will be able to come to the conclusion that Clinton has about a 21% lead (which is pretty much what you concluded from the polls yourself) or they could come to the conclusion that Clinton does in fact have a less than 10% lead because the poll that says so is more recent. Regardless, purposely leaving out information on the map adds WP:UNDUE bias and may be fraudulent. Prcc27 ( talk) 04:14, 2 August 2015 (UTC)
I fell like it's time to get rid of the percentages on the map. They are not used on the GOP primary article, and they were not used on the 2008 Democratic primary article even though in August 2007 Clinton had just as much of a commanding lead as she does now in 2015. She had a lead in 29 states compared to her current lead in 28 states. I think it would be smart to be consistent throughout the articles. If a map without percentages can be used in 2007 when she had a commanding lead- a map like that can also be used now since Clinton's situation is as similar then as it is now! We should only have three colors (unless other candidates start to lead in states): "Clinton lead", "Clinton/Sanders tied", and "Sanders lead". Since there are three states that Sanders leads or ties in- the percentages are no longer needed. The reason why the map had them in the first place is because Clinton used to lead in every single state (except for Vermont which was incorrectly colored in her favor). But now we have two candidates that are leading in states instead of one. Prcc27 ( talk) 04:05, 11 August 2015 (UTC)
Please update this page with the poll results here: http://www.wcvb.com/politics/new-poll-shows-sanders-ahead-of-clinton-in-new-hampshire/34665480
Please additionally update the count of states Bernie is winning in to TWO. — Preceding unsigned comment added by 74.107.74.186 ( talk) 03:30, 12 August 2015 (UTC)
To pre-empt any dispute on how these work, refer to this discussion Ramaksoud2000 ( Talk to me) 07:00, 13 August 2015 (UTC)
Poll results here: www.desmoinesregister.com/story/news/politics/iowa-poll/2015/08/29/iowa-poll-democrats-august/71387664/ — Preceding unsigned comment added by 74.107.74.186 ( talk • contribs) 22:26, 29 August 2015 (UTC)
The map states that there are 3 states shared between Hillary and Sanders. First of all, I don't believe that these states do exist. And if they do, they need to be reflected in the map so people wouldn't be guessing. Also, if these are Oregon, Iowa and West Virginia, then I would be very disappointed. Only Oregon qualifies, but the other two don't (Iowa and West Virginia). They are NOT within the margin of error as I explained before. In Iowa, Hillary is up by 7% and the margin is 4.9%. We have to be unbiased in Wikipedia. Finally, when a state is shared, the two results should be highlighted as blue to make it clear to the reader. All4peace ( talk) 05:19, 30 August 2015 (UTC)
http://www.wvgazettemail.com/article/20150826/GZ01/150829633
3.21% + or -, 36 for Hillary vs 32 for Bernie.
Please update the map to reflect this. — Preceding unsigned comment added by 74.107.74.186 ( talk) 17:49, 28 August 2015 (UTC)
Polls that are nearing 6 months old are not likely very accurate to the current opinions held in each state. Perhaps 6 months is too long a period to wait before removing polls from the map. 3 or 4 months may be more accurate. However, I do understand that the map will likely look very empty at this stage in the election cycle if we were to make this change. Maybe a significantly darker shade for older polls? Of course, this would also be unlike the current Republican polling map and I understand you wish to keep those similar. In any case, the last poll for Connecticut occurred 6 months ago (March 6-9), so that should be altered. 2ft7Ninja ( talk) 20:05, 9 September 2015 (UTC)
New poll from CBS today. Results: Bernie 43% Clinton 33% Biden 10%
Here: https://d25d2506sfb94s.cloudfront.net/cumulus_uploads/document/3zimki4xtx/Iowa_Release_20150913.pdf — Preceding unsigned comment added by 74.107.74.186 ( talk) 17:20, 13 September 2015 (UTC)
Could we change the map such as the gop map?
2016 U.S. presidential election | |
---|---|
Republican Party | |
Democratic Party | |
Third parties | |
Related races | |
| |
83.80.208.22 ( talk) 09:34, 12 August 2015 (UTC)
I was about to rage at the changed map, but I actually like the logic of it as long as the statistical ties are striped or whatever like the GOP map. The GOP map actually looks so much better without 50 colors in a few states. And adding a recentness to it is good. I remember the 2012 maps were like that. Good work, yall!-- Metallurgist ( talk) 14:00, 23 August 2015 (UTC)
http://www.wcax.com/story/30064232/poll-bernie-sanders-popular-in-vt-even-among-republicans
Sanders at 65%, Hillary 14%. Expected. Please update the map to show Sanders ahead in Vermont.
PS: Why is NH new shared again? That's one poll result going against a trend. — Preceding unsigned comment added by 74.107.74.186 ( talk) 22:39, 18 September 2015 (UTC)
Please hand NH back to Sanders. Full results: http://i2.cdn.turner.com/cnn/2015/images/09/24/cnnorcdemocrats2016newhampshire.pdf — Preceding unsigned comment added by 74.107.74.186 ( talk • contribs) 22:20, 24 September 2015 (UTC)
Full poll: http://i2.cdn.turner.com/cnn/2015/images/09/24/cnnorcdemocrats2016newhampshire.pdf — Preceding unsigned comment added by 131.215.141.159 ( talk) 22:31, 24 September 2015 (UTC)
Link: http://utahpolicy.com/index.php/features/today-at-utah-policy/7050- — Preceding unsigned comment added by 74.107.74.186 ( talk) 16:48, 21 September 2015 (UTC)
I updated the Utah section. The map needs to be updated still, to show it as a tossup. — Preceding unsigned comment added by 131.215.141.159 ( talk) 17:18, 21 September 2015 (UTC)
Utah map should have been updated to show Sanders has the state, not a tossup. Alternatively, since the Dan Jones polls include Republicans (that explains the ~50% that Don't Know), and these Republicans could have chosen Sanders to spite Hillary, this poll may be garbage for the purposes of this page, and should be ignored. I believe it is garbage because of the high disparity between it and every other Democratic poll in addition to the low number of respondents willing to back a candidate. I think the 604 person sample included all respondents on both questions for the GOP and Dem sides and are thus not representative of the theme of this page. Utah went 75-25 in the 2012 general election. The GOP question indicated 19% Don't Know. This lines up with the approximate proportion of voters who will vote Democratic in the general election, however the Dem question shows 34% Don't Know. This suggests to me that both questions were asked of all 604 respondents and that some Republicans attempted to spoil the result of the Dem question. I think we should either give Utah completely to Bernie, not striped, by following the results of the poll; or we should ignore all Dan Jones polls conducted for UtahPolicy.com thus far because they can reasonably be shown to be non-representative of the goal of this page. I make that distinction on the end because there is one Idaho poll conducted by Dan Jones that lists the preference for the Democratic candidates among all Idahoans and only among Democrats albeit without the sample size and MOE. @ SPQRobin: @ Prcc27: Jfaulk111 ( talk) 16:08, 25 September 2015 (UTC)
After rereading it, I see where this is correct. I swear I read the thing 5 times over. I missed the section at the bottom regarding only the Democratic numbers. Sorry for the oversight! — Preceding unsigned comment added by Jfaulk111 ( talk • contribs) 20:16, 25 September 2015 (UTC)
Bernie Sanders has won 4 of the last 5 polls. So let me get this straight - this page isn't using the average of multiple polls for some states and for others is relying on months' old polling data. This page is lacking integrity. JaskaPDX ( talk) 10:21, 26 September 2015 (UTC)
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-hillary-clinton-big-lead/22661331/ 83.80.208.22 ( talk) 11:45, 1 February 2015 (UTC)
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.pdf
A poll from Maryland was released a few days ago and a Idaho poll was released today
Idaho: http://idahopoliticsweekly.com/politics/636-poll-trump-and-sanders-lead-2016-field-in-idaho
Anonymous1706 ( talk) 03:11, 16 October 2015 (UTC)
Poll here: https://drive.google.com/file/d/0B6Ak7rw5iSECYjRCTjhjakxfb1U/view — Preceding unsigned comment added by 98.169.44.13 ( talk) 13:54, 16 October 2015 (UTC)
I propose that now that Joe Biden is officially not running, all polls must put preference over the results without Biden in them. In other words, if a poll has results with and without Biden, only post the results without Biden. Though, if a poll has no results without Biden, post them as is. Nitroxium ( talk) 19:40, 21 October 2015 (UTC)
Bernie Sanders is leading again in the latest NH poll: http://www.scribd.com/doc/286125782/BloombergPolitics-SaintAnselmNewHampshirePoll Please adjust the map accordingly. Thanks so much for keeping it up to date. 67.1.121.105 ( talk) 16:03, 24 October 2015 (UTC)
New CBS poll for Iowa and NH:
https://today.yougov.com/news/2015/10/25/clinton-surges-early-states-carson-levels-trump-io/
Iowa: Clinton 46%, Sanders 43% = TIED
NH: Sanders 54%, Clinton 39% = SANDERS AHEAD — Preceding unsigned comment added by 72.23.221.33 ( talk) 14:43, 25 October 2015 (UTC)
Clinton polling ahead in Maryland as of an October 2015 poll by the WashPost. Could someone update the map to color Maryland green?— Preceding unsigned comment added by Godwin1996 ( talk • contribs) 10:40, 29 October 2015 (UTC)
I saw some concerns about averaging polls earlier, but I'm going to speak up in support of it. The recent Monmouth poll has HRC jumping by nearly 20 points in at most 3 days after the CBS/YouGov poll. There's no way that that there's that big of a shift that quickly (other sources are attributing it to the different demographics polled by Monmouth). I'd like to see a rolling average used (maybe the last 7-14 days?) used instead of the most recent polls in Iowa and NH, in particular, instead of having the map colors swinging back and forth when trends have Sanders ahead in NH and HRC ahead in Iowa.
Now that I think of it, this does affect the MoE issue... Regardless, I really don't think the latest poll for any state should necessarily have the final say for the colors on the map. 23.25.224.97 ( talk) 18:24, 27 October 2015 (UTC)
Iowa caucuses are very hard to poll for. On Feb 1 2016, anybody over the age of 17 years and 3 months can show up at the precinct, and will be registered on site, and can take part in the caucus. Most polls are done based on registration data, and should normally result in an unbiased sample. In most years, they will. However,this election cycle could be like the 2008 election cycle, and because of the Sanders factor, many people who would not normally be expected to caucus will show up at the precinct. The only way to correctly poll is to sample the entire population over the age of 17, and get their likelihood to caucus and base the results on that.
Similarly in New Hampshire, you can register to vote on election day, and can request the ballot of either party. So the polling issues are similar to Iowa.
This is the reason, that IMO, the YouGov/CBS methodology is more correct (though they should be polling the entire population over the age of 18 as opposed to only registered voters)
These are things to consider when interpreting polling data. It is not simple to do, however, I do think that there should be a place where the caveats of the poll are easily accessible — Preceding unsigned comment added by 50.0.150.103 ( talk) 20:57, 27 October 2015 (UTC)
I've tried updated the map to reflect the fact that Washington has not had a poll in six months, but the image won't update. After re-uploading, the previous image was correct, but the second upload wasn't. I then reverted, but the revert is different than the original I was reverting to. Clearly, there is some sort of bug here. Does anybody know how to fix it/what's going on? — Preceding unsigned comment added by ArchPope Sextus VI ( talk • contribs) 22:31, 20 November 2015 (UTC)
We've been discussing margin of errors and statistical ties for a while now, but there's no consistency and no consensus. So can we reach a FINAL decision? I really don't care which approach the group prefers. I'm ok with ANY option as long as it is consistent.
Here are the three options, with Pro's and Con's of each one:
1. Ignore Margins of Error
Pros: No headaches. Simple. If a candidate is leading a state, the state will be shaded in his or her color.
Cons: Quick changes, and ignore very close races.
Today: Hillary would lead 32 and Bernie would lead 4
2. Difference is less than or equal to (one) Margin of Error (This is the current method)
Pros: It gives a better view of close races.
Cons: Not the most accurate method in determining Margin of Error in a poll.
Today: Hillary would lead 31, Bernie would lead 3 and one would be shared
3. Difference is less than or equal to twice the Margin of Error
Pros: It's the most accurate scientifically
Cons: The range can very wide (up to 14% difference) which make many states as shared, even if a candidate is 10 points above another candidate.
Today: Hillary would lead 29, Bernie would lead 2 and 5 would be shared
4. Difference of less than 5%
Pros: Not controversial and not unscientific. Easy to understand.
Cons: Ignores margins of errors and sample sizes.
My personal preference is option 4. But I really don't care. Whichever decision is finalized, I'm ok with it. Let's be consistent.
All4peace ( talk) 17:23, 24 November 2015 (UTC)
Poll source | Date | 1st | 2nd | 3rd | Other |
---|---|---|---|---|---|
LA Times/USC
Margin of error: ± 5.3%
|
Aug 29-Sep 8, 2015 | Donald Trump 24% |
Ben Carson 18% |
Ted Cruz 6% | Jeb Bush 6%, Marco Rubio 5%, Carly Fiorina 5%, Mike Huckabee 2%, Scott Walker 2%, John Kasich 2%, Rand Paul 2%, Chris Christie 1%, Rick Santorum 1%, Rick Perry 1%, Others 3%, undecided 20% |
@Everyone, please read the box, at right, and if you are willing, the
Wikipedia article, “Margin of error” from which it comes. It really explains how this works in a slightly clear way that is somewhat comprehensible. Quote: “The margin of error is usually defined as the "radius" (or half the width) of a
confidence interval for a particular
statistic from a survey.” i.e. two radii = one diameter = size of the confidence interval
Please stop saying (I don’t know if anyone has on this talk page, but people have left it as an edit summary on the map file on Wikimedia) other editors are using “twice the margin of error”. That makes as much sense as insisting that the circumference of a circle is equal to πr, and lambasting people who point out that the circumference of a circle is equal to 2πr as “doing it wrong” and “just doubling the number to make the result look better to themselves”.
The confidence interval equals one diameter. The margin of error equals one radius.
One radius, plus one radius, equals one diameter.
Notice, in the image at right, that the smallest margin of error represented is 2%, while the confidence interval (on the x axis) for that MOE ranges from 48% to 52%; id est, a difference of 4%! Why? Because the confidence interval is 2% in either direction on a number line; that’s what the symbol ± means.
I do apologise for speaking violently, no offence is intended. Please assume good faith if I am being overly harsh. 〜 Info por favor ( talk) 16:45, 28 November 2015 (UTC)
Oppose: The map already used to be like that and it caused too many problems especially with the margin of error. We don't need to over complicate things and it's best to keep the map consistent with the GOP map for clarity. Prcc27 ( talk) 21:42, 11 January 2016 (UTC)
With races that are very active, especially in the early states (Iowa/New Hampshire), the results from different can seem erratic. Once Hillary is leading by 10, then by 5, then down by 15 to Bernie in New Hampshire. It's a seesaw, and that's not very proper at displaying for the readers. Here is my recommendation:
1. Aggregation where there are multiple polling in the past month only, otherwise just show last poll. 2. Present it in a table, so people can have a quick look at the status of each state, instead of scrolling state by state. 3. MOE is no longer applicable. We'll have to set a hard percentage for the front runner. For this example, I used 5% (Appears to be the case of most MOE's anyways) 4. We should apply aggregation to the map as well.
Here's an example. I hope you like it.
State | No of Polls | Date | Hillary Clinton | Bernie Sanders | Martin O'Malley | Other/Undecided |
---|---|---|---|---|---|---|
Arizona | 1 | October 24-November 5, 2015 | 47% | 19% | 2% | 32% |
Iowa | 8 | November 28-December 17, 2015 | 52% | 36% | 5% | 7% |
New Hampshire | 5 | November 30-December 22, 2015 | 44% | 49% | 3% | 3% |
South Carolina | 3 | December 5-17, 2015 | 63% | 27% | 3% | 11% |
All4peace ( talk) 04:24, 27 December 2015 (UTC)
Who deleted the last Quinnipiac poll?— Preceding unsigned comment added by 93.149.243.179 ( talk • contribs) 20:06, 13 January 2016 (UTC)
Alaska is a closed caucus state, but if appears that you can register as a Democrat on election day. Should we include the Independents in the poll or only the Democrats? I lean towards only the Democrats since they are already eligible to vote in the caucus. Prcc27 ( talk) 06:22, 24 January 2016 (UTC)
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 | Archive 2 |
The map does not match with the polls, I think? 83.80.208.22 ( talk) 17:24, 19 February 2015 (UTC)
Is Vermont with 46% Clinton en 42% Sanders not shared? 83.80.208.22 ( talk) 12:00, 24 February 2015 (UTC)
Clinton has a weak lead in Washington according to the map, but the poll we're using to justify this was taken a day after another poll indicated Clinton has a strong lead. I think we should stripe states that have conflicting percentages within a 3 month time. We would probably need to use an svg text editable map which isn't actually a bad idea. Also, IMHO there only needs to be three shades of green at the most! I think we should consolidate the two darkest shades of green. Prcc27 ( talk) 16:03, 28 July 2015 (UTC) @ Nitroxium: Sorry to ping you and revert you again.. But per above the other map was problematic and inaccurate. the Public opinion of same-sex marriage USA map uses the striping system for conflicting polls just like the map I made. And the GOP map is way more cluttered than the Democrat map and I'm pretty sure the Democrat map won't get as bad especially since there are a lot of states that haven't even been polled/aren't polled that often. It's usually the early voting states that get polled the most. The reason why I picked the 3 month time period is because public opinion isn't likely to change too much in that amount of time and just because the most recent poll says something doesn't mean it's necessarily true. I might be willing to decrease it to 2 months- 1 month would be pushing it IMO. But the current system will not work because it's absolutely bogus to think that Clinton had a 33% lead in Washington one day, and then the very next day all of the sudden she only has a 9% lead. Prcc27 ( talk) 08:33, 29 July 2015 (UTC)
You already explained to me that both polls are accurate with each other within 1%, but that's if and only if Clinton has a 21% lead which the map does have Washington colored as. And it's not the polls that are conflicting, it's that the map doesn't take into account the margin of error and quite frankly a footnote isn't enough. If readers see Washington striped with two colors and they see the margin of error footnote- they will be more likely to scroll down to find out why Washington is striped and to find out what those margin of errors for Washington are. Then they will be able to come to the conclusion that Clinton has about a 21% lead (which is pretty much what you concluded from the polls yourself) or they could come to the conclusion that Clinton does in fact have a less than 10% lead because the poll that says so is more recent. Regardless, purposely leaving out information on the map adds WP:UNDUE bias and may be fraudulent. Prcc27 ( talk) 04:14, 2 August 2015 (UTC)
I fell like it's time to get rid of the percentages on the map. They are not used on the GOP primary article, and they were not used on the 2008 Democratic primary article even though in August 2007 Clinton had just as much of a commanding lead as she does now in 2015. She had a lead in 29 states compared to her current lead in 28 states. I think it would be smart to be consistent throughout the articles. If a map without percentages can be used in 2007 when she had a commanding lead- a map like that can also be used now since Clinton's situation is as similar then as it is now! We should only have three colors (unless other candidates start to lead in states): "Clinton lead", "Clinton/Sanders tied", and "Sanders lead". Since there are three states that Sanders leads or ties in- the percentages are no longer needed. The reason why the map had them in the first place is because Clinton used to lead in every single state (except for Vermont which was incorrectly colored in her favor). But now we have two candidates that are leading in states instead of one. Prcc27 ( talk) 04:05, 11 August 2015 (UTC)
Please update this page with the poll results here: http://www.wcvb.com/politics/new-poll-shows-sanders-ahead-of-clinton-in-new-hampshire/34665480
Please additionally update the count of states Bernie is winning in to TWO. — Preceding unsigned comment added by 74.107.74.186 ( talk) 03:30, 12 August 2015 (UTC)
To pre-empt any dispute on how these work, refer to this discussion Ramaksoud2000 ( Talk to me) 07:00, 13 August 2015 (UTC)
Poll results here: www.desmoinesregister.com/story/news/politics/iowa-poll/2015/08/29/iowa-poll-democrats-august/71387664/ — Preceding unsigned comment added by 74.107.74.186 ( talk • contribs) 22:26, 29 August 2015 (UTC)
The map states that there are 3 states shared between Hillary and Sanders. First of all, I don't believe that these states do exist. And if they do, they need to be reflected in the map so people wouldn't be guessing. Also, if these are Oregon, Iowa and West Virginia, then I would be very disappointed. Only Oregon qualifies, but the other two don't (Iowa and West Virginia). They are NOT within the margin of error as I explained before. In Iowa, Hillary is up by 7% and the margin is 4.9%. We have to be unbiased in Wikipedia. Finally, when a state is shared, the two results should be highlighted as blue to make it clear to the reader. All4peace ( talk) 05:19, 30 August 2015 (UTC)
http://www.wvgazettemail.com/article/20150826/GZ01/150829633
3.21% + or -, 36 for Hillary vs 32 for Bernie.
Please update the map to reflect this. — Preceding unsigned comment added by 74.107.74.186 ( talk) 17:49, 28 August 2015 (UTC)
Polls that are nearing 6 months old are not likely very accurate to the current opinions held in each state. Perhaps 6 months is too long a period to wait before removing polls from the map. 3 or 4 months may be more accurate. However, I do understand that the map will likely look very empty at this stage in the election cycle if we were to make this change. Maybe a significantly darker shade for older polls? Of course, this would also be unlike the current Republican polling map and I understand you wish to keep those similar. In any case, the last poll for Connecticut occurred 6 months ago (March 6-9), so that should be altered. 2ft7Ninja ( talk) 20:05, 9 September 2015 (UTC)
New poll from CBS today. Results: Bernie 43% Clinton 33% Biden 10%
Here: https://d25d2506sfb94s.cloudfront.net/cumulus_uploads/document/3zimki4xtx/Iowa_Release_20150913.pdf — Preceding unsigned comment added by 74.107.74.186 ( talk) 17:20, 13 September 2015 (UTC)
Could we change the map such as the gop map?
2016 U.S. presidential election | |
---|---|
Republican Party | |
Democratic Party | |
Third parties | |
Related races | |
| |
83.80.208.22 ( talk) 09:34, 12 August 2015 (UTC)
I was about to rage at the changed map, but I actually like the logic of it as long as the statistical ties are striped or whatever like the GOP map. The GOP map actually looks so much better without 50 colors in a few states. And adding a recentness to it is good. I remember the 2012 maps were like that. Good work, yall!-- Metallurgist ( talk) 14:00, 23 August 2015 (UTC)
http://www.wcax.com/story/30064232/poll-bernie-sanders-popular-in-vt-even-among-republicans
Sanders at 65%, Hillary 14%. Expected. Please update the map to show Sanders ahead in Vermont.
PS: Why is NH new shared again? That's one poll result going against a trend. — Preceding unsigned comment added by 74.107.74.186 ( talk) 22:39, 18 September 2015 (UTC)
Please hand NH back to Sanders. Full results: http://i2.cdn.turner.com/cnn/2015/images/09/24/cnnorcdemocrats2016newhampshire.pdf — Preceding unsigned comment added by 74.107.74.186 ( talk • contribs) 22:20, 24 September 2015 (UTC)
Full poll: http://i2.cdn.turner.com/cnn/2015/images/09/24/cnnorcdemocrats2016newhampshire.pdf — Preceding unsigned comment added by 131.215.141.159 ( talk) 22:31, 24 September 2015 (UTC)
Link: http://utahpolicy.com/index.php/features/today-at-utah-policy/7050- — Preceding unsigned comment added by 74.107.74.186 ( talk) 16:48, 21 September 2015 (UTC)
I updated the Utah section. The map needs to be updated still, to show it as a tossup. — Preceding unsigned comment added by 131.215.141.159 ( talk) 17:18, 21 September 2015 (UTC)
Utah map should have been updated to show Sanders has the state, not a tossup. Alternatively, since the Dan Jones polls include Republicans (that explains the ~50% that Don't Know), and these Republicans could have chosen Sanders to spite Hillary, this poll may be garbage for the purposes of this page, and should be ignored. I believe it is garbage because of the high disparity between it and every other Democratic poll in addition to the low number of respondents willing to back a candidate. I think the 604 person sample included all respondents on both questions for the GOP and Dem sides and are thus not representative of the theme of this page. Utah went 75-25 in the 2012 general election. The GOP question indicated 19% Don't Know. This lines up with the approximate proportion of voters who will vote Democratic in the general election, however the Dem question shows 34% Don't Know. This suggests to me that both questions were asked of all 604 respondents and that some Republicans attempted to spoil the result of the Dem question. I think we should either give Utah completely to Bernie, not striped, by following the results of the poll; or we should ignore all Dan Jones polls conducted for UtahPolicy.com thus far because they can reasonably be shown to be non-representative of the goal of this page. I make that distinction on the end because there is one Idaho poll conducted by Dan Jones that lists the preference for the Democratic candidates among all Idahoans and only among Democrats albeit without the sample size and MOE. @ SPQRobin: @ Prcc27: Jfaulk111 ( talk) 16:08, 25 September 2015 (UTC)
After rereading it, I see where this is correct. I swear I read the thing 5 times over. I missed the section at the bottom regarding only the Democratic numbers. Sorry for the oversight! — Preceding unsigned comment added by Jfaulk111 ( talk • contribs) 20:16, 25 September 2015 (UTC)
Bernie Sanders has won 4 of the last 5 polls. So let me get this straight - this page isn't using the average of multiple polls for some states and for others is relying on months' old polling data. This page is lacking integrity. JaskaPDX ( talk) 10:21, 26 September 2015 (UTC)
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-hillary-clinton-big-lead/22661331/ 83.80.208.22 ( talk) 11:45, 1 February 2015 (UTC)
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.pdf
A poll from Maryland was released a few days ago and a Idaho poll was released today
Idaho: http://idahopoliticsweekly.com/politics/636-poll-trump-and-sanders-lead-2016-field-in-idaho
Anonymous1706 ( talk) 03:11, 16 October 2015 (UTC)
Poll here: https://drive.google.com/file/d/0B6Ak7rw5iSECYjRCTjhjakxfb1U/view — Preceding unsigned comment added by 98.169.44.13 ( talk) 13:54, 16 October 2015 (UTC)
I propose that now that Joe Biden is officially not running, all polls must put preference over the results without Biden in them. In other words, if a poll has results with and without Biden, only post the results without Biden. Though, if a poll has no results without Biden, post them as is. Nitroxium ( talk) 19:40, 21 October 2015 (UTC)
Bernie Sanders is leading again in the latest NH poll: http://www.scribd.com/doc/286125782/BloombergPolitics-SaintAnselmNewHampshirePoll Please adjust the map accordingly. Thanks so much for keeping it up to date. 67.1.121.105 ( talk) 16:03, 24 October 2015 (UTC)
New CBS poll for Iowa and NH:
https://today.yougov.com/news/2015/10/25/clinton-surges-early-states-carson-levels-trump-io/
Iowa: Clinton 46%, Sanders 43% = TIED
NH: Sanders 54%, Clinton 39% = SANDERS AHEAD — Preceding unsigned comment added by 72.23.221.33 ( talk) 14:43, 25 October 2015 (UTC)
Clinton polling ahead in Maryland as of an October 2015 poll by the WashPost. Could someone update the map to color Maryland green?— Preceding unsigned comment added by Godwin1996 ( talk • contribs) 10:40, 29 October 2015 (UTC)
I saw some concerns about averaging polls earlier, but I'm going to speak up in support of it. The recent Monmouth poll has HRC jumping by nearly 20 points in at most 3 days after the CBS/YouGov poll. There's no way that that there's that big of a shift that quickly (other sources are attributing it to the different demographics polled by Monmouth). I'd like to see a rolling average used (maybe the last 7-14 days?) used instead of the most recent polls in Iowa and NH, in particular, instead of having the map colors swinging back and forth when trends have Sanders ahead in NH and HRC ahead in Iowa.
Now that I think of it, this does affect the MoE issue... Regardless, I really don't think the latest poll for any state should necessarily have the final say for the colors on the map. 23.25.224.97 ( talk) 18:24, 27 October 2015 (UTC)
Iowa caucuses are very hard to poll for. On Feb 1 2016, anybody over the age of 17 years and 3 months can show up at the precinct, and will be registered on site, and can take part in the caucus. Most polls are done based on registration data, and should normally result in an unbiased sample. In most years, they will. However,this election cycle could be like the 2008 election cycle, and because of the Sanders factor, many people who would not normally be expected to caucus will show up at the precinct. The only way to correctly poll is to sample the entire population over the age of 17, and get their likelihood to caucus and base the results on that.
Similarly in New Hampshire, you can register to vote on election day, and can request the ballot of either party. So the polling issues are similar to Iowa.
This is the reason, that IMO, the YouGov/CBS methodology is more correct (though they should be polling the entire population over the age of 18 as opposed to only registered voters)
These are things to consider when interpreting polling data. It is not simple to do, however, I do think that there should be a place where the caveats of the poll are easily accessible — Preceding unsigned comment added by 50.0.150.103 ( talk) 20:57, 27 October 2015 (UTC)
I've tried updated the map to reflect the fact that Washington has not had a poll in six months, but the image won't update. After re-uploading, the previous image was correct, but the second upload wasn't. I then reverted, but the revert is different than the original I was reverting to. Clearly, there is some sort of bug here. Does anybody know how to fix it/what's going on? — Preceding unsigned comment added by ArchPope Sextus VI ( talk • contribs) 22:31, 20 November 2015 (UTC)
We've been discussing margin of errors and statistical ties for a while now, but there's no consistency and no consensus. So can we reach a FINAL decision? I really don't care which approach the group prefers. I'm ok with ANY option as long as it is consistent.
Here are the three options, with Pro's and Con's of each one:
1. Ignore Margins of Error
Pros: No headaches. Simple. If a candidate is leading a state, the state will be shaded in his or her color.
Cons: Quick changes, and ignore very close races.
Today: Hillary would lead 32 and Bernie would lead 4
2. Difference is less than or equal to (one) Margin of Error (This is the current method)
Pros: It gives a better view of close races.
Cons: Not the most accurate method in determining Margin of Error in a poll.
Today: Hillary would lead 31, Bernie would lead 3 and one would be shared
3. Difference is less than or equal to twice the Margin of Error
Pros: It's the most accurate scientifically
Cons: The range can very wide (up to 14% difference) which make many states as shared, even if a candidate is 10 points above another candidate.
Today: Hillary would lead 29, Bernie would lead 2 and 5 would be shared
4. Difference of less than 5%
Pros: Not controversial and not unscientific. Easy to understand.
Cons: Ignores margins of errors and sample sizes.
My personal preference is option 4. But I really don't care. Whichever decision is finalized, I'm ok with it. Let's be consistent.
All4peace ( talk) 17:23, 24 November 2015 (UTC)
Poll source | Date | 1st | 2nd | 3rd | Other |
---|---|---|---|---|---|
LA Times/USC
Margin of error: ± 5.3%
|
Aug 29-Sep 8, 2015 | Donald Trump 24% |
Ben Carson 18% |
Ted Cruz 6% | Jeb Bush 6%, Marco Rubio 5%, Carly Fiorina 5%, Mike Huckabee 2%, Scott Walker 2%, John Kasich 2%, Rand Paul 2%, Chris Christie 1%, Rick Santorum 1%, Rick Perry 1%, Others 3%, undecided 20% |
@Everyone, please read the box, at right, and if you are willing, the
Wikipedia article, “Margin of error” from which it comes. It really explains how this works in a slightly clear way that is somewhat comprehensible. Quote: “The margin of error is usually defined as the "radius" (or half the width) of a
confidence interval for a particular
statistic from a survey.” i.e. two radii = one diameter = size of the confidence interval
Please stop saying (I don’t know if anyone has on this talk page, but people have left it as an edit summary on the map file on Wikimedia) other editors are using “twice the margin of error”. That makes as much sense as insisting that the circumference of a circle is equal to πr, and lambasting people who point out that the circumference of a circle is equal to 2πr as “doing it wrong” and “just doubling the number to make the result look better to themselves”.
The confidence interval equals one diameter. The margin of error equals one radius.
One radius, plus one radius, equals one diameter.
Notice, in the image at right, that the smallest margin of error represented is 2%, while the confidence interval (on the x axis) for that MOE ranges from 48% to 52%; id est, a difference of 4%! Why? Because the confidence interval is 2% in either direction on a number line; that’s what the symbol ± means.
I do apologise for speaking violently, no offence is intended. Please assume good faith if I am being overly harsh. 〜 Info por favor ( talk) 16:45, 28 November 2015 (UTC)
Oppose: The map already used to be like that and it caused too many problems especially with the margin of error. We don't need to over complicate things and it's best to keep the map consistent with the GOP map for clarity. Prcc27 ( talk) 21:42, 11 January 2016 (UTC)
With races that are very active, especially in the early states (Iowa/New Hampshire), the results from different can seem erratic. Once Hillary is leading by 10, then by 5, then down by 15 to Bernie in New Hampshire. It's a seesaw, and that's not very proper at displaying for the readers. Here is my recommendation:
1. Aggregation where there are multiple polling in the past month only, otherwise just show last poll. 2. Present it in a table, so people can have a quick look at the status of each state, instead of scrolling state by state. 3. MOE is no longer applicable. We'll have to set a hard percentage for the front runner. For this example, I used 5% (Appears to be the case of most MOE's anyways) 4. We should apply aggregation to the map as well.
Here's an example. I hope you like it.
State | No of Polls | Date | Hillary Clinton | Bernie Sanders | Martin O'Malley | Other/Undecided |
---|---|---|---|---|---|---|
Arizona | 1 | October 24-November 5, 2015 | 47% | 19% | 2% | 32% |
Iowa | 8 | November 28-December 17, 2015 | 52% | 36% | 5% | 7% |
New Hampshire | 5 | November 30-December 22, 2015 | 44% | 49% | 3% | 3% |
South Carolina | 3 | December 5-17, 2015 | 63% | 27% | 3% | 11% |
All4peace ( talk) 04:24, 27 December 2015 (UTC)
Who deleted the last Quinnipiac poll?— Preceding unsigned comment added by 93.149.243.179 ( talk • contribs) 20:06, 13 January 2016 (UTC)
Alaska is a closed caucus state, but if appears that you can register as a Democrat on election day. Should we include the Independents in the poll or only the Democrats? I lean towards only the Democrats since they are already eligible to vote in the caucus. Prcc27 ( talk) 06:22, 24 January 2016 (UTC)