Jump to content

Talk:Nationwide opinion polling for the 2012 United States presidential election

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Form

[edit]

Having the polls back in the regular form would be much better because it shows the all candidates, and provides a better. The provided does not and contains less information Please stop reverting.

-electoral jew2 —Preceding unsigned comment added by 71.167.72.140 (talk) 02:33, 19 July 2009 (UTC)[reply]

I agree with the above. Picking and choosing which information from each poll to include is a violation of WP:OR and potentially WP:NPOV. – Hysteria18 (Talk • Contributions) 14:04, 19 July 2009 (UTC)[reply]

my recent edits

[edit]

apologies for any confusion i might've caused there. i was trying to do too many things at once and wrongly reverted an edit which added a totally correct poll result. i've since fixed it, though i'm now wondering how best to make the distinction between the 'Republicans' and 'All voters' polls clear. Hysteria18 (talk) 18:06, 9 December 2008 (UTC)[reply]

I think you're missing a couple

[edit]

I'm not even sure how valid these polls are - every poll seems to show a different leader. But I think there were two polls that came out, can't remember from who, that showed Romney in the lead. One might have been a Wall Street Journal poll.

76.230.57.176 (talk) 02:20, 13 January 2009 (UTC)[reply]

No, of course they're very unlikely indeed to tell us anything currently of value. My personal justification for keeping this article is for historical reference when the 2012 election finally does roll around; it's certainly fascinating now to look back at the early polls at Nationwide opinion polling for the United States presidential election, 2008. I'm presuming that the Wall Street Journal poll you mentioned was the NBC-WSJ poll mentioned here and in various other MSNBC news stories, which I can't find a page dedicated to, but that link'll do I suppose. I'll add it now. — Hysteria18 (TalkContributions) 16:08, 13 January 2009 (UTC)[reply]
Scratch that, having just read through that in a bit more depth it doesn't strike me as worthy of inclusion. First, it's described not as a poll but as a 'survey', and second, it's apparently concerned with who 'should be the GOP's leader', rather than their next presidential nominee. — Hysteria18 (TalkContributions) 16:24, 13 January 2009 (UTC)[reply]
I saw polls also that show Romney in the lead and I actually saved the page on Read Later Fast. The polls from Rasmussen Tracking May 3-5 and April 27–29 both show Romney in the lead. I think people are purposely removing the polls with Romney in the lead. By people I mean Obama people. Whiteboycat(TalkContributions) —Preceding undated comment added 04:11, 22 May 2012 (UTC).[reply]
Polls are released at frequencies that vary among the polling sponsors. Rasmussen, for example, publishes tracking polls daily while many other pollsters were publishing polls only every month or less often for most of the campaign; the consensus for this Wikipedia article has been to include only some of the polls reported by the pollsters who were conducting very frequent polls to avoid giving undue weight to information from individual pollsters. It was decided in March that twice-monthly updates of Rasmussen polls would suffice (see Talk discussion "Should we include every daily Rasmussen poll"). My understanding was that it was decided to add Rasmussen and Gallup updates twice a month on Wednesdays (that day of the week chosen as a standard to avoid bias associated with picking and choosing individual polls out of the week based on how favorable, or unfavorable, they looked on a given day of the week). Wikipedia keeps a publicly-available record of all edits, so if you want to specifically check whether May 3 or April 27 Rasmussen polls were once in the article but subsequently removed, you can go to the article page, click on the tab toward the top right called "View history," then scroll down to around the dates that the polls would have been released by the pollsters (for example, May 8). Click on the link "prev," which will take you to a copy of how the article looked on that date at that time of day (the top of the page will show what edits were made at that date and time, while the bottom of the page will show what the page looked like following those edits). If you don't see that the poll you were expecting had been entered by that date, go back to "View history" and choose a date a few days later; you can keep repeating this all the way to the present, if necessary. When content is removed, the "View history" will show a negative number of bytes of data (for example, -965 for the second edit entered on May 9) -- assuming that more info was removed than was added with the edit (the + or - reflects the net change in number of characters entered or removed). Dezastru (talk) 17:20, 25 October 2012 (UTC)[reply]

CPAC straw poll

[edit]

The CPAC straw poll is unscientific and, as unscientific polls are not reliable sources, shouldn't be included here. Please don't restore it, or any other future unscientific polls, without commenting here first. Thanks. — Hysteria18 (TalkContributions) 21:28, 1 March 2009 (UTC)[reply]

Split Page?

[edit]

I'm sorry, but the way links and the like work on Wikipedia, having both general election and primary opinion polling on the same page makes it difficult to find the information specific to the Republican Primaries. Unless someone raises major objections, I'm going to move the information on the Republican primaries to a new page by this thursday. Mburn16 (talk) 22:12, 25 August 2009 (UTC)[reply]

Newsmax/Zogby poll

[edit]

Forgot to put this in my edit summary, but the article giving the results for the recent Newsmax/Zogby poll doesn't appear to give the date the poll was administered. Any thoughts on how best to deal with this? I've just left the cell empty for now, but there's probably a better way to do it. Thanks. – Hysteria18 (Talk • Contributions) 22:11, 26 January 2010 (UTC)[reply]

New Format

[edit]

I think this would look better, only it would include all the candidates:

Poll Source Margin of Error Sample Size Dates Administered Haley Barbour Newt Gingrich Mike Huckabee Bobby Jindal Sarah Palin Tim Pawlenty Mitt Romney Not Sure/Others
Gallup ±5% 455 July 10-12, 2009 2% 14% 19% - 21% 3% 26% ???

Please be open minded. I think this is a great idea.--Jerzeykydd (talk) 01:22, 26 March 2010 (UTC)[reply]

Agreed it looks much better, at least in the sample. However,I would oppose switching for two reasons.
  • It's more likely to be harder to read, requiring horizontal scrolling depending on browser settings.
  • If a new candidate emerges, adding new columns to wiki tables while maintaining proper alignment is more difficult, especially for novice users we'll probably pick up as the elections draw closer.
  • Which just prompted me to change this to three reasons, because even without adding candidates, maintaining alignment under a constant header is more error prone than typing a name and number on the same line.
I'd be quite happy with a shortened version of the two-way race format used for the general election; sort of a one-way race, eliminating the Democratic column, then listing the candidates covered in the poll alphabetically, each having their own row. Fat&Happy (talk) 02:25, 26 March 2010 (UTC)[reply]

Since I was asked, I find giving all candidates their own column too error prone. So I'd say either the old method should be used, or only the major candidates should be included. It would remain to be defined where "major" begins; I'd suggest getting 10% in the last 3 months or something like that.

Poll Source Margin of Error Sample Size Dates Administered Newt Gingrich Mike Huckabee Sarah Palin Mitt Romney Others
Gallup ±5% 455 July 10-12, 2009 14% 19% 21% 26% Haley Barbour 2%, Tim Pawlenty 3%

I hope I'm not making matters worse. But if people couldn't agree on a new method, I'd say let's just keep the old one. Ambi Valent (talk) 17:05, 27 March 2010 (UTC)[reply]

I'm fine with these proposals, only I support shading leaders Purplebackpack89 (Notes Taken) (Locker) 17:08, 27 March 2010 (UTC)[reply]

(edit conflict) I agree with most of Fat&Happy's points. By my count, there are 17 potential candidates who've been included in Republican primary polling. Combining this with at least five permutations of "other"/"none"/"no opinion"/"would not vote"/"too soon to say" etc. (which of these responses have the same meaning as others is itself ambiguous) would lead to a somewhat unwieldy table and excessive horizontal scrolling, something WP:ACCESS warns against. Also, as Fat&Happy points out, the current table is relatively easy for newer users to navigate in the edit window and to add content, something we'll rely on more as 2012 draws near. Under your idea, if one pollster decides to include, for instance, Gary Johnson in their poll, we'd need to create an entire new column for Johnson, which in a large table is a fairly time-consuming exercise in my experience. So I'm afraid I remain in favour of keeping the article as it is now, unless there's substantial evidence that these problems are solvable. – Hysteria18 (Talk • Contributions) 17:38, 27 March 2010 (UTC)[reply]

Better wikitable

[edit]
  • Well apparently most people have a big problem with the excessive horizontal scrolling. I think there is room for compromise as I think we could put in the top 10 candidates, based only upon polling frequency, and the rest can be put in the other column. An example of poll frequency is that Mitt Romney's name was in every poll, as Scott Brown was only conducted in one poll. I tried to shrink the table as much as possible:
Poll Source Sample
size
Date(s)
Administered
Haley
Barbour
Dick
Cheney
Newt
Gingrich
Rudy
Guiliani
Mike
Huckabee
Bobby
Jindal
Sarah
Palin
Tim
Pawlenty
Ron
Paul
Mitt
Romney
Others
Gallup 455 July 10-12, 2009 2% - 14% - 19% - 21% 3% - 26% None

In addition I wanted to mention that the problem with the current format is that it's difficult to see how gaining popularity and seeing comparisons in percentages between each candidate.

What do you think of this table?--Jerzeykydd (talk) 19:10, 27 March 2010 (UTC)[reply]

As stated before, I would shade leaders Purplebackpack89 (Notes Taken) (Locker) 04:32, 28 March 2010 (UTC)[reply]

Looking at the statewide polls shows that except for Utah (Romney before Huntsman) you find Romney, Huckabee and Palin as the top three in different orders. Other candidates are often left out, but there are patterns: Gingrich was included in 4 state polls, getting 12-13%, never beating any of the top 3. Paul was included in 4 polls, getting 4-8%, never beating any of the top 4. Pawlenty was included in 6 polls, getting 1-3%, never beating any of the top 5. And no other candidate named there was beating any of the top 6.

Of course, things will only turn serious after the midterm elections, and the primaries themselves are almost two years away, so there's still lot of room for game changers left. Ambi Valent (talk) 20:39, 28 March 2010 (UTC)[reply]

I think this format looks better and is more readable, so I support it. I also agree with the suggestion about shading leaders, as that would bring greater clarity.--Polly Ticker (talk) 18:17, 29 March 2010 (UTC)[reply]

I'm afraid I remain opposed, for most of the same reasons I articulated above. I don't think the readability problems of the current table are significant enough to justify using a table founded on arbitrary distinctions, which is difficult for newer editors to contribute to. (If consensus remains in favor of this version, I too support shading the leaders, and would propose also shading (perhaps in a lighter color) runners-up who trail within the margin of error.) – Hysteria18 (Talk • Contributions) 19:14, 29 March 2010 (UTC)[reply]

I support shading the leaders and shading runners-up who trail within the margin of error. I also would like Hysteria to clarify what he means by arbitrary distinctions.--Jerzeykydd (talk) 20:10, 29 March 2010 (UTC)[reply]

By "arbitrary distinctions", I mean a point or figure (e.g. having polled at 5% or having appeared in five polls) which is used as a benchmark across the board, with no consideration to factors specific to individual candidates or individual polls. Apologies for the late response, and for not making that clear originally. – Hysteria18 (Talk • Contributions) 18:05, 5 April 2010 (UTC)[reply]

Explanatory stuff

[edit]

In the Republican primary section: Is any of it really necessary? The table looks nice, is intuitive, and tells you everything you need to know. I don't see the point of the paragraph above and the statistics table. – Hysteria18 (Talk • Contributions) 17:52, 19 April 2010 (UTC),[reply]

I can sort of see the argument for the statistics table, but not the paragraph above. The only purpose of that seems to be to be to explain how to use the table which, as I say, is fairly intuitive anyway. – Hysteria18 (Talk • Contributions) 19:40, 19 April 2010 (UTC)[reply]

Running mate polling

[edit]

In response to this: this article conveys polling numbers for the 2012 primaries and general election. We don't include numbers only related to those elections (e.g. favourability, "would you re-elect?" questions, Obama v. generic Republican questions, and running-mate preferences), mostly because of the sheer volume of polling in each of these areas compared to the very little actual value it has to the reader given that none of those questions will ever appear on an actual ballot paper. – Hysteria18 (Talk • Contributions) 15:15, 3 October 2010 (UTC)[reply]

Newsmax/SurveyUSA polls

[edit]

Newsmax has put out a bunch of polls conducted by SurveyUSA pitting Obama against folks like David Petraeus, Hillary Clinton, Bill Gates, Warren Buffett and Donald Trump and various others. The problem is, it's never specified what party, if any, these challengers are coming from, so I don't know how we include them in this article. Any ideas? – Hysteria18 (Talk • Contributions) 21:39, 10 November 2010 (UTC)[reply]

I'd put Clinton as Democrat (of course) and Petraeus and Trump as Republican (since they're mentioned candidates on that page). And Gates and Buffett, not being speculated as candidates of either party, would be Independent.Ambi Valent (talk) 22:51, 25 November 2010 (UTC)[reply]
I have removed the Newsweek Clinton vs Obama match-up from the Democratic primary section as it was a poll about the general election, not the democratic primary. It could be included somewhere in the general election section, but as previously noted it does not really fit there either.Fimiliar (talk) 17:02, 4 December 2010 (UTC)[reply]

TOP 10: replacing Giuliani and Jindal with Christie and Daniels?

[edit]

Now more than a year has passed that Giuliani has been in a poll, and Jindal has only been in one poll in that time, reaching 1%. Of those not in the TOP 10 with their own column, Christie and Daniels seem to be the currently strongest.

Does anybody else think the TOP 10 columns should be changed?Ambi Valent (talk) 01:51, 26 November 2010 (UTC)[reply]

I think that John Thune is eligible for a column, he has been in many polls recently. ~Gosox(55)(55) 04:53, 10 December 2010 (UTC)[reply]

Remove Jeb Bush

[edit]

He's definitively denied he's thinking about running and he's only been on 9 polls and even then, not since last November. Imo, he should be moved to the "others" sections. Whether he's replaced with anyone else is a different matter. Trump and Bachmann have been polled a lot recently, but whether that will turn into consistent polling is unclear. Either way, Jeb Bush should be removed. Does anyone disagree? Tiller54 (talk) 15:11, 12 May 2011 (UTC)[reply]

New Bachmann Poll

[edit]

I'm not sure yet how to add a new poll to the table, but this one is interesting because it has Bachmann in the lead: [1] Difluoroethene (talk) 23:21, 22 June 2011 (UTC)[reply]

Break the article in two

[edit]

Should the article be broken in half between the primaries and the general election polling? I mention that because the article is a little long, and will only keep getting longer as it continues. Behun (talk) 21:35, 25 June 2011 (UTC)[reply]

I support this measure. Thunderstone99 (talk) 01:01, 26 June 2011 (UTC)[reply]

I think this really needs to be carried out now, the Republican part is getting larger and larger, so is the General Election PArt. Thunderstone99 (talk) 01:00, 15 September 2011 (UTC)[reply]

Done; the Republican part has been moved into its own article: Nationwide_opinion_polling_for_the_Republican_Party_2012_presidential_primaries. Difluoroethene (talk) 04:17, 15 September 2011 (UTC)[reply]

Give Perry a column?

[edit]

Seems pretty obvious to me. SOXROX (talk) 21:29, 16 August 2011 (UTC)[reply]

Me too. Difluoroethene (talk) 00:38, 17 August 2011 (UTC)[reply]

Table sortability

[edit]

It would be nice. Anna Frodesiak (talk) 01:06, 7 September 2012 (UTC)[reply]

What fields would you want to sort by? As far as I can see, the only advantage would be if it were possible to sort by the pollster (eg, get a chronological list of all the Gallup polls in the tablet, separate fromhe polls conducted by the other organizations). That would be useful, but it would take quite a bit of work. If you look prior to about mid April, you will see that the data fields are a little different. All of the fields for the whole table would need to be normalized for a sort to work optimally. The names of the pollsters would also need to be made uniform (CNN/ORC would be treated differently in a sort than CNN/Opinion Research, for example). The external links that are associated with each name would also probably need to be entered in a different way (I haven't actually checked the format recently, just going by memory) for sorting to work optimally. If there were a lot of editors willing to work on the conversion, it could be done without too much headache, but it doesn't seem that there are enough editors working on the page to make it happen, at least not over the next few months. Dezastru (talk) 01:28, 10 September 2012 (UTC)[reply]
Well, I thought maybe the % column, so we could see all-time highest figures. I just tried and the combined cells make it really hard to do. I gave up. I think you're right. It's not worth it. Thank you for the thoughtful reply. Anna Frodesiak (talk) 00:25, 11 September 2012 (UTC)[reply]

Margin of error and tie

[edit]

When the margin of error exceeds "Leading by", the "Leading by" column ought to say "tie". --Gerrit CUTEDH 16:14, 11 September 2012 (UTC)[reply]

The problem with doing that is that it would lend support to a misinterpretation of the reported margin of error, resulting in the table erroneously under-reporting what are true statistical ties. The American Statistical Association produced a pamphlet discussing this issue:

A misleading feature of most current media stories on political polls is that they report the margin of error associated with the proportion favoring one candidate, not the margin of error of the lead of one candidate over another.

So if a poll finds that 47% prefer Obama and 44% prefer Romney and the margin of error is reported to be 3%, that means that the margin of error is ±3% for the 47% AND ±3% for the 44% — it is not ±3% for the 3-point difference between the proportions favoring Obama and Romney. In other words, with a 3% margin of error, the proportion who prefer Obama could be anywhere in the range of 44–50% while the proportion preferring Romney could be anywhere in the range of 41–47%.
The margin of error of the lead should really be larger than the margin of error of each individual preference category:

In more technical terms, a law of probability dictates that the difference between two uncertain proportions (e.g., the lead of one candidate over another in a political poll in which both are estimated) has more uncertainty associated with it than either proportion alone. Accordingly, the margin of error associated with the lead of one candidate over another should be larger than the margin of error associated with a single proportion, which is what media reports typically mention (thus the need to keep your eye on what’s being estimated!).

Until media organizations get their reporting practices in line with actual variation in results across political polls, a rule of thumb is to multiply the currently reported margin of error by 1.7 to obtain a more accurate estimate of the margin of error for the lead of one candidate over another. Thus, a reported 3 percent margin of error becomes about 5 percent and a reported 4 percent margin of error becomes about 7 percent when the size of the lead is being considered.

Thus, if the poll has a margin of error (for the individual proportions) of 3.1%, the candidates would be statistically tied whenever the lead was less than or equal to 5.3 percentage points.
Given that consideration, it would be best to leave the way a tie is reported in the table as it has been thus far. Dezastru (talk) 22:27, 11 September 2012 (UTC)[reply]

Four-way race table before Two-way race table

[edit]

I would like to suggest leading with the Four-way race table, which is currently the best representation of the ballots that voters will see this fall. Leading with the Two-way race table can be construed as biased towards a two-party electoral system, which is inappropriate for this medium. — Preceding unsigned comment added by 72.187.61.208 (talk) 15:10, 13 September 2012 (UTC)[reply]

Column title issues and time division with regard to completion of Convention Nominating process

[edit]

Comment and Edit: 00:01, 14 September 2012‎ 142.103.225.99 (talk)‎ . . (194,181 bytes) (+2)‎ . . (→‎Two-way race: Opposing is the inccorect term used.)

Response and Further Edit: 02:52, 14 September 2012‎ 72.187.61.208 (talk)‎ . . (196,316 bytes) (+636)‎ . . (Separated General Election Polls before and after Convention Nominating process and corrected column heading)

"Opposing" may not be the best term, but "Republican" is certainly not an appropriate heading for a column that includes "Hillary Rodham Clinton" as data. Added "Since Convention Nominations" and "Before Convention Nominations" headings. Moved Four-way race and Two-way race tables data into "Since Convention Nominations" heading. Changed "Republican" back to "Opposing" under "Before Convention Nominations" heading. — Preceding unsigned comment added by 72.187.61.208 (talk) 02:57, 14 September 2012 (UTC)[reply]

Major Events

[edit]

On the Opinion polling for the next United Kingdom general election it lists major events that would affect the polls such as party conventions, debates, scandals e.c.t would it be a good idea to put things like this on these pages to explain sudden changes in attitudes?. C. 22468 Talk to me 11:54, 15 October 2012 (UTC)[reply]

Doing so would be a form of original research if it directly implied that the event "caused" the poll shift. Without citing an external source that had proposed and demonstrated a causative link, it would not be suitable for wikipedia.Duster (talk) 13:29, 30 October 2012 (UTC)[reply]

Margin of Error

[edit]

The colouring representing when a winner is predicted (as opposed to a tie) is incorrect. Anyone with a basic grasp of statistics and polling will be able to confirm that when a poll has, for instance, Obama 49%, Romney 46%, Margin of Error +/-4%, then that poll has not predicted an Obama win. It is as much a tie as a poll that has them both on 47%. The colouring/bold text that indicates a predicted winner should be changed to reflect this. Otherwise it is in error. Duster (talk) 13:27, 30 October 2012 (UTC)[reply]

This issue was discussed before. See the section above on "Margin of Error and Tie." What is your proposal? Dezastru (talk) 18:58, 30 October 2012 (UTC)[reply]
Above, you wrote Thus, if the poll has a margin of error (for the individual proportions) of 3.1%, the candidates would be statistically tied whenever the lead was less than or equal to 5.3 percentage points. -- I'll have to defer to your judgment since I have little grasp of statistics.
But then you continued: Given that consideration, it would be best to leave the way a tie is reported in the table as it has been thus far. -- Which I don't get. If your first point is correct, wouldn't that consequently mean that we should format polling results as ties whenever the lead is less than the individual margin of error x 1.7?
The problem with doing that is that it would lend support to a misinterpretation of the reported margin of error, resulting in the table erroneously under-reporting what are true statistical ties. -- Isn't the table currently under-reporting true statistical ties? That's exactly our concern here. Your logic appears to imply that we are right and many of the polling results should correctly be formatted as ties, which they currently aren't. So why would we leave the current, incorrect under-reporting of ties in place? --87.78.0.81 (talk) 16:29, 3 November 2012 (UTC)[reply]
If you want to go through the entire table article and change every instance in which the difference is less than the MOE x 1.7 to read "Tie," have at it. (And don't forget to come back and make sure that any new additions going forward also conform.) Dezastru (talk) 17:14, 3 November 2012 (UTC)[reply]
Ok, will do. But do you see my point, or am I missing something? I was hoping for an explanation of where my above reasoning may go off-track, or alternatively for confirmation that my point has merit. --87.78.0.81 (talk) 19:05, 3 November 2012 (UTC)[reply]
I've formatted the Since convention nominations / Two-way race section. On the three- and four-way race polls, I'd need help regarding the statistics. What rule would apply there? Still the same 1.7 rule? --87.78.0.81 (talk) 19:54, 3 November 2012 (UTC)[reply]
Your reasoning is logical. But you can't stop at having modified the since-convention table. You now need to adjust the pre-convention table as well. I'm not a statistician, so not sure how the figures work for more than two comparisons. Presumably it is still 1.7 x MOE for the comparison between the top two; but perhaps it should be a larger figure. You could try asking one of the Wikipedians who are experts on statistical analysis. Incidentally, if the problem was that you felt that using the term "tie" for cases in which the candidates had the same vote preference proportions, it might have been easier to just switch leading by "tie" to leading by "0." Dezastru (talk) 20:14, 3 November 2012 (UTC)[reply]
You now need to adjust the pre-convention table as well -- On it. I saved in between tables to prevent losing the edit (my Firefox is somewhat prone to crashes).
Presumably it is still 1.7 x MOE for the comparison between the top two -- Yes, that's my thinking as well.
You could try asking one of the Wikipedians who are experts on statistical analysis. -- Good idea. I'm off to WP:RD/MA, then, after I finish formatting the Before convention two-way table.
it might have been easier to just switch leading by "tie" to leading by "0." -- I was afraid it might confuse readers. "Statistical tie within the margin of error" would probably be the clearest, but it's awfully long for that narrow column. --87.78.0.81 (talk) 20:21, 3 November 2012 (UTC)[reply]

This new format of the page is making things look alot messier. I really prefer the previous way of doing it, it looks nicer, and is the same format as sites like RCP follow. By including a "margin of error" section, there is no reason why someone should get confused over who is actually ahead Green-Halcyon (formerly Aunty-S) (talk) 20:29, 3 November 2012 (UTC)[reply]

Formatting one party as the winner is clearly wrong when the lead is smaller than the margin of error. As pointed out, the lead needs to be bigger than 1.7 x the margin of error. If we keep formatting these statistical ties as not being statistical ties, it means that we're not accurately and truthfully reflecting the results of these polls. In fact, it puts us at odds with all of our core content policies. Which in turn makes "it looks nicer the other way" an exceptionally weak argument. --87.78.0.81 (talk) 20:34, 3 November 2012 (UTC)[reply]
We need to be clear. It's fine to show who got a higher polling number as long as we make it clear that it's within the margin of error. I've changed it back to the original format and added this clarity. Somedifferentstuff (talk) 10:55, 4 November 2012 (UTC)[reply]
The current format is sloppy. RealClearPolitics presents similar information without controversy. On that site, they have a column labeled "Spread," in which the candidate who is leading is named with an indication of the difference in the proportions by which that candidate is leading. The leading candidate's name is highlighted with the color associated with that candidate's party. So, for example, the entry for the 10/25-10/28 CBS News/New York Times poll shows in the "Spread" column "Obama +1" in blue.
There seems to be a misconception about this kind of presentation of data ("Formatting one party as the winner is clearly wrong"), as nowhere do we say that one party is the "winner" in a poll. We are simply indicating which candidate is leading in terms of the absolute proportions measured. Since the margin of error is noted for each poll, the reader can infer for herself whether a given poll suggests whether a candidate is "winning" or is tied. I suggested above that it would be simpler to change the term "Tie" to leading by "0" to address the concern that we were undercounting ties. While I still think leading by "0" would be the way to go if there is consensus for a change, an alternative would be to change the column header from "Leading by" to "Spread" and leave the previous usage of "Tie" as it was. Another option is that a general note could also be added to the article that would discuss how to factor in the reported margins of error when considering whether a poll suggests a statistical tie, rather than having to have editors make the calculation for each entry. This latter point, that you are asking editors to perform an extra, non-intuitive step, is really the main problem with trying to explicitly label each entry with regard to the results being statistical ties: most editors are not familiar with the concept and will not be entering the information consistently if they have to perform a calculation in order to be able to enter the data correctly.
The current state of this article, in which ONE table is formatted with one style of formatting while the other tables are formatted according to a different style of formatting is not helpful for readers and, worse, is highly misleading. If major changes are going to be implemented, they need to be worked out in the user's sandbox or on the user's page before being taken live onto the article page. We should also try to be consistent across Wikipedia, at least for a given election cycle. So the style used for the nationwide polling article (ie, this article) should be consistent with the style used at Statewide opinion polling for the United States presidential election, 2012. Dezastru (talk) 18:59, 4 November 2012 (UTC)[reply]

Suggest we post all remaining polls we can find, including Rasmussens and Gallups

[edit]

With just days remaining prior to the end of the election, it would be useful for post-election analysis to gather all the remaining posts, including tracking polls. Which polls will most accurately predict the election results? Dezastru (talk) 01:47, 2 November 2012 (UTC)[reply]

Good idea, lets do that Green-Halcyon (formerly Aunty-S) (talk) 20:23, 3 November 2012 (UTC)[reply]

Question: Nate Silver prediction

[edit]

In the New York Times Nate Silver gives the numbers: 80.9 Obama, 19.1 Romney

How does he come to this numbers? And why are they so far from the listed polls in the article? --RicardAnufriev (talk) 16:37, 2 November 2012 (UTC)[reply]

For that question, you should search his blog. There's also Fivethirtyeight#2012 U.S. presidential election. Dezastru (talk) 17:22, 2 November 2012 (UTC)[reply]

Should We Include Every Daily Rasmussen Poll?

[edit]

Yesterday's Rasmussen poll (3/12) is currently included in our list, but Rasmussen plans on doing daily polls of Obama vs. the Republican front-runners from now until the election and have done another one today as well. Obviously, the day-to-day fluctuations are kind of irrelevant and will make for a very, very long list. But if we do want to have as much data as possible, it might make sense to update this Wiki page every single day with the latest poll...otherwise, I'd recommend keeping the Rasmussen polls to a weekly thing. — Preceding unsigned comment added by 108.205.50.158 (talk) 16:09, 13 March 2012 (UTC)[reply]

I agree with your last point and think that only one Ramussen poll should added per week . All the poll names already have links to each individual poll's website, so if any readers want more polls or information, they can just click on said links. In addition, we should keep this in mind:
After Election night that year, Nate Silver of the New York Times blog FiveThirtyEight concluded that Rasmussen's polls were the least accurate of the major pollsters in 2010, having an average error of 5.8 points and a pro-Republican bias of 3.9 points according to his model. [1] Mauri96 (talk) 03:45, 17 March 2012 (UTC)[reply]
Once a week. Somedifferentstuff (talk) 10:01, 17 March 2012 (UTC)[reply]
Maybe we should make it once per two weeks? Aetherlur (talk) 22:34, 24 March 2012 (UTC)[reply]

There are several daily polls and average of polls. It makes little sense to include them all or update them daily. These would include Rasmussen, RCP, Gallup, and perhaps some others as well. I would suggest at most a once a week usage for Rasmussen and Gallup (Wednesday is my preference) and removal of RCP. I have nothing against RCP, but as a poll of polls it is probably doesn't add anything that isn't already obvious from stated polls. Arzel (talk) 00:58, 10 August 2012 (UTC)[reply]

I agree, the Rasmussen Polls should be added more regularly as it approachs the election, every week would be a good idea. RCP should be kept though, maybe only one average at a time Green-Halcyon (formerly Aunty-S) (talk) 19:50, 16 August 2012 (UTC)[reply]

Twice a month is already enough. Anything more frequent than that would be undue weight. Readers who need to see weekly or daily Rasmussen polls can click on the link to Rasmussen's website and pay a subscription to get their polling data fix. Also, maintaining that RCP Average at the top makes for a nice, clean way of presenting info that otherwise can be somewhat challenging for the reader to process quickly without a pocket calculator. Dezastru (talk) 09:33, 17 August 2012 (UTC)[reply]
Now that we are getting much closer to election day, and more polling organizations are releasing polling results with greater frequency, it might be worth considering weekly updates of tracking polls. As Mauri96 noted months ago, the polls of individual pollsters tend to skew a certain way. If the table includes weekly polls only from pollsters that tend to find results that skew in a single direction, it would have a negative impact on the balance of the table. But if a variety of polling organizations, covering the gamut of polling biases, are included, the table can, hopefully, remain relatively neutral. I haven't checked to see which organizations are now releasing weekly polls, apart from Rasmussen's and Gallup's even more frequent polls. Dezastru (talk) 01:38, 10 September 2012 (UTC)[reply]
Once a week at most for Rasmussen. Somedifferentstuff (talk) 12:50, 15 September 2012 (UTC)[reply]
Why is Rasmussen being included at all? Why not include the ones from "Unskewed Polls" while we're at it? Rasmussen has been so far off the mark this entire election, we can't take them seriously. I think it's time to take a good look at their methods and realize that including them at all is disingenuous to readers looking for solid information. — Preceding unsigned comment added by 76.180.161.206 (talk) 19:47, 28 September 2012 (UTC)[reply]
At this point, there are several other pollsters who are releasing frequent updates. I would not oppose inclusion of weekly updates of Rasmussen and Gallup, along with other pollsters. Dezastru (talk) 21:11, 1 October 2012 (UTC)[reply]
I would agree that for the last month of the race a weekly update of Gallup and Rasmussen on Wednesdays (after they post that day's numbers) would be fine. If we continue to do them every two weeks it would mean they get listed only twice more, then the election is over.--Mangoman88 (talk) 16:12, 2 October 2012 (UTC)[reply]

Last two weeks of the race. Anyone else think Gallup and Rasmussen should be added to the list once a week from now on? Green-Halcyon (formerly Aunty-S) (talk) 13:47, 22 October 2012 (UTC)[reply]

I'm fine with doing Rasmussen every Wednesday. Somedifferentstuff (talk) 16:47, 22 October 2012 (UTC)[reply]
I think Monday would be a better idea, can get 3 of them in before November 6th then. A poll on the day before the election would also be more informative Green-Halcyon (formerly Aunty-S) (talk) 19:39, 22 October 2012 (UTC)[reply]
I'm for every Wednesday. I think 2 more is plenty from Rasmussen. Let's wait to hear from some others. FiveThirtyEight currently has Obama's odds of winning at a little over 70% as can be seen here.[2] --- Somedifferentstuff (talk) 20:24, 22 October 2012 (UTC)[reply]
Hi, I havent done any editing on this article myself, but I do read it a good bit. I think every Monday is the better idea, though I can see how Wednesday would show public opinion following the debate tonight. IrishLad1916 (talk) 21:14, 22 October 2012 (UTC)[reply]

A couple of reviews have been published that have looked at the accuracy of the pollsters leading up to the 2012 presidential election.[2][3] Rasmussen and Gallup were among the least accurate in both studies. In one of the studies, that by Nate Silver, they were also judged among the most biased. Silver writes, "For the second consecutive election — the same was true in 2010 — Rasmussen Reports polls had a statistical bias toward Republicans, overestimating Mr. Romney’s performance by about four percentage points, on average.... In late October, Gallup consistently showed Mr. Romney ahead by about six percentage points among likely voters, far different from the average of other surveys. Gallup’s final poll of the election, which had Mr. Romney up by one point, was slightly better, but still identified the wrong winner in the election. Gallup has now had three poor elections in a row."[3] Dezastru (talk) 17:55, 12 November 2012 (UTC)[reply]

  1. ^ Silver, Nate (November 4, 2010). "Rasmussen Polls Were Biased and Inaccurate; Quinnipiac, SurveyUSA Performed Strongly". The New York Times.
  2. ^ Kall, Rob (November 8, 2012). "Study: Which pollsters most accurately predicted election". OpEdNews.com.
  3. ^ a b Silver, Nate (November 10, 2012). "Which polls fared best (and worst) in the 2012 presidential race". New York Times.

Help with another opinion poling article

[edit]

Hi all. Given your experiences in editing this article, I wondered if some here could offer some opinions on the opinion polling article for the next UK general election and a debate over possible OR. The key discussions are at Talk:Opinion_polling_for_the_next_United_Kingdom_general_election#Graph_to_show_tie_for_3rd_Place, with a summary at Wikipedia:No_original_research/Noticeboard#Description_of_opinion_polls_results_in_UK_w.r.t._UKIP. Bondegezou (talk) 11:51, 19 January 2013 (UTC)[reply]

Study Inclusion

[edit]

If somebody could include the nonpartisan 2012 NORC Presidential Election Study (September-November 2012), that would be nice. Data reports are on the right hand side of the following link: http://www.norc.org/Research/Projects/Pages/2012-norc-presidential-election-study.aspx — Preceding unsigned comment added by Userser (talkcontribs) 23:18, 18 March 2013 (UTC)[reply]

[edit]

Hello fellow Wikipedians,

I have just added archive links to one external link on Nationwide opinion polling for the United States presidential election, 2012. Please take a moment to review my edit. If necessary, add {{cbignore}} after the link to keep me from modifying it. Alternatively, you can add {{nobots|deny=InternetArchiveBot}} to keep me off the page altogether. I made the following changes:

When you have finished reviewing my changes, please set the checked parameter below to true or failed to let others know (documentation at {{Sourcecheck}}).

This message was posted before February 2018. After February 2018, "External links modified" talk page sections are no longer generated or monitored by InternetArchiveBot. No special action is required regarding these talk page notices, other than regular verification using the archive tool instructions below. Editors have permission to delete these "External links modified" talk page sections if they want to de-clutter talk pages, but see the RfC before doing mass systematic removals. This message is updated dynamically through the template {{source check}} (last update: 5 June 2024).

  • If you have discovered URLs which were erroneously considered dead by the bot, you can report them with this tool.
  • If you found an error with any archives or the URLs themselves, you can fix them with this tool.

Cheers.—cyberbot IITalk to my owner:Online 23:34, 1 March 2016 (UTC)[reply]