so according to the latest updates how many electoral votes would each have? we can keep a running tally if not a mpa.(Lihaas (talk) 21:51, 21 December 2010 (UTC)).
I don't think you can make that sort of forecast until there is a better idea of who the Republican nominee will be and data has been collected from all of the states. That probably won't happen within the next 15 months. Location (talk) 22:42, 21 December 2010 (UTC)
The page for statewide polling for the GOP nomination has a colour-coded map showing who is ahead in each state. Why not make one for this page as well? I would do it myself, but I don't know how to! Tiller54 (talk) 15:03, 10 November 2011 (UTC)
Done! I've created a blank map for this purpose. Blank Map I'm working on populate the map with the polling data for the different candidates. My current scheme for doing so is the following:
Is there a poll between candidate X and candidate Y? If no, then use no data.
Has there been a poll between candidate X and candidate Y since Dec 1? If no, use old data.
Does the most recent poll overlap at least one day with another poll? If so, average those overlapping polls.
Based on most recent poll (or average), does the winner win by less than the margin of error? If so, tie.
If a candidate wins outright the latest poll(s), highlight for that candidate.
If you have comments or questions, please speak up. If you don't know, this image is an SVG file, which means you only need a text editor to change it. Open the SVG file in notepad, and edit the section that starts with all the <use id="... Hopefully its pretty obvious. I'll post the images here as I get done with them, and then I'll integrate them all into the article at once. --Nick2253 (talk) 05:12, 2 February 2012 (UTC)
Obama vs. SantorumObama vs. PaulAnd now I'm done with the Santorum and Paul maps as well. I'm currently working on an infobox, and once everything is in place, I'll add it to the article (which should be by the end of the day) --Nick2253 (talk) 20:27, 2 February 2012 (UTC)
Good job! One suggestion I would make is that using faded colours for older polls was abandoned on the GOP primary page because it was too confusing and black and grey stripes were used instead. Other than that, it's very good. Tiller54 (talk) 14:31, 4 February 2012 (UTC)
The reason they got rid of the faded colors (or at least my understanding of that) was because certain colors don't "fade" well, and when you have (at the time) 6 candidates, that mean 12 different colors, plus the color combinations for ties (they weren't using stripes for ties at the time). The black/gray stripes were a compromise to prevent reader confusion with all the colors going on. In the case of a two-person contest, I think the shading works much better. That's obviously just my opinion, and if there is enough consensus, we can move to the stripe overlay --Nick2253 (talk) 21:45, 5 February 2012 (UTC)
Errors with the Maps
The map on the main page should be based on state-by-state forecasts for more accuracy. For example, Georgia and Oregon consistently show in state forecasts as two of the safest states in the nation, but this page shows "within the margin of error." All states have multiple sources for state forecasts, including Wyoming, which the current map lists as gray. As of October 31, 2012, only Colorado, Florida, Iowa, New Hampshire and Virginia have mixed result forecasts. The only state forecasts as of 10/31/2012 with less than 95% consensus are: Colorado, Florida, Iowa, Nevada, New Hampshire, North Carolina, Ohio, Pennsylvania, Virginia and Wisconsin. I tried updating this information, but had my update immediately reversed. — Preceding unsigned comment added by Foggynotion (talk • contribs) 19:52, 1 November 2012 (UTC)
I'm adding this section in order to keep a central list of problems with one or more of the maps. Hopefully, if people notice a problem, they will post here instead of ignoring it. Or, if you're like me, and you notice a problem when you don't have the capacity to fix it at the moment, post what errors you see, and then somebody (hopefully you when you have the time) will fix it. If you notice a problem, add it to the list, and if you fix it, please strike though the problem. --Nick2253 (talk) 19:47, 13 February 2012 (UTC)
These maps are great, but I notice you are still using December 1, 2011 as the threshold for "old data". Is there something special about that date? If not, it should probably be updated. When the maps were first created, that was 2 months old. The same age now would give a threshold of around February 1, 2012. I suggest declaring any poll from before then as "old". --Tango (talk) 03:14, 25 March 2012 (UTC)
Thanks! I think I chose Dec 1 because other similar articles (like the Republican Polling) used that. But I like the idea of a X month cut-off. At this point in the game though, I'd suggest using 3 months as opposed to 2 months, since many states don't update their polling, and the political climate is generally stable enough that 3 month-old polls aren't significantly less accurate than 1 week-old polls. Unless anyone has any problems with that, I'll implement that next week. There's a bunch of problems with the map and count as they stand right now, so I'll probably take an hour or two and go through and update everything (unless someone else wants to do it for me!). --Nick2253 (talk) 01:05, 31 March 2012 (UTC)
I noticed that the map for Mitt Romney has him wining Arkansas despite there being no poll for Arkansas. While I believe Mitt Romney will likely win Arkansas it seems to be the only place that is marked for a candidate with no corresponding poll. Is there a reason for this? — Preceding unsigned comment added by 24.225.64.67 (talk) 19:05, 1 April 2012 (UTC)
You are correct. That is an error. These maps are simply supposed to show polling, not other kinds of predictions. --Nick2253 (talk) 17:47, 2 April 2012 (UTC)
Section 'Obama v Republican' - Delete/Keep?
Obama will not be up against a non-named Republican, he will be up against one of those in the main body of the article. So I cant see that this section adds anything to this article. (Which is already large and will only get larger as the vote gets closer). I propose to delete this section unless anyone has good grounds for keeping it. If so, please post them here. OldSquiffyBat (talk) 15:29, 25 February 2012 (UTC)
Agree: I think there isn't any value added here. The only possible thing I could think of is if the Republicans actually have a brokered convention (as improbable as that is, though), then this section would be the best info at how such a contest would turn out. --Nick2253 (talk) 18:59, 12 March 2012 (UTC)
Maine and Nebraska split their electoral votes by congressional district, so we should be taking that into account. Right now every poll in these two states is counted for all their electoral votes. But statewide polls should really only count for 2 electoral votes in each state, since they only award 2 votes to the statewide winner. We would need to see polls of individual congressional districts to determine how the other electoral votes will go. From what I can tell only NE-02 has been polled separately so far, in a recent PPP poll, and I don't think other districts will be polled. So here are two solutions:
- We allow statewide polls in Maine and Nebraska to count for all the electoral votes excluding congressional districts that have been polled separately. In Maine this would mean statewide polls would continue to count for all 4 electoral votes. In Nebraska statewide polls would count for 4 out of its 5 electoral votes, since NE-02 with its 1 electoral vote has been polled separately.
- We only let statewide polls count for 2 electoral votes in Maine and Nebraska. Congressional districts that haven't been polled would be grey for no data.
Historically, both Nebraska's and Maine's Congressional District's electoral votes have always gone with the state as a whole. '08 was the first time in history that either Nebraska or Maine split because of electoral votes being allocated to the Congressional District. However, Obama is far less popular now than he was in '08, so the likelihood that Nebraska will split is greatly reduced. There is some value is showing those electoral votes separately from the state proper, but at this stage, we have only one poll for one District, so I personally think it's a bit hasty (and likely very unnecessary) to split them on the map. --Nick2253 (talk) 01:14, 31 March 2012 (UTC)
I think it makes sense to at least split NE-02, since the poll there shows a statistical tie, not a solid Republican lead like the state as a whole. --Noname2 (talk) 20:24, 31 March 2012 (UTC)
Remove all but Romney v. Obama map.
Now that Santorum has dropped out of the race, Gingrich is effectively out of the race, and Paul doesn't have a realistic chance of winning, I vote that we have only the Romney v. Obama map. Does anyone have any reasons why we should keep the other three? --Nick2253 (talk) 23:18, 10 April 2012 (UTC)
I concur. I was wondering if a better title would be more helpful. Example: "Electoral Vote opinion polling for US Presidential election, 2012". The current one is consistent with previous elections, but the electoral vote map is new. Just a thought but not too important for me. --Kihiu (talk) 18:15, 11 April 2012 (UTC)
Swing state section
How about a seperate swing state section, with the most up to date polling?
This would be good for gaining a quick overview.
I think there's a big enough difference between the two to warrant separation; this would also help people better understand just how many electoral votes are in the (close) swing states. 128.120.173.89 (talk) 02:20, 9 May 2012 (UTC)
Three-way race represented in maps
Some of the results in the three-way race would give a different map than those in the two-way race section. For example, Romney wins outright in Montana in a two-way race, but with Gary Johnson (who is definitely running) as an option, Obama and Romney are in a statistical tie. Should the map be edited to account for these differences, or left as is? --Platypus2012 (talk) 16:17, 30 May 2012 (UTC)
The following discussion is closed. Please do not modify it. Subsequent comments should be made in a new section.
Article is huge (Input on what to do about it requested)
Is there a way we can cut down on the size of the article here? Here are some proposals I have that could work:
Make seperate articles for each state that has alot of polls on this page present (Florida would be a good choice for example).
Make one article for all of the older polls for the states.
Make an article for the swing states (A swing state defined by multiple reliable sources).
Delete some of the polls that havent gained much since the one before it. (If Obama or Romeny gained one or two points but was far in the lead already or behind is an example)
I would love to get consensus on this, if nobody objects I will be WP:BOLD and choose an oiption as this article is a long mess that takes forever to load and edit, id rather hear input than going bold first however. - Knowledgekid87 (talk) 04:54, 1 June 2012 (UTC)
I like the idea of separate articles for each state, but that seems like it would be a ton of work to maintain fifty articles. Trying to split off swing states sounds like it could become contentious, and would we have to worry about moving states back and forth between the "regular" article and the swing state article depending on changing polls and how sources reported them? I also think picking and choosing which polls to keep and which to remove would be potentially contentious and cause issues with WP:WEIGHT. I think the easiest thing to do would be to split off all polls from before 2012 and move them into a separate article, for starters. We can see how this main article grows in size, and later on, it would be relatively simple to again split off older polls, perhaps into something like the first half of 2012. —Torchiesttalkedits14:45, 1 June 2012 (UTC)
Would it be possible to limit this page to just the poll results reflecting Obama vs. Romney? As in, removing the results between Obama and the candidates who have dropped out? Most of the polls have Obama vs. at least two or three other candidates. Or, would this not make much of an impact on the article's size? --Andy 1One (talk) 23:08, 1 June 2012 (UTC)
Splitting off the older polls into a seperate article seems to be the theme so far here, Im fine with the split, the older polls and the ones with canadates no longer in the race can be moved. All of the 3 race ones im not too sure of as there is a third party canadate running (Gary Johnson). Gary should remain in the article, other than him the others can be moved. I will give this discussion a few more days to gather more opinions then we can go from there. - Knowledgekid87 (talk) 00:23, 2 June 2012 (UTC)
Sounds good to me. One final thing I would suggest is leaving all the three-way polling in this article, since there aren't enough of them to mess with dividing up. If you plan on creating the new article, go for it, or I will get it going later today. —Torchiesttalkedits16:35, 4 June 2012 (UTC)
Well seeing that a new article was made and a solution reached, im closing the discussion. If you have other opinions there is always the talk pege here and the one on the new article =) - Knowledgekid87 (talk) 18:12, 4 June 2012 (UTC)
The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.
Define Shared Lead More Precisely and follow the definition
Some states are moved from Obama to Romney or vice-versa based upon just the most recent poll (even when a poll from the previous week favors the other candidate and the margin of victory is single digits). Other states just hang out in "shared lead" indefinitely. I propose that any state with a tied or contradictory polling result within one month of that state's most recent poll stay or be placed in "shared lead" and any state for which one candidate leads all polls that are no more than one month older than the most recent poll for that state be assigned to the leader. — Preceding unsigned comment added by128.113.122.48 (talk) 17:50, 19 June 2012 (UTC)
Three-way race article?
I think a good article can be made out of the three way polls here, one three way poll addition to this article equals Three regular polls betweem Mitt Romney and Obama in terms of size (Adds to the article) My proposal is to split the three way race from this article. Feedback? - Knowledgekid87 (talk) 22:47, 20 June 2012 (UTC)
I am going to go ahead and make the article then if nobody objects, I dont see any harm in doing so as it is not going to effect the quality of the article. - Knowledgekid87 (talk) 15:30, 21 June 2012 (UTC)
Oh, I meant to comment on this yesterday, but got sidetracked. I am slightly against the idea. Can you explain what you mean about a three-way poll being three times the size of a two way poll? —Torchiesttalkedits15:36, 21 June 2012 (UTC)
In terms of size adding a 3 way poll on average is almost 900KB or so while the addition of a 2 way poll adds about 400KB, while yes now I see it isnt quite doubled the numbers add up in terms of the article's size. I see a possible article for the 3 way race polls that could also include brief backround information about the 3rd part canadates. In short im preparing for the future of the article before things become huge again down the line. - Knowledgekid87 (talk) 15:42, 21 June 2012 (UTC)
I just did some testing in my sandbox, and while I guess creating the initial table is a little bigger, the actual difference in adding a new poll to an existing poll is very minor. It was 374 bytes (not KB), to add a row to the Obama/Romney table, and 416 byte to add a three way row to a table. I copied all the raw text from the article into MS word and let it count the characters: 160,979. Removing the entire three-way race section, it brings the total number of characters down to 149,429. So, that's only 11,550 characters for the three-way section. It doesn't seem like it would make that much difference to remove it to another new article; definitely nothing like how helpful it was to pull tons of old polls out a while back. —Torchiesttalkedits16:00, 21 June 2012 (UTC)
Well it has been 3 months now and the polls have grown, I estimate that around 20,000 characters will be freed now if make into a seperate article. I bring this up again as the early statewide polls article is at 200,000 characters and this one is as well, its not much when it comes to freeing up room but as long as it makes the article here more readable it makes sense to me (WP:SIZERULE). - Knowledgekid87 (talk) 00:52, 19 September 2012 (UTC)
No, there is no need for a separate article. The fact that some pollsters did some polls with two candidates and others with a third makes no difference in their content. Rather, these should be merged into their relevant state sections. The three-way races for Colorado should be organized with the two-way races for Colorado. Readability is in no way a problem here. The whole article is just tables cleanly sorted into many sections; just because they are wikicode-intensive with colors, links, and alignments does not make the list any more difficult to read. Your size rule link is irrelevant as it only applies to prose, not markup. There is absolutely no reason why a reader should have to go to a separate article for the exact same contents.Reywas92Talk02:09, 19 September 2012 (UTC)
Yeah, I'd still prefer moving old polls to the "early" polls article instead of removing 3- or 4-way polls. I did just notice something we're doing that is really bulking up the article size significantly. We're writing margin of error and sample size for every single poll. We should just change those into two new columns. It would cut thousands, maybe tens of thousands, of characters from the page size. —Torchiesttalkedits14:42, 19 September 2012 (UTC)
I would agree with that. Any poll with Santorum, Gingrich, or Paul is outdated and meaningless in this article. The particulars about a poll are still good info, though. Reywas92Talk19:48, 19 September 2012 (UTC)
They're tough to find because all the tables are collapsed, and the formatting is pretty messed up on most of them, but I found at least four states that have them integrated into the main lists: Colorado, Montana, New Mexico, and Pennsylvania. —Torchiesttalkedits23:04, 19 September 2012 (UTC)
Is Florida really outside the margin of error?
The margin of error is above 2, and the difference is 4. From Wikipedia...
In a plurality voting system, where the winner is the candidate with the most votes, it is important to know who is ahead. 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.[9][10] 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. The survey results also often provide strong information even when there is not a statistically significant difference. 108.18.242.227 (talk) 03:11, 23 June 2012 (UTC)
According to the poll Obama holds a 1.6 lead over Romney when you factor in the Margin of error, while yes not alot this still puts Obama ahead. As for the map ot reflects the most recent polls and not them overall. Now we could make this like the Huffington post to reflect the overall change (Florida is a tossup there) but I am not sure how that would work here and where you define the start date. - Knowledgekid87 (talk) 04:11, 23 June 2012 (UTC)
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)
Yeah, this is pretty irrelevant. Rasmussen was one of the worst of the major pollster's by Nate's rankings in 2000, and it was one of the best in 2008. Until we know the actual results of the election, any conversation on "house effect" is irrelevant. Rasumssen's tendencies to show a pro-Republican bias may be just that, a bias, or it might point to them capturing or accounting for an actual voter dynamic that other pollsters fail to pick up on. The purpose of this article, in my opinion, is to report the polls, plain and simple, without any modification or weighting. If you want that kind of an analysis performed, I'd recommend following any of the major scholars in this area, like Nate Silver or Larry Sabato. --Nick2253 (talk) 21:34, 19 July 2012 (UTC)
Margin of Error
I don't think we are currently dealing with the margin of error correctly here (see Florida discussion above). Let's say a candidate receives 45% of the vote, for example, with a margin of +/- 5%. The margin is the 95% confidence margin. What this poll is telling us is that 95% of the time (or 19 polls out of 20) a poll will actually capture the true public support for that candidate within 5% of the surveyed support. In this case, that would be somewhere between 40%-50%. If a second candidate receives 51% of the vote, even though the margin is greater than the margin of error, that candidate's true support would be captured (95% of the time) between 46%-56%. In other words, they could be tied, even though the margin is greater than the margin of error. States should only show for one candidate or another if the polling margin is greater than twice the margin of error. --Nick2253 (talk) 20:40, 19 July 2012 (UTC)
I've gone ahead and corrected the map as per above. This produces some counter-intuitive results (see TN and SC), but the most recent polls in both states are done by smaller pollster outfits with small sample sizes, so it's not really like they mean a whole lot. However, I don't think we should do anything to "correct" these results. As I've said before, our job on this page should be to simply report the polls and their status, not to analyze or speculate. --Nick2253 (talk) 04:54, 20 July 2012 (UTC)
Looking at the sources Washington Post, [1], [2], [3] Noplace do I see states like Montana and South Caroline as being tied (Swing state status) we should follow closer to the reliable sources rather than going on our own here. - Knowledgekid87 (talk) 23:44, 20 July 2012 (UTC)
I'm not sure about that. Those pages are reporting the results from other polling organizations. If we go directly by what the organizations are reporting, and use their information, we're not doing anything but following the same statistical analysis they are. But the MoE problem is ubiquitous, so it's not an easy problem. —Torchiesttalkedits23:54, 20 July 2012 (UTC)
The Huffington Post source was last updated on: Friday, July 20 5:43 pm ET I use that source when it comes to placing polls on the page here as when you click on a state info from the poll sources is presented. Here is an example: Huffington Post (Montana), Montana in the example is considered a strong Romney state. - Knowledgekid87 (talk) 00:42, 21 July 2012 (UTC)
The problem with using the Huffington post (or any other similar site, such as RealClearPolitics) is that isn't the point of this map. They provide analysis; we simply report polls. Montana is almost undoubtedly going to Romney, and it's actually because of that certainty that few polls have been conducted in MT. Nevertheless, the polls that have been conducted in MT have been polls with small sample sizes, which leads to large margins of error. Even though the poll shows a pretty substantial Romney lead, and its results mesh with existing analysis, the margin of error effectively "invalidates" that lead. --Nick2253 (talk) 02:17, 22 July 2012 (UTC)
Types of Polls
I was looking through the polls here, and a number of polls seem to be labeled inconsistently. I think it's clear to use RV for registered voters and LV for likely voters, but what about polls that survey all possible voters (even those not registered to vote)? PV? Should we include those polls? Is there a standard used elsewhere on Wikipedia? --Nick2253 (talk) 20:44, 19 July 2012 (UTC)
I've checked them about ten times and came to the conclusion that there was some kind of mistake. The number of electors from states "tied within the margin of error" was by 20 points too low (probably Tennessee and South Carolina were ignored in the count because they have always been "red") and because of that the number of Obama's electors 20 points too high. That's why I corrected them.
89.12.42.228 (talk) 14:25, 27 August 2012 (UTC)
Who in the world came up with this map? I mean Georgia, South Carolina and South Dakota listed toss-ups? That's just flat out preposterous. — Preceding unsigned comment added by Thismightbezach (talk • contribs) September 4, 2012
It's pretty simple and objective. We need actual polls that show a lead greater than twice the margin of error. For instance, in the SC poll used for the current map, Romney's lead is 6 percentage points, but needed to be at least 7 percentage points to overcome the margin of error. For SD, again the lead is 6, but needed to be 9 to overcome the MofE, due to the lower sample size. But who's going to do a new, larger poll in SD? --Spiffy sperry (talk) 15:33, 21 September 2012 (UTC)
Map needs an update
Not much but the map needs an update, as of 9/14/2012;
Thanks as always for your work here, I made these updates to the map and infobox, with the exception of Ohio. Let's wait for another poll there, or at least 2 of the 3 most recent, to show a lead greater than the plus and minus margin of error. I suspect the NBC/WSJ/Marist Poll is just showing Obama's post-convention bump there, and it will return to being within the margin.-- Patrick, oѺ∞23:28, 18 September 2012 (UTC)
Okay, will do. I notice that a second Ohio poll, from Fox News, has Obama above the margin of error. So though I still expect in to return to being within the margin of error, I'll change that too now.-- Patrick, oѺ∞23:48, 20 September 2012 (UTC)
District of Columbia (3) - 80% Vermont (3) - 37% Hawaii (4) - 27% New York (29) - 23% Maryland (10) - 22% Massachusetts (11) - 22% Rhode Island (4) - 21% Illinois (20) - 19% California (55) - 16% Connecticut (7) - 15% Washington (12) - 14% Maine (4) - 12% New Jersey (14) - 12% Delaware (3) - no polls
The source for the map is this entire article, and is discussed above in this talk page (1 and 2). Any mismatch between the map and its caption is due to the complexity (believe it or not) of adding up 51 numbers, some of which are constantly changing categories, and the same issue would crop up with the table you propose. Also, updating the map is non-trivial, while updating the caption is easy, so there is sometimes a lag between their updates.
The difference between the map and your proposed table is the one-size-fits-all approach for the table's categories. The map is based on margins of error for each individual poll. If the sample size is large enough, a 5% lead may be enough to declare a state red or blue, but if the sample size is smaller, a 9% lead may be necessary for such a declaration. However, in the table, everything with a >5% lead is in the "Likely" category, regardless of margins of error.
So in sum, the map and its electoral vote counts are based on the data in the underlying polls. The proposed table and its electoral vote counts, while insightful, are based on arbitrary percentage cutoffs. --Spiffy sperry (talk) 16:33, 27 September 2012 (UTC)
Though I like the idea of an overview, you'd need to frame the overview in terms of the margin of error to be useful, not just an arbitrary percentage lead. However, that won't necessarily add anything to the article that the map doesn't already provide. To be more detailed, you could break the categories out as follows: Tied within one MofE, tied within two MofE, leading outside of two MofE (or something like that).
Furthermore, classifying Delaware, Alaska, and Wyoming as "Safe" one way or another is clearly WP:OR. There's no polling in those states and, though I will personally agree with your assessments, those assessments fall outside the scope of this article. --Nick2253 (talk) 01:00, 26 October 2012 (UTC)
It would seem that Washington DC has been coloured in blue despite the fact that there haven't been any polls there. Obviously, Obama is going to win the district, probably with over 90% of the vote, so this isn't an issue. There are now only 3 grey states left - safe red Alaska and Wyoming and safe blue Delaware. If DC has been coloured blue why not colour these three in as well? We all know how they will vote and having them left out of Obama and Romney's electoral votes totals is odd. Tiller54 (talk) 15:39, 24 October 2012 (UTC)
Ah well that makes sense then! I hadn't seen anything about that poll. It still looks odd for solid red and blue states to be greyed out though... Tiller54 (talk) 21:52, 24 October 2012 (UTC)
I agree I think we should color the remaining states, Wyoming is one of the most conservative states in the country for one, second the majority of sources have them colored in as solid Romney/Obama. - Knowledgekid87 (talk) 22:01, 24 October 2012 (UTC)
I'm reluctant to color anything we don't have a source for. Wikipedia isn't a prediction site, this is just the facts as reported by the polls. Let people do the math themselves.-- Patrick, oѺ∞23:53, 24 October 2012 (UTC)
Out-of-date tag for map
An editor has tagged the page (twice) as out-of-date due to Oregon's status on the map. To clarify how this map works, there are some things to consider about the margin of error (MofE), the first of which has already been discussed on this talk page:
MofE is a measure of 95% (typically) confidence for each candidate's percentage, not the difference of percentages. If you represent each candidate's percentage as a range using the MofE, then they will overlap (thus be counted as "Tied" on this map) unless the lead is twice the MofE. This is why a 7 percentage point lead in Oregon is showed as Tied when the MofE is 4.2%. This technique is not original research, due to WP:CALC. If you don't like this result, then one remedy is to wait for a poll with a larger sample size and thus a smaller MofE.
There is a more difficult statistical analysis for comparing the differences of percentages (which is only partially covered at Margin of error#Comparing_percentages). Using this method, the lead doesn't quite have to be double the MofE to call a state for either candidate at the 95% confidence level; and then Oregon (Dem), Georgia (Rep), South Carolina (Rep), Tennessee (Rep), and maybe Arizona (Rep) could currently be changed to solid colors. The map on this page does not delve into that kind of statistical analysis due to WP:OR. If you want to see a map that uses this method, you can visit the various polling sources cited in the External Links section or your favorite news source.
- The article should have a section with snapshots (map and electoral votes) from August, September and after the major events in the campaign (TV debates, Sandy...)
- The article could have an evaluation of the poll sources' last result and the actual election result.
