2012 Presidential Election Polls
 


“Road to 270″ Enhancements

July 11, 2012

Below the electoral map on our site, there’s a feature you may have seen called “The Road to 270″.     This feature calculates the number of ‘critical path’ combinations to 270 that remain for either party based on the undecided states in your map, as well as any possible tie combinations.   It updates automatically each time you change the status of a state.

By critical path, we mean those combinations that are available to cross the required 270 electoral votes threshold.   To take a simple example, let’s say Obama has 262 electoral votes on your map, and the only two undecided states are Virginia (13 electoral votes) and New Hampshire (4 electoral votes).   While there are two overall combinations there (VA or VA+NH), only one of them is relevant to our goal of 270.  Put another way, if you were managing the Obama campaign in this scenario, how much resource would you devote to winning New Hampshire?

The new features include the following:

  • Must Win for 270:    Displayed separately, these are any states that, if lost, would give the other candidate to 269 or more electoral votes.   Therefore any states in this category will be required in every critical path to 270.
  • Not Needed for 270:   Displayed separately, these are states that are not required in any critical path, like New Hampshire in the earlier example.
  • Filtering:   For all the states not fitting into one of the above two categories, we now display the number of combinations that include that state.  You can check one or more states to filter to only those paths that include all the checked states.
  • Totals:  The total electoral votes associated with each combination are shown
  • Probability:  The combinations are ordered by probability of occurring (based on polling averages).   Note that probability doesn’t necessarily mean “probable”.  For example, if Obama has 268 electoral votes on your map, and the states remaining are Montana and Kentucky, Montana will appear higher than Kentucky on the list, but we’re talking about “improbable” in the case of Montana, and “extremely unlikely” in the case of Kentucky.    Note also that when dozens of combinations remain, with multiple states in a combination, almost any single specific path is going to have a low probability, particularly if it involves toss-up states.

The Road to 270 also calculates tie scenarios that are possible with your map.   Every remaining state is involved in a tie scenario, of course.   The Ties page shows each possible tie scenario and how the states break out between Obama and Romney.   These tie scenarios are not in any particular order at this time.

The Road to 270 feature appears when 12 or fewer states remain undecided on your map and neither candidate has reached 270 electoral votes.

  • http://www.examiner.com/libertarian-in-charlotte/matthew-reece Matthew

    Could you add Gary Johnson to the electoral map? It would be nice to be able to look at what happens if he can take off in the polls and win a few states.

    • Allan

      It would actually be a major overhaul to the site to add a 3rd party. If a candidate like Gary Johnson becomes viable (in the context of winning states), we’ll figure out a way to make the change, but absent that there are other projects on which we’ll be spending our limited resources this year.

      • Duey

        Gary Johnson and other 3rd party might be the spoilers in this years very tight elections. Most polls have the 2 candidates at just some where above 45% with 5 to 6% going to other. Right now it is in the realm of possibility of a 266 266 and 4 to a 3rd party. It only takes one small state to make a difference. Shame on you, do the work and put the 3rd party candidates on your poll.

        • Chris

          There’s almost no way that Gary Johnson or other 3rd parties affect the electoral map. They may swing some states one way or the other, but they won’t win any states. This was demonstrated in 1992 when Ross Perot won over 20% of the popular vote and ended up with a grand total of Zero electoral votes.

  • Rick

    Just out of curiousity, is there any way to make Maine an Indy in the 2012 senate map to reflect a King victory?

    • Allan

      We’re trying to come up with something.

      • Rick

        Just out of curiousity, can’t you just do whatever you did for Crist in 10? And no offense, I realize your funds, etc. are limited, but I agree that there appears to be a pretty strong tilt towards Obama. I’m not sure how any of this programming, scripting, etc. works, but it seems to be a bit out of date, though I don’t like the way mark worded it.

  • mark

    With the staggering amount of new polls out, it would seem that this website is seriously outdated. i don’t often subscribe to conspiricies but this may be an attempt to bias the election by showing Obama with a big lead. Why no updated map for such a long time? Is it just laziness on part of the website operators? or is it bias. As I stated, numerous polls have come out since any updates on the electoral map. My question is simple. Why is 270towin not updating at all? If you are afraid to show Romney is in a dead heat with Obama just say so and start a lemonade stand. Your time will be better spent that way. I hope you guys get your act together and soon.

    • 270toWin

      Thanks for the feedback. If you ran a website like this, you’d want the race to be as close as possible. It would be poor business to favor anybody. The dead heat polls are national; this site limits itself (and always has) to state-level polling and the electoral vote, as we don’t have a national popular vote election. The state polls haven’t moved much; Obama remains even or ahead in most battleground states. You may disagree with the approach, but please don’t equate one candidate or another being in the lead with bias.

      • mark

        “you’d want the race to be as close as possible” and “poor business to favor anybody” your quotes. why not have an open and honest site about where the race actually stands? most polls are weighted heavily for democrats at my estimation averaging dem +6 I have not seen this website mention that at all. Oversight or intended?

        • mark
        • Edwin

          mark: the goal of this site (which I find admirable) is to use statewide polls from all reputable sources and create probabilities for the upcoming election. There is no attempt by the creators of the website to manipulate the data and no attempt to give a preference to one candidate or the other.

          The fact that your candidate is not doing well in the Electoral College (a fact corroborated by many, many other non-partisan political analysts) is not evidence of bias. Romney *could* be ahead in the national poll yet lose the election by a large margin (remember Gore/Bush in 2000?).

          As for your claims about poll weightings being biased, I doubt it. The polling companies are the same ones that predicted the last few elections exceptionally well. But even if the companies ARE doing it, that is not the business of 270towin. Their obligation is to use all reliable statewide polls — not selectively accept only the polls they think are on the mark. Indeed, if they did this, they would be creating bias.

  • Alex

    Mark the website you linked is a conservative biased site to begin with. That automatically puts any polling numbers they post as suspect (the same is also true with liberal websites). I am a registered member of the Republican Party living in a ready state, and consider myself to be a political Moderate.

    270towin.com has stated that they use state polling data in their calculations. The staff here use multiple state polls as opposed to national polls as the state polls are a more accurate representation on how the residents of a particular state are leaning politically.

    The problem with using national polling data is that, unless your polling sample is large enough, the data you get will not be an accurate representation of who voters are likely to vote for.

    The poll provided in the link you shared had a sample population of 100. What the poll doesn’t tell you is where the 100 people were from. If this was a true national poll then that means 2 people from each of the 50 states were sampled. There is no way that you can establish any hard data relying on just two voters from one state.
    Can 2 people in Flordia really provide enough data to establish which way the state is leaning politically? The answer is of course no. Had the poll been a state poll surveying 100 people in one particular state, the data it produced would be more valid. The data would not be perfect, but it would be a more accurate representation.

    Now had the link you provided shown a national poll sample of 50,000 with 1,000 respondents for every state then you could take those numbers to be more accurate.

    But with conducting a national poll of 100 with what was probably calls to random phones across the country, you cannot get a clear picture. For all we know 40 people in Texas are respondents, or 33 hail from Virginia. There is just no way to no as the website did not put forth their polling methodology.

    Also you seem to be attacking this website claiming that they are biased towards Obama. I can assure this is not the case for the following reasons:

    1) It is not in their best interests to favor one candidate over the other. They could stand to lose credibility or ad revenue should this occur. This website is still relatively young when compared to similar sites. This along with limited resources inhibits what upgrades they can make and how much time the operators can spend on the site. It’s my guess that 270towin.com is not their sole source of revenue in life. They probably have a real life job they have to attend to while maintaining a wesbite that only sees a high volume of traffic once every four years or so.

    2) I’ve been using this site since 2004 and have noticed no biase to any one political party. They showed Kerry no favortism in 2004 nor Bush in 2004. The same can be said for Obama and McCain in 2008. I’ve also seen no hint of bias this year either.

    3) I’ve independently checked to see if the polls they show for each state are accurate from their sources. So far I have never seen this website fake any poll data for any election. They have an 8 year track record of running a fair unbiased website. If you want data, then that is the number you should focus on.

    4) This website is free for all to use and even comment on. Heck you don’t even gotta sign up to post comments or use any features. But if you feel that this website is spreading deceit for one party or candidate, you do have the choice to stop using this site. This is a free product, and you are receiving a free service from it. Now I don’t want you stop using this website as I enjoy a good political discussions. But I understand if you refrain from using this site in the future given your opinions about its credibility. Which leads me to:

    5) You have made unsubstantiated claims against this website which you have failed to backup with any facts. You also come across as being biased against Democrats as well given your statements. This hurts your personal credibility here on the site. Your personal attacks on the staff of this website have been made without any empathy in regards to their real life situation. You never bothered to ask if real life issues such as a job load, lack of free time, personal health, etc… have kept them from updating the site to your satisfaction. Heck I don’t even know how big the staff of this website is, for all we know it could be 1 person trying to manage all this. Try to show a little more compassion is all I am saying. I do apologize if you are offended by any of this, but this is the way I see your actions. Others may interpret your actions differently than I.

    Alex

    • mark

      Not offended at all. I read your post and you make a couple good points. however you failed to address the overwhelming over sampling of democrats. I quoted the website itself . and since when is CBS/NEW YORK TIMES a right leaning poll? I enjoy the website, but surely you can agree that it needs to be updated as several new STATE polls have emerged recently….thank you for the good discussion

  • Alex

    Yes I would definitely like to see every state poll be as up to date as possible. However I also realize that the operators may not have the time or personal to check for new polls every day. Perhaps the 270towin.com staff can tell us how big their staff really is and how they operate. I think this would help shine a clearer light on your questions.

    In regards to the oversampling of Democrats, I’m not sure what you mean and where you are getting that data. Are you saying that 270towin.com is over sampling democrats or just polls across the country in general?

    The poll you linked showed that 6 more democrats responded than Republicans, but I cannot say that one poll shows a national trend.

    Oversampling one party over the other in polls is going to happen in some states. Pennsylvania for example has 1 million more registered Democrats than Republicans. So the chances for a larger amount of respondents being Democrats is greater. So you have to take numbers like that into account as well as luck of the draw, or luck of the phone call in this case.

    • mark

      Presidential elections are only every 4 years…staff size is a lousy excuse for out dated polling. And the major pollsters dont rely on who ever answers the phone….they set the parameters before hand. Example: latest CBS/NYT poll had a Dem +11. and their result was a huge lead for Obama of course…they poll to get the result they want. its a setup to drive down voter participation. And if you can provide me with ANY poll that has oversampled Republicans, I would like to see it.

      • mark

        Mitchell Communication just released a state poll of michigan showing romney leading. Realclear politics has Michigan as a dead heat, yet this website has michigan solid Blue. Time to update now?

        • Allan

          There was another poll today that had Obama +14 in Michigan, and a Rasmussen poll yesterday that had Obama +6. Lots of variability. All Michigan polls we’ve recorded this cycle can be found at http://www.270towin.com/2012-polls/Michigan/. You can find polls for any state by visiting http://www.270towin.com/2012-polls/ and clicking the state name. That page also lets you sort on a variety of other elements. We also have an electoral map based on polls, which you can find at http://www.270towin.com/2012-election-polling-map/obama-romney/. Michigan is a toss-up on this map, and actually has been since June 14.

          On the home page, we present a variety of starting view options for users to create their own maps, including a ‘toss-up’ states map. The methodology for that map can be found here: http://www.270towin.com/blog/2012-election/toss-up-states-june-1-update. The two should converge as the election nears and polling (which is just a snapshot in time) better incorporates how people are really thinking as they get ready to vote (i.e., the economic and political conditions in late October will decide how people ultimately vote).

          I don’t think we’ll be able to convince you we’re not biased, and that’s fine — we get emails saying we’re biased for Romney as well — but wanted to share with you some of the content on our site that you may not have seen. Hopefully, you’ll be able to use the site for its primary purpose, which is to let you create (and share) your own prediction maps.

          • mark

            Just a little more food for thought.. A survey by Mitchell Research & Communications showed the race is a statistical dead heat between President Barack Obama and presumptive Republican challenger Mitt Romney, with Romney leading, 45 percent to Obama’s 44 percent. …

            Also last month, the Mitchell survey showed Obama in front, 47 percent to 46 percent.

            “Mitt Romney’s home state continues to look as though it is going to be a battleground state this year,” said Steve Mitchell of Mitchell Research & Communications.

            The results are consistent with other polls showing Romney closing the gap since becoming the presumptive GOP candidate. Obama once had a commanding lead in Michigan, as much as double figures in one May poll.

          • mark

            I dont think it’s bias. I think the site could go more in depth about shady polling practices. I enjoy the website.

  • mark
  • Jeremy

    Shouldn’t michigan really be changed to a swing state on the 2012 electoral map?

    • mark

      YES IT SHOULD…

  • Jeff

    Will this site eventually have a 2012 House election map?

    • Allan

      There isn’t yet a commercially available map that is adaptable for our configuration that includes the new District borders. From speaking with map vendors, it is possible that there won’t be one until 2013.

  • Alex

    http://news.yahoo.com/blogs/ticket/obama-leads-romney-three-battleground-states-poll-shows-103942709.html

    excerpt:

    President Barack Obama holds an edge over Mitt Romney in Ohio, Florida and Pennsylvania, a new Quinnipiac University/New York Times/CBS News poll out Wednesday shows.

    When likely voters in the three battleground states were asked whom they would vote for if the election were held today, here’s the way it broke down:
    •Florida: Obama, 51 percent; Romney, 45 percent.
    •Ohio: Obama, 50 percent; Romney 44 percent
    •Pennsylvania: Obama, 53 percent; Romney 42 percent

    The polls were conducted by telephone from July 24-July 30 among 1,177 likely voters in Florida, 1,193 likely voters in Ohio and 1,168 likely voters in Pennsylvania. The margin of error was plus or minus 2.9 percentage points in Florida and Pennsylvania and 2.8 percentage points in Ohio.

    • Allan

      There were several other polls today as well. All polling info available on our site at http://www.270towin.com/2012-polls/.

    • mark

      Some bad polling as usual. the crosstabs on this “poll” is weighted by Dem +8 in every swing state. Nothing to see here..Move along

  • Alex

    Mark where are you seeing the data that says +8 Dem in every swing state?

  • mark

    No one on this site can provide ANY poll that ever oversamples republicans…that is called bias plain and simple. When EVERY SINGLE POLL is using an absurd Dem +6, 8, or even 11 in the final product.

  • Alex

    Mark I’m going to be honest and say that I ignored the link you posted to the breitbart.com website. That website and its namesake are extreme conservatives. I did however follow the links in that article to Pew Research and Nate Silver. Those two websites do seem to say that Dems are oversampled when surveys are of Registered Voters by 4-7%. So the data from those articles does seem to back up your claim of all polling been biased towards the Democrats. I don’t personally agree with that claim as I find it hard to believe that every poll in the country favors Democrats over Republicans. I think seems to be highly conspiratorial and illogical especially when random chance is involved. However I cannot provide any data to refute what the data shows in those two articles.

    However both websites agreed that Polls taken of likely voters tend to be the most accurate. The Pew Research article (link here: http://people-press.org/files/2011/03/UnderstandingLikelyVoters.pdf) which was found in the breitbart.com article you linked stated that when LikelyVoters are surveyed Republicans are favored slightly more than Democrats. So likely voters seem to be the least biased polling sample.

    With that said, the poll data I shared was made up of likely voters. So that seems to suggest the numbers are accurate given that the bias was supposed to be for the republican candidate.

    Alex

    • mark

      Glad you took the time to fact check. as far as Breitbart is concerned.. their reporting is spot on…even if they are very conservative…my opinion

      • Jonathan

        Mark, thanks for your honesty. You definitely are entitled to your own opinion. If you trust Breitbart, I don’t honestly think you are going to trust any other source that shows any positive news about Dems. I salute you and praise you to carry on with your fight for the glory of the Republicans!

    • FthisState

      “So the data from those articles does seem to back up your claim of all polling been biased towards the Democrats. I don’t personally agree with that claim as I find it hard to believe that every poll in the country favors Democrats over Republicans. I think seems to be highly conspiratorial and illogical especially when random chance is involved. However I cannot provide any data to refute what the data shows in those two articles.”

      So you did your own research and the data told you that polling was biased towards the democrats, but you simply find it hard to believe the data? What a ridiculous claim. This is why our world is in the state it is in today. People won’t admit they are wrong and cling to their ideas regardless of the facts presented to them.

  • mark

    Latest poll with a dem +6 http://hotair.com/archives/2012/08/13/politicogwu-poll-shows-romney-obama-in-virtual-tie/dvantage. I am making it my mission to help others see the truth about these bought and paid for polls…Your welcome.

  • Alex

    http://hotair.com/archives/2012/08/13/politicogwu-poll-shows-romney-obama-in-virtual-tie/

    The above is a working link to the poll mark posted, for whatever reason the link he provided does not work.

    The follow link provides access to two links to the data retrieved from the poll:

    http://www.politico.com/polls/politico-george-washington-university-battleground-poll.html?hp=l1_b3

    The links you want to click on for the data are both PDF and titled “Complete Poll Results”
    http://images.politico.com/global/2012/08/politico_gw_bg_48_questionnaire.pdf and
    “Polling Data” http://images.politico.com/global/2012/08/politico_gw_bg_48_tables.html

    Mark,
    According to the Data collected this poll was conducted of a sample size of 1000 voters who identified themselves as likely voters. Page 13 of the Complete Poll Results link I provided shows a table which breaks down party affiliation. According to the data 40% of respondents identified as Republican. Independents made up 15% of respondents. 44% identified themselves as Democrats. 1% identified themselves as something else. This is hardly the +6 Dem you keep advocating for. Please keep in mind that according to the pollsters there is a 3.1% margin of error.

    In the “Polling Data” link I provided above, on page 408 there is a table which breakdowns political mindset. 57% identfied as Conservative, 6% identiied as Moderate, and 37% identified as Liberal.

    This is one of the most comprehensive polls I’ve seen and the amount of data is quite staggering. THE TARRANCE GROUP, INC. / LAKE RESEARCH PARTNERS are two seperate entities which conducted the poll. The Tarrance Group, INC is a national Republican Polling Firm. Lake Research Partners seem to me to be more closely tied to the Democrats than Republicans based on what I’ve read. However it should be noted that they have also worked with Conservative groups like Planned Parenthood.

    Alex

    • mark

      So you have to admit the poll is skewed. It over samples dems and under samples Independants…..And since when is Planned Parenthood a conservative group? Thats a new one I must say…Next?

  • Alex

    I don’t think the poll is skewed at all. The way these things are conducted, it is whoever answers the phone first and participates. In this polls case 4% more Democrats answered than Republicans. This is just random chance and not some continued conspiracy as you want to believe. The pollsters have go with the first 1000 people who participate in order to maintain the parameters of the polling. This is standard methodology of polling. Would it be better to make each party have equal representation? Yeah probably so. Would the data be accurate? I don’t know. Also the true number of pure (non-partisan) Independent voters is around 22% according to data I have seen.

    Alex

    • Jonathan
    • mark

      So to be clear….In every single poll more democrats answer the phone? Absurd. Your argument would hold water if you could give a single example or poll in which Republicans outnumbered Democrats…But isnt it funny how you cant cite not a single poll…HHMMMMM

      • Jonathan

        Why don’t you ask Karl Rove the same question? He doesn’t have a problem with these polls.

  • Andrew

    Romney is going to win every state that McCain won in 2008. Of the states that Obama won in 2008 that Romney will win in this election will be New Hampshire, Virginia, West Virginia, North Carolina, Florida, Ohio, Indiana, Wisconsin, Iowa and Colorado, giving Romney the victory. Obama will most likely keep Pennsylvania, Michigan, New Mexico and Nevada.

  • Alex

    Obama didn’t win West Virginia, heck that would be the biggest blow ever to dealt to conservatives outside of him winning Texas.

    • Edwin

      Utah would be more shocking than Texas.

  • mark

    http://hotair.com/archives/2012/08/22/ap-poll-narrows-obama-lead-to-one-point-4746/ With a Dem oversample of +6 of course. no denying the bias of these polls any longer. The only purpose of these skewed polls is to bias the turnout. there is NO OTHER EXPLANATION.

  • mark

    Latest poll sample follies. Let’s just cut to the chase. What do the partisan splits in the samples look like? Let’s lay out all three states and compare the D/R/I of this poll to 2008 and 2010:
    •Florida: 34/28/32; 2008 37/34/29, 2010 36/36/28
    •Ohio: 34/26/34; 2008 39/31/30, 2010 36/37/28
    •Wisconsin: 32/28/33; 2008 39/33/29, 2010 37/36/28

    Once again, we have a significant under-representation of Republicans in all three states, especially against the 2010 model. And yet, Barack Obama doesn’t seem to be faring too well even with the boost. It would be nice for the American voting people to get the real state of the race with accurate polling.

    ….

    • mark

      This is from the latest CBS/NYT/ and Quinnipiac pollsters….or I should say fraudsters..

  • mark
    • Carlos

      Just want to let you know, there are more registered democrats than Republicans, more Democrats vote than Republicans, so of course the polls reflect that more Democrats respond. However, Democrats don’t vote along party lines as much as Republicans do (just slightly) to account for this difference.

      • mark

        Sorry Carlos. these latest polls are likely voters…not registered voters.

  • Alex

    Mark you missed Carlos’ point I think. He is right by saying that more Americans are registered as Democrats than as Republicans. I’ve seen numbers that say Democrats outnumber Republicans by anywhere from 10-33 million. There is no way to get an exact numerical total that I am aware of. But the consenus I have seen from multiple websites is that more Americans are Democrats than Republicans.

  • mark

    Polls are supposed to be based on the last national elections of 2010 in which republicans won in a wave election. The dem/rep/Ind break down was almost even with the dems getting 36% and the reps got 35% Ind. made up the rest. Hardly the +6, 8, even 11 that the “pollsters” are using now to over sample democrats…

    • Edwin

      Why should they be based on 2010? State demographics have changed since then. And, as you admit, it was a “wave” election — and waves do not typically follow each other. Sometimes, but not commonly.

      So… the more likely scenario is that this will not be a republican wave election year (again, not impossible… but unlikely). Given that, what responsible polling organization would use political party demographics from 2010 — except as a starting point.

  • Justvote

    Who cares about polls anyway, when it really comes down to it you’ve got to make a choice based on your own viewpoint of life, not someone elses.

    • Edwin

      For political junkies, it gives us our “fix” — especially the simulation feature. It’s too bad the states cannot be “lumped” together to avoid odd changes (like NJ red but VA blue)… but that would take some really messy programming for relatively little benefit.

  • Alex

    Why would you base polls on data from two years ago? Justvote has it right, individuals will decide this election. The American people have always shown sound judgment when choosing their leader out of the candidates provided.

    • mark

      The American people deserve to know the real state of the election…not biased polls trying to depress turnout. And Americans in order to make a sound decision need reliable information…..Americans dont always use sound judgement…remember Jimmy carter? The guy was a disaster!

    • Edwin

      I disagree with how sound the judgement of the American people is, but I appreciate your optimism.

  • Alex

    Most Americans don’t even pay attention to the polls. They’ll make up their minds based upons soundbites, news clips, tv ads, and the debates. Carter wasn’t a disaster of a President. His failure was letting the Iranian hostage overshadow the remainder of his term and the 1980 election.

  • mark
    • Ed

      Mark, dude:

      There are more Democrats than Republicans:
      http://en.wikipedia.org/wiki/File:U.S._party_affiliation.png
      (I hope that doesn’t make you feel too left out)

      To add to that, southern democrats tend to be less likely to vote “Democrat” in national elections. Polling bias has nothing to do with it. I don’t cry “Bais!” every time a poll comes out where they “throw away” more Democratic voters than Republican ones because they are less likely to vote.

      http://fivethirtyeight.blogs.nytimes.com/2012/08/19/aug-18-obama-leads-big-among-those-least-likely-to-vote/

      Nate Silver (who runs the above site) is an Obama supporter (or was, last I heard) and his polling has been excellent; he works hard to be sure there is no bias in his predictions (his 2008 and 2010 results are spot on) and he actually as a de-biasing factor for each poll, in case you are interested as to what the REAL bias of these polls are, and NONE of them are skewed by anything near 11 points:

      http://fivethirtyeight.blogs.nytimes.com/2012/06/22/calculating-house-effects-of-polling-firms/

      • Ed

        P.S.:

        I said Nate’s “polling” has been excellent, which is an error; he doesn’t do any polling. His predictions, based on weighted polling averages, has been excellent, accurate, and any other positive adjective you can think of.

  • Magnum Serpentine

    I was wondering if you had any safe guards against a bunch of people coming on the site and filling out a map to reflect a win by Mitt Romney. I think the greater GOP community has gotten thousands to sign onto this site with the sole purpose of skewing the results and showing a Romney win. I remember when this site first started and the results on the user maps reflected the current real world trends in the 2012 elections,

    Thank you

    • Allan

      This project is a first-time experiment, so there’s no way of knowing until the election what predictive value it will have as a forecast. The underlying theory is that by offering prizes for accuracy, people will be motivated to predict what they think will happen, as opposed to what they’d like to happen. If it ends up being wrong, so be it.

      While it can’t be ruled out that some of what you say is happening (although we do have some protections against invalid entries), keep in mind that the last couple weeks have been relatively positive for the GOP, with a popular selection (within the party) for Romney’s running mate and their convention this week. A bump would not be unexpected, and in fact we are seeing that in some of the ‘real’ pollster activity out there.

      Let’s see how it plays out over time.

      • Magnum Serpentine

        I believe you will find that those who come on here for the sole purpose of attempting to use this site as Propaganda to show Romney winning really don’t give a flip about winning anything except seeing Romney win the Presidency. And I do not know what poll you are using but I use Real Clear Politics which shows President Obama 47 to Governor Romney 46 ( as of 28 August 2012AD). Real Clear Politics is an average of polls. Out of 12 polls, Only Fox and ABC show Romney ahead.

  • cu

    I just read a fascinating book on equality in America. Turns out the Red states are quite different. The northwest and north middle states are more equal than many blue states. They have fewer social problems as well. A couple of these states are really red % wise. It is the south, Texas and Arizona where the Red states are the worst in problems.
    Making assumptions about America is foolish. Too many different currents going on.Lots more going on than a standard conservative or liberal view.

  • Dan

    I have created a 2012 election game for anyone who’s interested. Sorry, there is no ability to play as Gary Johnson :)

    http://www.americanhistoryusa.com/campaign-trail/

    • Duey

      Your game is horrible. I can’t play as Romney and use all the answers to the questions or Obama and use none. :)

  • Edwin

    Thank you for creating this website. I wish I were teaching Stats this fall — I could have my students analyze combinations.

    A useful variant might be a “percentage shift” — a dial (or whatever) that arbitrary moves ALL state probabilities by up to +/- 10% (at the user’s discretion). This sort of modifier could reflect potential October surprises and/or changes in demographics unaccounted for by pollsters. It would be very interesting to see if Obama would still have an same apparent edge if, say, polls dipped by 5%.

  • Edwin

    I have been thinking about the oversampling claim by Matt, and I believe I have something to add to the conversation.

    Most professional polling organizations carefully gather demographic data about the people who answer their polls. Why? Because they KNOW that certain types of people are more/lass likely to answer the phones than actually vote. So they gather data by category and merge the categories together to create a best estimate.

    If the pollsters are oversampling democrats, that could mean one of two things (I don’t know which from what I’ve read here):
    1) they might have found democrats more willing to answer the phone and talk to them — or even that more democrats tend to be at home during the times the polls are taken, OR
    2) the oversampling means they took the percents for various parties and gave more weight to the democrat numbers.

    Why would they do the second thing? A LOT depends on demographics. We can look election after election and see what percentage of actual voters belong to each party. If populations were stable, this percentage is the BEST predictor of future voting — not the sample percentages. So polling organizations skew their data to make it fit prediction models. This is actually NOT bias — it is a more complex AND more accurate form of statistical modeling, and it traditionally leads to MUCH better predictions.

    Unfortunately for pollsters, state demographics change — sometimes a lot. Virgina, for example, is trending much more democratic than in past years. Likewise for North Carolina. Other states may be trending the other way. Population age also makes prediction more complicated — young voters are *sometimes* likely to vote, but often they don’t bother. How do you predict this? They have cell phones, and most polling companies can’t legally call cell phones with a survey (at least not on purpose). So how do they adjust data to predict what they can’t directly measure?

    It is most like that THIS is what the polling organizations are doing when it says they “oversample” democrats. The goal is to get a CORRECT PREDICTION of the upcoming election, not simply report the data. The companies that do this sort of data manipulation are generally VERY ACCURATE when they predict — that’s how they stay in business. So whatever arcane statistical wizardry they do with the numbers — whether they take the democrat numbers and subtract two thirds of the number of people with silly voices, adding in the number with the letter “G” in their name, or not… — if the companies have been successful predictors in the past, it is most reasonable to assume their goal is not to ‘spin the numbers’ towards any sort of propaganda, but rather to predict the actual election outcome without any political bias.

  • Jeff

    Thanks for the 2012 House map! Terrific addition! I will be using this site often during this election!

  • mark

    http://www.examiner.com/article/mitt-romney-would-lead-eight-unskewed-data-from-newest-cnn-orc-poll More evidence of polling bias…..If you deny it now…you are living under a rock!!!

    • Ed

      Well, looking at the link I posted above, I see CNN has a 0.4% Democratic lean, so that 6% becomes a 5.6%. Does that make you feel better, Mark? If we “debounce” the convention effect, based on http://fivethirtyeight.blogs.nytimes.com/2012/09/10/sept-9-call-it-as-you-see-it/#more-34347 the net bounce as of today would be about 1.6 percent, so Obama’s lead becomes 4.0%. If this isn’t low enough for you, I’m sorry, ‘cuz it ain’t getting any lower.

      • mark

        Enjoy your Barack flavored kool-aid Ed. The middle east is on fire. Obama and his foreign policy is going to drive down his polls…even with the liberal bias thats built in to them..

      • mark

        I see you are using Nate silver as your source….you do know he is a staunch liberal right?

  • Alex

    Ed’s use of Nate Silver is no different than your continued use of conservative based websites like breitbart.com

    The link you posted mark http://www.examiner.com/article/mitt-romney-would-lead-eight-unskewed-data-from-newest-cnn-orc-poll gives us an article associated with the website http://qstarnews.net/ which as you can see by my link is incredibly biased to the conservatives. Also Dean Chambers who wrote the article you linked, is with a Conservative Based Newspaper called Arlington Conservative Examiner. This is obviously not a unbiased source of information. You seem very keen on saying all polls should be ubiased, to which i agree, but you continue to use conservative based websites to back up your claims.

    Also if you look at the links provided in the link you shared, you’ll see this:

    And coming soon… http://www.liberalracism.com – a web site exposing the blatant racism of liberals, the ones who loves to labels others racist.

  • Alex

    http://newsbusters.org/about-newsbusters-org
    About NewsBusters.org

    Welcome to NewsBusters, a project of the Media Research Center (MRC), the leader in documenting, exposing and neutralizing liberal media bias.

    In August of 2005, with the assistance of Matthew Sheffield of Dialog New Media, the MRC launched the NewsBusters blog to provide immediate exposure of liberal media bias, insightful analysis, constructive criticism and timely corrections to news media reporting.

    Taking advantage of the MRC’s thorough and ongoing tracking of liberal media bias, including a wealth of documentation and an archive of broadcast and cable news recordings dating back to 1987, NewsBusters plays a leading role in blog media criticism as a clearinghouse for evidence of liberal media bias. It combines this formidable archive and its compilers inside the MRC with the contributions of already-established netizen watchdogs.

    At least Newsbusters.org admits that they are a conservative based website. The Executive of NewsBusters is also conservative biased http://twitter.com/mattsheffield

    From the link about Heritage Action:

    Heritage Action for America is a unique combination of top-notch conservative policy analysis, a widely respected government relations team and dedicated grassroots activists that advance conservative policy.

    I’m all for revealing biased polls, but using biased websites that always seem to say one party is discriminated against is not the correct way to do that. I should also point out Mark that I referenced Nate Silver in a previous post when pointing out voter polling data here on 270towin:

    Alex August 7, 2012 at 10:08 pm

    Mark I’m going to be honest and say that I ignored the link you posted to the breitbart.com website. That website and its namesake are extreme conservatives. I did however follow the links in that article to Pew Research and Nate Silver. Those two websites do seem to say that Dems are oversampled when surveys are of Registered Voters by 4-7%. So the data from those articles does seem to back up your claim of all polling been biased towards the Democrats.

    Your response to that post:

    mark August 9, 2012 at 9:37 am

    Glad you took the time to fact check. as far as Breitbart is concerned.. their reporting is spot on…even if they are very conservative…my opinion

    You did not seem to voice an objection to Nate Silver when your claims were supported by that there was a poll sample bias in the data that was collected by Nate Silver.

  • Alex

    My link about Heritage Action seems to have been lost in the formatting. The Heritage Action group is associated with newsbusters.org

    http://heritageaction.com/about-heritage-action/

    Also forgot to mention:
    http://newsbusters.org/about-newsbusters-org
    I found the following links at the bottom of the page:

    The MRC’s other Web projects

    Media Research Center, “the leader in documenting, exposing and neutralizing liberal media bias.”
    TimesWatch, a site dedicated to “documenting and exposing the liberal political agenda of the New York Times.”
    Business & Media Institute, “advancing the culture of free enterprise in the media.”
    CNSNews.com, the CyberCast News Service, where you get “The Right news. Right now.”
    Culture and Media Institute, “advancing truth and virtue in the public square.”
    Eyeblast.tv, where the conservative movement shares videos, photos, and opinions.
    Read more: http://newsbusters.org/about-newsbusters-org#ixzz26qMfHUA1

  • mark

    About Nate Silver…..a broken clock is right twice a day as well..

    • Ed

      Polling firms oversample Democrats. Do you have any evidence that it has led to errors compared to election outcomes?

      Re: the broken clock thing: Nate’s right 49 to 50 times out of 50 on election nights ;)

      To quote me: “…Obama’s lead becomes 4.0%. If this isn’t low enough for you, I’m sorry, ‘cuz it ain’t getting any lower.”

      • mark

        Rassmussen was the pollster who nailed the 2008 election..not Nate silver..

      • mark

        On election night? hahahaha anyone can look at polling returns and get it right…..pollsters are supposed to know before the fact…and only Rassmussen does that on a regular basis….next?

  • Alex

    The only results I trust are the one’s on election night.

  • mark

    The so called simulator based on state polls? I tried it..and 7 out of ten times Obama wins Tennessee….Obama has NEVER had a poll there showing him winning. and the Obama advertisement on the same page is very telling . I had hoped this website was honest…I have serious doubts. If they have a poll showing Obama taking Tennesse….lets see it!

  • Greg

    I have also written a simulator for the US election but instead of basing my probabilities on poll data, I have used current betting odds for each state which I feel are more reliable. The results of my simulator closely mirror the results on this site with Obama winning in 9435 of 10000 simulations, Romney winning in 521 cases and a tied election in 44 scenarios.

  • mark

    The national polls of the presidential race remain skewed and this is an issue that remains important for following the political process and truly understanding what is taking place. The Gallup tracking poll, which has been over-sampled Democrats in the past, has released its latest numbers today showing President Obama leading 48 percent to 45 percent for Mitt Romney. But the non-skewed uses a sample weighted by the expected partisan makeup of the electorate, the QStarNews Daily Tracking poll, shows Romney leading over Obama by a 53 percent to 45 percent margin. The difference is the sample or the weighting, the latter result is based on a weight of 37.6 percent of the electorate being made up of Republican voters while 33.3 percent is made up of Democratic voters. The Gallup survey likely bases it’s numbers on an expect turnout, or an “over-sample” of Democrats by a four to six percent margin.

    At the most basic level, this issue is important because the desire is to believe the polls are reasonably accurate if the polls are to have any meaning or provide any understanding about what is truly happening in the political process. Polls and surveys have a margin of error and can not be expected to be perfectly accurate. However, when polls taken at about the same time claim or three or four percent margin of error but show results from one poll to another that vary far more than that, one questions the accuracy of the polls. When Gallup shows Obama leading by three and QStarNews shows Romney leading by eight in today’s tracking polls, that is a difference of 11 points that exceeds the margin of error of the polls.

    View slideshow: The skewed polls issue

    .
    A graphic showing different weightings that could be used in polls of the presidential race,.

    Photo credit:

    QStarNews.com
    . .
    The difference is the weighting and/or sampling of Democrats, Republicans and independents in each survey. Some tests of different weightings, assuming the same breakdown of independents, shows how variations in the weightings of Democrats and Republicans produce varying different final results between Obama and Romney. Most recent polls have shown Romney leading among independents by 10-15 percent, so for this testing of weighting it will be assumed Romney gets 56 percent to Obama’s 44 percent among independents on election day. With this, the weightings of Democrats to Republicans will produce varying results. The first graphic in the slide show with this article shows the five tests of weightings.

    Using Rasmussen’s 37.6 percent Republicans to 33.3 percent Democrats, Romney would win the election with a 54.10 percent to 45.90 percent margin.

    Using a weighting that assumes Republicans and Democrats will be 35 percent each in the actual electorate that votes in the election, Romney would win the race 51.8 percent to 48.2 percent.

    If the electorate is Republicans by two percent, meaning 36 percent Republicans to 34 percent Democrats, Romney would win 50.70 percent to 49.30 percent.

    If the electorate is Democrats by two percent, meaning 36 percent Democrats to 34 percent Republicans, Romney would win 52.90 percent to 47.10 percent.

    Democrats will have to have a four percent edge among the voting electorate, meaning 37 percent Democrats to 33 percent Republicans, for Obama to win the popular vote by a very narrow 50.40 percent to 49.60 percent majority.

    The recent Pew Research Poll showed Obama leading 48 percent to 43 percent based on a sample that over-sampled Democrats by 3.8 percent and under-sampled Republicans by 7.0 percent. Unskewed, this poll’s data indicated Romney leading 54 percent to 44 percent. The second picture in the slide show illustrates the skewed sample used in this poll.

    The recent NY Times/CBS News poll showed Obama leading 49 percent to 46 percent based on a sample that over-sampled Democrats by 10.7 percent and under-sampled independents by 11.2 percent. Unskewed, this poll’s data indicated Romney leading 51 percent to 44 percent. The third picture in the slide show illustrates the skewed sample used in this poll.

    A recent NBC News/Wall Street Journal poll release last week showed an Obama lead of 50 percent to 45 percent using a sample that under-sampled Republicans by 12.6 percent and over-sampled independents by 8.8 percent. Unskewed, this poll’s data indicated Romney leading 51 percent to 44 percent. The fourth picture in the slide show illustrates the skewed sample used in this poll.

    The infamous CNN/ORC poll showed Obama leading by six percent, 52 percent to 46 percent, almost the margin by which he won in 2008. But the survey sample was doubly-skewed, over-sampling Democrats 17.1 percent and under-sampling independents 25.0 percent. Unskewed, the data from this poll would reveal a 53 percent to 45 percent Romney lead. The last picture in the slide show illustrates the skewed sample used in this poll.

    Those are just four examples of recent mainstream media-commissioned presidential election polls that have been skewed of many such polls to have produced such results. The importance of this is whether or not the polls are reasonably accurate and tell us what is really taking place in the presidential race. To the degree these polls are believed or not believed is the degree to which campaign strategies are evaluated and analyzed and the news coverage and analysis perceives how the campaign is unfolding. These polls have created a media narrative that Mitt Romney is losing the race and running an ineffective campaign while President Obama is winning and running a more effective campaign. The media coverage of this analysis and resulting narrative could paint the picture of this race that voters see, influence how they judge the candidates, and possibly have a profound effect on the election itself

    So many of these polls are skewed and showing results unrealistically favorable to Obama that they are leading to a perception among voters of the inevitability of Obama’s victory in the election in November. In fact, every recent survey asking voters which candidate they believe will win show them predicting that Obama will win. What if all these polls really are wrong and Romney is, perhaps, running four or five percent ahead in the race right now? How differently would the media and voters view the politics of this presidential race right now if the conventional wisdom was that Romney is winning? Wouldn’t accurate information be better than having information that is heavily skewed for one side or the other?

    The Real Clear Politics average of polls, made up mostly of heavily-skewed polls, shows Obama leading by 3.8 percent. The UnSkewed Average of polls at UnSkewedpolls.com released today shows Romney leading by 7.8 percent. An average of the two averages would indicate a 1.8 percent Romney lead in the average of both. Which averages or polls are more accurate? Ultimately the real answer to that question come on election day in November.

    • Ed

      “Poll Averages Have No History of Consistent Partisan Bias”
      http://fivethirtyeight.blogs.nytimes.com/2012/09/29/poll-averages-have-no-history-of-consistent-partisan-bias

      I really like how my prediction is panning out so far: “…Obama’s lead becomes 4.0%. If this isn’t low enough for you, I’m sorry, ‘cuz it ain’t getting any lower.”

      • mark

        Sorry…it is getting lower….by the minute! Next?

      • mark

        You have run out of ideas…sir

        • Ed

          Obama’s lead has gotten lower than 4%? Oh wait, yes it’s at 3.9%. I’m SoooooooOOOoooo sorry about that. :)

  • http://urbanreview-dmero3834.blogspot.com David Mero

    With all of the end of the world talk coming from the far right, I wonder what amazing stories they [far right wing nuts] will come up with after Obama is re-elected. It’s a shame a group of people can create a slew of falsehoods and their sheep will follow this unfounded foolishness to the slaughter. I’ve heard so much craziness from these people based on fairy tales and plain ole mindless ramblings that it makes you wonder if these people can collect an independent thought without any help

  • Kevin Sun

    obama won 273 to 203

Previous post:

Next post:


Copyright © 2004-2014 270towin.com All Rights Reserved