Princeton Election Consortium

Innovations in democracy since 2004

Outcome: Biden 306 EV (D+1.2% from toss-up), Senate 50 D (D+1.0%)
Nov 3 polls: Biden 342 EV (D+5.3%), Senate 50-55 D (D+3.9%), House control D+4.6%
Moneyball states: President AZ NE-2 NV, Senate MT ME AK, Legislatures KS TX NC

Joe Biden’s dawn?

October 19th, 2012, 7:30am by Sam Wang

VP Biden as ApolloThe tyranny of online political writing is that one often feels under pressure to interpret events before enough information is available. I am now wondering if I wrote off the VP debate a bit prematurely.

Last week, President Obama crash-landed spectacularly, ending up with a bare electoral lead over Mitt Romney. The Popular Meta-margin (defined as how much swing it would take to create an electoral vote near-tie) was Obama +0.6%, the lowest his lead has been all season. At the time I thought it had started to turn around before the VP debate, so that Biden’s performance was like the rooster greeting the dawn – not a contributing cause. Is it time for me to eat crow?

Take a look at this, which is from the state-poll Meta-analysis:

Biden stops the fall?

Keep in mind that it takes a few days for fresh state polls to even start getting into the Meta-Analysis. For example, until the rightmost data point on October 18th, this graph made use of no polls taken after the second Obama-Romney debate. And a shock to the race takes 10-12 days to become fully apparent (see the effects of Debate #1). Therefore any changes you see might be caused by debate #1 or the VP debate – but not by debate #2.

Here is what I see:

  • The Vice-Presidential debate came precisely when the Meta-Margin (and of course the EV estimator) was bottoming out. It is not possible to give Biden credit for stopping Obama’s fall.
  • But look a few days later. Today the Meta-Margin has crept up to Obama +1.5%, almost a full point higher than last week. Although the President is well below his post-convention high of +6.2%, he has made up ground. The only preceding event is the Vice-Presidential debate.
  • The effects of Debate #2 are just starting to hit. Look for changes today, probably favoring Obama.

The rise is small, but its timing makes Biden more like Apollo, the god who pulled the Sun up every morning in his chariot. More tongue-tied than the original…though he was still generally perceived as winning the debate.

Biden was good for Democratic morale…but was the reversal his doing, exactly? Today’s media herd galumphs off after one story at a time. I don’t know that the VP debate moved opinion directly. What it did was change the story away from “Obama was nearly comatose” to some mix of “Obama could learn from Biden,” “Biden thrashed Ryan,” and even “Biden smiled too much.” It doesn’t matter. It really could have been anything. The news cycle is naturally one week long. In August, the addition of Paul Ryan to the VP ticket led to a bounce – which was stopped when Todd Akin opened his mouth about “legitimate rape.” And on to the next news cycle. Three more cycles to the election.

History of electoral votes for Obama


Finally, some commentary on recent national polls.

Despite the newspapers’ fascination with the most extreme surveys (Gallup comes to mind), these are exactly the ones you should ignore. Instead, use the middle one (a.k.a. median) of all available surveys. Doing so gives Romney +0.5% +/- 0.8% (n=10, Oct. 11-18 survey dates).

During that same period, President Obama has been ahead in the Electoral College, with 277-290 electoral votes and a Meta-Margin fluctuating around Obama +0.6 to +1.5%. This means that Obama currently has a structural advantage in swing states of 1-2% compared with national polls. In other words, Obama could lose the popular vote yet win the election. Something similar happened in 2000 when George W. Bush lost the popular vote, yet became President. Live by the sword, die by the sword.

I’m on National Public Radio today, Talk Of The Nation’s Science Friday, at 2:15pm with Flora Lichtman and Nate Silver. Tune in!

Tags: 2012 Election · President

85 Comments so far ↓

  • Matt McIrvin

    Anecdotal chatting-to-people data suggests to me that Biden’s performance actually repelled some people: they thought he was rude and felt sorry for Paul Ryan.

    But whether that hurt the ticket is another story, I guess, since pity is probably not the feeling you want people to be having when you’re running for Vice-President.

  • Ralph Reinhold

    I believe the more reliable polls right now are the ones where the same people are interviewed over and over (RAND and YouGov, for example.) I think the ‘likely voter’ rules are problematic because of the polls results are not within each other’s margins of error. If you look at the likely voters, they are pretty much within the margin of error. If you add the +.7 – .9 in house errors presented by the Emory site, they are all within the margin of error.

    The likely voter results are bazaar. PPP, which should show an Obama +.7 bias, shows a lower Obama margin than Rasmussen, which should show an Obama -.9 bias in one state, at least.

    • wheelers cat

      bizarre, unless you are shopping for results.

    • Matt McIrvin

      One problem with those polls is that, while they can show changes with great precision, you really do need to worry a lot about the MOE when trying to figure the absolute state of the race, since the random sampling error on them is not reduced by repeated sampling.

    • Craigo

      That’s my concern too, Matt. Any mistakes in the initial randomization are locked in for the rest of the study.

    • Matt McIrvin

      …And the other problem with them is that their panel presumably gets more and more divergent from the general population by the simple fact that someone is polling them over and over.

    • Craigo

      Heisenberg strikes again.

    • wheelers cat

      Still, RAND has solved the non-responder problem, which is more critical IMHO than respondents becoming educated. Isn’t being aware and educated a good thing?

      that is why there is Wheeler’s Cat.
      much luckier than Schrodinger’s Cat.

    • Craigo

      @wheelers cat: If your sample becomes significantly more high-information than the population, then methodologically speaking it’s a bad thing.

    • wheelers cat

      im not so sure.
      i think the internet and social media are radically changing approx half the electorate and the polls are missing it because they only measure enthusiasm.
      Who is to say a more educated and aware electorate isnt the underlying truth?

    • Craigo

      If the sample and population converge and become equally high-information, then my statement would not apply. I’m just skeptical that the average likely-voter is going to be as high-information as the average RAND panelist when all is said and done.

  • MAT

    Per your ideas in the post, would maybe it be more useful to plot events on the meta margin graph by when polls whose mid dates are after the event first hit the calculations?

  • MAT

    From rooster to Apollo. That’s quite the promotion :-)

  • wheelers cat

    My debate fractal disappeared in RAND with a move back toward Romney yesterday. Flocking behavior, jitter, or asymmetrical enthusiasm pushback?
    Heres my anecdote.
    The conservatives I know have become even more virulent in their disdain for Obama post debate2. Its like they finally accepted ownership of Romney and will defend him tooth and nail.
    They have committed.
    You are correct about the media, Dr. Wang, but it isnt random. For example the media really NEEDED a Romney bounce at debate1. He was circling the drain before the debate, and a one horse race doesnt sell. So there was asymmetrical coverage of Romney the debate-god. The media is driven by market forces and the need to sell product mostly. I think Obamas success in debate2 actually forced indifferent conservatives to commit in stead of depressing enthusiasm.
    Here is a great picture of the state of the campaign.
    Its actually Josh Romney who says he wants to take a swing at the president for calling his dad a liar, but he is prevented by the secret service cordon.

    • Olav Grinde

      Imagine the headlines if Malia or Sasha Obama had said she wanted to take a swing at Mitt Romney!

      Fox News would have been apoplectic.
      But then again, that doesn’t take much…

    • Ralph Reinhold

      The media flocking is shown by the number of one day polls following debate 1 (3 or 4 per day for 3 days) vs debate 2 (I am aware of none)

    • Ralph Reinhold

      Josh Romney said that he had been slugged by Tadd a number of times and can assure the President that he had nothing to worry about.

  • wheelers cat

    In RAND that what is we see.
    Debate2 increased enthusiasm for both red and blue phenotypes, blue phenotypes by more, because they were more depressed.
    So that is what happened to the fractal– asymmetrical enthusiasm effect.

    • xian

      phenotypes? why the pseudoscientific babble?

    • wheelers cat

      Its just the standard vernacular for red/blue genetics and neuropolitics.
      Are you about to call me a eugenicist?

    • wheelers cat

      Its not “pseudo-scientific babble”.
      Its just the standard individual descriptive used in red/blue genetics and neuropolitics.

      Are you about to call me a eugenicist?

    • wheelers cat

      What the hell?
      phenotype is pseudo-scientific babble?
      where did you go to school?
      BYU? Oral Roberts?

    • Fred Fnord

      wheeler (and his cat): You seem upset. Perhaps you should consider meditation.

      And I just did a quick google search. You are the only one talking about ‘red and blue phenotypes’ that google knows about.

    • wheelers cat

      xian said:
      “phenotypes? why the pseudoscientific babble?”

      phenotype is a scientific term.
      where did you go to school?

    • Sam Wang

      OK, enough, guys…

  • PM

    Sam, I love, love, love the site, but it’s wrong to say that the only event that happened between now and then was X. Lots and lots and lots of things have happened. It’s true that the only formally defined political ritual that happened between now and then was the VP debate, but think of the ad strategies that have shifted, the shocks that have taken place because of the baseball playoffs and college football games (see Healy and Malhotra in PNAS for more), fluctuations in the real economy, news out of Europe, Romney’s cancellation of the view, etc. etc. etc.

    To put it another way: the meta-margin is not set up to do identification at this resolution. Afterward, with finer-grained data and with more cautious hypotheses, perhaps. But not at this margin, and not without knowing much more about the race than we do. Just because the debates are easily labeled doesn’t mean they’re the only events.

    • Craigo

      Sam has mentioned this before – was the relative lack of a GOP convention bounce due to the event itself, or did Ryan’s announcement suck up all the oxygen?

    • Sam Wang

      Mostly agree. I have heard this called the “attribution problem” by political scientists. The best use of polling meta-analysis is to identify when a change happened. Then that narrows the candidates.

      I tried to qualify my statement. Possibilities: (1) voters and media forget what they saw in the first debate, (2) the VP debate itself. You are suggesting (3) something else, which is possible. However, I will note that your alternatives had smaller audiences, making the mechanism somewhat less obvious. For instance, in the US, attention to Europe is low even on good days.

  • Ms. Jay Sheckley

    I’m grinning in drunken, senile joy at ApolJo! Of course, here you gots yer propter hocs all over again: A rooster stood next to Leto as she birthed Apollo & bro. Longterm, true as a PEC map/chart set, being “cocky” or a sun god’s just sides of the same coin. [ ]
    Evidence notwithstanding, people may always argue that old debate saw: Who’s the race’s first, the biden or the yegg?

  • bsk

    Continued from the previous thread, I still have two outstanding methodology questions (shamelessly copied and pasted):

    First, can anyone explain to me why the meta-margin isn’t at exactly Obama +2% right now? As far as I can tell, he’s up at least 2% in states worth over 270 EV, according to the “power of your vote” bar.

    Second, why was Ohio still at 2% yesterday evening (18/10/12)? Taking all polls on Pollster with sample end dates from the 11th onwards gives Obama +3.5%. Does Sam use the date of publication rather than the sample dates?


    • bsk

      Thinking about it again, I may have the answer to the first question: if it’s the swing in all polls that would hypothetically cause the median _simulated_ EV total to hit 269, then that makes more sense.

    • badni

      bsk – I don’t know the answer to the first question, but it looks like for purposes of the Jerseyvotes everything is rounded to the nearest 0.5%, which is presumably not the case when doing the full calculations to determine MM. Maybe it’s as simple as that?

      I had also speculated that your explanation may be the case, but the math didn’t seem to me to work out (though, with the guesstimation style of phony math I was using, I wouldn’t rely on that).

    • Froggy


      1. The MM isn’t exactly 2% because it takes into account that there is some margin of error in the predictions for each individual state. An O+2 is not treated as exactly O+2, but rather as a distribution centered around O+2. At least that’s how I see it. (I think that there are other factors as well.)

      2. The seven days is calculated from the latest polling date, not the date when the calculation is made. The polling date is taken as the average of the starting and ending dates (fractions dropped), rather than the polling end date. The latest OH poll is 10/17, and there are four polls from 10/10 onward: O+1, O+3, O+5, and O+1. The median of these is O+2. (Note that both O+1 polls are from Ras, which is holding back the Ohio margin.)

    • bsk

      Froggy and badni: Thanks for the replies.

      Froggy: your answer to 1 seems plausible, and 2 makes perfect sense.

  • JimmHck


    It appears to me that your prediction is introducing a bias. The Bayesian prior is assuming that the race is most likely to revert back to its mean value over the election instead of wandering randomly. Is there any real justification for this? It seems to me that if real fundamental changes have occurred in the campaign, then past polling data is not necessarily a good predictor of future polling data.

    It seems your model expects Obama’s polling numbers to improve simply because they are at a record low in the campaign. I don’t see any reason why they couldn’t continue to fall further however.

    • Sam Wang

      It’s an interesting point. Generally you are not right – a campaign’s history *is* predictive of its likely future range. I set that prior pretty broadly. But in the home stretch it may be appropriate to remove that assumption. Thanks for the input.

    • Matt McIrvin

      As I said just before the numbers dropped, it’s a hypothesis, which is being tested.

  • MAT

    Results after first day of early voting in NC: 47.3%D, 34.8%R, 17.8%I. 223,385 total ballots cast.

    • Pat

      Yes, that sounds good, but it’s actually very bad compared to early voting for NC in 2008: 51.4%D, 30.2%R.

    • Pat

      The Florida numbers for early voting are also much worse than in 2008:
      2008: 45.6%D, 37.3%R
      2012: 39.6%D, 44.9%R

      Certainly, this is only the beginning: in 2008, 51.8% of people in FL voted early. Currently, we are only at 6.3% of the total 2008 vote. However, the trend is certainly nothing to feel very excited about.

    • Craigo

      It’s the same in Iowa, Ohio, and Virginia, Pat. The good news (?) for Obama is that his margins everywhere but Virginia and North Carolina were relatively strong, so he can afford some slippage, especially if his in-person GOTV is strong.

    • J-NC


      Remember to account for the phase of early voting (ie, where were we at this time in 2008?).

      For NC, it’s been one day of early voting, so I admittedly don’t have a good picture of it.

      But for FL, remember that in-person early voting hasn’t started yet. We’re looking at numbers for absentee only. So far, the Dems are pretty happy from what I’ve seen. The last numbers I saw were +4R versus +16R for absentee balloting in 2008.

    • Matt McIrvin

      Historically, absentee tends to be really good for Republicans, in-person early voting for Democrats. I’m not sure why that is (a component is probably military families and elderly voters in the case of absentee, African-American church campaigns for in-person early).

  • RDT

    My guess is that one of the big factors in “bounces” is just the news chatter refrain of either “wasn’t that spectacular” or “wow, what a loser.”

    So I think part of the ante-hoc nature of the VP debate (and other predictable-event bounces) is that the news chatter begins to change leading up to the event. News shows talk less about how bad Obama was and more about Biden and Ryan, and that in itself stops some of the bleeding.

    It might be interesting to compare at meta-margin changes around scheduled events (conventions, debates) versus changes related to unexpected events (Akin, the 47% tape).

  • Olav Grinde

    Rasmussen has Romney increasing his lead in Missouri, to +11% on 10/17 from +3% on 10/02 . Sorry, but I just don’t believe those numbers.

    By extension, when the latest Rasmussen poll pegs North Carolina at +3%, I think Obama has a decent shot at taking that state.

    • Froggy

      Reports today are that the Romney campaign is shifting its staff out of NC. I’d say this is a good move, even if they’re not certain of carrying NC — it’s highly unlikely that NC is going to decide this election (not even in the Jerseyvote top 10).

    • Olav Grinde

      Froggy, I suspect it’s vitally important for the Romney-Ryan Campaign to exude confidence.

      I do, however, believe Obama should make an effort to take North Carolina. It’s not just about getting a victory in the Electoral College. Obama’s mandate also matters — and carrying one, two or three Southern states is of immense importance with regards to pushing through his policies in the second term. The better argument the GOP has for saying “the country is divided”, the more obstructionist they can be.

    • Matt McIrvin

      Virginia should be higher priority than NC. But, certainly, keep the ground game going in both.

    • Froggy

      Mandate, shmandate — just make sure you win the dang thing first before you worry about having the right kind of win. The GOP was very able to be obstructionist when Obama won 365EV, and holding NC this time won’t make them change their minds.

      But I don’t have a problem with Obama staying in NC as long as he has a chance to win — his campaign has to exude confidence as well.

    • ML

      I think the proper way to interpret Romney pulling resources from North Carolina is two-fold: (1) as others have pointed out, NC is unlikely to be a tipping-point state, and (2) the Romney ground operation is generally inferior to Obama’s.

      As a result, Romney has to engage in triage – pull some staff out of PA and move them to OH, some out of NC and move them to VA, because their ground game is simply inferior in those pivotal states. The necessity of triage is never positive, whether its because the state seems to be trending toward or away from the campaign.

    • Brian

      ML: I’m shocked Romney has anybody left in Pennsylvania. I would have thought he’d have given up on PA after the conventions.

  • Brian

    Fellow nerds: I’ve updated the PEC Google Chrome Extension. It now displays the current Meta Margin right there on the icon so you don’t even have to click. It auto-updates every 2 hours. And when you do click, it now shows you the current electoral map.

  • Michael Worley

    And the MM trend reversed to 1.32…

    • Froggy

      The drop in the MM appears to be a glitch in the data stream — CO dropped from O+3 back to O+0.5 for no reason I can discern. O+3 is right given what is posted on Pollster.

      There are a couple of polls that haven’t made it Pollster yet that will move the MM when they do. One is the VA R+3 Ras poll, which will move VA from R+1 to R+1.5.

      The other is a NV O+8 poll from the Mellman Group, which should shift NV from O+3 to O+5.

    • Olav Grinde

      Glitch — I like the sound of that.
      Obviously this is a glitch with a Republican bias. ;)

    • Michael Worley

      But PPP NH is R+1

    • Froggy

      Michael, the PPP NH wasn’t in the last update. And it won’t change the MM much in any event. The most it will do is shift NH from Tied to R+0.5. (I can’t be sure yet because I haven’t seen the polling dates.)

      In addition, the just-released Ras R+5 in FL won’t shift the FL margin, so will not (by itself) impact the MM.

    • Matt McIrvin

      New Hampshire has been hard to figure out all year, and I speak as someone who spends a lot of time there. There are strong regional variations within the state. It’s traditionally libertarian/conservative. Nashua is basically a Boston suburb and tends to swing Democratic, but the outer Boston suburbs have a right-wing element in themselves by virtue of being suburbs.

      Obama obviously doesn’t have as much of a lock on NH as he did in ’08 but it’s hard to predict what will actually happen.

  • Ram

    Here are Obama’s swing state net change in RCP Composite between Oct 3 and Oct 19:-

    State Change RCP (Oct 3) RCP today
    OH (-2.4) O: +5 O: +2.5
    FL (-4.5) O: =2 R: +2.5
    VA (-3.5) O: +3.5 Tie
    CO (-3.3) O: +3.1 R: +0.2
    NC (-6.4) R:+0.8 R; +5.6
    NV (-2.2) O:+5.2 O: +3
    NH (-5.2) O: +6 O: +0.8

    The real firewall has been NV and to a lesser extent OH. But I am not sure OH will withstand the resources Romney (RNC) campaign is pouring.

  • Jon V

    Without any scientific basis, I intuitively find it difficult that Romney can win in Colorado with legalized Marijuana on the ballot

  • Jon V

    Ram, I am concerned about Ohio as well, however. Obama has a huge advantage on the ground, and many people have already voted, and most of them have voted Dem. I think Obama’s 4 year lead on the ground in the swing states will save him…. at least I am hoping

  • Olav Grinde

    Ram, let’s say Romney wins Ohio. (unlikely)

    Obama holds Colorado + Nevada + New Hampshire.
    That gives 272 Electoral Votes and re-election for President Obama.

  • Jon V

    Olav, right now Obama is polling stronger in Ohio than he is in Colorado. I would think if he looses Ohio, Colorado would also be gone. However, I do not think he will lose Ohio. Even after Debate1 his numbers held. people are already voting, and his ground game is 4 years ahead of Romney’s. I am nervous, but I am optimistic about Ohio, and the remaining swing states

  • Ram

    The race is virtual dead heat and this comes after almost three-four months-long period of stability in which Obama seemed to be leading this race by 1.5%-4%. What caused this shift?

    The recent polls (Pew, WP/ABC) indicate movement of some independents and women voters probably reassured by Romney’s debate performance.

    • Olav Grinde

      There are few state-level polls after Debate # 2.
      As they arrive, we’ll have a truer picture.
      Ram, I hope you’ll post then as well.

    • ML

      The Meta-Margin and Nate Silver’s model both suggest that Romney’s first debate bounce has reached its zenith and begun to recede. Over the last day, day-and-a-half, 5 out of 7 national polls have shown Obama with the lead, anywhere from .5 points to ~6 points. The 6th, Rasmussen, shows a tie (with Gallup basically parting ways from the consensus). Median and average of those polls point to a small but slowly-growing Obama lead, right in line with the Meta-Margin and 538 analyses. That also tracks what PPP is saying about its tracking breakout by day (last 2 days have been stronger than day 1), and what little we have seen from state polls the last few days.

      It’s possible that this Obama resurgence is transitory, and soon enough Romney will start to gain ground again. But right now, the polls point to a small-but-growing Obama lead, both nationally and in the swing states.

    • Reason

      And there goes Ram’s disconnect from reality again. What part of O’s moving up slowly are you not getting? And where do you get these numbers for women? We get it. You are an R supporter.

    • wheelers cat

      Romney has dropped 10 points in Silvers forecast.
      Beware the revenge of Son of Poll Truthers.

    • Matt McIrvin

      People are noticing the persistent weirdness of Gallup’s likely-voter screen now.

      The thing is, this isn’t a new thing. Back just before the conventions in the (shallower) depths of the Paul Ryan bounce, there were a lot of articles about Obama’s enthusiasm problem and Obama as “leader of the unlikely voters”. Around that time, I remember Josh Marshall fretting about the gigantic size of the difference between Gallup’s RV and LV numbers.

      I mentioned it here, and Sam responded that as far as he could tell, LV screens really didn’t make that much difference. And I looked at some of the state numbers over on RCP, and sure enough, they really didn’t make that much difference.

      I got really puzzled about it, and started talking about popular-electoral splits and such, just like people are doing now. But in hindsight, I think the issue was at least 75% just Gallup and their unusual likely-voter screen.

      The difference is that now their tracker is actually using the screen; I think it wasn’t then.

  • Craigo

    A quick note on random sampling: In the last thread, someone asked a question about how a sample of 500 can possibly be representative of an electorate of over 100 million.

    The answer of course is the law of large numbers and equal probability of selection, with a little help from demographic weighting. The first two have been mathematically proven, and the third is a necessary evil given that the population being sampled – adults with phone access – largely contains but is not identical to the electorate, our target population.

    A random sample of 500 will, 95% of the time, capture the true value of the population, plus or minus 4.38%. (5% of the time the true value will lie outside that confidence band. God gave us a window, but sometimes he likes to slam it shut on our fingers.)

    But one reply implied that because national polls are targeting a much larger population than state polls, they are necessarily less reliable. That is absolutely not true – sampling works regardless of how large the population becomes, provided that it is truly random. It’s the size of the sample that matters.

    • RDT

      However, since the opinions of much of the country don’t matter for the outcome of the elections, it’s not really clear that national polls have a lower “jersey vote” uncertainty than the state polls.

  • chris

    Can anyone make sense of the new poll from PPP in iowa today R is now up by 1 ? does anyone think that is an outier?

    • Craigo

      Yes, random error happens. It’s best not to get bogged down in it.

    • Matt McIrvin

      The thing I’m noticing is that Talking Points Memo has developed a strange tendency to headline the most frighteningly pro-Romney polls they can find.

      It’s almost comical, and makes Sam’s point about the difference between the parties. A conservative news site would never do that; they’d… headline the most wonderfully pro-Romney polls they can find.

  • Jen

    Yes, what the heck is that about with Talking Points Memo? The headlines are ridiculous. What are they trying to prove?

  • Ram

    Some more Obama’s swing state net change in RCP Composite between Oct 3 and Oct 19:-

    State, Change, RCP (Oct 3), RCP today
    WI (-4.8) O:+7.6 O: + 2.8
    IA (0.2) O: +3.5 O: + 3.3
    PA (-3) O: +8.0 O: + 5.0
    MI (-5) O:+10 O: +5.0

    I am looking at the data just to understand the race. But I can see some people here think I am a R supporter.

    I think NV, FL, and NC are all off the swing state map. I suspect O is doing better in NV than what we see in polls. Remember Harry Reid’s election. Thus O needs to put away WI, IA, NH next week to be safe on Nov 6. My view is O’s national numbers must continue the upward trend we are seeing today.

  • wufwugy

    I like how the “news” media’s incessant adherence to sensationalism and false controversy to generate ratings has provoked CNN’s latest poll in FL to cut out nearly 1/3rd of their registered voters just so they could turn O +7 into R +1

    The overwhelming majority of polls showing Romney leading since the first debate have awful methodology and aggregators should treat them like the bad science they are

    • Olav Grinde

      Wufwugy, do you have a source and preferably a reliable analysis that supports your conclusion?

      As far as I know, four Florida polls were released yesterday. While SurveyUSA had Obama +1%, the three other pegged Romney with a 1–5% lead (CNN, Rasmussen, Fox News). Not unexpectedly, the GOP-leaning outlier is Rasmussen. Any way you look at it, Florida seems a tight race.

      While I too question some of the recent Florida polls, I have not seen good arguments for claiming they are skewed or for dismissing them.

      In the meantime, I’m simply waiting for additional fresh polls — from Florida and other swing states.

  • wufwugy

    @Olav, Rasmussen is not a pollster. I am baffled that people think he is, but the fact is that he cannot be considered a legitimate player until he releases methodology at the very least. The fact that he has been accurate to some degrees in the past is irrelevant. This is supposed to be a science

    Similar for Fox. They do release methodology, but are pathological liars with a clear agenda and should never be taken seriously. When glancing over their recent FL methodology, I see their race categories are “white” and “non-white” and they have Obama receiving <50% of the non-white. So this means we can throw this poll in the trash because when blacks and non-Cuban Hispanics are included, there is no way in hell he's anywhere close to that low. What we have here is a fine example of Fox doing what they do best: lying. They probably purposely made up most of their "non-white" category into Cubans

    Regardless of the fact that CNN's primary objective is drumming up sensationalism and controversy for ratings, their FL poll shows them cutting out an ungodly amount of registered voters to reach their likely voter model. We'd be better off ignoring their likely voter list and applying the historical trend of registered voters leaning about 2 to the left (IIRC), so then the CNN poll shows that Obama is approximately +5, not -1

    Historically, SurveyUSA is a legit pollster, but they need to weight more heavily towards cell users since they're finding differing results for them and 23% is not high enough representation of the demographic. Granted, I don't know how much is, but there is potential that approximately 50% of FL doesn't answer a landline for polling, and if this is true then Survey's results are skewed towards Romney. They also need to add the Espanol option. Maybe they already do, but I haven't seen it listed in their methodology

    FL is probably a tight race, but we don't do ourselves any favors by taking the pollsters' word for it. Their agenda and methodology needs to be accounted for, and then we can decide if they're empirically relevant

  • Nancy Cadet

    I responded to a Rasmussen automated phone poll last Fall (pre election season) and it featured a series of hair raising conservative myths posed as questions: ie. Do you favor redistribution of wealth ? Shocking and instructive, it seemed to be a kind of propaganda tool…I said yes to the question above, btw

Leave a Comment