Poll Analysis: Romney slips a bit

ObamaRomney
99.9% probability of winning 0.1% probability of winning
Mean of 326 electoral votesMean of 212 electoral votes

My previous analysis of state head-to-head polls showed President Barack Obama leading Romney with a mean of 323 to 215 electoral votes. Obama had a 99.5% probability of winning; Romney had a 0.5% chance.

Six new polls have been released since then. I’ve also fixed a few minor errors (largely in older polls) thanks to some more help from Sam Minter:

startendsample%%%
stpolldatedatesizeMOEORdiff
CALA Times17-May21-May10023.55637O+19
COPNA/Keating21-May24-May6014.04844O+4
MIPPP24-May27-May6004.45339O+14
MOPPP24-May27-May6024.04544O+1
WAStrategies 36024-May27-May5004.45140O+11
WIMarquette23-May26-May62551.243.0O+8.2

That’s a rather blue collection of new polls.

Both of the current California polls have double-digit leads for Obama.

In Colorado, Obama squeaks out +4% over Romney, whereas the previous poll had them tied. Overall the trend in Colorado polls looks favorable for Obama:
ObamaRomney30Apr12-30May12Colorado

Michigan gives Obama a remarkable +14% lead over Romney. One has to go back eight polls, to November 2011, to find a poll in which Romney is leading. Obama has managed to turn Michigan around, from toss-up to solid Obama, over the past six months:

ObamaRomney30Apr12-30May12Michigan

In Missouri, Obama has a meager +1% lead over Romney. Up to now, Missouri has given a small edge to Romney over Obama. My hunch is that further polling would put the state back into Romney territory. Here are the polls to date:

ObamaRomney30Apr12-30May12Missouri

The new Washington poll puts Obama at +11% over Romney. Both current WA polls give Obama double digit leads.

We have six current polls in Wisconsin. And all but one goes to Obama. With today’s poll, the trend cannot be considered good news for Romney:

ObamaRomney30Apr12-30May12Wisconsin

Now, after 100,000 simulated elections, Obama wins 99,941 times and Romney wins 59 times (including the 13 ties). Obama gains three electoral votes for an average of 326 to Romney’s 212. If the election was held now, Obama would have a 99.9% probability of beating Romney, based on the polling data.

Electoral College Map

AlabamaAlaskaArizonaArkansasCaliforniaColoradoConnecticutDelawareFloridaGeorgiaHawaiiIdahoIllinoisIndianaIowaKansasKentuckyLousianaMaineMarylandMassachusettesMichiganMinnesotaMississippiMissouriMontanaNebraskaNevadaNew HampshireNew JerseyNew MexicoNew YorkNorth CarolinaNorth DakotaOhioOklahomaOregonPennsylvaniaRhode IslandSouth CarolinaSouth DakotaTennesseeTexasUtahVermontVirginiaD.C.WashingtonWest VirginiaWisconsinWyoming

Electoral College Map

GeorgiaAlaskaArizonaArkansasCaliforniaColoradoDelawareConnecticutFloridaMississippiAlabamaHawaiiIdahoIllinoisIndianaIowaKansasKentuckyLouisianaMaineMarylandMassachusettsMichiganMinnesotaMissouriMontanaNebraskaNevadaNew HampshireNew JerseyNew MexicoNew YorkNorth CarolinaNorth DakotaOhioOklahomaOregonPennsylvaniaRhode IslandSouth CarolinaSouth DakotaTennesseeTexasUtahVermontVirginiaD.C.WashingtonWest VirginiaWisconsinWyoming

Here is the distribution of electoral votes [FAQ] from the simulations:

Ten most probable electoral vote outcomes for Obama:

  • 313 electoral votes with a 4.75% probability
  • 322 electoral votes with a 4.40% probability
  • 312 electoral votes with a 4.12% probability
  • 323 electoral votes with a 4.03% probability
  • 314 electoral votes with a 2.88% probability
  • 315 electoral votes with a 2.31% probability
  • 328 electoral votes with a 2.24% probability
  • 342 electoral votes with a 2.22% probability
  • 316 electoral votes with a 2.20% probability
  • 324 electoral votes with a 2.07% probability

After 100,000 simulations:

  • Obama wins 99.9%, Romney wins 0.1%.
  • Average (SE) EC votes for Obama: 325.7 (18.0)
  • Average (SE) EC votes for Romney: 212.3 (18.0)
  • Median (95% CI) EC votes for Obama: 323 (295, 363)
  • Median (95% CI) EC votes for Romney: 215 (175, 243)

Each column of this table shows the electoral vote total aggregated by different criteria for the probability of winning a state (Safe=100%, Strong=90%+, Leans=60%+, Weak=50%+):

ThresholdSafe+ Strong+ Leans+ Weak
Safe Obama160
Strong Obama120280
Leans Obama3232312
Weak Obama101010322
Weak Romney111216
Leans Romney4747215
Strong Romney124168
Safe Romney44

This table summarizes results by state. Click on the poll count to see the individual polls included for the state.

00EC#Total%%ObamaRomney
84VotespollsVotesObamaRomney% wins% wins
AL91*754 37.8 62.2 0.0100.0
AK30*(0)(100)
AZ1121329 45.6 54.4 1.2 98.8
AR61*679 36.8 63.2 0.0100.0
CA5521728 58.3 41.7100.0 0.0
CO91552 52.2 47.8 77.0 23.0
CT71*1460 58.9 41.1100.0 0.0
DE30*(100)(0)
DC30*(100)(0)
FL2932976 49.3 50.7 29.0 71.0
GA162952 43.7 56.3 0.4 99.6
HI41*517 64.8 35.2100.0 0.0
ID40*(0)(100)
IL201*546 61.5 38.5100.0 0.0
IN111*447 45.0 55.0 6.6 93.4
IA611086 55.4 44.6 99.5 0.5
KS61*442 45.0 55.0 7.2 92.8
KY81*528 45.5 54.5 7.0 93.0
LA81*542 41.1 58.9 0.2 99.8
ME21552 54.3 45.7 92.4 7.6
ME111*488 64.8 35.2100.0 0.0
ME211*421 53.7 46.3 85.7 14.3
MD101792 62.4 37.6100.0 0.0
MA1121015 62.7 37.3100.0 0.0
MI1621060 55.6 44.4 99.4 0.6
MN101467 58.0 42.0 99.4 0.6
MS61*717 40.0 60.0 0.0100.0
MO101536 50.6 49.4 57.4 42.6
MT31428 46.3 53.7 12.9 87.1
NE21460 42.4 57.6 1.2 98.8
NE111*389 45.5 54.5 9.9 90.1
NE211*252 49.6 50.4 46.5 53.5
NE311*284 35.9 64.1 0.1 99.9
NV61480 54.2 45.8 90.0 10.0
NH411093 56.4 43.6 99.9 0.1
NJ1422077 56.5 43.5100.0 0.0
NM51*494 57.5 42.5 99.1 0.9
NY291720 60.7 39.3100.0 0.0
NC1542125 48.8 51.2 22.5 77.5
ND31*480 41.3 58.8 0.4 99.6
OH1832702 52.4 47.6 96.0 4.0
OK71448 30.4 69.6 0.0100.0
OR711327 52.0 48.0 84.7 15.3
PA2021058 54.0 46.0 96.6 3.4
RI41*495 59.4 40.6 99.9 0.1
SC91*1833 51.7 48.3 85.1 14.9
SD31*442 44.3 55.7 4.9 95.1
TN111654 46.0 54.0 7.0 93.0
TX381460 38.9 61.1 0.0100.0
UT61*688 33.0 67.0 0.0100.0
VT31528 67.8 32.2100.0 0.0
VA1321863 53.0 47.0 96.3 3.7
WA122936 56.9 43.1 99.8 0.2
WV51*373 40.8 59.2 0.6 99.4
WI1063282 52.5 47.5 97.9 2.1
WY30(0)(100)

* An older poll was used (i.e. no recent polls exist).

Details of the methods are given in the FAQ.

The most recent analysis in this match-up can be found from this page.

Comments

  1. 1

    Politically Incorrect - who has been banned over at soundpolitics.com spews:

    It’s like I said many times – Obama will get re-elected but he’ll have a hostile House and Senate to deal with. So, nothing much is gonna happen in the next approximately 5 years.

  2. 2

    Roger Rabbit spews:

    @1 “So, nothing much is gonna happen in the next approximately 5 years.”

    Then that will be the GOP’s fault.

  3. 3

    Roger Rabbit spews:

    @1 (continued) I think you’re excessively optimistic about how the GOP will fare in congressional races. I foresee a smaller GOP majority in the House and a smaller Democratic majority in the Senate.

    To expand upon my comment @2 above, our country is in a crisis,
    and if nothing is done in the next two years because of Republican obstructionism, Obama can count on a friendly House and Senate for his last two years.

  4. 4

    Puddybud spews:

    With the slobbering liberal press hammering Romney over little things, while giving Obummer a pass over everything, it’s a wonder Romney has any traction anywhere.

    We see the slobbering liberal press won’t report specific findings about Obummer’s connections to Vulture private equity firm Vestar, and private hedge funds billionaire Jeffrey “I’m not a sexual predator, I’m an ‘offender’”Epstein. Eventually this will come out and when it does we’ll see.

    The slobbering liberal press won’t discuss http://www.committeeforisrael.com/ findings.
    Or Haaretz

    Obama also stressed he probably knows about Judaism more than any other president, because he read about it – and wondered how come no one asks Speaker of the House of Representatives John Boehner or Senate minority leader Mitch McConnel about their support to Israel.

    William Kristol, founder of The Weekly Standard, criticized Obama’s comments at the meeting with Conservative rabbis, saying that the “reason no one asks John Boehner or Mitch McConnell about their support for Israel is because they really do support Israel,” adding: “The reason people ask Barack Obama about his support for Israel is because his support for Israel has been equivocal.”

  5. 5

    Liberal Scientist is a slut who occasionally wears a hoodie spews:

    With the slobbering liberal press hammering Romney over little things, while giving Obummer a pass over everything, it’s a wonder Romney has any traction anywhere.

    Teh STOOOPID! It BURNS!! AAHHHARGGHHH!!!

  6. 7

    Liberal Scientist is a slut who occasionally wears a hoodie spews:

    Why don’t we make Israel the 51st state and just be done with it.

  7. 8

    Puddybud spews:

    Another reason DUMMOCRAPTS get over is

    without lies, Democrats could never get elected. And they’ve become accustomed to getting a pass from the media on their lies, so that anyone who calls them on their constant, routine and habitual lying must be destroyed as an existential threat to the Democratic Party.

    For example Politifact’s Lie of the Year was not reported by the slobbering liberal press! Yet leftist piheads Paul Krugman, Steve Benen, and Matthew Yglesias all lied in their attack on Politifact. And of course you leftist pinheads joked about throwing grandma over the cliff like it was real.

  8. 9

    Liberal Scientist is a slut who occasionally wears a hoodie spews:

    WRT Electoral calculations…all the nefarious nastiness by Gov Skeletor in florida might come to naught – if Obama only keeps Mich, Ohio, Pa and Va – not too tall an order, along with solid blue states – he wins.
    The coalition of the Confederacy, farmers and Mormons that Mittster is going to win will not be enough to put him over the top. See here. This allows him to lose Colorado and Nevada, too, which he won’t do anyway.

    No matter what the national polls say, the action is in the states, and if Mitt cannot win any of his ‘homestates’ (Mich, Mass, NH), or the industrial states that are coming back under Obama (Mich, Ohio, PA) or the purple->blue Virginia, he’s cooked.

    One question that struck me the other day: has an ex-governor ever won the Presidency without winning the state he governed?

  9. 10

    Liberal Scientist is a slut who occasionally wears a hoodie spews:

    Any sentence that begins…

    For example Politifact’s…

    …is discredited before it begins. Peddle your ‘politifacts’ elsewhere, puddles – they, like you, have no credibility with rational, critically thinking people.

  10. 11

    Puddybud spews:

    Peddle your ‘politifacts’ elsewhere, puddles – they, like you, have no credibility with rational, critically thinking people.

    Hey ylb, looks like your use of Politifact is not well received by Lib da moron!

    So if Goldy, Darryl, Lee or Carl uses Politifacts in the future I’ll remind them of the legend in his own mind commentary!

  11. 12

    Liberal Scientist is a slut who occasionally wears a hoodie spews:

    Oops – linky above doesn’t seem to work. I was using ’270towin.com’ – cool interactive Electoral College calculator.

  12. 13

    Liberal Scientist is a slut who occasionally wears a hoodie spews:

    So if Goldy, Darryl, Lee or Carl uses Politifacts in the future I’ll remind them of the legend in his own mind commentary!

    You’ll do whatever crazy-ass idea *pops* into your pointy little “computer industry” head.

    I find Politifact utterly uncredible. See here. And here. And here.

    It may have been a good idea, once – but bowed to the pressure to be ‘fair and balanced’ in an environment overwhelmed with right-wing lies, it, were it truthful, would be criticizing the right more than the left, and for awhile did that, before succumbing to the pressure. It’s now a parody.

  13. 14

    bob spews:

    Glad to see Darryl’s finally shifted NC to the red column. Now, SC……

    NBC/Marist has a tie in IA.

    NBC/Marist has Obama up only one in CO.

    NBC/Marist has NV up only two in NV.

    I’m looking forward to another poll in PA and in OH after Obama’s ‘Polish Death Camps’ gaffe.

  14. 15

    Puddybud spews:

    Or the incessant Tea Party stories while OWS freakazoids are plotting to blow up bridges and buildings and computer hacking. Barely covered in the slobbering liberal press.

    Thanks to the Democrat-endorsed Occupy Movement, Islam is no longer our only serious terror threat. Occucommies are making the inevitable transition from defecating on police cars to attempting terror attacks:

  15. 16

    bob spews:

    DOMA just struck down by an appeals court. An odd chorus of cheers from the gay-rights crowd, and from the Tenth Amendment Matters crowd.

    This one needs its own thread. On top of ACA, it will be a pretty big deal in October. States’ rights. Apparently someone thinks they matter. What a concept.

  16. 17

    Puddybud spews:

    Bwaaaaaaaaaaa haaaaaaaaaaaa haaaaaaaaaaaa haaaaaaaaaaa haaaaaaaaaa haaaaaaaaaa using Rachel Maddow Lib the moron? The one who claimed the US Constitution doesn’t have a preamble? Really?

    So let’s review…
    In 2009, Politifact’s Lie of the Year was “Death Panels.” Y’all applauded the choice.

    In 2010, Politifact’s Lie of the Year was “Government Takeover of Health Care.” Y’all applauded the choice.

    Golly you didn’t scream at all this Tuesday when your crazed deranged debazed databaze bud ylb placed his Politifact link “claiming” Obummer was a wise spender because it made Obummer falsely look good. Skipped right over that one. It was the first in the thread. Or are you saying you don’t pay attention to his crap?

    Oh the burn of “false righteous indignation”!

    Whatamoron!

    SMACK!

    SMACK!

    SMACK!

  17. 21

    Puddybud spews:

    @18,

    If Politifact is fair and balanced in their evaluations and not libtard nuanced like the are most of the time, they are useful. When they swing from the libtard side, libtards like Lib da schmuck go apoplectic!

    Typical so typical!

  18. 22

    spews:

    Bob @ 14,

    “Glad to see Darryl’s finally shifted NC to the red column. Now, SC……”

    No…I can’t take credit or blame for the color shift in NC.

    The color of a state is determined from polling data. In this case, the four current polls (defined as polls covering the last 30 days) favor Romney and suggest he would have about a 77% chance of taking the state if the election was held now.

    For South Carolina, unfortunately, the most recent poll is from last December. That was a very large poll and it happened to give Obama a small edge (+3%) edge. That translated into an 85% probability of winning the state. That stands until there is new evidence in the form of a new poll….

    Regarding the three new polls you mention…they came out today, as well as a new New York poll (with Obama leading) and a new Ohio poll (with Romney at +2). All of them (and more) will be incorporated into the next analysis.

  19. 23

    Puddybud spews:

    Facebook stock continues to drop like a rock (down $1.25) and the slobbering liberal press will not discuss the love-in between DUMMOCRAPTIC loving FB head Mark Zuckerberg and the Obummer Sadministration starting April 2011.

    How many DUMMOCRAPTIC 99%ers lost their shirts over Obummer’s love of FB???

  20. 24

    bob spews:

    @ 21

    It’s desperation. Obama out-raised McCain 3:1 in 2008 and that stupefying advantage is not going to be there for him this year. It’s going to be about his record, and everything that can’t be spun his way hurts.

    “…sharp deceleration of employment in the springtime months,”

    http://www.cnbc.com/id/4762794.....in_May_ADP

    With the new definition of recession, if Q2 comes in poor like Q1 did, Obama will be facing the fact that a recession occurred under his watch, in an election year.

    Kinda hard to fundraise while claiming that is 2012 recession is because of GWB43 in 2008. Although I’m sure Axelrod will try.

    Fear.

  21. 25

    Puddybud spews:

    I realize Massachusetts is a DEEP BLUE state but this could affect the US Senate race.

    When the issue first surfaced last month, Warren said she only learned Harvard was claiming her as a minority when she read it in the Boston Herald.

    and…

    Warren has not met any of those standards. Though she continues to consider herself Native American, she has not provided any genealogical evidence.

    So this continues to be true

    without lies, Democrats could never get elected. And they’ve become accustomed to getting a pass from the media on their lies, so that anyone who calls them on their constant, routine and habitual lying must be destroyed as an existential threat to the Democratic Party.

  22. 26

    bob spews:

    @ 22

    We’ve been through this before, Darryl. You have your criteria and that’s fine.

    It’s just that someone needs to counter the YLBs and the Rujax!s by pointing out that common sense dictates that Obama does not have a 99.9% likelihood of re-election, your analysis notwithstanding.

    Thanks for the info about OH, as I have not yet seen that poll result.

  23. 28

    spews:

    Puddy, Bob,

    This is not an open thread. A discussion of Facebook stock or OWS really doesn’t belong here. Take it to an open thread.

    Likewise, Bob, discussions of DOMA belong elsewhere. Please review the HA Comment Policy

    I’m happy to see a broad and meandering conversation take place in these election analysis threads, but if you want to open up a discussion of an unrelated topic, it belongs in an open thread.

  24. 29

    Liberal Scientist is a slut who occasionally wears a hoodie spews:

    I tire of smacking puddles around – yesterday wore me out – it really isn’t worth more than a bit of diversion, and morbid fascination, like watching a slow-motion car wreck. There’s not a rational actor behind the screen, merely a twisted, angry Dominionist with poor control over his id.

    I’ll be spending most of my time today on more constructive pursuits…got a mancave to build.

  25. 31

    bob spews:

    @ 28

    Noted. I do think it will play into the election, and this is a thread about predicted election results. Otherwise it goes….where? A two-day old thread?

    As long as we’re pointing fingers about adherence to policy, how many electoral votes @7 are you going to grant to Israel in your next poll, Darryl?

  26. 32

    spews:

    Bob,

    “It’s just that someone needs to counter the YLBs and the Rujax!s by pointing out that common sense dictates that Obama does not have a 99.9% likelihood of re-election, your analysis notwithstanding.”

    I’m not sure how your comment “counters the YLBs and Rujax’s, especially considering that neither of them have commented in this thread…[edit: or hadn't when you wrote your comment].

    In the absence of functioning crystal balls, it would be pretty silly to pin a specific number like 99.9% to Obama’s chances in November. However…

    The polls are a very specific form of evidence, and they DO suggest (with quite strong evidence) that Obama would almost certainly win a hypothetical election held now.

  27. 33

    Puddybud spews:

    Darryl,

    If the press discussed the true tie-in between Obummer’s sadministration and Facebook, maybe some of those close state percentages would flip the other way. That’s why I posted it here. How many peeps bought into FB because they are Obummerites? No press coverage whatsoever!

  28. 34

    Puddybud spews:

    So Darryl,

    You don’t think a true press discussion of the Democrat-endorsed Occupy Movement to the same level as their Tea Party screeds won’t affect some independents in swing states? No press coverage whatsoever, just a blatant press cover-up!

  29. 35

    spews:

    ylb placed his Politifact link “claiming” Obummer was a wise spender

    Yawwwnnn.. That’s a bald faced lie.. Never claimed that at all..

    Lie upon lie upon lie.. Another day in Puddybud’s delusional universe.

  30. 36

    bob spews:

    @ 32

    I don’t have the comment-searching prowess that YLB has to back me, but while you say you’re not sure, I believe you have a pretty good idea.

    ‘Silly’ is an excellent word.

    Things change quickly, and using month-old data as part of your poll criteria is a big problem. I wonder, if you just threw all of the deep blue and deep red electoral votes to the appropriate candidates and focused only on the states absolutely or potentially in play, and then used poll data only from the previous 10 days, what your Monte Carlo analysis would show? You would still have the SC problem although you could use a trend of voter party identification in, say, 2004, 2008, and 2010 to give a ‘mulligan’ to the GOP on that state, rather than adhering to a December poll.

    I’m just saying that relying on criteria that, by your own admission, results in a ‘silly’ predictive numerical value seems, well….silly.

    And in your last sentence, I take issue with ‘almost certainly’. Hold that election on, say, June 7-14, after WI results are known and the real money starts to flow in to the PAC coffers, and you’re somewhere between ‘possibly’ and ‘likely’. But I digress.

  31. 37

    spews:

    Puddy @33 and @34,

    “You don’t think a true press discussion of….”

    I don’t think you introduced either of those topics in the context of the election. So they belong in an open thread.

    Seriously, those comments would have been perfect for an open thread; they didn’t quite fit in the conversation going on in this thread.

    So…let’s terminate this conversation here (follow-ups should go to an open thread) and get back to election stuff. Thanks!

  32. 39

    spews:

    Bob @ 36,

    “…but while you say you’re not sure, I believe you have a pretty good idea.”

    Seriously…I have absolutely no idea what you are getting at.

    “Things change quickly, and using month-old data as part of your poll criteria is a big problem.”

    Actually…with a few exceptions, things don’t tend to change that quickly. It is easy to misinterpret the inherent variability that arises from sampling as “change”. One of the advantages of doing the analyses the way I do is that the entire distribution of polling results is considered. Hence, even though Obama had a smallish lead in S.C., he only won S.C. in ~85% of the 100,000 simulated elections.

    “I wonder, if you just threw all of the deep blue and deep red electoral votes to the appropriate candidates and focused only on the states absolutely or potentially in play, and then used poll data only from the previous 10 days, what your Monte Carlo analysis would show?”

    Wonder no more. Just for you, I re-ran the analysis using an 18 day window and a 10 day window. (BTW: One of the “arbitrary” decisions made in my analyses is the size of the “current poll” window. Right now the window is 1 month, but as the frequency of polling increases, the window will shrink.)

    With an 18 day “current poll” window:

    • Obama wins 99.8%, Romney wins 0.2%.
    • Average (SE) EC votes for Obama: 321.9 (19.3)
    • Average (SE) EC votes for Romney: 216.1 (19.3)
    • Median (95% CI) EC votes for Obama: 321 (287, 363)
    • Median (95% CI) EC votes for Romney: 217 (175, 251)

    With a 10 day “current poll” window:

    • Obama wins 99.9%, Romney wins 0.1%.
    • Average (SE) EC votes for Obama: 340.9 (18.8)
    • Average (SE) EC votes for Romney: 197.1 (18.8)
    • Median (95% CI) EC votes for Obama: 342 (299, 374)
    • Median (95% CI) EC votes for Romney: 196 (164, 239)

    “You would still have the SC problem although you could use a trend of voter party identification in, say, 2004, 2008, and 2010 to give a ‘mulligan’ to the GOP on that state, rather than adhering to a December poll.”

    Yes…there are many arbitrary things I could do to “fix” things that defy how your or I “think it should be.” But I prefer to set a very specific and impartial rules before hand and adhere to them. The fact is, last December there was solid evidence that Obama would take the state (the results were well within the margin of error because the sample was over 2000 people).

    “I’m just saying that relying on criteria that, by your own admission, results in a ’silly’ predictive numerical value seems, well….silly.”

    Oh? I suggest that the results are silly? In fact, because the results are driven by polling results I think they are far less silly (and far more objective) than any analyses based on how you or I “feel” things should be.

    “And in your last sentence, I take issue with ‘almost certainly’.”

    I know…you just don’t “feel” that the evidence is quite right. But that is the best evidence we have.

    “Hold that election on, say, June 7-14, after WI results are known and the real money starts to flow in to the PAC coffers, and you’re somewhere between ‘possibly’ and ‘likely’. But I digress.”

    Yes…you do. I don’t “feel” the WI election will change much in the state head-to-head polling, but I’ll let myself actually be informed by the new polling.

  33. 40

    Liberal Scientist is a slut who occasionally wears a hoodie spews:

    @39
    The differences between left and right thinking, in sharp relief.

    I’m stunned at the difficulty with grasping the primacy of data and observable fact, and the drive to ‘tweak’ to produce a desired or expected result. I suppose there is a reason that I have met so few Republican scientists.

  34. 42

    bob spews:

    @ 39

    Regarding rapidity of change, look at what happened to Lugar in the last two weeks of his primary.

    Look at what has happened to Romney’s favorability numbers with women in the past two weeks or so.

    Look at what happened to Gerald Ford after a bad debate.

    Look at what happened to Reagan/Carter in the weekend just before the ’80 election.

    When there are wide gaps, things don’t change much. When it’s close, a lot can happen in a very short period of time.

    A big thank you for the numbers work in @39. Way beyond the call of duty. Way beyond.

  35. 45

    spews:

    Bob @ 41,

    Ummm…still makes no sense.

    Above you stated your reason: ““It’s just that someone needs to counter the YLBs and the Rujax!s by pointing out that common sense dictates that Obama does not have a 99.9%”

    In the comment you link to, YLB does not claim Obama has a 99.9% chance of being reelected. Rather, he made the simple, factual statement that, according to my analysis, your claim that obama is “currently in” “deep shit” is not supported.

  36. 46

    bob spews:

    @ 45

    I do not accept a 99.9% likelihood of an Obama victory given current reality. I think it’s noteworthy that people like Silver, Trende, and Sabato don’t spew any number remotely approaching that. In fact, I don’t see Monte Carlo analyses anywhere else (would appreciate links to other sites that use it for political purposes, tho). I don’t have a statistical background to support my suspicion, but my suspicion is that Monte Carlo is not an appropriate analysis to use for this purpose. Yes, Lib Sci, it’s an uninformed opinion I hold here. Feel free to point it out, and to despair in doing so.

    No one in their right mind would take Obama on a monetary bet if those were the odds.

    My point was that people like YLB and Rujax! tend to give undue credence to your results, and the fact that YLB shortly after your post referred to your post in another thread sort of reinforced my point.

    I’m variously referred to as a troll, a wingnut, and am frequently described as being paid to comment here. At minimum, I’m not being paid. The others, I suppose, are open to interpretation.

    Without people like me, and others on the right commenting here, people on the left have no reason to try to adhere to reality. And my sole point is that 99.9% is not reality.

    Does Romney have a better chance than Obama at winning? No, not right now. Does Obama have a 99.9% chance at winning? Absolutely not. And everyone here knows it, despite what they might be wishing will happen.

  37. 49

    spews:

    Bob @ 42,

    “Regarding rapidity of change, look at what happened to Lugar in the last two weeks of his primary.”

    Well…you can believe there was rapid change if you want. But since there were only two polls taken in the race, one month apart, and using pretty small sample sizes (especially the first), chances are, you are confusing variability from sampling with “rapid change”.

    “Look at what has happened to Romney’s favorability numbers with women in the past two weeks or so.”

    This is a classic case of narrative outstripping actual data. What you cite is based on one poll, and one subclass (women) within that poll that someone happened to notice. In fact if you look at Romney’s favorability overall, it changes rather gradually, with a large scatter of point estimates about the line. In other words, poll-to-poll differences are largely sampling variability. I think this favorability graph does a great job of showing how variable the polls are, but how gradual the aggregate trend is.

    “Look at what happened to Gerald Ford after a bad debate.”

    Not true. The Poland gaffe had absolutely no measurable effect on the polling. The debate was in early October, and the polls were very stable through the election.

    “Look at what happened to Reagan/Carter in the weekend just before the ‘80 election.”

    Again…narrative outstripping actual evidence. There were so few polls taken in the last week, that the claim cannot be supported over normal sampling variability. Take a look at the polling data.

    “When there are wide gaps, things don’t change much. When it’s close, a lot can happen in a very short period of time.”

    Sure…things _CAN_ but empirically, it rarely happens. More often, people make up narratives about poll results that are not statistically supported. This particularly happens when the “numbers flip” in a close race. The flip is considered “real,” and “meaningful,” when it is quite likely a result of sampling variability.

    For example, if you look at the Colorado polling at the top of this page, you could imagine the narratives as each of the last three polls came out:

    Last PPP poll: “OH MY GAWD! OBAMA IS RUNNING AWAY WITH COLORADO!!!!”

    Last Purple Poll: “OH MY GAWD! OBAMA HAS SUDDENLY TANKED IN COLORADO!

    Last PNA/Keeting Poll: Oh…okay. So Obama has a modest lead in Colorado.

    Statistically, there is no good evidence that anything has changed in the past three polls. In fact, the polling data in-total, supports the hypothesis that Obama has had a constant +3% lead in the state over the entire period. (More complex hypotheses are possible, as well, but parsimony is a good principle to use for these things.)

    “A big thank you for the numbers work in @39. Way beyond the call of duty. Way beyond.”

    Piece of cake! The computer does all the work… I just wait about 5 minutes for the analysis to complete and then copy/paste the table.

  38. 50

    spews:

    Bob,

    “I do not accept a 99.9% likelihood of an Obama victory given current reality.”

    Okay…but, once again, you seem to be making an elementary error: I am not predicting a 99.9% likelihood of a November victory for Obama. Rather, I am giving the current “score”. I go out of my way to point out that the probability is that of winning a hypothetical election now (not November).

    “I think it’s noteworthy that people like Silver, Trende, and Sabato don’t spew any number remotely approaching that.”

    I wouldn’t know, I don’t really read any of them. But, are they giving a current score? Or is this, again, your confusion in attributing my results to a November victory?

    “In fact, I don’t see Monte Carlo analyses anywhere else (would appreciate links to other sites that use it for political purposes, tho).”

    There were about 1/2 dozen of us doing Monte Carlo analyses in 2008. I haven’t really looked for 2012. But Nate Silver did do Monte Carlo analyses for the 2008 election. He started in February 2008, as I recall (I started in Oct. 2007). The problem with his analyses was the he added “stuff” other than polls into his analyses. Consequently, his final results (mean electoral votes) were pretty far off. I was off by one vote. I wrote about it (and why) here.

    “I don’t have a statistical background to support my suspicion, but my suspicion is that Monte Carlo is not an appropriate analysis to use for this purpose.”

    Hmmm…I do have a statistical background (in fact, I teach graduate level applied statistics at a major research university), so…I don’t share your suspicion. In fact, a Monte Carlo analysis, if done properly, IS the appropriate way to estimate the _probability_ of each side winning for the period that the polls cover. (There is a non-Monte Carlo combinatorial technique that yields the same results.)

    Note that most analysts do not talk about the _probability_ of winning….because that takes a Monte Carlo analysis.

    “Yes, Lib Sci, it’s an uninformed opinion I hold here.”Feel free to point it out, and to despair in doing so.

    You missed his point. There is a place for opinion, feelings, reading between the lines, etc. But when looking at a jar of mostly green marbles, don’t deny they are mostly green, simply because you don’t like that color.

    You might argue that the _assumptions_ are wrong (greens are only on the _outside layer_, or it could be an artifact of the lighting, or they are “mood marbles” that will turn orange when the room warms up, etc.). But please, try to offer something rational, rather than simple feeling-motivated denial.

    “No one in their right mind would take Obama on a monetary bet if those were the odds.”

    ??? If those were the odds, I rather think lots of people would be willing to bet on it. But I think your problem is a misinterpretation of the meaning of the “99.9%”. The election happens in November, not now. The number refers to an election held “now.”

    When watching a basketball game, a 66 to 10 lead ten minutes into the game doesn’t mean the team with 66 will win the game. But the score is still considered highly useful information. That’s what I am doing these elections for–to keep a running score through the election.

    “My point was that people like YLB and Rujax! tend to give undue credence to your results, and the fact that YLB shortly after your post referred to your post in another thread sort of reinforced my point.”

    YLB and Rujax (both of whom know me) accept that my analyses are completely objective, data-driven, and appropriate. (And they are right!) They accepted my analyses even during the long periods when Obama was not on top.

    You don’t accept the analyses because (1) you don’t understand them, (2) you don’t like ‘em, and (perhaps) (3) you don’t trust that they are objective.

    “Without people like me, and others on the right commenting here, people on the left have no reason to try to adhere to reality. And my sole point is that 99.9% is not reality.”

    Again…this is a simple error on your part of believing this is a prediction for November. It isn’t. The 99.9% is reality, in that that’s what the data suggest for an election held now. My methods are well described and repeatable for anyone with computer programming skills; I make the data available for each analysis. The results are the results whether I personally like them or not. The one thing that isn’t reality is that the numbers are for a hypothetical election.

    “Does Romney have a better chance than Obama at winning? No, not right now. Does Obama have a 99.9% chance at winning? Absolutely not. And everyone here knows it, despite what they might be wishing will happen.”

    Hopefully, you are now able to identify the errors you’ve made in this statement. :-)

  39. 51

    Steve spews:

    “Without people like me, and others on the right commenting here, people on the left have no reason to try to adhere to reality.”

    My, you wear your narcissism so proudly, Bob, as though it was a cheap flag pin or something.

    Reality, as perceived by Bob, a dumbed-down racist troll who shares with us his visions of an America ruled by a Republican Caesar and his overlords, with white militias roaming the dark streets of American cities. Yeah, right, Bob. You and reality.

  40. 52

    spews:

    Puddy @ 47,

    Thank you for pointing out the Rasmussen Reports web site.

    As it happens, I am familiar with the site and their polls. In 2008, I collected 365 of their state head-to-head polls in the Obama–McCain race (I believe it was every one of them). This season, I’ve collected all 37 of their state head-to-head polls in the Obama–Romney race.

  41. 53

    Liberal Scientist is a slut who occasionally wears a hoodie spews:

    Darryl, you do yeoman’s work. Thank you for your intriguing analyses, and your remedial work with our right-ward wingnuts and their troubled relationship with data.

  42. 55

    Doc Daneeka spews:

    Bless you, Darryl.
    I sincerely doubt you find yourself having to summon such patience with your grad students.

    Bob, Darryl’s giving you a very valuable education for free. I know I appreciate it, and I already had to learn most of this stuff decades ago. Yet I have to admit that human bias is often more potent than a quality education.

    Plus also (too), the marbles should be M&Ms cause they’re delicious.

  43. 57

    bob spews:

    Crosstabs in WI:

    Obama and Mitt Romney were tied at 46% among likely voters in the Marquette Law School poll released Wednesday.

    Obama led Romney 47% to 46% among likely voters in the Public Policy Polling survey released Tuesday – a virtually identical result.

    http://www.jsonline.com/blogs/.....ewestfirst

    Same poll, and yet instead of using the numbers that reflect the group of voters most like to actually show up and, you know, VOTE, instead numbers showing an 8 point Obama margin rather than a tie are used.

    All lessons aside, one must wonder how many of the polls used in this analysis are polls in which data less representative of ballot box participation were used rather than more predictive likely voter data contained within the same polling result.

  44. 58

    spews:

    Bob @ 57,

    “Same poll, and yet instead of using the numbers that reflect the group of voters most like to actually show up and, you know, VOTE, instead numbers showing an 8 point Obama margin rather than a tie are used.”

    Nope…you are mistaken on a number of fronts.

    Rather than speculate in ignorance, why don’t you simply look at the data? I make it super easy to do. Go to the state table above and click on the poll count for Wisconsin.

    If you had done that, you would have learned that:

  45. There were six separate polls included in my analysis for Wisconsin
  46. The Marquette poll released Wednesday did NOT have Obama and Romney tied. You were confusing it with an older poll. In Wednesday’s poll:
    • Obama led Romney by 52% to 40% among registered voters
    • Obama led Romney by 51% to 43% among likely voters
    • I used the likely voter results (I always do when both are given).
  47. There was a second Marquette poll included that is a couple of weeks old. That is the Marquette poll mentioned in the JS article you linked to. In that one:
    • Obama led Romney by 46% to 44% among registered voters
    • Obama was tied with Romney, 46% to 46% among likely voters
    • I used the likely voter results (I always do when both are given).
    • In fact, I used the raw numbers which had Romney leading Obama 281 to 278, so my chart, in fact, shows Romney leading Obama by 0.5%
  48. The PPP poll mentioned in the JS article is included in my analysis.
  49. The JS article you link to discussing the polls is from May 17th, and there have been three new polls taken since then. There is also one older Rasmussen poll included because it falls within the one month “current window”.
  50. In aggregate, the Wisconsin results include a sample of 3,599 people.
  51. “All lessons aside, one must wonder how many of the polls used in this analysis are polls in which data less representative of ballot box participation were used rather than more predictive likely voter data contained within the same polling result.”

    False. One need never “wonder” about this, because one can get off one’s ass, click through to see the polls I include, and then click through to the poll sources….

  52. 59

    Old Midwesterner spews:

    I don’t think there’s much chance you’ll read this, but I’ve been meaning to tell you about this error for quite some time.

    You have a small error on your section for Nebraska. It was Nebraska CD-2 (Omaha metro) that went for Obama in 2008, not CD-1 (Lincoln and eastern fifth of Nebraska, minus Omaha).