On Those New Polls in the Murray—Rossi Race

As I mentioned here and here, we have been treated to two new polls today in the race between Sen. Patty Murray (D) and Dino Rossi (R).

A SurveyUSA was a robopoll taken on 678 likely voters (3.8 MOE) from 24-27 October and has Rossi and Murray tied up at 47% each. The KCTS/KPLU/Washington Poll poll used live interviews of 500 registered voters (4.3 MOE) taken from 18-28 October, and has Murray leading 49% to 45%.

In some sense, both of these polls are older than yesterday’s Rasmussen robopoll of 750 likely voters (4.0 MOE) taken completely on the 26th of October that has Rossi up +1% (48% to 47%). Therefore, the only reasonable thing to do is combine all three polls into one meta-poll and do a Monte Carlo analysis.

The combined 1,928 “votes” are split 916 for Murray (47.5%), 904 for Rossi (46.9%) and 108 (5.6%) for neither candidate. From a million simulated elections at these proportions and sizes, we learn that Murray wins 579,294 times and Rossi wins 414,495 times. In sum, these three most recent polls support a Murray victory with a 58.3% probability and a Rossi win with a 41.7% probability. This is very close to a tie (statistically, it is a tie), but Murray has a slight edge.

WA_SUSA_Rass_Oct
My usual near-election practice is to analyze all polls taken in the past two weeks. There are six such polls:

StartEnd%%%
PolldatedateSizeMOEMurrayRossiDiff
Rasmussen26-Oct26-Oct7504.0 47 48R+1
SurveyUSA24-Oct27-Oct6783.8 47 470
WA Poll18-Oct28-Oct5004.3 49 45D+4
Rasmussen17-Oct17-Oct7504.0 49 46D+3
Marist14-Oct17-Oct5894.0 48 47D+1
PPP14-Oct16-Oct18732.3 49 47D+2

The resulting meta-poll has a total of 5,140 “votes,” of which 2,484 go to Murray (48.3%), 2,406 go to Rossi (46.8%), and 250 go away. From a million simulated elections, Murray wins 785,190 times and Rossi wins 211,969 times. Thus, from the evidence found in polls taken over the past two weeks, we find that Murray has a 78.7% probability of winning and Rossi has a 21.3% probability of winning.

6LateOctPolls

Statistically, the race is a tie because the probability of winning is under 95%, but the odds do favor Murray a little bit.

Comments

  1. 1

    Mr. Cynical spews:

    Darryl–
    Here is an interesting article from a National perspective. It says of those who already voted, R’s & D’s are about equal…yet it also discusses the enthusiasm gap. I guess you’d have to look at which states voted early to draw any conclusions…but it’s kind of a mixed message. It seems to boil down to the huge losses by Democrats of INDEPENDENTS….whereas Obama carried them…he has now lost the same folks by a HUGE margin. In addition, there are more Independents and less R’s & D’s…so the Independent groups has become vital

    http://www.foxnews.com/politic.....566102765/

  2. 2

    Glenno spews:

    Pyrite…

    Now come on…would a bank let you average the last 3 time you checked your balance and average them to let you know what your balance is? In the real world the answer would be NO.

    I guess the answer would be yes if it received a bailout from Murray & Obama.

    So come Tuesday are ready to post Congradulations Sentor Elect Rossi?

  3. 3

    BeerNotWar spews:

    Rasmussen is one of the least accurate and pro-Republican pollsters out there. As a Democrat I’m deeply comforted that Ras shows this race tied. Robo polls also tend to undersample Democrats, so SurveyUSA is likely also off by a small (about 2%) margin.

    On top of that I wager the “Enthusiasm Gap” narrative is skewing these pollsters’ likely voter models even further. I think actual voters will be more Democratic and Murray is going to win, and it won’t be that close.

  4. 4

    czechsaaz spews:

    @2

    Welcome new troll. The Pyrite gloss is pretty funny actually, but you might notice that there are a few posters here at HA. Darryl = Pyrite is just stupid on your part.

    As to your quibbling about Darryl’s methodology, the larger a sample, the better chance of an accurat prediction. Let’s assume the non-robo polls used multiple humans to make the calls. The methodology you don’t like is no different than combining the results of all the calls of all the poll workers and determining results.

    In the real world, if you had bank balances in three different banks, one has $100 in it, one has a million and one is overdrawn by 1,000,100 dollars, by your logic you’d tell everyone that based on one account, you’re a millionaire.

  5. 5

    spews:

    BeerNotWar,

    After doing these type of analyses on a nearly daily basis from Oct 2007 through election day 2008, and on and off since then, my experience is that Rasmussen’s head-to-head polls have NOT been inaccurate or biased.

    [However, Rasmussen daily presidential tracking polls have been systematically biased toward Bush and against Obama. (By "bias", I mean, simply, shifted. It doesn't make the poll useless, but it does mean you'd want to correct for the bias.)]

    That said, a lot of pollsters, pundits, and poll analysts alike have noticed that this year there is a systematic difference between robopolls and live interview polls, with Republicans doing better in robopolls. The bias may be “enthusiasm gap” or “cell phone bias” or something else, but the effect seems to be real. What we don’t know is how it will play out relative to the election results.

  6. 6

    Mr. Cynical spews:

    3. BeerNotWar spews:

    Rasmussen is one of the least accurate and pro-Republican pollsters out there. As a Democrat I’m deeply comforted that Ras shows this race tied. Robo polls also tend to undersample Democrats, so SurveyUSA is likely also off by a small (about 2%) margin.

    And your proof of this claim is______________???
    Nothing, right?
    Outta yer butt and onto the Ass!!

  7. 7

    rhp6033 spews:

    Of course, and discrepencies between the polls and the ultimate result in the race can be dismissed afterwards as “well, we measured it accurately peoples opinions AT THE TIME THE POLL WAS CONDUCTED…”

    But of course, that’s not how polls are used – by anyone.

  8. 8

    Xar spews:

    @6 Cynical: read any of Nate Silver’s many analyses of Rasmussen at fivethirtyeight.com.

    And before you scream about Nate having a liberal bias, read a little more of his site and realize that he’s predicting massive Republican gains, more than many, and that he uses empirical mathematical methods.

  9. 9

    Puddybud identifying useless Moonbat!s since 2005 and identifying rujax as an arschloch! spews:

    Rasmussen is one of the least accurate and pro-Republican pollsters out there.

    And where do you post facts for that claim beeronthebrain? Puddy been posting the polling accuracy on this blog for years. Puddy knows you HOPE Rasmussen is wrong but Scott Rasmussen has been the most accurate pollster for the last 5+ years.

    Suck on that sucka!

  10. 10

    Xar spews:

    @9:

    Puddy, the facts don’t so much bear that out, as much as you’d love for them to. Check out fivethirtyeight.com’s numbers.

  11. 11

    JC spews:

    Darryl, that analysis ignores momentum, although the write-in factor makes momentum less important as I’m guessing the majority of voters have already mailed in their ballots. Whoever wins, I doubt it will be by more than 3%

  12. 12

    Daddy Love spews:

    Deep thought:

    Democrats hate blacks so much that we nominated one for president and then elected him.

  13. 14

    Daddy Love spews:

    6. Mr. Cynical spews:

    ‘Rasmussen is one of the least accurate and pro-Republican pollsters out there…Robo polls also tend to undersample Democrats, so SurveyUSA is likely also off by a small (about 2%) margin.’

    And your proof of this claim is______________???

    In the thread about the SurveyUSA poll, I posted a link to Nate Silver’s NY Times column explaining exactly the reasons why.

  14. 16

    Daddy Love spews:

    Rasmussen:
    1. Robo polls
    2. Doesn’t call cell phones
    3. Calls only at certain time of day
    4. Does not randomly select a person in the household to respond

    2 and 3 and shrink the pool of respondents in a non-random way. The result is a more Republican sample.

    The problems with their “likely voter” screen are fodder for a differnt post.

  15. 19

    Puddybud identifying useless Moonbat!s since 2005 and identifying rujax as an arschloch! spews:

    Xar,

    When an academic (you know leftist leaning type) similar to Darryl, sez it’s true, it’s gotta be.

    Slate 2004 – Rasmussen most accurate

    Daily Kooks 2006 –

    It would be interesting as day goes by in other comparisons to other outfits like Mason-Dixon or Survey USA on close they came but even with some of its bias, the Rasmussen polls are as probably as good as any other out there for accuracy.

    Politico 2008 – Rasmussen most accurate

    Even Josh Marshall of the The Potty Memo begrugingly gave him kudos.

    The toplines tend to be a bit toward the Republican side of the spectrum, compared to the average of other polls. But if you factor that in they’re pretty reliable. And the frequency that Rasmussen is able to turn them around – because they’re based on robocalls – gives them added value in terms of teasing out trends. But the qualitative questions, in terms of their phrasing and so forth, are frequently skewed to give answers friendly toward GOP or conservative viewpoints. All of which is to say that his numbers are valuable. But they need to be read with that bias in mind.

    Just because Nate Silver looks at support for the poll doesn’t make it wrong. It shows the libtardo press harping like pussies over the fact they like their polls like Research2000. Oh yeah they are worthless huh? Didn’t Kooks sue Research2000 after trusting them all that time. Made him look more useless than he already is!

    Also Nate doesn’t like robopolls

    This results in a slight adjustment toward the Democratic candidate in most (although not all) states because the robopolling firms poll with much more frequency, and our models are designed to reward quality rather than quantity.

    Yep, Nate doesn’t like the fact Rasmussen is the most accurate pollster with his robocalls.

    Hey Curtis “Daddy” Love, da facts are da facts! Too bad you can’t handle the truth!

  16. 20

    Puddybud identifying useless Moonbat!s since 2005 and identifying rujax as an arschloch! spews:

    DUMBOCRATS have outspent Republicans $1.074 Billion to $677 Billion and they are worried over Patty’s seat with all the smears and lies generated by the DUMBOCRATIC machine.

    Puddy loves this headline direct from HA

    Research 2000: Murray 52%, Rossi 41%

    Now what happened to Research2000 again boys and girls. Oh yeah… the Kookmeister got upset and threw a tantrum over Research 2000. Damn that was funnnnnnnnnnnnny!

    1 – You have web sites like Gawker
    2 – You have web sites in Soros’ control. Now add NPR to that list
    3 – You have the slobbering libtardo MSM
    4 – You have blogs like this one who repeat the crap from 1-3
    5 – Then you have the slobbering libtardo MSM repeating the crapola you read in the kook blog for example. Then, when called on it, there’s never a retraction. Witness the tepid cumbacks from Rachel Madcow when she’s been caught factless. Puddy placed a Google link up yesterday displaying her antics.

    And you have all the unions putting out their automatons for erection day. Then you have the chronically unemployed like the HA dirt stain ylb telling others what to do and how to vote. What a hoot. Well maybe he hawks stuff like Tim Eyman.

  17. 22

    drool spews:

    Still waiting for Cynical to put words into my mouth so I can tell him to go fuck himself.

  18. 23

    spews:

    We will see just how far off landline robocalls are compared to the actual results in Cal and Tx.
    Hispanic cellphone only homes not showing up in the polling data, but at the polls, should show us something. Can’t wait for the Pew research data to show up.

    Survey of landline only users shows a Republican bias.

  19. 25

    spews:

    ylb telling others what to do

    Nope. I promote democracy.

    If you’re born in this country or are naturalized you have a right to express your will through the ballot box.

    Even if one is an idiot like you Puddybud.

    I leave it to the right wing to order people around. That’s the ultimate fulfillment of their miserable lives.

  20. 26

    Medicaid-Motorchair Conservative spews:

    Darryl: Is there any way to mathematically weight polls that are taken for the express purpose of influencing the results?

    1- How would one determine whether the poll was real or fake?
    2- How would one mathematically determine the effect?

    If you can quantify this sort of human cupidity, you would have done us all a mighty service.

    And we could have more respect for the Monte Carlo technique.

  21. 27

    spews:

    MMC @ 26,

    There may be polls that are just trying to influence outcome, but those are probably rare exceptions. For the most part, pollsters live or die by how accurate their results are for those who purchase them. Even Republicans (who generally exist outside the realm of of reality) are able to appreciate that an accurate poll is necessary to know where to invest money and resources. Being fed bullshit wastes their money.

    But there is an important form of selectivity bias in which polls, taken on behalf of partisans, are selectively RELEASED. That is why I generally exclude polls released by campaigns and parties. Using such polls biases results unless you know about ALL the polls the campaign/party has done. This scenario probably comes closest to your concern. My solution is to completely ignore them in serious simulation results.

    But pollsters like Rasmussen, SurveyUSA, PPP, Marist, etc. release all of their head-to-head polls, so even if they are paid for by partisans, the (basic) results are open. No issues arise from using such polls.

    Occasionally we see “pollsters gone bad”. Puddy seems to be saying something about Research 2000 (although I am not sure who the hell can follow his prose). R2K was caught by some academics in cooperation with Kos faking their tracking poll. They were, essentially, committing fraud against DailyKos, and were caught. Even so, I’ve seen no evidence to suggest they were faking their head-to-head candidate polls, so the example Puddy gives from the Murray–Rossi race is entirely irrelevant. R2K is now extinct as a serious polling firm. Likewise, former right-wing pollster Strategic Vision was recently (a year or so ago) caught faking polling data on their head-to-head polls and they are now extinct as a polling firm.

    I hope this has answered your questions.

  22. 28

    ld spews:

    You do not have to look past the polling on all the initiatives in this state to see what side of the fence the voters are on.

    They don’t trust government with more of their money, taxes, or regulation, and they prefer many parts of government be reformed or outsourced.

    Look at New Jersey if you want to see a state that loves its governor, and what is he doing, he is taking the unsustainable government and reforming it.

  23. 29

    spews:

    a state that loves its governor,

    Yawwnn.. A good part of that state is already sick and tired of Tony Soprano meets Rush Limbaugh.

    And the romance the rest are having will wear off soon enough.

    The spectacle that obese jerk made of himself at that Meg Whitman meet and greet was only the beginning.

  24. 30

    Mr. Cynical spews:

    Chris Christie is the most articulate, strong leader of any American Governor. He is clear about the unsustainable government pensions & benefits….and doing something about it.
    He is not mean-spirited…just pragmatic.
    And if other R’s get into office, any office, they need to be just as pragmatic and show leadership…not hot rhetoric like the Obama Regime.
    Obama blew his chance at leadership.
    Yeah, he go a couple big items passed like ObamaCare…but at what cost?
    And will ObamaCare be likely gutted or repealed in the next 3 years? You bet. It’s not where this country is headed. Big Government is not the answer…free enterprise is.

    So Obama and his idea of “progressivism” will merely go down as an ugly mark in history. An example of over-reaction and being a hollow “hope & change” which was really nothing more than a undercover Socialist Movement. The covers are off.
    Emporer Obama has no clothes.

  25. 31

    Mary Plante spews:

    Washington voters have a history of supporting Eyman’s initiatives and simultaneously voting for Democratic legislators. The fact that Washington voters plan to roll back taxes does not indicate that they will reject Democratic candates. In fact, a good many must be supporting Dems and supporting these initiatives. There is no other way to explain the leads Dicks, McDermott, Larson, Inslee and Smith have in the polls.

  26. 32

    Mr. Cynical spews:

    Mary–
    Dicks will win…but not even close to the margin last election. And Larson?? We’ll see.