Marist, You Magnificent Bastard!

Early voting is creating a unique problem for polling organizations this year  in that the results will skew in favor of the party with the higher early turnout, in this case the Democrats. This built in early voting bias to polls greatly diminishing the polls actual value since you know up front one party’s partisans are over-sampled. Since Democrats tend to vote early, you see the Democrat candidate typically leading by wide margins in early voting according to many polls. When it comes to polling results, all voters who said they already voted make it through the likely voter screen and end up in the final results. This means a sizable pro-Democrat segment of those polled are guaranteed to make it through the likely voter screen. This inherently over-samples Democrats which practically guarantees a favorable result for Democrats. This is how a poll consistently shows Democrat turnout levels at or greater than the best in a generation turnout Democrats enjoyed in 2008 despite mountains of evidence saying otherwise. Of course, Marist has magnificently achieved these outrageous party IDs well before early voting which just goes to prove the old axiom: foolish consistency is the hobgoblin of feeble minds.

Final thoughts on Marist before moving on to the states: I’d argue Marist has been the absolute worst polling outfit this election cycle.  Completely in the tank with Pro-Democrat turnout models arriving at unrealistic results in nearly every survey.  The race for much of the last month has been a dead heat across the Battlegrounds with Romney arguably pulling slightly ahead.  But I saw a statistic this morning that in the dozens of Battleground state polling done by Marist for NBC and the WSJ Mitt Romney led in only one of those polls.  If true and Romney wins the election, no one should ever pay for, read or blog a Marist poll again.  A truly disgraceful showing.  But this is nothing new for Marist.  As I reminded readers two weeks ago thanks to Jay Cost at The Weekly Standard, Marist has a fairly bad track record of over-sampling Democrats.  Immediately before the 2010 mid-terms they released a national survey claiming that among likely voters the country was split right down the middle 46 to 46 voting between the Democrats and Republicans up for Congress (~60% of the way down). As history showed, the election results were quite different from what Marist was seeing. Republicans won the popular vote 52 to 45 netting 63 seats in the House of Representatives.  As the Washington Examiner’s Michael Barone wrote “you could argue that this is the best Republican showing ever.”  Marist?  Nice knowing you.


President Obama leads by 3-points, 49 to46 with 2% voting third-party and 3% Undecided

Party ID is D +5 (Dem 34, Rep 29, Ind 35).  This compares to 2008 of D +6 (Dem 39, Rep 33, Ind 29) and 2004 of R +3 (Dem 35, Rep 38, Ind 27) in 2004.  A very aggressive turnout in favor of the President comparable to his 2008 performance which seems highly unlikely.  Same party ID as their survey a week ago. Another early vote phenomenon favoring Democrats. According to NBC’s First Read, “25 percent say they have already voted or will do so before Election Day, and those voters are breaking to Obama by a 59 percent to 39 percent clip.” This is consistent with yesterday’s Marquette Law School poll (that somehow I missed — totally hiding behind the Hurricane Sandy excuse for as long as I can btw) showed Obama leads among early voters, 56-36%. Survey too many Democrats and you get a Democrat leading, not much more to it.  At the same time, IF Team Obama mobilizes his ground troops to repeat the 2008 turnout advantage, congratulations on your re-election.  I simply believe the overwhelming evidence that shows 2008 was the exception and not the rule for party turnout.

New Hampshire

President Obama leads by 2-points, 49 to47 with 1% voting third-party and 3% Undecided

The party ID is D +1 (Dem 27, Rep 26, Ind 47). In 2008 it was D +2 (Dem 29, Rep 27, Ind 45) and in 2004 it was R+7 (Dem 25, Rep 32, Ind 44). This still strongly shades toward Democrats but quite honestly anything is possible for New Hampshire in my book.  I never know how to read this electorate and I’m always pleasantly surprised when the GOP does well in the state.  It’s just my deep blue New England bias that always makes this state so surprising to me. Objectively though this is a turnout result strongly favors Democrats and Obama only leads by 2 so all-in-all not the worst poll for Romney.


President Obama leads by 6-points, 50 to44 with 2% voting third-party and 4% Undecided

It cracks me up how quickly the Obama surrogates disclaim these large Iowa leads. Obviously they are worried about over-confidence but when both sides say a poll is way-off, it’s not worth spending time simply saying “we agree.”

The party ID is D +3 (Dem 34, Rep 31, Ind 34). This compares to 2008 of D +1 (Dem 34, Rep 33, Ind 33) and 2004 R +2 (Dem 34, Rep 36, Ind 30).  A highly unlikely scenario considering every metric between voter registration, early voting proclivity and enthusiasm dramatically favors Republicans versus the 2008 comparison.  This is a state with aggressive early voting and Democrats dominating so this is again one of the ways where you end up with screwy party IDs that greatly diminish the polls overall value as indicative of state sentiment. According to the First Read write-up, “In Iowa, according to the poll, 45 percent of respondents say they have already voted early or plan to do so, and Obama is winning those voters by nearly 30 points, 62 percent to 35 percent.”


  1. Brad
    Posted November 1, 2012 at 9:53 am | Permalink | Reply

    Rass. numbers out today are troubling. 49-49. Suggests President picked up some ground from Sandy.

    • Brad
      Posted November 1, 2012 at 10:01 am | Permalink | Reply

      Apologies! Those are WI numbers…not national. National not out yet…let’s hope I don’t prove prophetic or anything. 🙂

      • stuckinmass
        Posted November 1, 2012 at 10:02 am | Permalink

        Thanks, I can breathe again!

        49/49 in WI is good news, no?

      • kyle
        Posted November 1, 2012 at 10:03 am | Permalink

        you jinxed it!!! ha

      • jeff
        Posted November 1, 2012 at 10:06 am | Permalink

        Still good numbers in WI. If Romney is tied in a blue state that probably means hes ahead. Also shows +3 in CO.

      • Rick
        Posted November 1, 2012 at 10:06 am | Permalink

        Good numbers for WI…very much in play for R/R

      • Brad
        Posted November 1, 2012 at 10:09 am | Permalink

        Yeah. Rass’ last poll was 50-48 in WI, but a one point move is just noise to me and still a positive.

      • JGS
        Posted November 1, 2012 at 10:12 am | Permalink

        National is now out, Rasmussen still has it Romney +2.

      • TJ
        Posted November 1, 2012 at 10:12 am | Permalink

        You scared the heck out of me…

      • Brad
        Posted November 1, 2012 at 10:13 am | Permalink

        Good. No Sandy bump.

    • John
      Posted November 1, 2012 at 10:09 am | Permalink | Reply

      Not good indeed. As much as it pains me to say it, most of the polls out in last few days show Obama gaining. Some here and elsewhere have poo-pood the “Christy effect” but the pictures and words demonstrated a bi-partisan tone Obama had never shown before during the campaign and it may be causing the few remaining undecides to break toward the President. Christy may end up being “October surprise” nobody expected. I see nothing Romney could’ve done to prevent or change this. The storm was a freak of nature and the press has played it heavily in favor of Obama. Up until today my gut feeling was 60% chance Romney would win. Now I see even chances at best and dropping.

      • Brad
        Posted November 1, 2012 at 10:12 am | Permalink

        John, read my mea culpa. National numbers just came out and are unchanged. Sorry to get everyone riled up….I guess that’s why Keith gets the big bucks around here.

      • John
        Posted November 1, 2012 at 10:13 am | Permalink

        So I did jump the gun – yeah! I’m not a subscriber and assumed the top post had access to the Ras nationals but they were just for WI. Nationals remain 49-47 – Whew I can breath again with Obama staying stuck at that highly referenced number of 47…

      • John
        Posted November 1, 2012 at 10:14 am | Permalink

        P.S. – I still don’t like how Christy handled it. He could’ve been perfectly polite/professional/courteous without fawning over the support Obama was lending. It amounted to an unpaid ad for Obama played over and over on the MSM.

      • Guest
        Posted November 1, 2012 at 10:24 am | Permalink

        Well, the theme of common-sense, uniting, rather than divisive leadership is one that Team Mitt needs to continue hammering, even as they build up his positives on character, values, and policy.

      • UncleFred
        Posted November 1, 2012 at 10:28 am | Permalink

        It’s a dual effect. First you can’t expect Christie to let presidential politics prevent him and the NJ government from cooperating closely with the administration in the face of what’s happened. Second, you are witnessing the power of the press when they are able to spin this “bipartisan” effort to the max. There will be some fallout, but I suspect it won’t be large in the key swing states which have not been hit. It may put Pa out of reach, but these kinds of disasters are very local in their impact.

      • CledusSnow
        Posted November 1, 2012 at 11:13 am | Permalink

        The image of a polarizing president and a polarizing governor coming together in a time of crisis is worth noting, but I just don’t see any undecideds saying to themselves “Oh look at how great Obama is at reaching out to a Republican governor” all the while completely ignoring the past 4 years of his entirely one sided style of running the country. Making political hay out of a terrible natural disaster might even seem tasteless to some indies.

      • TTK
        Posted November 1, 2012 at 11:47 am | Permalink

        John. The pictures or videos of Obama and Chris Christie together will not change any voters’ mind. I would say 90% of the likely voters have already made up their mind.

        The pictures/videos of Obama + Christie is not important to most people. It’s the pictures of Obama + Unemployment that will matter, and on Tuesday the economy will speaker louder than Sandy.

    • Brad
      Posted November 1, 2012 at 10:11 am | Permalink | Reply

      Whoo hoo! National numbers from Rass. are unchanged, 49-47!

      • Posted November 1, 2012 at 10:59 am | Permalink

        Thing to remember about Rasmussen is he’s using D+2. Indicators of an even split or R+1 would dramatically change the total. And could widen the Romney advantage up to six points at the peak level.

  2. Bryan
    Posted November 1, 2012 at 9:57 am | Permalink | Reply

    Extra points for the Patton reference!

  3. stuckinmass
    Posted November 1, 2012 at 10:07 am | Permalink | Reply

    I’m confused? Are Dems winning the early turnouts? What about these polls showing Romney up in early voting?

    Or are Dems just winning the early turnouts in the critical states?

  4. Posted November 1, 2012 at 10:07 am | Permalink | Reply

    Rasmussen has Romney up 3 in Colorado.

    I’m also reading that the private sector added the most job in 8 months in October…Yeah I’m sure.

    • JGS
      Posted November 1, 2012 at 10:10 am | Permalink | Reply

      Letting the federal government release jobs data 4 days before the election is sort of like letting O.J. have control of the critical piece of evidence in his murder trial.

    • Guest
      Posted November 1, 2012 at 10:12 am | Permalink | Reply

      I suppose Obama will run on a slogan that “you’re better off than you were eight months ago”.

      • Guest
        Posted November 1, 2012 at 10:15 am | Permalink

        Does ADP have seasonable adjustment though?

        If not, the BLS numbers will be significantly lower.

  5. rick in virginia
    Posted November 1, 2012 at 10:13 am | Permalink | Reply

    love that title from Patton.

  6. Posted November 1, 2012 at 10:16 am | Permalink | Reply

    Keith (or anyone) I have a question: In the Wisconsin poll, what causes Marquette to change the party ID for Reps and Inds down from their previous poll, especially less than one week before the election? This seems to fly in the face of logic and history.

    • Mike
      Posted November 1, 2012 at 10:31 am | Permalink | Reply

      Marquette and many other pollsters don’t adjust for Party – they’ve said that numerous time. They do, however, adjust for demographics and likely voters. So, what they’re saying is that they randomly called X number of voters, applied their likely voter screen and then adjusted for demographics (age, sex, race, etc.) and voila, they expected a D 5+ (or whatever Marquette had) electorate.

      What Keith is trying to explain is that their likely voter screen is flawed because it allowing more Democrats to pass thru because of the early voting. I too agree that pollsters are (slowly) figuring out how to handle the early voting phenomenon.

      I don’t normally like to jump on the Rasmussen bandwagon, because I do believe he has a R lean, but what he does is a better approach. He runs separate polls to determine the partisan make-up of a said electorate and then based on this split and he’s (hopefully) educated guess, he comes up with what he believes is going to be the electorate turnout. Rasumussen believes it’s a D 2+ (as opposed to D 8+ in 2008). He then adjusts his polls for this split and other demographic factors.

      • kenberthiaume
        Posted November 1, 2012 at 10:34 am | Permalink

        If only 10% or so of people even answer the phone if one group is even a little more likely to answer (republican or democrats at a given age/area/sex) then it will throw the poll off. I guess that’s true even if many people answer the phone but at 10% it seems like people are using caller ID to screen their calls. So who does that more? If republicans answer the phone 9.5% of the time and dems 10.5% of the time, it will screw up the whole poll and you’d get results like we see. But I have no proof of that. Married people with kids probably aren’t in the mood to talk to a pollster as they’re putting kids to bed, and they tend republican…that alone could screw up the poll, I think.

  7. John
    Posted November 1, 2012 at 10:22 am | Permalink | Reply

    Can someone explain Keith’s opening sentence, “Early voting is creating a unique problem for polling organizations this year in that the results will skew in favor of the party with the higher early turnout, in this case the Democrats” in light of the Gallup and Pew polls both showing a strong Romney lead in national early voting. Is this just a state-by-state issue or are a large number of Dems voting for Romney which would support both Dems leading in ballots cast and Romney leading in actual early votes. I hope the later is true.

  8. JGS
    Posted November 1, 2012 at 10:22 am | Permalink | Reply

    Excellent article today on the differences in polling assumptions being used by the two parties. In short, Republicans expect reversion to 2004 or 2010. Democrats see inexorable increase in minority voting percentages as they become a greater share of the electorate, and an inexorable increase in Presidential election turnout. The article doesn’t quite put it this way, but it looks like Democrats expect a registered voter electorate, while Republicans are using a “likely voter” model that is whiter, older, and more Republican-leaning than in 2008. Definitely worth a read — agree that one side is way wrong, and hopefully it’s the D side.

  9. kenberthiaume
    Posted November 1, 2012 at 10:24 am | Permalink | Reply

    Iowa breakdown looks a little off. The site I saw had dems leading about 44-32 with the rest independents in early voting. So basically republicans get nothing from independents, assuming the crossovers cancel. That’s the only way you wind up with 62-35.

    Althought that’s people who HAVE voted early (about 1/3 of the total 2008 vote), not people who HAVE or WILL vote early (Marist has an extra 12% who haven’t voted early yet but will). I suppose it’s possible that late “early voters” will vote 90-10 obama or something.

    Confusing things further, there are about 130K absentee ballots outstanding and according to the total ballots early voted, absentee returned, or absentee outstanding it’s 43/31/26 for d/r/i. The total ballots early voted, absentee returned or absentee outstanding is 660K out of a total vote last time of around 1.54M or 43%.

    But I still don’t see how you can get 62-35 when total ballots outstanding are 43% of 2008 votes and are split 43/31/26 and assumed 45% of eventual 2012 vote.

    • kenberthiaume
      Posted November 1, 2012 at 10:25 am | Permalink | Reply

      Also that’s the sec of state’s site as of 10/30 (not sure if it’s after voting for the day or before)

    • Evan3457
      Posted November 1, 2012 at 10:54 am | Permalink | Reply

      USEP report says last updated 11/1 (meaning through yesterday) for Iowa:
      Party breakdown is 43.3/32.2/24.4 D/R/I compared with Marist’s 62/35, implying the independent early voters are going 7 to 1 for Obama, according to Marist (dubious, to say the least), and early voters comprising 36% of the electorate, not 45%, as per Marist. Compared to 2008, Democrats nearly identical, Republicans up 2% of the early votes submitted.

      The bad news? Obama won by 9.5% in 2008.

      • kenberthiaume
        Posted November 1, 2012 at 11:50 am | Permalink

        well my point was that if they have an obama plus 6 poll and 45% of the group is off by 10+ points of spread (62-35 instead of 56-44 or whatever), then it casts doubt on what they’re doing. If Romney narros the gap from 27 to 12, that’s 15 point improvement with 45% of the population and it would move their survey about 7 net points in Romney’s direction. The wild card is the extent of cannibalization of the election day voters….

  10. billyboy
    Posted November 1, 2012 at 10:39 am | Permalink | Reply

    unemployment rate, every article I have seen says 100-140k jobs and unemployment rate either 7.8 or 7.9

    • jeff
      Posted November 1, 2012 at 11:02 am | Permalink | Reply

      Unless a jaw dtopping report either way womt affect the race.

  11. MikeN
    Posted November 1, 2012 at 10:57 am | Permalink | Reply

    A point about the Wisconsin voter breakdown. Wisconsin voted Democrat while it had the R+7 in 2004. And has been voting Democratic for a while. Perhaps this is just people who are Republican because of the state’s long historical lean to that party, but the reality is these have always been Democrats. So perhaps it’s not a turnout boost, but disgust with George Bush that caused us to see an official change of ‘status’ to D.

    Wisconsin does not do registration by party so I can’t confirm this.
    If this theory is valid, then the R+7 in 2004 is perhaps closer to R+1. At R+3, Dems run 3 points ahead of party identification in 2004 and 10 points ahead in 2008.

    • MikeN
      Posted November 1, 2012 at 11:11 am | Permalink | Reply

      I confused NH and Wisconsin numbers. So Dems outperformed their ID by 8 in 2008 and 3 in 2004. That is more reasonable.

      • UncleFred
        Posted November 2, 2012 at 12:50 pm | Permalink

        Obama won independents by a large margin in Wisconsin in 2008. Romney is winning them now. It will all come down to turnout, both the living and the dead.

  12. UncleFred
    Posted November 2, 2012 at 12:43 pm | Permalink | Reply

    The NH poll is using very unrepresentative demographics. Independents do NOT make up 47% of the electorate. Republicans are the largest demographic by 1%-2%, followed by Democrats, followed by independents and libertarians. Libertarians run somewhere between 3%-6% and generally get lumped in with independents by various pollsters.In past elections, if the vote is close, between 50%-75% of the libertarian vote for president goes to the Republican. Gary Johnson may reduce that this year. So a sample that looks like 32%R, 31%D,30%I,4%L is probably reasonable. Allow +-2% for the major parties. Using Marist sub groups you get 32/31/34. About 34% of the NH electorate has college degree or better, this poll has 54% of respondents with a degree. Planning to vote absentee in NH is not the same as having voted. At this point if a voter does not have a ballot (not an application but an actual ballot) they have to go to the county clerk and get the ballot in person. There is not enough time to complete the process by mail. So many of the people who “plan” to vote absentee will either not vote or vote Nov. 6th. In NH you have to have a reason you can’t vote at the polls to get a ballot.

    I would suggest that you disregard this poll. The over sample of college grads alone is enough to skew the numbers, without the oversampling of “independents”.

One Trackback

  1. […] It bears repeating what I wrote on early voting in my last Marist undressing: […]

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: