Blog analysis predicted Iowa; something happened in NH

I don't know exactly how kos predicted the results of the Iowa caucuses. Maybe it was Bushgut, maybe Jupiter was aligned with Mars, or maybe he is just ubersmart. I suppose there may be pollsters who got it pretty close. But there are also now professional blog-sentiment-miners who seem to have a pretty good replica of kos' neural programming. They spot and analyze posts and take into account the poster's track record...if you have read "The Tipping Point" you will recognize the term "mavens."  

So these people atCollective Intellect somehow grab all these blog posts and put it through their sausage grinder (along with the appropriate spices) and out comes a score, which seems to be very closely related to the percentages of the total vote.

Their results in NH emphasize the point that something strange seems to have happened there, perhaps very late in the game. Bradley effect? Tearjerk effect? Big Dog effect? old ladies liberated by good weather? whatever it was, it was not there until the last few days.

I am going to keep an eye on these guys at Collective Intellect and see what they come up with for the next few primaries. (They mostly do work for corporations who need feedback on their products, corporate image, and the like.)

What you blog matters in more ways than you think, especially if you tend to blog intelligently.

disclosure: I was tipped to this by a family member associated with Collective Intellect.

•  tips/kickbacks n/t

  by seesdifferent on Wed Jan 09, 2008 at 06:37:37 PM PDT

•  Maybe a bunch of New Hampshire voters

  just decided to vote for Clinton instead of anybody else and it was enough to put her over the top.

  - Albert Einstein

  by Walt starr on Wed Jan 09, 2008 at 06:38:44 PM PDT

  •  well, they did it at the last moment

    and without giving much of a clue. Or something.

    by seesdifferent on Wed Jan 09, 2008 at 06:41:41 PM PDT

•  Yes, What Happened in New Hampsire Was This...

  

  by EmilyD on Wed Jan 09, 2008 at 06:40:18 PM PDT

  •  Live Free or Die Bold

    The state motto.

    by FishOutofWater on Wed Jan 09, 2008 at 06:43:10 PM PDT

  •  WRONG!

    It was the space aliens who kidnapped fat Elvis and hold the keys tp the grassy knoll.

    - Albert Einstein

    by Walt starr on Wed Jan 09, 2008 at 06:46:53 PM PDT

  •  Troll-rated for spreading lies.

    New Hampshire does not use electronic voting.

    Optical scanning with a paper trail, yes. Very different animal.

    by Elwood Dowd on Wed Jan 09, 2008 at 06:49:30 PM PDT

    •  ???

      What am I missing here? Seriously, I'm asking...

      Even BradBlog was all over this.  

      What You Should Actually Be Concerned About

      However, for those in the media interested in actual concerns about the integrity of today's NH Primary, we'd strongly recommend the following video primer. In it, you'll learn that the Diebold optical-scan system used across most of New Hampshire (some 40% of precincts still count paper ballots by hand, at the precinct), can be easily rigged by insiders, and only if the paper ballots are properly audited by hand, in some fashion, can the results of the op-scanners be trusted in any way, shape or form.

      Link to the video at...the link.

      If you're upset about the actual picture posted here, fine--you can make that known without a TR, can't you?

      by o the umanity on Wed Jan 09, 2008 at 07:11:10 PM PDT

      •  There is a cadre of CTers

        who are deliberately trying to confuse people here.

        They use the term 'electronic voting' when they are talking about paper-trail scanners.

        It's been going on all day. It is a disinformation campaign.

        TRs are called for.

        by Elwood Dowd on Wed Jan 09, 2008 at 07:17:30 PM PDT

        •  I disagree

          to say it's a "disinfo campaign" sounds a little CT itself, elwood.

          Sorry; wrong picture posted or not, this doesn't deserve a TR--poster is a low UID, and has quite a few diaries that do not seem CT-themed. I'd like to respectfully request that your remove the TR.

          You don't have to--YMMV--but I'm uprating it.

          by o the umanity on Wed Jan 09, 2008 at 07:22:40 PM PDT

          •  I respectfully will not.

            I rate comments, not commenters.

            And yes, there is a disinformation campaign with the same deliberately misleading language from Bev Harris' organization being posted across many blogs.

            by Elwood Dowd on Wed Jan 09, 2008 at 07:30:24 PM PDT

            •  No problem

              I just think you're being way too harsh on this particular poster, especially considering the subject at hand.

              Yes, there's been a lot of kookery here and elsewhere today. IMO, this particular comment ain't part of it. But, suit yourself...

              by o the umanity on Wed Jan 09, 2008 at 07:32:41 PM PDT

    •  Optical Scanners can be manipulated

      So what's the freakin' difference?

      Diebold makes machines that can be hacked. Period.

      by posse comitatus on Wed Jan 09, 2008 at 07:22:15 PM PDT

      •  That is True

        In fact here in my county of California, where we use non-Diebold optical scanners with paper ballots the scanner still has an insertable chip, and the Inspectors actually pick the machine up from the warehouse the day before the election and store it in their home overnight before bringing it to the polls. How do I know this? I witnessed it with my own eyes!

        by EmilyD on Wed Jan 09, 2008 at 07:36:08 PM PDT

  •  You've been here a long time

    and are not a troll.  So why are you posting bullshit?  NH does not use touch screens.  ALL ballots are on paper. The only difference is some are hand-counted, and others are optically scanned.

    by dhonig on Wed Jan 09, 2008 at 08:18:45 PM PDT

•  For a long time...

  ...New Hampshire was in the Clinton column.  It was only after Obama won Iowa that the state was even considered in play.  I think it was just that people went back to their original votes after riding the Obama euphoria for a few days.

  by DH from MD on Wed Jan 09, 2008 at 06:41:43 PM PDT

  •  But it was such a FEW days, I mean

    even Bill Clinton complained about the fact that the date was moved up and gave no time....how many Obama events can you go to in that time?

    by ksh01 on Wed Jan 09, 2008 at 06:44:05 PM PDT

  •  Huh. That's too easy. Let's go CT

    Let's not give Hillary's voters credit for getting out in record numbers.

    by FishOutofWater on Wed Jan 09, 2008 at 06:45:16 PM PDT

  •  yes, they went home to their wife nt

    by Marlboro Lite on Wed Jan 09, 2008 at 06:50:55 PM PDT

•  my crazy assed question that never got answered

  maybe because it's crazy, but here's from an earlier post:

  because it occurred to me while reading Zogby's article about what went wrong with polling New Hampshire....but I'll add a disclaimer first:  I'm an Edwards supporter who has given money (very little) to Edwards, Obama, Dodd, Clinton, but I'm detached from the primary process.  I'm not a Hillary hater, I'm just intrigued about how so many polls could be so wrong.

  Zogby wrote:

   

  We expected that Obama would receive the lion's share of independents and drain the Republican primary of these voters. It now appears that, perhaps with a sense that Obama had a lock on the Democratic side, independents felt free to vote on the Republican side and reward their hero, John McCain.

  Mebbe....I personally can't imagine large numbers of people (like more than 5) who would vote for McCain because Obama seemed to have it wrapped up.  I could, however, imagine large numbers of republicans  same-day-registering as dems to vote for the candidate they believed is more beatable in the general.  Has anyone (any of the analysts) mentioned that as possible?  It's just so freakin' weird that ALL polls, save one, would be so far off....just a crazy thought....I'm not dissing Clinton, I'm just puzzled.

  by ksh01 on Wed Jan 09, 2008 at 06:41:56 PM PDT

  •  NH law doesn't permit.

    You cannot be registered as a Republican and re-register on voting day. You cannot change party affiliation.

    by Elwood Dowd on Wed Jan 09, 2008 at 06:57:18 PM PDT

    •  but you can same day register,

      no?

      N.H. REV. STAT. ANN. § 654:34 says you can register the same day as an election if you're previously unaffiliated.

      654:34 Change of Registration. –
         I. Change of registration of a voter whose party membership has been previously registered.
            (a) Any legal voter whose party membership has been registered may change such registration by appearing in person before the supervisors of the checklist for his town or ward any time they meet, except as prohibited by paragraph IV, and stating to them under oath or affirmation, if required, that:
               (1) He intends to affiliate with and generally supports the candidates of the party with which he offers to register, in which case he shall be registered as a member of such party; or
               (2) He does not wish to be registered as a member of any party, in which case his party designation shall be removed from the checklist.
            (b) He may also change such registration at any primary, upon making oath or affirmation to the same effect, but he shall not be permitted in such case to vote the ballot of any party at such primary.
         II. Change of registration of a voter who has been registered as an undeclared voter.
            (a) Any legal voter who has been registered as an undeclared voter may register as a member of the party of the voter's choice by appearing before the supervisors of the checklist for the voter's town or ward any time they meet, except as prohibited by paragraph IV, and stating to them, under oath or affirmation, if required, that the voter intends to affiliate with and generally supports the candidates of the party with which the voter offers to register, in which case the voter shall be registered as a member of such party.
            (b) A voter may also register as a member of a party at any primary by requesting to be registered as a member and voting the ballot of the party of the voter's choice. A voter may also register once again as an undeclared voter after voting in a primary as a registered member of a party by returning the card provided for in paragraph V to the person at the polls designated by the town or city clerk to accept the card. All such cards shall be in the possession of the supervisors of the checklist at the close of the polls on election day.
         

      by ksh01 on Wed Jan 09, 2008 at 07:17:05 PM PDT

      •  I'me very familiar with election law here.

        I spent 12 hours yesterday registering new voters.

        Registered Republicans had two choices: take a Republican primary ballot or not vote.

        by Elwood Dowd on Wed Jan 09, 2008 at 07:19:08 PM PDT

        •  Yeah, I'm not challenging you on your knowledge

          but it does seem like you can same day register if you're  a new voter or previously unaffiliated.  I'm only saying, it may be plausible that some folks crossed the line for less than forthright purposes.

          by ksh01 on Wed Jan 09, 2008 at 07:24:14 PM PDT

          •  ??

            Crossed what line??

            If you are already on the checklist as a Republican, you are SOL.

            by Elwood Dowd on Wed Jan 09, 2008 at 07:28:11 PM PDT

            •  ok, maybe I'm not being clear or

              you're being cranky.  Whatever.  I'm talking about people who haven't registered as republican but are or first time voters who would be republican but voted dem.  It's a little strained, but I think Zogby's idea that a person who planned to vote for Obama would vote for McCain is more farfetched.

              by ksh01 on Wed Jan 09, 2008 at 07:42:32 PM PDT

              •  Well, if that's what you mean -

                Of course it happened!

                And there were others among the 44% who are registered as Undeclared, who generally vote Republican but voted for Hillary - or Obama - this time.

                And others who are Undeclared who generally vote Democratic, who took a Republican ballot to vote for Ron Paul or against Mitt Romney.

                All expected.

                by Elwood Dowd on Wed Jan 09, 2008 at 07:48:26 PM PDT

                •  It IS an interesting issue and question

                  Should party primaries be open at all?

                  I think the effect is to pull both parties toward "the center." There is a "moderate" bloc of voters up for grabs for McCain v Bradley in 2000, or McCain v Obama or Hillary in 2008.

                  Is that healthy? Worth a diary of its own.

                  by Elwood Dowd on Wed Jan 09, 2008 at 08:05:18 PM PDT

    •  They do have same-day registration, however.

      So the theory could (basically) work for new repub-leaning registrants to register as dems and monkey with the outcome....

      maybe.....

      by nicejoest on Wed Jan 09, 2008 at 08:02:25 PM PDT

      •  That's not the issue.

        We let Undeclared voters - registered but not party-affiliated - vote in either primary. 44% of our electorate is unaffiliated.

        Same-day registration is a small factor in comparison.

        by Elwood Dowd on Wed Jan 09, 2008 at 08:11:13 PM PDT

        •  Okay. So right-leaning folks could register as

          unafilliated if they wanted to affect the other race.

          Theoretically.

          by nicejoest on Wed Jan 09, 2008 at 08:16:55 PM PDT

          •  Yes,

            Just as in other states they could register as Democrats and do the same thing.

            by Elwood Dowd on Wed Jan 09, 2008 at 08:19:58 PM PDT

            •  Yes. For example, Wisconsin has a sort of

              tradition of doing this.

              by nicejoest on Wed Jan 09, 2008 at 08:46:31 PM PDT

•  How about the turnout effect

  It works every time.

  by FishOutofWater on Wed Jan 09, 2008 at 06:42:18 PM PDT

•  I am hoping that finally there is a rising

  against the MSM push candidate!

  I think the voters said they had enough the press choosing the candidate!.

  Several polls are run, adminstered and calculated by colleges, they are not perfect.

  ANd the older voters whether women or men said enough of the youth push.

  by America08 on Wed Jan 09, 2008 at 06:44:39 PM PDT

•  BTW, I was pretty close

  Right here

  Democrats

  Obama 37
  Clinton 35
  Edwards 19
  Richardson 8
  Kucinich 1

  Repubclicans

  McCain 38
  Romney 23
  Huckabee 16
  Paul 14
  9u11iani 8
  Grandpa 1

  - Albert Einstein

  by Walt starr on Wed Jan 09, 2008 at 06:46:07 PM PDT

•  The pollsters are going to

  dig very deep to find out what happen.  They've fixed the exit poll problems, but this is something else.  Primaries are supposed to be much easier to poll than caucuses.

  by bugscuffle on Wed Jan 09, 2008 at 06:47:12 PM PDT

  •  Rasmussen has their initial

    analysis up.  here.

    They cite several factors.

    by LithiumCola on Wed Jan 09, 2008 at 07:03:32 PM PDT

    •  Here's another cool anyalysis from

      Keith Olbermann.

      by nicejoest on Wed Jan 09, 2008 at 08:17:50 PM PDT

•  I read today

  that looking at the demographics of likely voters nationally the nomination will probably come down to black women. If Obama wins them by massive margins, his advantages among men,indies(in soime states) and young/well educated people will be enough to overcome Hillary's massive margins among older white women. Remember going forward as culinary said today it's no more "whitebread" Obama needs the same kind of turnout among black people overwhelming the pollsters predictions that Hillary got among older white women in NH.

  by nevadadem on Wed Jan 09, 2008 at 07:03:29 PM PDT

•  Diebold?

  Why is nobody mentioning this? There are some very fishy discrepancies between the hand counted ballots and the machine counted ballots. I predict a lot more to come on this.

  by aClem on Wed Jan 09, 2008 at 07:03:40 PM PDT

  •  stop with that

    the ballots represented towns where Obama ran better....

    by nevadadem on Wed Jan 09, 2008 at 07:06:25 PM PDT

  •  I don't understand why it hasn't

    gotten more traction; maybe I just don't understand what they're doing.

    by seesdifferent on Wed Jan 09, 2008 at 07:06:48 PM PDT

    •  i think because the

      actual results matched the exit polling.

      by Marlboro Lite on Wed Jan 09, 2008 at 07:23:53 PM PDT

•  BO's arrogant and condescending comment in the

  debate.. words to the effect -- you're ok
  Hillary.

  It was diaried by Kos.

  Coupled with Hillary using her own voice!!

  by ExStr8 on Wed Jan 09, 2008 at 07:04:20 PM PDT

  •  it was the tears

    monday polling would have picked up on the debate stuff....

    by nevadadem on Wed Jan 09, 2008 at 07:07:33 PM PDT

    •  Actually it was Obama's callous remark...

      ...that turned the tide. In one fell swoop he insulted Hillary, her supporters, and women voters who think she is a very likeable person indeed.

      by Psychotronicman on Wed Jan 09, 2008 at 07:26:43 PM PDT

      •  something like "you're likeable enough"? n/t

        by DrReason on Wed Jan 09, 2008 at 09:21:00 PM PDT

    •  Oh look

      It's McCain supporter nevadadem.

      Didn't you promise us you were going to leave the site?

      by taylormattd on Wed Jan 09, 2008 at 07:33:07 PM PDT

•  This says it all:

  From CI's website:

  Note: Ron Paul has been left out of these projections simply because in the past, the massive amounts of posts, re-posts and links from his online supporters have skewed our data far more than any other candidate's, and it has been demonstrated recently that this online support is not transferring much to the polls.

  So your model is flawed, and instead of fixing it, you just throw out the "bad" data.  How do we know if the data is "bad"?  By observing the results and then comparing your prediction to the results.  

  Find a way to exclude the outliers before seeing the results, and you may be on to something.  Otherwise, you're basically peddling a statistical gimmick.

  Also, you need to compare your predictions based on percentages so they are more easily compared to the actual results.  The fact that you don't do this suggests that you're trying to hide innacuracies, and the motive for doing so is that your trying to sell your services.  Advertising these services on a diary here at dKos is really low class.

  by Subterranean on Wed Jan 09, 2008 at 07:36:22 PM PDT

•  My theory:

  (warning: not very deep)

  Independents decided the election. Some of these people were thinking about voting for McCain, but got caught up in the Obamamania and told pollsters they were voting for Obama.

  Then, they heard the media talking nonstop about how Obama was going to win by a landslide---meanwhile, they heard the Republican race was very close... who knows, Romney may actually win.

  Well, these independents know Romney from neighboring MA and they don't like him. So, they figure, "Why waste my vote adding to Obama's huge margins when I can give my old buddy McCain a lift over Romney the scumbag?"

  After all, folks (given the choice) would rather vote in a close election than in a blowout.

  So that's my idea.

  by nicejoest on Wed Jan 09, 2008 at 08:25:35 PM PDT

•  Bradley effect

  I only heard about this a few days ago (from one of you other there in Kosistan). Here's a blurb from Wikipedia:

  there have been instances in which statistically significant numbers of white voters tell pollsters in advance of an election that they are either genuinely undecided, or likely to vote for the non-white candidate, but those voters exhibit a different behavior when actually casting their ballots. (...) This reluctance to give accurate polling answers has sometimes extended to post-election exit polls as well.

  How you'd distinguish this from outright vote theft is beyond me.

  Researchers who have studied the issue theorize that some white voters give inaccurate responses to polling questions because of a fear that they might appear to others to be racially prejudiced. Some research has suggested that the race of the pollster conducting the interview may factor into that concern. At least one prominent researcher has suggested that with regard to pre-election polls, the discrepancy can be traced in part by the polls' failure to account for general conservative political leanings among late-deciding voters.

  So, plenty of speculation.

  by DrReason on Wed Jan 09, 2008 at 09:20:07 PM PDT