Thursday, October 11, 2012

October 11, 2012 Analysis

Polls included:
O+1 - Rasmussen Daily Track
R+1 - Fox News
R+5 - IBD/Tipp
O+1 - Battleground
R+4 - Pew
Even - Wash Times/Zogby

R+1.33% - Current RCP Average
R+0.55% - Average using the 2008 turnout model
R+5.26% - Average using the 2010 turnout model
R+5.27% - Average using the 2004 turnout model
R+6.96% - Average using the Rasmussen Party ID turnout model

With the CNN poll now out of the average, Romney leads in all turnout models, including the 2008 model.  A bit surprising that Rasmussen flipped to Obama +1 today, clearly he had a good polling day yesterday, which is odd since nothing good for him occurred yesterday.  Rasmussen has been gradually moving his sample to be more Democrat, and is now at D+5.  I'm starting to wonder about his new weighting methodology in his daily track, since he is now 7 points out of sync with his monthly partisan ID poll.


  1. Rasmussen is making me nervous, let's be honest, he probably have the best track record of all the polls you listed, and they have Obama up by one point.

    My own sense is that Romney is winning and if the election were tomorrow, he'd win solidly. But then again, I'm the equivalent of a Fantasy Football player, and Rasmussen is an actual professional that has an enviable track record.

    Is Rasmussen off his rocker with a Democrat +5 model? Is he trying to gain credibility because liberals have essentially written him off as a Republican pollster?

    I just don't see how Romney doesn't win this election when he's winning independents by well over double digits, that looks like the recipe for a landslide to me.

    1. I wouldn't be. For whatever reason, he is coming up with a D+5 sample in his poll. The electorate won't be D+5.

      Actually, I know why he is coming up with D+5. I think he has a problem in his methodology. He will need to fix it after the election.

    2. Thanks Dave.

      I saw your post at Ace of Spades regarding this. I have to agree with your conclusion:

      One of his figures has to be wrong, either his R +2.6 for party id tracking or his D+5 for the Presidential election. They completely contradict each other.

      I'm in the camp that thinks the final makeup will be R0 to R+1, but that's a more optimistic scenario.

      Still, it hurts that Rasmussen was what I regarded as one of the few "honest" pollsters in a sea of MSM push polls that were clearly being used to try and influence the election.

      Now, the MSM polls with their fantasy Democrat turnout models are showing Romney winning? What alternate universe is this?

      Regardless, even if Rasmussen is "right" and the Democrats are able to get close to where they were in 2008 in terms of turnout, it still looks like a narrow Romney victory.

    3. You know why and you're not sharing?!? :)

      I ask because I'm confused as hell as to what his sampling algorithms are. Do they even use the partisan ID as an input?

    4. Uriah, good question. DaveinFla, I have heard Ras is using +3 now you say its +5. How do you know that? Is that speculation? Does he publish that?

    5. He does use partisan ID weighting, but it from inside the same poll. What he does is he uses the partisan ID split of the respondents from the previous 7 days, then weights his 3 day average using that split.

      He is getting a +5 split because over the last 7 days 5% more Democrats have responded to his poll than Republicans.

      My personal theory (unproven) is that Democrats are more motivated this year to respond to polls, while Republicans are more motivated to hang up. Poll response rate does not correlate to turnout. I think his methodology is severely flawed in this one poll.

    6. Thanks for the insight.

      That's a logical way to sample for a tracking poll as it captures any movement in the race. The dynamics are captured over the 3-day window and the noise is filtered/stabilized by using a larger 7-day partisan avg.

      If true, the sum over all the daily states should be his monthly partisan index?

  2. 39 Philip Klein ‏@philipaklein
    RT @ReutersPolitics: Mitt Romney pulls ahead of President Barack Obama 47-44 percent in Reuters/IPSOS tracking poll

  3. It could be a business decision for Rasmussen to skew Dems. He can gain some cred and create some buzz on the left for a change. Now dems are citing Razz? lol.

    The Ras poll is basically unchanged for months, maybe he likes it there. I suspect he will start moving it, and end up with a D+1 going into November.

    1. I honestly believe that could be some of his motivation, just to get some street cred with the MSM.

      I've worked on campaigns where polls have been commissioned, it's a business like any other, we wanted a poll to refute another poll that was clearly engineered to push public opinion, and they basically gave it to us. It's not like polling firms take some sort of sacred oath to be in the business they are, many of them are basically just PR firms that serve their clients' wishes first.

      I firmly believe Romney is winning right now, and by far more than most MSM polls are projecting. But the election is not tomorrow, and I will admit Rasmussen keeps me from absolute certitude that it's Romney's race to lose.

  4. I wouldn't get too hung up on day to day movements in tracking polls, which seem to defy conventional wisdom most of the time. A good day for Romney drops out, but Ras moves to Romney....a good day for Obama drops out, it moves back to Obama!

    More likely, Romney was 47.5 and Obama 47.4 or something and it's moved 0.1 either way to go from 48-47 to 47-48.

    The general trend of moving a couple of points towards RR and it being tied up or RR+2-3 is confirmed.