Midterm forecasts: big losses for the Dems

Paul Gronke

I just got back from Washington DC and the annual meeting of the American Political Science Association. Because we were in DC, and because it is a midterm year, the biennial panel on election forecasting got more coverage than usual.

The predictions are pretty grim. John Sides summarizes the results pretty well over at the MonkeyCage. John explains how the forecasts were obtained.

This graphic provides a quick overview of the forecasts:

  • (Show?)

    Since the academicians have spoken, I guess we don't need to bother voting.

  • (Show?)

    And at this point in the 2008 election, John McCain had a lead over Barack Obama.

    Robert Creamer over at The Huffington Post has a good counter-argument to the prevailing "wisdom" on the mid-term elections.

  • (Show?)

    Bill, there are a lot of reasons why these models may be wrong. I'm happy to talk about them.

    But it's impossible to deny that the "fundamentals," which is what these models reflect, look very bad for the Dems right now. Unemployment is high and the President's approval rating is low. Add in the typical midterm seat loss by first term presidents, and this is not shaping up to be a good year.

    This is why the DNC just announced it is pulling out of some races and focusing its resources on those races (Baird and Wu to name two of them) that are in danger but can be salvaged.

  • (Show?)

    Mitchell, I don't know where you get the claim about McCain/Obama, but here are the trial heats:


    For one week, polls showed a McCain lead, only during his post convention bounce. For the weeks before and all the race afterwards, he was losing.

    That week says nothing about the validity of the forecasts above.

    • (Show?)

      Ah, I see! Mitchell you made the same fallacy as Creamer did, focusing on the ONLY week where a McCain lead was shown, and using that as to claim that all pre-election polls and pundits are wrong.

      Bad analysis (as is the rest of the Creamer piece). Yes "every race is between two candidates" but there are 435 races. And here's the thing, if these models didn't predict so well...then they'd not predict so well!

      You can't just wave your hands like Creamer does and cite 10 cases of contests that he thinks 'really' can be won. You have to claim that the whole system has shifted in some fundamental way so that voters don't care anymore about 10% unemployment and a presidential approval rating of 43% is not low.

      I don't know when that seismic change in American politics occurred. I sure didn't feel it.

      • (Show?)

        Except it wasn't one poll, so it is hardly a "fallacy". From your own polling reports link for the same basic time frame from the election where we are curently, there were 9 polls showing a McCain lead:

        Franklin & Marshall 9/15-21
        American Research Group 9/13-15
        GWU Battleground 8/7-11
        Democracy Corps 9/8-10 AP-GfK 9/5-10 FOX/Opinion Dynamics 9/8-9
        Ipsos/McClatchy 9/5-9
        ABC/Washington Post 9/5-7
        CNN/Opinion Research 9/5-7
        USA Today/Gallup 9/5-7

        I don't disagree that we have factors where the wind is in our face and not at our back, but as others have noted, polling this far out has little predictive value and almost all the doom and gloom are based on generic ballot polling and not on individual race polling data.

        We certainly have our work cut out for us, but it is far from over.

  • (Show?)


    I appreciate the reporting that this mid-term election may result in some Republican pick-ups.

    I'm not sure what you're implying in that comment about our local races, though.

    Cook still has the OR 1st CD in the "Likely D" column. Are you reporting that the DNC is shifting resources to shore it up?

    And by highlighting that congressional race in Oregon instead of Schrader's, are you suggesting that the 1st is more of a battle than the 5th? Cook has the 5th only as "Leaning D", a tier below the likelihood of re-election predicted in the 1st.

    WA 3rd (Baird isn't seeking re-election) has been a toss up from the get-go. Is it now trending R?

    I chased the link to the Monkey Cage, but they only discussed models based on reverse engineered correlations, not specific discussions about why any particular race was in danger but can be salvaged.

    Any other info you can pass along to clarify?


  • (Show?)

    Paul, any sense how predictive these kinds of models are at the two-months-out period? We've seen a fair amount of movement in the last two months, and I'm wondering what effect time has on a model. Of course, it may get worse. Two months ago, these models would have predicted fewer GOP pick-ups.

  • (Show?)

    Why are those numbers displayed as dots on an X-Y chart? Unless studied closely, the chart gives the impression that there's something on the decline -- when the Y-axis merely represents independent categories of data.

    The chart could just as easily be flipped vertically, which might imply that something's on the increase.

    A bar chart would be more accurate here. Or better yet, a table of numbers, with a median or mean calculated.

  • (Show?)

    Folks, while there is breath, there is an opportunity to pull it out. Focus on what you personally can do to produce a positive outcome. The GOP is the wrong party to have in power. I do see that we are splintering into more than two parties and that is fine as far as I am concerned. One thing we all agree on, however, is that the GOP is the wrong party! The more you study them, the more nauseated you get. Keep your eyes on them and remind people of who they are and what they stand for.

  • (Show?)

    Well apparently there is great disagreement that the Democrats are the correct party to remain in power. I would not be surprised to see a tidal wave shift at the state and many local levels. It remains to be seen if the voting block of 6-7 counties in the northern part of the state, along with AFSCME and SEIU can pull out a win for the former governor

  • (Show?)

    The Multnomah County Democratic Party has endorsed Oregon Measure 74 and other county Democratic parties will soon be following suit. Medical cannabis on the ballot helps Democrats at the ballot box. Demonstrating compassion for patients while also creating jobs for Oregonians will aid Oregon Democrats this year and every year.

connect with blueoregon