Thursday, October 18, 2012

Senate Update - October 18, 2012

Some recent polling is out there that is causing me to shift my percentages on individual Senate races:
  • Indiana 90% - Rasmussen has recently polled this race and Mourdock is up 5 and almost at 50%.
  • Nevada 95% - Heller has this pretty much sewn up.  Two recent polls, including Rasmussen have Heller up at least 6.  Rasmussen has Heller at the magic 50% mark.
  • Wisconsin 60% - The Marist poll that give Baldwin +4 is Dem +5.  Marquette has Thompson up by 1.
  • Montana 50% - This is starting to look tough.  Rasmussen has it at a 48-48 tie.
  • Massachusetts 45% - Polling is starting to look bad for Brown, with Rasmussen showing Warren at 49 and leading by 2.
  • Virginia 50% - Allen is polling much better now WeAsk has him at +5, while Rasmussen has Kaine up 1.
  • Florida 50% - Mack is starting to look good.  Rasmussen has Nelson up 1 and even though PPP has Nelson up 11, Nelson still is at 45.  Nelson isn't even close to 50% in any poll.
  • Ohio 65% - Rasmussen has Brown at +1, and the Survey USA poll is heavily over sampled with Democrats.  Brown is short of 50% as an incumbent.
  • Pennsylvania 35% - Casey's lead has dropped in all polls, and Smith is in striking distance.  Casey is not at 50%.
  • Connecticut 35% - Murphy is still not hitting 50%, there are a lot of undecideds.
Odds of at least a tie in Senate = 90.4%
Odds of winning the Senate = 78.8%


  1. This is more good stuff. I don't think many people realize the work that goes into these posts. We just see a few paragraphs and some numbers and it's like "oh wow, cool, Dave's posted again".

    So, excuse me for another dopey question, but do you think a "wave" election would make a difference here? Purely hypothetically, because I'm counting absolutely no chickens... at a certain point (say, Romney by 8 points nationally), would you have to adjust these figures with some kind of moderate built-in boost?

    1. Absolutely. The odds I give are assuming each election is a separate probability event. But a wave election will either help all of our candidates or cause a lot of them to lose close races (if Dems were to turnout in unexpected numbers).

      All I'm using is straight probability theory.

  2. A nice, handy summary. I had just been wondering about this stuff right before I clicked on over.

    To put it simply, it looks like we keep NV and IN, lose ME, and add NE and ND. That gives us the status quo of 47.

    But then there are also roughly 50/50 shots at 6 states: WI, MT, MA, VA, FL, OH. Expect to pick up 1/2 of them; that gets you to 50.

    Seems like the max odds should be at 50, though, with somewhat better chances for greater than 50 (given the outside chance at a couple of other seats), than for lower than 50. 79% and 90% seem high to me, but I'd guess at LEAST 40% and 60%.


    1. It's the magic of the binomial equation. It "seems" high, but that is actually where the math takes you. The brain wants to add a 50% chance to a 50% chance and say it's a 50% chance. It isn't, it is a 75% chance.

      Quite a bit more complicated here, since I'm using the odds of winning 7 out of 16, but it's the same idea.

    2. Your "7 out of 16" might explain it. You're only listing 10 here, so perhaps you have 6 more probabilities you're figuring in that have been shown previously. I was assuming you were starting with 45, and adding the 10 races given from there.

      Doing the full math using 45 and your 10 probabilties, I get this pdf:
      47: 0.9%
      48: 4.6%
      49: 13.5%
      50: 23.9%
      51: 26.8%
      52: 19.3%
      53: 8.6%
      54: 2.1%
      55: 0.2%

      51+: 57.0%; 50+: 80.9% (So yes, better than 40%, 60%, but less than 79%, 90%.)

      Still not bad odds.


    3. Yes, you need to go look at the previous Senate post for the full list. I'm only talking about changes from last week. I don't mention NE and MO above, because they didn't change.