Sunday, October 14, 2012

October 14, 2012 Analysis

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

R+1.00% - Current RCP Average
O+0.43% - Average using the 2008 turnout model
R+4.52% - Average using the 2010 turnout model
R+4.54% - Average using the 2004 turnout model
R+6.30% - Average using the Rasmussen Party ID turnout model
R+2.30% - Average using the D+3 turnout model

All of the tracking polls are beginning their weekend shift toward Obama.  This is a normal event and will start being corrected by Tuesday.

The loss of the Battleground poll from the average eliminates the one piece of good news that Obama had in the average.  All polls now are taken from after the first debate.

For another day, the race is remaining stable with Romney holding a strong lead nationally.  The next debate on Tuesday will be the first opportunity for a change in the dynamics of the race.

10 comments:

  1. Ras Sunday is still +2 Romney... Impressive.

    Are you going to put the D+3 between 2008 and 2010 to keep the rough ordering?

    ReplyDelete
  2. As the days pass, I'm more and more convinced that Bambi's ceiling is baked into the cake. Romney can't play it too safe at the debate, but he has a lot more chance to drive people away than Bambi does to grab any. All he needs to do is be consistent.

    That said, I'm pretty concerned about the Senate. Everything was fine until they blew up with Akin. If the Demons retain control, they will use it to destroy Romney, and there will of course be no chance of dumping BambiCare. Would love to see some continued in-depth analysis as you get more info. Is the state polling any good?

    ReplyDelete
  3. I'll try to take a hard look at the Senate soon. I've heard some things that make me more bullish on Allen.

    Even with my pessimism on individual races, I'm still showing a 76% chance that we take the Senate.

    The Dems are just defending too many seats.

    ReplyDelete
  4. Thanks, Dave. I'll take 76%. I remember how psyched I was in '08 they'd be defending so many in 10 and especially 12. It would be criminal if we couldn't get there.

    ReplyDelete
    Replies
    1. I think we will get there. The DSCC is picking some specific Senators who they want to save, like Bill Nelson, and spending gobs of money in those races. They aren't doing that for other races where they aren't defending one of their own, like in Virginia.

      I'm pretty bullish that we will hold NV, MA, and IN, then pick up Nebraska, Montana, and North Dakota. Then we just need one of the close races (MO, OH, PA, CT, VA, FL) to take an outright majority. Also, if we win 50, there is a good chance the new ME Senator caucuses with us, or someone switches party.

      Frankly, if we are in a wave election, I'm pretty sure we take at least MO, OH, and VA.

      Delete
  5. Here's some numbers from Cuyoga county that I put together.
    This is the # of absentee ballots requested through Oct 13 of each year

    year Dem Rep Oth Total
    2012 119459 47761 62581 229801
    2010 110693 46323 39571 196587
    2008 112961 32136 67500 212597

    What this means to me is R vote went way up in 2010 and is staying up. That's consistent with depressed base in 2008.

    Spreadsheet file:
    http://tiny.cc/ik96lw

    ReplyDelete
    Replies
    1. So a couple very interesting points from this. In 2008, the Republicans in Cuyahoga were 15% of the response rate, in 2012 they are up to 20%. A 5% jump.

      Independents were 31% in 2008, they are now 27%.

      Democrats are holding about the same, down about 1%.

      Nationally in 2008 Dems were up 2% over 2004, GOP down 5% over 2004 and Independents went strong for Obama.

      This is Obama's BEST county in Ohio, and it is showing a return rough return to 2004 numbers.

      Ohio in 2004 was R+5.

      Delete
  6. These results are encouraging, but do have any insight into the details of the electoral college? Specifically, here's what I wonder about:

    1) It's assumed that Romney will get all of McCain's states. Do you happen to know just HOW safe those states are?

    2) It also seems to be assumed that IN is safe. How safe is that?

    3) NE has that oddball allocation of EVs. Any chance that one or two might go the wrong way (like last time)? (I thought I had heard Romney might get 1 of ME's the same way - any truth to that?)

    4) The story seems to be emerging that Romney's road to the WH would be by winning FL, NC, VA, and OH, plus any other. You've talked about FL and OH, IIRC, but how real is the NC situation, and do you have a handle on how these are all doing right now?

    5) AoS was counting CO for Romney. Do you agree? How are the other "swing states" doing? Even if they're all 50/50, Romney's almost certainly in if he's got the 4 in my #4. Some say PA is virtually even - do you think that's really true?

    6) How well do the key 4 swing states track with the variations we see in the national polling? It seems like it's a practical impossibility to win if you're more than a point or two shy of having the popular vote, but you never know.

    7) What do you think your MOE is? Usually 3 or 4% is quoted, but if you're averaging a few that would drop some. Or is even more averaging going on ("tracking")?

    8) How well would you methods have worked to give the results in '08?

    9) Just how biased might ALL these polls be, due to the challenges of getting people to actually answer the surveys?

    10) Oh, and why is there a weekend "bump" for Obama, and how big is it usually (maybe I "tuned in" late?)

    I guess you could say I'm curious. I don't trust the poll results I hear, and yet a number of reasonable pundits seem to use them as gospel.

    ReplyDelete
    Replies
    1. I can't answer the rest, but 9 & 10 I can answer:

      9- They're all terrible. In fact, public polling is the worst way to get a read on public sentiment, except for all the other ways that have been tried.

      10- The theory for the weekend 'D' bump is that people who vote 'R' are more likely to be doing things where they just turn off the phone, or whatever. Basically, while they never get a very good sample to begin with, it's even harder to get a good sample on the weekend. It's not that Democrats are more likely to pick up the phone and answer questions, but that Republicans are less likely to do so.

      As for how big it is? As far as I know no one has done the research to determine that. My gut says its a point or two, but I don't have any concrete data on that.

      Delete
    2. Wow, quite a few questions here...

      1) They are a stone cold lock. The only one the is even questionable is Missouri, and Romney is leading by 5 or 6 there.

      2) Romney is leading in IN by 11. Is that safe enough for you?

      3) We might get ME-2, they redistricted and made that more red, so it could follow NH. If Romney takes NH, which looks pretty solid right now, then ME-2 is probably 50/50. If we have a wave that drops swing states to Romney left and right, then ME-2 will come along for the ride.

      4) NC is gone for Obama. Everyone keeps asking why RCP keeps it in their tossup list. Obama is pulling funds out of NC and Suffolk U won't even poll there any longer.

      5) I think CO is probably like NH now. A solid, but small Romney lead. As long as nothing changes, and the GOTV efforts are successful, CO will go for Romney. PA on the other hand I don't think is tied. I think Philly will have a larger influence in PA than Cleveland will have in OH. I think Obama is still slightly ahead there. Romney should still focus on Ohio, since he leads and you don't want to take the eye off the ball.

      6) The swing states track the national polling exactly. When Obama was ahead by 3, Florida and Virginia we tied, Ohio had Obama ahead 3, and NC was a slight Romney lead. Once the national polls swung to Romney +1, Florida, Virginia, and North Carolina fell off the table for Obama, and Ohio moved into a tie. Personally, I don't believe the Ohio tie. I think Ohio is "tied assuming a D+3 turnout, but we are more likely to get a 2010 turnout".

      7) I am averaging polls, to theoretically I should be having a lower MOE than the individuals polls. Unlike RCP, I'm comparing apples to apples. I would guess I'm probably carrying an MOE around 1 to 1.5. Keep in mind, a lot of the poll's MOE comes from getting the demographics of the actual turnout wrong. Since I'm offering up 5 different turnout models, I have a lot higher probability of being right. And you can be sure, I will be touting my accuracy against the actual turnout, if I am right.

      8) I haven't run the numbers, but I expect they would have done pretty well. The Rasmussen Party ID poll was showing that huge Dem turnout, so that turnout model in particular would have been very accurate. That might be an interesting thing to do, if I get some time. Run the 2008 final polls through my model. Obviously, my model would have been very wrong with the 2004 and 2006 turnout models.

      9) Allen answered this, but that is the real question this cycle. How much is the refusal rate skewing the polls? I know some of these pollsters are reporting their actual results, but when they are getting a 9% response rate you have to wonder if that is a primary contributor to the over sample.

      10) Allen also nailed this one. Historically about a 2 to 3 point bump in Democrat numbers on Saturday and Sunday. Rassmussen won't even release daily track numbers on the final weekend.

      Delete