Thursday, October 18, 2012

October 18, 2012 Ohio Analysis

Since Survey USA and PPP have recently released Ohio polls, I thought it would be good to take a look at the state again.  There are a lot of nervous people out there worrying that Romney can't win Ohio and agitating for him to go after Pennsylvania or Michigan.

Don't panic, Romney is fine in Ohio.

The panic is being induced by some silly weighting samples in the Ohio polls.  They don't match anything approaching reality.  Survey USA is using a D+7, ARG has a D+9, Gravis a D+6, and NBJ/WSJ a D+11.

Since it came out last night, let's take a look at the Survey USA poll.  It uses a D/R/I sample of 39/32/26, which is almost identical to the 2008 turnout in Ohio.  With this sample they are noting Independents favor Romney by 8 points, and Obama leads by 3 points 45-42.

However, if we apply the various turnout models to these results we get the following:

R+9.04% - Using the 2004 turnout model
O+0.56% - Using the 2008 turnout model
R+5.78% - Using the 2010 turnout model
R+5.06% - Using the 2012 registration model
R+2.85% - Using a D+3 turnout model

So in this "Obama +3" poll, Obama really only has a slight lead in a 2008 turnout scenario.  Romney leads comfortably in all others, including the magical "D+3" scenario that is so popular this cycle.

Here are the results from all of the current polls in the RCP average (including PPP):

Polls included:
O+3 - Survey USA
R+1 - ARG
O+1 - Rasmussen
R+1 - Gravis
O+5 - PPP

O+2.43% - Current RCP Average
R+7.35% - Average using the 2004 turnout model
O+2.25% - Average using the 2008 turnout model
R+4.09% - Average using the 2010 turnout model
R+3.37% - Average using the 2012 registration model
R+1.16% - Average using the D+3 turnout model

Even if Obama is able to get Democrats to turnout at a rate 3% higher than Republicans, he is still losing by over 1 point.  Romney is clearly leading in Ohio right now.


  1. I hope you are right. It still worries me that the most heavily D weighted Ohio polls you mention show Romney winning, while the more realistic polls show Obama ahead. There is so much static in these competing polls - someone is clearly wrong.

  2. I should have said that two of the heavily D weighted polls are the two showing Romney ahead. Gravis and Arg don't seem to match the rest.

  3. Love your thoroughness with all the different models! I may make it through election day with your help. Thanks!

  4. I think it's all going to come down to Ohio, and I hate having just one state determine the election.

    I think it's almost certain at this point Romney will win the popular vote, but we could have a situation like we did in 2000.

    What worries me about Ohio is all the shennanigans from early voting, absentee ballots, etc.

    I think Romney will pull it through, and have heard some good anecdotes from Ohio residents that the entire state seems locked and loaded for Romney.

    1. I consider Ohio providing the margin of victory to be a worst case scenario. It is much more likely that we will win OH, WI, IA, NV, and OR.

      Ohio is the "one state" only because it is the fulcrum. If Obama holds it, then GOP turnout isn't high enough to win. If Romney wins it, he likely wins everywhere.

  5. These "turnout models" in the first part of your post are national turnout models, not state-level turnout models, right? What was Ohio's turnout in 2004 and 2010?

    1. No, for Ohio I use the turnout model for Ohio in the specific year. I don't use the national models.

      2008 D+8
      2010 R+1
      2004 R+5

      There is a previous Ohio post where I give the exact percentages for Ohio.

    2. Wow, I didn't realize. This is amazing analysis at a time when we all need amazing analysis. Thanks, Dave.