Monday, October 1, 2012

October 1, 2012 Analysis

Polls included:
O+3 - Rasmussen Daily Track
O+6 - Bloomberg
O+2 - ABC/WaPo
O+2 - Battleground
O+5 - Fox News

O+2.25% - Current RCP Average
O+3.14% - Average using the 2008 turnout model
O+0.17% - Average using the 2010 turnout model
O+0.13% - Average using the 2004 turnout model
R+0.68% - Average using the Rasmussen Party ID turnout model

With the AP/GfK poll dropping out of the RCP average, all of the polls now included have a defensible D/R/I sample and are pretty close to each other.  This is likely the best view of the current race we have seen in a long time (if ever).

As can be seen above, the race is tied under all turnout models, except the 2008 model.  If the election were held today, it would come down to which party turns out their base in a higher proportion.  Also note that when you get into the <0.5% range you reach the point where the popular winner may not win the election.  This could be bad for Obama, with him running up large leads in states like California and New York, only to see Ohio (and the presidency) slip away.


  1. What are you using for Ras P/A #s? August, or some blended months?

  2. I discuss it in the methodology post. I average the last three Rasmussen Party ID polls for a total of 45,000 samples. So no, I'm not using just the August result. He will be releasing September shortly, so the average will change. I will be sure to make a post about the update.

  3. The guy @ polls2012 blogspot is predicting a September # of .55. How would that change your results?

  4. I'm sorry, I don't understand what .55 represents. The August numbers were D/R/I of 33.3/37.6/29.2.

  5. Defensible isn't good enough. Why are there no polls with a GOP skew? Have you EVER seen an R+6 poll, let alone R+16?

    My advice is: ignore the topline, ignore the DRI split. Ask just one question: who is winning independents?

  6. R + .55 won't change things much. I'll be rolling June off the average, which was R + 1.4.

    Given that September included the post Democrat Convention bounce, I would be unsurprised by that result.

    Honestly, the real Rasmussen Party ID is probably about R + 1.5, halfway between the September and August results.