View Single Post
Old
11-06-2012, 11:38 AM
  #78
Saugus
Ecrasez l'infame!
 
Saugus's Avatar
 
Join Date: Jun 2009
Location: Connecticut
Country: United States
Posts: 97,441
vCash: 500
Send a message via AIM to Saugus
Quote:
Originally Posted by BenedictGomez View Post
The 99% thing is way overdone. This stems from Nate Silver's 92% analysis, he's become something of a liberal hero this cycle, but I consider him much a clown. I view him like that guy that claims he speaks to dead people.

On the turnout, that is everything.

Some of the biggest polls are sampling Democrats in a proportion ≥ that of the historic 2008 Dem wave election. Tough to believe.
Incomplete analysis of this. He also ran the numbers if the 2004 turnout numbers were used, and in that case Obama's advantage in many swing states diminishes, but he is still ahead using that assumption.

Pollsters like PPP do oversample Democratic voters, and his model acknowledges that house bias and weights it accordingly. Similarly, Rasmussen's methods tend to heavily oversample Republicans (he doesn't call cell phones and doesn't call more than once if he doesn't get a response, also his definition of a likely voter is more stringent than average), and his polls have been consistently 2 or 3 points to the right of the consensus polling average. The model is said to correct for that too.

We'll see. With several state polls (even the Republican leaning ones) showing consistent movement toward Obama over the last week, Romney's only chance is if the pollsters are all oversampling Democrats to an absurd degree.

Saugus is offline