Monday, October 8, 2012

Pew Poll - October 8, 2012

Likely Voter, 1081 sample size, Romney leads 49-45, 6% undecided.

This poll will make Democrat heads explode (and already is).  Suddenly people who thought partisan ID samples were no big deal are freaking out about a poll that has an R+3 split.  Chuck Todd is having a hissy fit on Twitter.

The really weird thing about this poll is that the likely voter screen almost exactly matches the 3 month average for Rasmussen's Partisan ID poll.

Rasmussen D/R/I - 33.8/36.4/29.8
Pew D/R/I -            33.4/36/3/30.3

This tells me that this poll is a confirming event for Rasmussen's Party ID poll, and represents a good view of what the turnout would be if the election were held today.  Ironically, because of this sample, under all of the turnout models the results are worse for Romney than the top line number.

I can live with that.

R+4 - Current result
O+1.2 - 2008 turnout
R+3.8 - 2010 turnout
R+3.7 - 2004 turnout
R+5.7 - Rasmussen Party ID

The fun thing is that this poll must be included in RCP, since they just average without weighting.  They can't change their model at this time.  So the RCP just tanked for Obama.


  1. And considering undecided voters break heavily for the challenger on Election Day, if this poll is accurate, it means Romney would win by a landslide.

    My own sense is that the Partisan make up of the Pew poll is overly optimistic since Republicans have never had that much of an advantage in any post-War election (although that doesn't mean it can't happen) but it's definitely a LOT closer to reality than the nonsense D+8 polls we've been getting nonstop from the MSM on a regular basis.

    Preference Cascade, here we come!

  2. Rasmussen's party I'd confirms it. Pew is pushing out "accurate" polling now that we are getting closer to the election. I fully expect to see an even turnout up to a R+3 turnout.

  3. While I too am conservative by nature in all my statistical work, there is accumulating evidence to support the Rasmussen case for a R+N partisan split. Not only is Rasmussen's data damn near identical to Pew's sampling, but it fits with the registration data that's trickling out which to this point has been very favorable. There is also historical data matching to the previous midterm, ala. Michael Barone, which I consider weaker.

    Unknown, I too am reluctant to jump on the Rasmussen partisan wagon, but take a look through their results and compare them to CNN's exit poll data -- he's been right on.

    1. Yeah, look at that September 2008 number. Almost dead on for the final turnout in 2008.