Tag Archives: new poll

Yours is Askew

So some brave men and women (and bored?) have taken all the polls of the Obama-Romney race and “re-skewed” the “skewed” polls that oversample Democratic turnout and assume a 2008-or-higher level of Democratics. Let’s look at some of the outlets, what they say, and what “un-skewed” has to say.

POLLING FIRM               SKEWED                 UNSKEWED

Reuters-Ipsos                   O48-43                     R54-44

NBC/WSJ                         O50-45                      R51-44

NYT/CBS                          O49-46                      R51-44

Fox News                          O48-43                     R48-45

ABC/WashPo                   O49-48                     R52-45

CNN/ORC                        O52-46                      R53-45

Here is an example of how they “un-skew” a poll.


(1) take the sample of the poll (33D-24R-36I), re-calibrate it to match something more likely to be the turnout (this is based on Rasmussen Party ID numbers) – (33.06R-31.75D-28.19I)

(2) Now take the survey breakdown found in the poll – 90% D to Obama, 90% R to Romney, and 60% I to Romney. There’s your new poll numbers.

Whether you use Rasmussen Party ID or Gallup Party ID or if we assume the election is going to be roughly D+2 (which is what I think)….I see something more like 50.6Romney-49.4Obama. (90% of each party goes to their nominee, Romney has a 10-point lead on indies at 55%).

%d bloggers like this: