Making Sense of the Georgia Polls

As Nate Silver noted in remarkable detail, the polls showed a Democratic bias this time around -- and also in Georgia. While the reasons are still being debated (I wrote about it here), let's look at some of the Georgia polls and see how well they predicted the U.S. Senate race between Michelle Nunn and David Perdue.

The Georgia "winner," just glancing at some of the latest polls, seems to be WSB-Landmark. It had Perdue at 47 percent on Oct. 25, at 48 percent on Oct. 30, and then at 50 percent on Nov. 3. It was the only poll near the election that put Perdue so high (his actual vote stands, at the moment, at 52.9 percent). That poll had Nunn at 46 percent (she stands now at 45.2) and had Swafford, the Libertarian, at 2 percent compared to the actual vote of 1.9 percent. Not too shabby.

Below I list the major polls conducted from Nov. 1 until the election and how much they predicted Perdue would receive and the "bias" compared to Perdue's 52.9 percent of actual votes (so far).

	Perdue Poll	Perdue Bias
WSB-Landmark	50	-2.9
Insider/Advantage	48	-4.9
NBC/Marist	48	-4.9
SurveyUSA	47	-5.9
PPP	46	-6.9
YouGov	44	-8.9

I'll do a more in-depth analysis of this once time allows, such as including a column on a poll's methodology (robo-poll, etc.).