To the Polls!

| October 25 2012
Greg Conterio

So what are the guys over at Unskewed Polls saying?

As some of you will recall, a few weeks ago, an outfit on the web called unskewedpolls.com gained some attention for their theory explaining why all the major polls seemed to be at variance with reality.  Their explanation was really very simple, and in fact is something we have all known for many years: many pollsters oversample Democrats, sometimes dramatically.  They actually explained it very well, pointing-out that the actual electorate is composed of a known percentage of registered Democrats, Republicans and independents respectively.  Let’s say for example the makeup of your congressional district is as follows:

Democrats          34%

Republicans        29%

Independents     37%

But what pollsters often do is use assumptions about turnout which are different from the actual makeup of our imaginary district, assuming there will be more Democrats and fewer Republicans than are actually present.  This naturally skews the polls in favor of the Democrats.  Unskewed tries to remove this bias from the polling process, and show what the results would be if they followed the actual breakdown by party affiliation instead of the assumptions by the pollster.

 

Here is what Unskewed has to say about the Presidential race as of today, October 25:

Poll Date Sample MoE Skew ind. voters Obama(D) Romney(R) Spread
UnSkewed Avg. 10/15 – 10/23 +3.5 D +7.9 R 45.8 50.0 Romney +4.2
ABC News/Wash. Post 10/19 – 10/22 1382 LV 3.0 +5 D +8 R 49.0 51.0 Romney +2
CBS News 10/17 – 10/20 790 LV 4.0 +0 D +4 O 47.0 47.0 Romney +0.4
NBC News/WSJ 10/17 – 10/20 816 LV 3.4 +6 D +11 R 47.0 50.0 Romney +3
Politico/GWU/Battleground 10/15 – 10/18 1000 LV 3.1 +2 D +13 R 46.0 51.0 Romney +5
Monmouth/SurveyUSA/Braun 10/18 – 10/21 1402 LV 2.6 +4 D +19 R 43.0 50.0 Romney +3
IBD/TIPP 10/15 – 10/20 938 LV 3.5 +7 D +6 R 44.0 48.0 Romney +4
Rasmussen 10/19 – 10/21 1500 LV 3.0 +3 D +13 R est 44.0 52.0 Romney +8
Gallup 10/17 – 10/23 2700 LV 2.0 +1 D +2 O 46.0 51.0 Romney +5

Of course the best polling-data is that paid for by the respective campaigns themselves.  They don’t pay for data that has been spun.  But Unskewed’s numbers reflect what others are saying about how the campaigns are acting.  Of course we’ll find out in about 12 or 13 days, but for now things still look to be going solidly Romney’s way.

0 comments
Sort: Newest | Oldest