Colorado
FiveThirtyEight Projections Dem Rep Margin
Polling average 47.8 46.5 Obama +1.3
Adjusted polling average 48.0 46.9 Obama +1.1
State fundamentals 48.1 46.7 Obama +1.4
Now-cast 48.0 46.8 Obama +1.2
Projected vote share ±3.5 50.1 48.9 Obama +1.2
Chance of winning 63% 37%
Voter registration drives in Colorado, Florida, Iowa, Nevada, New Hampshire, North Carolina, and Virginia all favor Democrats. While Andrew Sullivan and his ilk were having a meltdown over the first debate, the Obama campaign was quietly registering hundreds of thousands of voters. Granted, not everyone they registered are supporters of the President. But the campaign has identified and made likely to vote over a hundred thousand new registrants--because if somebody takes 10 minutes to fill out a voter registration form, they're going to vote...whether or not Gallup or Marist deems them "likely voters."
Campaigns pay for better databases than the cheap press. They pay for cell phone surveys. Polls--like the utterly pathetic house poll here called "PPP"--that do not poll cell phone voters drastically understate the Latino vote, a vote which is breaking 3 to 1 for Barack Obama because Latinos have the highest cell phone usage rate of any ethnic group. Simply put, no matter how many call backs you do, if the person only has a cell phone, and you're only calling landlines, you'll never count that voter, and people like her who only use cell phones. This group is disproportionately young, minority, and highly mobile. They're also generally pretty educated and likely to vote. Sound like the core of Barack Obama's coalition? It is. And any poll that doesn't count these individuals is not worth the paper it's written on.
Simply put, likely voter screens and outdated polling methods make this election one where the polls are inaccurate. We saw the confluence of bad likely voter models and a failure to call cell phones combine to show Sharon Angle ahead by five in 2010, while Harry Reid actually won by five. Harry Reid's pollster--and the campaigns pay for great internal polling--had it spot on because they invested in proper, and 21st rather than 20th century polling methods.
And so it is now, the polls show a "tied race among likely voters," with registered voters breaking decidedly for the President. The same thing that caused the polls to be wrong in 2010 will cause them to be wrong in 2012--they're using outdated methods that don't track to changes in social behavior, phone usage, and voting patterns.
http://www.dailykos.com/story/2012/10/21/1147834/-Likely-Voter-Models-Understate-Democratic-Vote
WeAskAmerica 10/30 - 10/30 1246 LV 2.9 50 47 Obama +3
ARG 10/25 - 10/28 600 LV 4.0 47 48 Romney +1
Purple Strategies 10/23 - 10/25 600 LV 4.0 47 46 Obama +1
PPP (D) 10/23 - 10/25 904 LV 3.3 51 47 Obama +4
NBC/WSJ/Marist 10/23 - 10/24 1128 LV 2.9 48 48 Tie
Rasmussen Reports 10/21 - 10/21 500 LV 4.5 46 50 Romney +4
PPP (D) 10/16 - 10/18 1000 LV 3.1 50 47 Obama +3
WeAskAmerica 10/15 - 10/15 1206 LV 2.9 47 48 Romney +1
Denver Post/SurveyUSA 10/9 - 10/10 614 LV 4.0 47 48 Romney +1
Gravis Marketing 10/5 - 10/11 2089 LV 2.2 48 46 Obama +2
Rasmussen Reports 10/7 - 10/7 500 LV 4.5 49 48 Obama +1
ARG 10/5 - 10/8 600 LV 4.0 46 50 Romney +4
CBS/NYT/Quinnipiac 10/4 - 10/9 1254 LV 3.0 47 48 Romney +1
University of Denver 10/4 - 10/5 604 LV 4.0 47 43 Obama +4
Gravis Marketing 10/3 - 10/4 1438 LV 2.8 46 49 Romney +3
McLaughlin/ACU (R) 9/30 - 10/2 300 LV 5.7 46 50 Romney +4
WeAskAmerica 9/25 - 9/27 1273 LV 2.8 49 46 Obama +3
Gravis Marketing 9/21 - 9/22 765 LV 3.4 50 46 Obama +4
PPP (D) 9/20 - 9/23 940 LV 3.2 51 45 Obama +6
Purple Strategies 9/15 - 9/19 600 LV 4.0 48 45 Obama +3
= new reply since forum marked as read
