Registered Voter Polls More Accurate than Likely Voter Polls

Markos wrote up an encouraging post about poll accuracy and determined that actual vote results in the past couple of elections more closely matched the registered voter polls.

Among registered voters, most of those polls gives President Barack Obama decent advantages. Thus, the operating theory is that a significant portion of Democrats will sit out the race because of a lack of intensity.

So for example, the latest NBC/WSJ poll yesterday had the presidential race deadlocked 47-47 among likely voters, but gave Obama a comfortable 49-44 lead among registered voters.

Different pollsters have different ways to filter out those they deem unlikely to vote. But bottom line, the registered voter screens have historically been more accurate.

Meanwhile, turnout in North Carolina is way up among African Americans and women:

In North Carolina early voting, 18-29 turnout is up 39.7 percent compared to four years ago. At this point in 2008, 144,331 African Americans had voted early in the state, or 29.03 percent of all voters. This year, 200,766 African Americans have voted, or 30.99 percent of all voters (source). Finally, while 277,714 women had voted at this point in 2008, or 55.86 percent of the total, that number is now 359,576 or … 55.5 percent of voters.

There simply isn’t a dropoff in core Democratic support in North Carolina.

All of that said, this thing is crazy tight right now. Frankly, the polling is excruciating to follow. So the most important takeaway is the Obama team needs to get out the vote in a big way if we want to see the Registered Voters theory work out.

Print Friendly
This entry was posted in Election 2012 and tagged , . Bookmark the permalink.
  • http://www.politicalruminations.com/ nicole

    This is why I try desperately not to follow the polls too carefully.

    Good piece by Markos, and Nate Silver did one on Gallup’s polling (outlier, outlier, outlier!) a few days ago.

    What’s really scary is this.