The issue is of sample size. The pollster should have ask whether you “consider yourself” a Democrat, Republican or Independent. Not “are you registered”. When achieving polling targets, self identification is far more important than part registration. Simply put, the pollster cannot verify this.

In looking at past elections the FINAL ADJUSTED exit polls in 2004 and 2010 told us that that their were as many self identified Republicans as Democrats. The ONLY exception is 2008 where this gap became an 11 point Democrat advantage. Also noteworthy is that the Democrat advantage in registration in 2012 on a percentage basis is not only lower than 2008 but 2004 when Bush won easy. One of the quickest way to see a trend in a battleground state (I don’t concede NC is one) is the trend in registration.

Now in fairness this pollster does ask what party are you registered, ….and the pollster correctly targets a +12% Democrat advantage. But this is not how to poll the electorate in NC; there are a number of voters, specifically older voters who may “think” they are Republicans who never bothered to changed their registration. Let me put it to you another way, Democrats have traditionally enjoyed a significant party registration advantage, yet over the past 30 years they have not faired nearly as well at the ballot box!

This is why the more experienced (and unbiased) pollsters, do not ask for “party registration” but rather “Party preference”.

If I were running a poll I would target a +2 Democrat self-identified advantage (adjusting such based on historical trends in party registration). For example if in October the party registration trends diminish, I would adjust the sample to +1 Republican.

Because of raw Registration NC is a tricky state, but make no mistake, NC is not in play. NC voters can take some relief that the advertising in October will diminish significantly as Obama will likely pull his money out of NC in early October. Obama needs to worry more about holding PA than what is a luxury in holding NC.