We just reported two polls that show the NC Marriage Amendment up by 14 and 20 points respectively. Why the disparity?

The most recent PPP poll overestimates 18-29 year olds by about 4 points, and, more importantly, underestimates voters over the age of 66 by about 10 point:

This alone is worth 3-4 points for the marriage amendment, not counting the "7 point effect."

What is the "7 point effect"?

Patrick Egan, a political scientist at NYU, issued a paper (PDF here) in 2010 that analyzed an abundance of polling data from 1998 until 2009 on a number of marriage-related ballot measures. Among other things, he found that the measures under-poll by an average of seven points.He did not find that either "social desirability bias" (fear of saying how one will actually vote) or voter confusion explained this gap, and left the explanation as an open question.

Regardless of the reason, there is consistent evidence that these marriage protection measures under-poll, and that the projected level of support is a "lowball" figure.

Now, this is obviously no reason to let up, but it is worth bearing in mind as we enter the final week of the North Carolina campaign.