Problems with Elon polling on NC marriage referendum

Previously, I have cited a major problem with polling produced by Elon University.  They poll only RESIDENTS — i.e., whoever answers the phone — instead of the more reliable REGISTERED VOTERS or LIKELY VOTERS.   So, Elon poll respondents could be your goofy 16 year-old son or Aunt Bertha from Birmingham suffering from Alzheimer’s.

Well, Public Opinion Strategies, a respected DC-based polling firm, has released a brief that raises more questions about the credibility of Elon’s polling work:

In conjunction with a pro-marriage amendment group in North Carolina, Public Opinion Strategies has released a brief critique of Elon University’s polling on same-sex marriage. A recent Elon poll concluded that 54 percent of North Carolinians oppose the amendment, even while other polls (from both left-of-center and right-of-center groups) show support around 60 percent.

The POS brief lists two problems with the methodology of the Elon poll. First, it doesn’t filter for registered or likely voters. Second, the survey doesn’t test the actual ballot language (“Marriage between one man and one woman is the only domestic legal union that shall be valid or recognized in this state”). Instead, it asks whether respondents support “an amendment to the North Carolina constitution that would ban same-sex marriage.”

Both factors contribute to producing a result more favorable to opponents of the amendment.