In my last blog post I warned readers to be careful with the latest Elon poll. The methodology is unorthodox for political polling. The results are legitimate but using it as a comparison to other polls or as analysis of the 2008 elections can be dangerous. You are comparing apples and oranges.
On Friday two friends told me they heard on the radio that John Edwards was behind in North Carolina. I’m not sure who reported that, but it was probably either WPTF or WUNC.
According to the Elon poll, Edwards was behind, sort of. He was behind according to an open-ended question asked to all adults, not just voters—or even likely voters—or even likely primary voters.
So any thought that Edwards’ campaign was faltering in his home state would be an overreaction to these poll results. Newspapers and TV did a better job not hyping an anti-Edwards theme. Examples here, here, here and here. But I did find one station, WXII, that ran this headline on an internet news story “Poll: John Edwards Trails in NC.”
I would have preferred that news sources also explain how this methodology was different from other polls too. For example, the approval ratings of President Bush and Senator Elizabeth Dole are a much different measure than those from Civitas and PPP polls, which interview only likely voters.
Random thread for this question, but is there a significant difference between how you and Civitas define likely voters?
ReplyDeleteIm especially interested in the question for the primaries.
Also, do you guys ever get calls from the media about other polls that come out?
I'm not sure how Civitas defines likely voters. I don't know if they use list-based sampling or random digit dialing.
ReplyDeleteFor the primary questions, however, I believe they talk to all Democrats and all Republicans that participated in their survey. In that case, they wouldn't have a likely primary voter definition.
If someone from Civitas is reading this we'd appreciate some clarification.
As to the media, we get calls on our own polls, but never about other polls. When it comes to analysis in news articles, one quote from a pollster is probably enough. I'd rather here from the candidates or campaigns about what they think of the poll numbers.