“the margin of error?”

Back in 2000 (the last election I covered at CNN as an “unbiased” member of the media) we always mentioned the MoE, which was usually 4-5 points. For example, we’d say, “Bush is ahead 47-45, which is within the poll’s 5 pt. margin of error.”

These days I hear polls reported as if they didn’t have any margin for error.

“Polls show Jim Webb is ahead 52-44 in Virginia.” He is? The poll shows him with an eight point lead — but what’s the MoE? And note that the previous polls listed in the story are all very exact — Allen up two, for example.

Well, the point of a MoE (also called “sampling error”) is that the poll isn’t ever going to be exact. You’d think, since polls have proven unbelievably unreliable in recent years, they’d want to expand, not contract, the MoE.

And by the way, this explains why the polls were actually correct in 2000. They showed a slight lead for Bush going into Election Day, but within the MoE. So Gore’s narrow win in the popular vote was actually predicted by polls that showed Bush winning. Odd, yet true.

–Rich Tucker

This entry was posted on Tuesday, November 7th, 2006 at 3:12 pm and is filed under elections, politics. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.