Inadequate disclosure of methodology, as Pollster.com has reminded us, is one common problem with political polls. But another is in how poll results are reported.
CNN provides a good example today, in a story headlined: “Giuliani Has Caught Up With Romney in New Hampshire,” based on a new CNN/WMUR poll of the Granite State conducted by the University of New Hampshire. The underlying data is that Mitt Romney’s nine-point lead over Giuliani in UNH’s July poll is now down to just one point.
So Rudy’s surging in NH, right? Well, not exactly. In July the numbers were Romney 34, Giuliani 20, Thompson 13 and McCain 12. Now they are Romney 25, Giuliani 24, McCain 18 and Thompson 13. So Rudy’s “surged” by four points, in a poll whose margin of error is 5.5%. The real news in the poll is a decline in Mitt’s support, and the most dramatic gainer was McCain, not Rudy.
From a pure horse-race perspective, the CNN story is accurate. But it’s also misleading unless you look at the actual numbers and particularly the margin of error. Still, I’m sure Rudy’s campaign is happy to take the gimme.