Update: The last poll to come out before the SC primary showed Huckabee pulling ahead by a comparable margin to what CI’s measurement of the blogosphere had projected. McCain ended up with a victory. Perhaps this has something to do with older Evangelical voters being unable to come to the polls because of the weather, or perhaps it’s one of many other reasons. Regardless, CI’s projections have consistently been at least as accurate as traditional polls, which speaks to the power of the so-called “collective intellect.”

CI correctly projected Clinton’s win in Nevada. We used two different methodologies for measurement in SC and Nevada, and it looks like averaging results from both state blogs and national blogs is so far the most accurate measure. We’re 2 for 2 using this method, so this is how we will proceed in all future projections.

We used two different tactics for measuring the Democratic Nevada caucus and the Republican South Carolina primary. For Nevada, we combined a sampling of Nevada blogs with an across-the-board measurement of sentiment and activity about each candidate. For South Carolina, we combined across-the-board blogs with ones which we have identified as “right-leaning.” We tracked activity from Jan. 15 to Jan. 18.

We’re still showing Clinton with a slight lead in Nevada, but the late-breaking good-for-Obama news definitely has a possibility of changing the landscape.

The latest poll shows McCain with a 7-point lead in South Carolina, but our measurement of the blogosphere shows Huckabee pulling out a win–although it’s by less than 1%. Regardless, we’re showing a tighter race than what the polls predict.

For details on the methodology CI used to calculate these results, please see these previous posts:
Iowa/NH
Michigan

(For the first time, I included Ron Paul in our projection. Since the New Republic story broke last week, his online followers appear to have decreased their spamming efforts, making measurement more possible. We’ll see whether or not this comes back to bite us.)

Tags: Analyzing Consumer Generated Media, Mass Media/Culture //