Still on the political thought-train... but I'm thinking of using this article in my writing classes. Holland does a good job of working not just with evidence, but also with the vagaries of probability. Since so many of my students want to just put up some percentages or polling data in support of their argument, I'm thinking this might 1) show them how it's done and 2) open up a space for talking about why Holland chooses to make clear the limits of his speculation.
Beyond that, he's probably right. The media was caught up in Obamamania. Women turned out in greater numbers than in 2004 (57 v. 54 percent) and coupled with a switch in Biden voters (I can attest that in my own caucus precinct, we nabbed over 50% of Biden voters largely on the theory that supporting a clear front-runner in our precinct, Obama, might help erode a third-place show so that Biden might make up ground elsewhere in the scramble for that #3 spot. Risky and ultimately it lost, but that was what the Biden precinct captain advised, so I let him go with it.