FiveThirtyEight actually gives Romney a better chance of winning the Electoral College while losing the popular vote, in part because many prolific state pollsters have Democratic-leaning house effects. There is a high evidentiary burden for demonstrating that any candidate holds a structural advantage in the Electoral College. The Electoral College almost always follows the popular vote, and even when the popular vote winner fails to secure the necessary electoral votes, it isn’t necessarily apparent in advance. Heading into Election Night 2000, the fear was Gore winning the Electoral College and Bush winning the popular vote. The exact opposite happened only a few hours later. In an extremely close national election, deviations of only a few percentage points in the closest few states can complicate even the best gamed electoral scenarios.
Cohn doubles-down in a later post.