Getting a handle on which way the races are trending is at once easier and harder than it used to be. Easier, because in this wired world everybody is offering up predictions; harder, because many aren't reliable--they're uniformed, overtly partisan, or fly-by-night hucksters. There's no Good Housekeeping Seal of Approval.
But James E. Campbell, a political scientist at SUNY Buffalo, has done something very useful and informative. He went and checked the House race predictions of one venerable prognosticator, the Cook Political Report, going back to 1984 and measured them against the election outcomes to see how good they were. It turns out they were very good. Across 11 elections (1984-2008) the Cook Report correctly called races rated "solid, likely or leaning to the Republicans" a whopping 97.5 percent of the time. On the Democratic side, the Cook Report was right 97.8 percent of the time.
Now you might be thinking to yourself, "Anybody can make a good prediction a week from Election Day." And that's what I think is most noteworthy about Campbell's study: he only measured the ratings issued before Labor Day, which is to say, all of these calls were made a good two months before the election.
The skeptic in me wondered if the numbers might look so good because of gerrymandering. Maybe there were a ton of "solid" races (easy calls) that masked a mixed or poor record on leaners (tougher calls). But Campbell broke out those races and the predictions were still very good (85.7 percent). Even the Cook Report's "toss up" category was accurate: Republican-held seats rated "toss up" stayed Republican 55 percent of the time, Democratic toss ups, 49.2 percent of the time.
Campbell's study seems especially salient because--brace yourself, Democrats--on Tuesday (the day after Labor Day) the Cook Report came out with its latest House ratings and, for the first time, they showed Republicans picking up a minimum of 40 seats. Republicans only need 39 seats to win back the House. Based on the Cook Report's track record, you have to believe there's an extraordinarily high likelihood that control is going to switch in November.
You have to pay to subscribe to the full Cook Report, but most of the general overview stuff is free. The individual ratings, and ratings changes, are pretty widely disseminated, though. One easy place to keep up with them is on Twitter, where you can follow the Senate expert, Jennifer Duffy (@jennifereduffy), and the House expert, David Wasserman (@dave_wasserman).
We want to hear what you think about this article. Submit a letter to the editor or write to email@example.com.