In my
diary last weekend, I released some spreadsheets I made up to try to analyze how well the pundits predictions would hold up against the actual results. Now that the results are all in, I thought I'd check back and see how well they did. Here are the spreadsheets that list the results for each
House and
Senate race as well as the election predictions by
Congressional Quarterly,
The Cook Political Report,
Stuart Rothenberg and the Washington Post's
Chris Cillizza (last updated Monday evening).
First, I'd like to send out a big
CONGRATULATIONS to Reprentatives-elect Dave Loebsack (IA-02), Carol Shea-Porter (NH-01) and John Hall (NY-19) for beating the expectations game in a big way. None of the four prognosticators I tracked predicted that any of these races might even be tossups. Way to go guys! And the same goes for candidates Larry Kissell (NC-08) and Ciro Rodriguex (TX-23) whose races are still to be decided by recount and runoff, respectively. By bringing in a competitive result, you have won, even if you end up losing the seats.
I got the election results from CNN and made two modifications: I'm listing FL-13 as "tossup" since the recount is still in progress and I'm listing LA-02 as "Democrat" since the results of the runoff will only change the name of the winner, the party will remain the same. All other races that have yet to be called by CNN I'm listing as "tossup":
| House | Senate |
Rep | 195 | 49 |
Dem | 230 | 51 |
Tossup | 10 | |
A race is said to be "miscalled" in any of the following cases:
prediction = Dem; result = Rep
prediction = Rep; result = Dem
prediction = Dem or Rep; result = tossup
It's important to note how many races each pundit predicted as tossups, since one could produce an error free prediction set merely by calling all races as tossups. That said, let's get to the results:
| Senate Races | Senate Tossups | House Races | House Tossups |
| Miscalled | Predicted | Miscalled | Predicted |
Congressional Quarterly | 0 | 4 | 13 | 23 |
The Cook Political Report | 0 | 7 | 6 | 39 |
Stuart Rothenberg | 0 | 1 | 9 (5) | 19 (40) |
Chris Cillizza | 0 | 4 | 15 | 20 |
Stuart Rothenberg's results were a bit hard to classify for the House, since he included the categories "tossup-lean Democrat" and "tossup-lean Republican". The first set of numbers in his row treat those predictions as straight "leaning" predictions and the numbers in parentheses treat them as "tossup".
Based on this election outcome, I'd have to rate Rothenberg as the best predictor. However, keep in mind, these results are based on the last minute predictions. As kos pointed out, Rothenberg was the guy who thought Chris Bowers was crazy in Jan. 2005 for wanting 80 competitive races. Also, this is only the result from this one election. One data point does not a curve make .
For details on which races each pundit missed, see the second sheet on the House spreadsheet.