Last month I published a piece suggesting that while the odds of a Republican takeover of the Senate were not high, the possibility could no longer be ignored. My article was not well received. Critics argued that (among my many sins), I had cherry-picked surveys, given credence to the (allegedly) fatally flawed Rasmussen results, and worst of all, ignored Nate Silver’s superior methodology.
In the ensuing four weeks, a number of articles arguing roughly what I did have appeared. The publication today of a new survey from the conservative group American Crossroads offers an opportunity to revisit the issue.
Because the American Crossroads poll uses samples of only 100 from each of thirteen key Senate races, we can learn little about the individual contests. Of more significance is the overall finding, which essentially replicates Stan Greenberg’s, that respondents find various versions of the generic Republican narrative more persuasive than the competing Democratic arguments. The Republicans this year are campaigning with the wind in their sails. If some Democrats in tight races survive, it will be because they have managed to tack skillfully in these adverse conditions.
So what of the individual races? The overall picture is this: Democrats begin with 44 safe and 4 likely seats, for a total of 48, while Republicans begin with 34 safe seats and 4 likely. That leaves 14 seats in play that will decide the control of the Senate. If Democrats win just 2 and Joe Lieberman continues to caucus with them, Vice President Biden’s tie-breaking vote will enable them to maintain control. Conversely, Republicans need to win at least twelve out of 14 (with a complication I’ll return to).
The following table summarizes where we are now as we await the results of the Colorado primary contests.
It’s hard to imagine that Republicans won’t prevail in the first five states listed, which moves their total up to 43. That leaves nine truly pivotal seats, of which Republicans would need to win eight. Four are open—Pennsylvania, Florida, Ohio, Illinois. The remaining five—Nevada, Colorado, Washington, California, and Wisconsin—all feature hard-pressed Democratic incumbents. Each is averaging well below 50 percent support—a parlous position for any incumbent.
Yes, dear readers, I’ve treated Rasmussen surveys on all fours with the others. To justify this, I turn to … Nate Silver (his site’s modest motto: “Politics Done Right.”) Silver ranks 63 surveys for the accuracy of their results. He ranks Rasmussen in the top quartile—15th, to be precise. Using his preferred metric of “pollster-induced error,” he finds that Rasmussen missed the mark by an average of 1.74 percent—a bit worse than NBC (1.53), Pew (1.60), or Gallup (1.66), but better than either CNN or CBS/New York Times (both 1.94).
So what are the odds of a Republican takeover? Basic probability theory says very low—if these races are truly independent events. But history suggests that they aren’t. Recent elections in which control of the Senate shifted—namely those in 1980, 1986, and 2006—featured a number of close races, nearly all of which narrowly went to the eventual new majority. The reason is simple: while candidates and state issues matter in Senate races, so do the underlying national tides, which can pull even weak challengers to victory (remember John East).
Now for the promised complication. I’ve assumed that Rubio would have to beat Crist in Florida. But it’s entirely possible that despite a bitter break with his party, a lifetime Republican like Crist would end up caucusing with them. He could also adopt a purely independent stance and withhold his vote from either party. If so, a 50-49 Republican edge would allow them to organize the Senate. The Vice President can break ties, but he can’t make them.