Earlier this week we examined Lisa Murkowski's chances as a write-in candidate for the Alaska Senate race (conclusion: it's a long shot). Now comes word that Mike Castle is toying with a similar tactic in Delaware. And it's not hard to see why: his chances look somewhat better.
Castle, a 70-year-old congressman and leading GOP moderate, was upset by Tea Party darling Christine O'Donnell in last week's GOP primary, despite polls showing him a near certainty in the November race against Democrat Chris Coons.
Yesterday, he conspicuously refused to rule out a write-in candidacy and acknowledged he was still "hurt and stunned" to lose the bitter primary in which O'Donnell parlayed a late flood of Tea Party money and the support of Sarah Palin into a victory, making insinuations about Castle's "manhood" along the way.
He now has until Sept. 30 to launch a write-in bid. GOP leaders are pouring cold water on the idea, and Castle himself has said he doesn't "necessarily want to interfere with Republican chances." But what if he concludes his candidacy actually increases Republican chances?
O'Donnell, on early indications, isn't going anywhere. Polls show her lagging Coons by 15 points, a weak result given the growing vulnerability of Democratic Senate candidates in other deep-blue states such as California, New York, and Connecticut. With the revelation this week that she dabbled in witchcraft, and TV host Bill Maher threatening to dribble out more embarrassing snippets from her 1990s appearances as a studio guest, her campaign already borders on a circus, despite some impressive fundraising.
So what might Castle's back-of-the-envelope calculations look like? In analyzing the Alaska Senate race, FiveThirtyEight's Nate Silver set up the following framework. First, what would be Murkowski's chances in a three-way race against the Democrat and the Republican if her name were actually on the ballot? Second, by what factor should those chances be discounted due to the leakage from being a write-in candidate? This might arise from a voter intending to support Murkowski then screwing up by, say, grotesquely misspelling her name, forgetting to color in the circle next to the write-in space, or simply deciding it's all too hard and ticking a name actually on the ballot paper.
Let's apply this to Delaware. As yet, there is no public polling of a three-way race among Coons, Castle, and O'Donnell. But a useful baseline is this week's CNN/Time poll. Coons leads O'Donnell, 55 to 39 percent. However, in a hypothetical race between Coons and Castle, the numbers flip amazingly, with Castle ahead, 55 to 37 percent. Castle thumps Coons because he wins roughly 60 percent of independents and 33 percent of Democrats.
By contrast, O'Donnell gets thumped by Coons because she wins only 43 percent of independents, attracts negligible support from Democrats, and loses up to 20 percent of Republicans—presumably Castle supporters so appalled by her nomination that they are prepared to defect to the Democratic candidate.
Next, what is the composition of the electorate in Delaware? Let's assume Democrats make up 45 percent of the electorate, Republicans 30 percent, and independents 25 percent (obviously rough, but probably not too far off the mark based on exit polls from the 2008 election).
From here, it's easy to construct a scenario in which Castle, hypothetically on the ballot, gets to, say, 42.5 percent of the vote. Assume, for example, that he continues to receive one third of Democrats (15 of their 45 percent) and half of independents (12.5 of their 25 percent), and he and O'Donnell split Republicans (Castle winning 15 of their 30 percent). Coons might get an overall 40 percent vote share based on winning two thirds of Democrats and most of the remaining independents, while O'Donnell, largely reduced to her base of Tea Party Republicans, would struggle to crack 20 percent.
These are obviously assumptions here. But an important point is that Coons would have a floor of at least the mid-30s due to his strength with Democrats, the biggest component of the Delaware electorate. So the race would boil down to who could best compete for the loyalties of Republican voters and independents wanting to vote for a Republican. The better O'Donnell does, the less likely Castle would be to creep past Coons.
Now to part two of Silver's framework. What might be the extra "penalty" suffered by Castle as a write-in candidate? Using the proxy of a 2006 House race involving a write-in candidate named Shelley Sekula-Gibbs, Silver noted that she performed 10 points worse on Election Day than polls had predicted. There are good reasons to believe that Castle would not suffer so great a penalty. The guy is an institution in Delaware, having represented the same statewide congressional district for 17 years. His approval ratings are solid. His name is a relative cinch to spell. If his "penalty" were even 5 points worse than the polls indicated, his vote would be in the high 30s, giving him a real chance.
As for the rivalry with O'Donnell, Castle's best shot would be to aim for a bandwagon effect, convincing Republicans that O'Donnell was unviable and that they would be better off switching their vote tactically.
It's not a sure thing. It never is, as a write-in candidate. And, to be frank, Castle, at his age, may not have the energy for it. But there is a critical difference between Alaska and Delaware. In Alaska, the GOP's Joe Miller was on track for an easy win—and still seems to be so, even with Murkowski in the race. Whereas in Delaware, the GOP's O'Donnell is on track to lose convincingly—even to an average candidate in Coons in a shocking year for Democrats. Castle, the potential write-in candidate, has unique and proven appeal to Democrats and independents. For Republicans, such a person is not so much a spoiler as he is a savior.
In short, if there was a case for Murkowski to launch a write-in bid, there is surely a better one for Castle to do so. At the very least, it draws Democrats away from Coons and scrambles a race the GOP would otherwise be almost sure to lose.