Minnesota’s Canvassing Board this afternoon completed the bulk of its review of challenged ballots. The Canvassing Board ruled upon 1,325 challenges, according to numbers prepared by the Star Tribune, including 852 challenges brought by the Coleman campaign and 472 brought by the Franken campaign. Among these 1,325 ballots, 758 were allocated to Franken, 319 to Coleman, and 248 ballots were assigned to third-party candidates or deemed to be illegal. This resulted in a net gain of 439 votes for Franken, giving him a nominal lead of 251 ballots.
Franken’s lead is almost certain to diminish once the Canvassing Board reviews more than 5,000 withdrawn challenges, and defaults them to the rulings originally made at the county level. Franken withdrew considerably more challenges than Coleman: 2,806, by my count, to Coleman’s 2,525. If all withdrawn challenges are assigned to the opposing candidate, this would result in a net swing of 281 votes to Coleman, putting him back in front by 28 ballots.
Not all withdrawn challenges, however, can safely be added to the opposing candidate’s total. Fundamentally, there are two types of challenges:
Type 1: You can challenge an opponent’s ballot that you were hoping to get excluded;
Type 2: You can challenge one of your own ballots that you were hoping to get included.
The first type of challenge will add to the opponent’s total if and when it gets rejected, but the second type will not. The question, then, is the relative frequency of the different types of challenges in each candidate’s withdrawn pile.
Suppose, for instance, that for both candidates 90% of the challenges were Type 1 and 10% were Type 2. If this is the case, Coleman would gain 2,525 votes from withdrawn Franken challenges, and Franken would gain 2,273 votes from withdrawn Coleman challenges. That would produce a net swing of 253 votes for Coleman, giving him a victory by 2 (!) ballots.
In practice, however, it appears that Franken’s pile contains a relatively higher instance of Type 2 challenges. There are at least three reasons to conclude so:
1. Although this isn’t necessarily meaningful, Franken had a higher proportion of Type 2 challenges among those challenges that were not withdrawn.
2. Prior to the challenge phase beginning, the Franken campaign claimed that it would gain ground if all challenges were rejected, pulling 4 ballots ahead of Coleman. A necessary implication of Franken’s claim is that he had a higher proportion of Type 2 challenges. If I try and work backwards from the Franken campaign’s math, I show that Franken now has a lead of something like 66 votes after accounting for withdrawn challenges.
3. The Star Tribune has a more direct way to account for withdrawn challenges. Specifically, they have projected them out based on the votes of their readers, thousands of whom sorted through each ballot individually on the Star Tribune’s website. The Star Tribune projects a Franken win by 78 ballots based on these reader estimates, fairly closely matching my back-of-the-envelope math.
There is undoubtedly some uncertainty in these numbers. (Moreover, we are uncertain about how much uncertainty there is). It appears likely, however, that Coleman did not connect on his challenges at a high enough batting average, and that Franken will emerge with some sort of solid, double-digit advantage exiting this phase of the process.
Coleman, then, has two additional processes by which he couple hope to reverse that result, as well as a couple of hail marys.
Issue #1. Rejected Absentee Ballots. Minnesota’s Supreme Court yesterday mandated that the campaigns must work with the counties in attempting to identify and count absentee ballots that may have been rejected in error. The process the Supreme Court established is unusual, in that each of the campaigns would have to mutually agree that a ballot is valid before it gets counted.
The problem for the Coleman campaign is that counting more absentee ballots will probably benefit Franken. Democrats made a push nationwide for early and absentee voting, and at least one pre-election survey also had Franken doing better among absentee voters. The behavior of the respective campaigns, of course, has been perhaps the strongest signal that such votes are likely to help Franken.
But now that he’s (probably) no longer ahead, Coleman has conflicting objectives on the absentee ballot front. On the one hand, he might want to gamble and count as many of them as he can — he has little to lose, and has to pick up votes somehow. But on the other hand, he knows it’s more likely than not that a plurality of such votes will be for Franken. The dilemma is a bit like that facing the gambler who, having lost his shirt at blackjack, puts his last few chips on ’00’ on the roulette wheel hoping to get even.
Issue #2. Duplicate Ballots. The other legal issue is that surrounding duplicate ballots, which we addressed at length earlier today. In brief: the Coleman campaign believes that some ballots have been double-counted, and wants the Supreme Court to require that the counties check for any such instances and amend their counts appropriately.
Here too, however, there is no particular reason to conclude that identifying duplicate ballots would work to Coleman’s benefit. The Coleman campaign presented more than 200 alleged duplicate ballots to the Canvassing Board. The Franken campaign, however, indicated to the Canvassing Board late today that it had more than 300 such ballots in its possession, but had withdrawn them from its stack of challenges. (EDIT: The ‘200’ and ‘300’ numbers reflect all types of ‘incident’ ballots, including but not limited to potential duplicates, although the duplicates make up a large fraction of these).
From my vantage point, it’s pretty much a 50:50 proposition which candidate will benefit if the double-counted ballots can be identified (something, by the way, which is much easier said than done). If one campaign is making a lot of noise about the issue, it’s because that campaign is losing and needs to gamble. In this case, that would be Coleman, who only today filed a petition with the court to address the issue. (Earlier in the recount process, in fact, the Coleman campaign had been admonishing the Franken campaign for challenging duplicate ballots).
Regardless, if Coleman trails Franken by something like 70 votes, it seems unlikely that there are enough potential double-counted ballots to allow him to make up his deficit. Collectively, the campaigns were apparently sitting on something like 550 potential duplicate ballots. Suppose that all 550 such ballots can be identified as duplicates, and that all are otherwise legal ballots cast for either Coleman or Franken. Coleman would need Franken to be identified on 56.4% of such ballots to make up a 70-vote deficit with with him. If such ballots are equally likely to help Franken and Coleman, the odds of this happening through chance alone are remote — about 1,100-to-one against. And if Coleman has instead to make up, say, a 120-vote gap because Franken gains ground from the counting of absentee ballots, then good luck to him. It is possible that there is some sort of systematic reason why Franken is more likely to have been named on a duplicate ballot than Coleman, but as yet I am unaware of it.
Hail Marys. If the above processes fail to push Coleman past Franken, Coleman can always contest the election. There are nearly as many reasons for a challenge as there are lakes in Minnesota: Coleman could challenge the Canvassing Board’s standards for counting challenged ballots, or the Canvassing Board’s decision to give Franken credit for missing ballots in Minneapolis, and there might be other legal paths open to him on either the absentee ballot or duplicate ballot front. If the votes ultimately aren’t there for him, however, Coleman is just treading water — he could win his court case and lose the election.
Finally, Coleman could ask for a re-vote — but the only agency with any sort of statutory authority to grant this is the Democratic-controlled U.S. Senate. Coleman would have better odds of filling Amy Poehler’s shoes on Saturday Night Live.