Hold Your Horses, Minnesota

The State of Minnesota's Canvassing Board has finished adjudicating challenges brought to the Board by Al Franken. They will begin evaluating challenges brought by Norm Coleman's campaign tomorrow.

The Star Tribune has logged 391 challenges made by the Franken campaign. Among these, 225 ballots were counted for Norm Coleman (57%), 56 for Al Franken (14%), and 110 (28%) either for a third-party candidate or for nobody. (The Star Tribune did also identified dozen or so challenges apparently made by the Coleman campaign that had somehow found their way into Franken's pile; these challenges are not included in the totals above).

There seems to be something of a consensus that these results represent good news for Al Franken, as a fairly high number of his challenges were successful. For a variety of reasons, I am less than certain about this, and think that we haven't yet learned very much about the ultimate outcome of the recount.

For one thing, the information provided by the Star Tribune is incomplete. We now know something about how the challenges ended up, but we don't know anything about how they began. That is, we don't know what the initial ruling on the ballot was at the county level, nor the reason for the challenge.

To tackle this question, I evaluated 155 challenges brought to the Board by the Franken campaign, as available on the Star Tribune's website. (Why 155? Because this is about all I could handle before becoming deathly bored). We can classify these challenges into four groups:

I. 87 ballots (56%) appear to initially have been counted for Norm Coleman, and the Canvassing Board upheld that ruling upon review. These represent unsuccessful challenges.

II. 18 ballots (12%) were initially counted for Coleman, but were ruled as nonvotes by the Canvassing Board. These represent successful challenges.

III. 31 ballots (20%) were initially counted as nonvotes (or for third-party candidates), but were deemed to be Franken ballots by the Board. These also represent successful challenges.

IV. Lastly, 19 ballots (12%) were initially counted as nonvotes (or for third-party candidates), and the Canvassing Board also deemed them nonvotes upon its review. These represent unsuccessful challenges.

In total, about three in ten Franken challenges -- everything in Groups II and III -- appear to have been successful. This is a reasonably high success rate. However, we should have expected the Franken campaign's success rate to be reasonably high, because it withdrew the vast majority of his challenges. Taken as a percentage of the 3000 or so ballots that the Franken campaign challenged initially, its success rate is more like 4 percent.

A couple of other salient points from these statistics. Firstly, quite a decent fraction of the challenges did not concern potential Coleman votes that the Franken campaign was trying to get thrown out. Instead, approximately one-third concerned potential Franken votes that the campaign was trying to get included.

This distinction is important because, when the Franken campaign challenged a legal Coleman ballot, this resulted in a temporary deduction from Coleman's total in the Secretary of State's running count (essentially, such ballots were treated as guilty until proven innocent). By contrast, when Franken challenged a nonvote that he wanted to have counted for himself, this produced no immediate adjustment to the state's totals. Long story short, the greater percentage of challenges that were made to nonvotes rather than to Coleman votes, the better off Franken is relative to the state's accounting.

In addition, the different types of challenges were associated with vastly different success rates. From among the 105 Coleman ballots that the Franken campaign challenged, 18 challenges were accepted, or 17%. By contrast, from among the 50 nonvotes that Franken challenged, more than half (31 or 62%, to be precise) were counted for him by the Canvassing Board.

Even these numbers, however, are hard to place into context, because we don't know anything about the nature of the thousands of challenges that Franken withdrew. The state does not appear to have added the withdrawn challenges back into its totals; in fact, it does not appear to have done anything with them. We don't know, among other things, how many of these withdrawn challenges were made to Coleman votes (in which case, Coleman will now get credit for them) as opposed to nonvotes (in which case, they're basically irrelevant). The Coleman campaign has also withdrawn the majority of its challenges, although fewer than the Franken campaign has (in fact, yesterday Coleman unwithdrew about 200 of his challenges).

With these withdrawn challenges existing in a state of limbo, we have very little idea about where we stand. We might know now something about many votes a candidate has gained during the challenge process, but we don't really have a good idea of where he stood before the challenge phase began.

We also know nothing so far about the nature of the Coleman challenges. Since Coleman withdraw fewer of his challenges, does this mean that the challenges he retained are less likely to be successful? That seems highly probable -- but, we can't be certain. Also, will Coleman's challenges have a different typology from Franken's? For instance, will a significant fraction of them concern nonvotes, as opposed to Franken votes the campaign is trying to have excluded? We simply don't know.

I do consider Franken the favorite in the recount, but that is because it appears that erroneously rejected absentee ballots will be considered by many counties; these ballots could easily tip a net of 100 or so votes to Franken. Without those absentee ballots, the recount still appears to me to be too close to call. Until we learn more about (i) Coleman's challenges and (ii) the withdrawn challenges, I would be suspicious of overly-specific claims about the status of the recount that you might see elsewhere.