I have studied the numbers in the Minnesota race between Al Franken and Norm Coleman very closely, and while I was always confident that Franken did have a chance, I would have had a hard time saying that things were looking bright for him these past couple of weeks. Tonight however I feel confident that Al Franken is the odds on favorite to win after the canvassing board decides on the challenged ballots.
The
http://senaterecount.startribune.com/">Star Tribune website has a "Franken Coleman ballot Challenge" in which you can view many of the ballots that have been challenged by both campaigns and vote on who they should be counted for. I looked at a 10% sample of the 6,655 challenges that were made by the two campaigns and I found that there seems to be significantly more votes for Franken in the pile of challenged ballots than there are for Coleman.
Now I will make no secret of the fact that I want Franken to win this thing, but I do have credentials for analyzing this recount in an unbiased manner. I work in the survey research field, every year my department captures literally millions of pages of surveys many of which are setup in similar fashion to our ballots. I have spent literally thousands of hours looking at survey data, and much of that time involved figuring out the intent of the person who filled the survey out. In addition to my experience in survey researh, I also have experience with working on election related issues. As a college student I was heavily involved in the Minnesota Public Interest Research Group (MPIRG) as the Democracy Task Force Leader which meant my focus was on election reform, and I have quite a good understanding of Minnesota election law as a result of that work.
I analyzed the ballots based off both Minnesota law and my own experiences in judging the intent of those filling out forms. I held both Franken and Coleman votes to the exact same standard, and I have confidence that the judges will agree with at least 90% of my decisions as on the vast majority of ballots the intent was clear. Now there were a small number of ballots in which the intent is questionable and in those cases the judges on the canvassing board may rule differently than I would, but considering that there were questionable ballots on both sides and I used the same standard no matter the candidate I don't believe this will make a huge impact on the margins.
One thing that I can not be entirely sure of however is whether my sample was a good representation of the state as a whole. I do know one thing however, and that is through looking at the Congressional races on each ballot that my sample was not all coming from one part of the state. There were ballots from every Congressional district in the state all mixed together so at the very least we know that the ballots were not all concentrated in one part of the state, which probably means a more representative sample although we can not be certain of that.
That being said however Franken was doing very well in the ballots I analyzed. Coleman started with a lead of 192 votes, but as I looked at the ballots I saw Franken was getting more ballots in the votes that I looked at. As I examined more ballots the lead started jumping back and forth between Franken and Coleman despite the fact that Franken had started with a significant deficit, but by the time I finished the 10% sample the numbers were looking very good for Franken and very bad for Coleman.
With 10% of the challenged ballots examined (interestingly out of 6655 challenged ballots, that makes a sample of 666), I found 279 votes for Coleman, 319 for Franken, and 68 which would not be counted for either of those candidates. The Star Tribune says that gives Franken a 208 vote lead with a 214 vote margin of error. That certainly looks very good for Franken, in fact it looks a little too good and I would be pretty surprised if that margin does not get tighter once all the ballots are examined. While in the end it may be tighter however, if the Star Tribune gave me a good representative sample with those 10% of the challenged ballots I looked at then things are looking very good for Franken, and I am much more optimistic about his chances tonight than I was when I woke up this morning.