The Long Winding Road of the Minnesota Senate Recount

There has already been a lot of controversy in the Minnesota Senate recount, but the finding of 171 new ballots in Ramsey county has generated real concern. The Star Tribune wrote that "Ramsey County's recount problem caught many by surprise." The Star Tribune noted the new ballots were one of two pieces of good news "boosting the prospects of DFLer Al Franken" when the vote recount was looking grim for him.

Sen. Norm Coleman's campaign said it was "skeptical about [the ballots'] sudden appearance."

All the previously uncounted ballots were discovered in Maplewood Precinct 6. Of the 171 votes, 91 went to Franken, 54 to Coleman and 26 to other candidates. The new votes increased the total votes for the Senate candidates in the precinct by over 12 percent. The percentage of votes Franken got from these new votes in this precinct were statistically significantly different than the lower percentage that he obtained from the ballots that were first counted in that precinct.

When Karen Guilfoile, Maplewood’s director of city services, was asked about whether she was surprised that such a large percentage of the ballots could be over looked she told FOX News that "We are still trying to figure out what happened. We are going back and trying to figure out exactly what went wrong. Fortunately, we have a paper trail." One possible explanation is that a ballot counter that kept jamming was replaced, but the ballots were apparently not run through again after the counter had been replaced.

While Guilfoile defended the vote-counting process in the precinct and said that she believed that votes that had not been originally counted were secure, the Star Tribune reports that Coleman observers objected that there were now 31 more ballots than voters who had signed in on Election Day. Guilfoile told FOX News that she was "not certain" but the difference in numbers "appeared at first blush" to be due to errors in recording absentee ballots on the voting rolls on Election Day.

It is striking how different these votes were from the ones that had already been counted for the precinct. Of the votes that were counted, Franken received 45 percent and Coleman 39 percent. For the new ballots, it was 53 percent to 32 percent, respectively. Compared to the percentages that Franken and Coleman had received from the previously counted ballots, Franken had a net gain of 26 votes, a number that is close to the 31 unaccounted for votes, but that might simply be a coincidence. Guilfoile had “no comment” when asked about these numbers.

Given the vote shares for the counted votes, what are the odds that you could get such different results between the initial votes and the new votes in this precinct? There were 1,170 votes already counted for the two major party candidates and 145 among the new ballots. Comparing these two sets of ballots, there is just over a one in a hundred probability that these differences in the percentages could have happened by chance.

If you compare the variation in Franken’s and Coleman’s shares in the other Maplewood precincts or the precincts for the rest of Ramsey County or even the entire state, the seven or eight percentage point differences are even more unlikely to be due to chance.

In such a tight race, where 32 absentee ballots had been held in a Democrat election judge’s car (also in Ramsey County), these votes begin to add up.

At the end of the recount on Monday night, Coleman had a 370 vote lead. By Tuesday night, with the new results from Ramsey County and other changes, his lead had been cut to 303 votes. But there were also 6,003 ballots challenged by the two candidates, with virtually the same number challenged by each of them.

As a rough experiment, four students were hired to look at the sample of 599 challenged ballots at the Star Tribune’s website. The students with different political orientations agreed on the intent of the voter 85 percent of the time. If that rate holds true for all 6,003 challenged ballots (as of Tuesday night), that means that there will be about 900 contested votes (300 involving over votes and 600 involving undervotes).

If these 900 contested votes are equally divided between Coleman and Franken challenges, reducing Coleman's lead well below 400 at least makes mathematically possible for Franken to win. A 450-vote lead for Coleman, for example, means that even if Coleman loses all his challenges, Franken would have to get all of his challenges to win.

Secretary of State Mark Ritchie's decision to have county's separate out absentee ballots by the reason that they have been rejected so that Franken can challenge their rejections helps keep that route alive.

Without the votes from Maplewood Precinct 6, the other found absentee ballots, and the ability to count absentee ballots that been rejected, Coleman’s lead would be very close to a statistically impossible hill for Franken to overcome.

*John Lott is the author of Freedomnomics and a senior research scientist at the University of Maryland.