Nonprofit, nonpartisan journalism. Supported by readers.

Donate

FiveThirtyEight recounting: Franken ‘slight’ favorite

It’s the end of a long day, and I’m not going to get too deep in the weeds with polling guru Nate Silver’s second analysis of Minnesota’s U.S. Senate recount today. At least it’s shorter.

Nut grafs:

If Al Franken in fact wins anywhere near 52.5% of the undercounted ballots, it is quite likely that he will prevail, even given what I would consider to be fairly pessimistic assumptions about the number of correctable errors. You could halve my estimate of the number of recounted ballots, for instance (to 5,623) and Franken still projects to prevail around 69% of the time. If, on the other hand, Franken only wins say 51% of the undercount, then the precise number of correctable errors is more important.

I hesitate to say this, but I think the evidence points on balance toward Franken being a slight favorite to win the recount.

You can also learn about all our free newsletter options.

Comments (4)

  1. Submitted by Pete Anderson on 11/10/2008 - 05:40 pm.

    where are these “undercounted” ballots coming from? Why would there ever be a discrepancy of more than a few votes? It really doesn’t make a whole lot of sense to me.

  2. Submitted by David Koski on 11/10/2008 - 06:55 pm.

    What I have learned over the past week is that the vote count is never truly accurate the first go round. But if there is a statistical advantage of one candidate which is greater that the possible amount of undervotes or incorrect precinct tallies, there is no need for a recount. Thus the less than 1/2% triggering an automatic recount. The process of vote certification is always done and it is being done now. Regardless there will be a recount in this election, because it will remain close enough. When the election is close, the count is more accurate and undervotes (may not have been scanned) are then manually counted.
    The statistical theory is that mistakes, should be equally possible for any candidate.

    Herein begs the real question, Why does Norm Coleman want to stop the recount? Because the likelyhood of him losing is greater than Franken, since of the undervotes, more voted for Obama than McCain.

    Of course, the Coleman camp is using the tired old accusation that Democrats do not know how to vote correctly and thereby should not have their vote counted. I have a general sense that the election could very well have been rigged through the voting machines, but due to bad karma, an idiot for the Republicans alllowed the election to be too close. If this is true, does the Coleman campaign cover the Republicans parties ass, so other voting irregularities are not held in question?, If so , how is that going to be done?

    Any Republicans not liking what I just said, should get over it.

  3. Submitted by Paul Brandon on 11/10/2008 - 06:57 pm.

    The main source of undercounts is ballots that are marked in a manner where the intent is clear but not readable by the electronic scanner, such as putting an X in the box, or circling a name.
    Minnesota law requires that the recount be done by hand and take into account the voters’ intent when it is clear.

    Also, voting places are a zoo, with a lot of pressure to get the votes counted and get out of there to the post election parties ;-).
    It’s easy for a packet of absentee ballots to get misplaced and found a day or two later.
    Shouldn’t be, but it is.
    The whole problem is that the voting system is a human enterprise, with the inevitable human errors.

    Of course, a cynic might say that there have simply been a number of very clumsy attempts by Republicans to hide ballots in DFL-leaning precincts, but of course that would never happen, would it?

  4. Submitted by Tony Wagner on 11/10/2008 - 07:20 pm.

    If I may add:

    Pete Anderson’s intuition in comment #1 is correct: there generally is not a discrepancy of more than a few votes. However, that’s in each precinct, and there are 4,130 precincts in the state. You would be hard-pressed to count 2.8 million of anything across 4,130 separate locations without some level of error.

    Normally it doesn’t matter — even a 1% advantage for one candidate would mean a rather insurmountable lead of 28,000 votes — but here we have only a ~200 vote difference, which is less than one-hundredth of one percent!

Leave a Reply