The folks at fivethirtyeight.com are doing a bang up job of doing the math with some informed and fairly objective analysis. They have some fascinating analysis about the upcoming Franken-Coleman Minnesota Senate vote recount.
Determining the winner of a vote recount depends on three primary factors:
- the original margin (221 for Coleman),
- the fraction of total votes that weren’t originally counted but will be counted in a hand recount (what they call Correctable Error Rate), and
- the distribution of newly counted votes
If the distribution of newly counted votes is 50/50, Coleman wins, as Franken can’t gain any ground. The higher the fraction of previously uncounted votes are, the more potential there is for Franken to make up new ground.
At this point, a reasonable range of estimates of these parameters allows us to assess the odds.
Votes don’t get counted because they’re overvoted (the machine decides a ballot selected two candidates) or undervoted (the machine decides a ballot did not select any candidates.) Overvotes are almost always unintentional; undervotes are usually intentional (leave an X instead of filling in the oval, etc.) Probably, less than one percent of votes will be reclassified from over- or under-voted to a definite vote during the hand recount; about 15% of those will go to Dean Barkley. This by itself would not be enough for Franken to make up the 221-vote difference. But together with the other factor, the distribution of new votes, Franken could come out ahead.
Why? Because undervotes and overvotes occur predominantly among the elderly, low-income and low-education voters, and first-time voters. (The article also mentions minority voters… I think this part of the reasoning applies only to those for whom English isn’t their primary language.) In this election, the elderly vote is split, but the youth, first-time, and “minority” vote have gone to Franken.
Now the author can crunch some numbers. The highlights:
- If the uncounted fraction of votes that will be counted is 0.90% (25,000 votes), and Franken takes 50.50% of those votes, the odds are even that Franken will win.
- If the uncounted fraction of votes that will be counted is 0.50% (about 13,900), and Franken takes 51% of those votes, then Franken has a 57% chance of winning.
- If Franken captures 53% of the new votes, then the number of new votes counted has to be less than 0.25% (less than 7,000 votes) for Coleman to have a chance.
So, if you’re for Coleman, you want to diminish the number of new votes, and hope that they count more of the middle- and upeer-class. If you’re a Franken supporter, you want lots and lots of new votes, on the theory that the newly-counted votes represent the “vulnerable vote.”
If you’re a Barkley supporter, you’ll need a time machine set to 4 November 2002.
None of this, of course, takes into account any legal wrangling that may ensue. How a municipality can allow their voting booth to put an inaccurate date and time stamp on a ballot is beyond my comprehension.
No matter what, stay tuned to www.fivethirtyeight.com for some reasonably objective analysis.
Did you enjoy this article? Please subscribe to my blog to receive updates.
You must log in to post a comment.