The Coleman-Franken Senate Race, or Why Nate Silver is The Man

You may have heard that the MN Senate race has gone into recount. Currently, wardrobe-assisted incumbent Norm Coleman leads Democratic challenger (and alter ego of Stuart Smalley) Al Franken by a whopping 221 206 votes out of over 2.8 million. By MN law, Franken has a right to a recount. (Coleman, of course, thought that Franken should just concede for the sake of the greater good. This is rich, coming from a dude who, among other things, accepted expensive suits from a wealthy contributor. But never mind.)

The question is, in the abstract, is it really a good idea to blow $86,000 on such a recount? Nate Silver, of course, has the goods. The short answer is, of course it is. The long answer is far more interesting, especially to a geek like me.

There are 3 parameters mainly at work here. The first is the difference between Coleman and Franken. Silver computes that the recent pre-recount reduction of the difference to 221 votes was crucial; had it been at the original 700 votes, then the probability of a change would be too small for the recount to make any difference.

The second parameter is what Silver calls the Correctable Error Rate (CER); says Silver: "This is the percentage of ballots that were not counted originally, but which will be counted given a hand recount."

The third parameter is Percentage of Ballot Errors favoring a particular candidate (we'll say Franken). Silver, again:

There is substantial evidence that undervotes and overvotes are significantly more common among what we might call vulnerable voters -- in particular, minorities, elderly voters, low-income and low-education voters, and first-time voters. A 2001 study for the House Committee on Government Reform, found that undervoted ballots were more than twice as common in minority-heavy, low-income precincts than in predominately white, upper-income precincts -- even when using the relatively reliable, precinct-based optical scanning system that Minnesota uses. (The discrepancies are significantly higher when using less reliable technologies like punch cards.)

Assuming that the difference stands at 221 votes at recount time, Silver produces the following matrix:

The yellows represent what Silver feels are the most likely ranges. As you can see, the darkest yellow has ranges from "no chance" to "of course". Another way of saying anything an happen. Silver points out that the average of the dark yellow is 44.3%, but that's not the point.

The point is that Coleman has no business in telling Franken to abandon his campaign at this point. Game remains on.