New Analysis of the Franken-Coleman Senate Recount

The current situation is this: We are waiting for a final decision next week on the outcome of the race, and most likely this outcome will start numerically with Franken’s current 49 vote lead and be adjusted by the addition of several hundred absentee ballots. The exact number of these ballots to consider is still being fought out.

Following the board’s decision next week, probably Wednesday, one of the candidates will be declared the winner (this is the job of the canvassing board …. to declare the winner) and sent to Washington. There may well be further court fights over this, but at some point the Senate seats its Senators and it is difficult to imagine how a court can overrule the Senate in this matter, as long as it has acted Constitutionally.

At some point, the candidate who loses will have to realize that there is not a legal “recourse” for losing an election.

Given that the absentee ballots are up for grabs and constitute the most important thing happening right now in the recount, I thought you might enjoy a brief analysis of what the possible outcomes might be. First, have a look at this graph:


Let me explain how to read this graph first, then I’ll tell you how it came to be and what it really means.

This is a Worst Case Scenario graph which can be used for the Franken campaign.

First, you find the number of absentee ballots along the x-axis that are going to be counted along the bottom. The total number is likely to be less than 1300, possibly and possibly fewer than 1000 ballots will be counted.

Then you find the corresponding number along the y axis. This is the chance of the “bias probability required” to not change the race’s current outcome.

So, if 400 ballots are looked t, there is an eighty percent chance that the outcome will not be changed under Wort Case Scenario conditions. As the number ob ballots goes up, the WCS for Franken gets more and more likely to happen. If the number of ballots counted stays under 49 (which it won’t) then the probability of the outcome not changing is 100%

This is based on a simple model that includes known variation in the voting preferences by precinct across the state. Think of it as a simulation model. It is the Worst Case Scenario simulation for two reasons. First, the variation is assumed to go against Franken 100% of the time. Second, it assumes a higher level of actual variation than probably exists because variation is estimated at the inter-precinct level rather than at the inter-county level.

This second WCS factor is probably small, so ignore it. The first WCS factor … the bias going against Franken all the time … is not as unlikely as it sounds. The null model, of the bias being roughly fifty-fifty, is clearly not correct given the tricks the Coleman campaign is using in rejecting absentee ballots. Therefore, it is probably a good idea to use this model rather than a 50-50 model, if one wants to be conservative.

Share and Enjoy:
  • Twitter
  • StumbleUpon
  • Facebook
  • Digg
  • Yahoo! Buzz
  • Google Bookmarks
  • LinkedIn

Leave a Reply

Your email address will not be published.