MN Sen Recount (UPDATE: Stuart Smalley certified winner, lawsuit forthcoming)

[emailking]

You can see 4 challenged ballots here (click "View full slideshow" on the right)

http://minnesota.publicradio.org/display/web/2008/11/19/recountbegins/

[emailking]

Franken's people on Silver's site are claiming that with the 37 votes added to Franken's number from the 171 "found" allots yesterday, they are now only trailing by 13 votes.

Can anyone tell me how they arrived at this number?

http://www.fivethirtyeight.com/2008/12/frankens-good-day-shifts-odds-on.html#comments

First of all, the discussion was with Talking Points Memo. 538 is just reporting on article. The campaign says that if all challenges are rejected they are 50 behind, assuming all remaining counting happens exactly as it originally did. This was before the 37 votes came to light. So adding those in, and taking on faith the campaign claims, that leaves Franken 13 behind at the moment.

[emailking]

Unless every Franken challenge of an uncounted ballot made yesterday is legitimate, there is simply no way he made up 59 votes yesterday alone.

The claim is that he is ahead by 22 assuming all challenges of every type are rejected. Thus their claim is not dependant on adding new votes to his total from Franken challenges (unless they challenged votes counted for Franken!). It does mean they are assuming maasive pickups from Coleman challenges being rejected.

I don't know where you get +59 from. They had said -50, and now +22. But 72 - 37 = +35. Obviously the +37 from the found ballots are a special case.

[emailking]

I'm just saying you can't count it in sort of the typical variations you get between the machine and hand recount. It's not voter error or machine error or anything. It's just a batch of ballots, valid or not, that were never addressed in the first place.

[emailking]

Nothing I said is incorrect because I'm only saying what they are saying. I thought you were taking two statements from the past few days and claiming they must be assuming some of their challenges are going to go their way.

Your flat assumption could be wrong of course. Perhaps Coleman is challening non-votes less frequently than Franken. On the other hand, the campaign assumption that all challenges will be rejected could be wrong.

Certainly both assumptions are wrong in an absolute sense, but then it's a matter of which one gives the better picture of what has happened / will happen.  

[emailking]

A lot of dems like ron paul. He knew Obama would win MN and wanted to write in Ron Paul. He knew the Senate race would be close and preferred a Dem in the seat to a Republican. Makes sense to me. We have first place only voting in this country for senate races. You have to vote strategically sometimes. If there were IRV maybe he would have put Barkley first.

[emailking]

One of the more creative challenges I've found so far.  Follow the logic (not that I think it's successful btw)

http://senaterecount.startribune.com/media/ballotPDFs/cottonwood_noprecinct_8.pdf

He's kicking people off the Island...except for the write-ins and not for the Mayor on Windom.

[emailking]

If you actually read Nate's "prediction" of 27 votes, you will, see he clearly stated that the error bars were large and there was a good chance Coleman could win. This is also the case with his latest predictions...the error bars being the hundreds of votes. Thus he originally said Franken was a slight favorite and now says Coleman is a slight favorite. It's no different than a poll that puts a candidate ahead 1 or 2 points with a 3-4 point MOE.

[emailking]

It's just a recommendation that the absentee ballots be counted. They acknowledged the counties cannot be forced to do it.

[emailking]

It's just a recommendation that the absentee ballots be counted. They acknowledged the counties cannot be forced to do it.

Of course, one would imagine every county would want to count them. After all, if you're a Republican county, your ballots probably favor Coleman; if you're a Democratic county, your ballots probably favor Franken. To refuse to count your county's ballots would be to hurt your favored candidate. (Assuming that there are no or very few counties with local governments radically different from their national voting patterns.)

But before some of the counties were refusing to address the issue while others were doing it voluntarily. Ramsey counted wouldn't count them, for example. I think the logic of the officials is that they have already done their job with the absentee ballots and that should be it? Perhaps it will change now, but I don't know what's different really.

[emailking]

This is an entirely separate matter from the voting system used. In a run-off, the result could end up this close. With IRV, it could be this close on the final iteration. Such a situation wouldn't be any less likely either than with first-place plurality voting. The point in using IRV on a run-off system is to better gauge the true preferences of the voters, not to avoid close elections.

[emailking]

The possibility for a separate runoff producing an equally close result exists, but it is not probable.

That's not the issue. You're looking at it the wrong way. It doesn't have to be close to begin with. Let's say we use a runoff system. Here's the initial count.

Candidate A: 45%
Candidate B: 30%
Candidate C: 25%

Here, A and B go into a runoff. It turns out that candidate C voters split such that the final result is:

Candidate B: 50.001%
Candidate A: 49.999%

Now we're locked in endless recounts as Candidate A fights to win and everyone is saying how nice it would be if we had just used a plurality system as then the result would have been clear in this case.

[emailking]

I'll agree with you that run-off is better than no run-off but close results are just as likely. That's the point I'm trying to make. The voting method should be a separate consideration from how to deal with close elections. Because we will have these rare, close elections under any system.

[emailking]

I'll agree with you that run-off is better than no run-off but close results are just as likely. That's the point I'm trying to make. The voting method should be a separate consideration from how to deal with close elections. Because we will have these rare, close elections under any system.

What's less likely is two consecutive close races. The assuming equal probabilities at each stage, the combined probability decreases as the square of the probability.

In a plurality system you can't have two consecutive races, so technically it's greater (since it's greater than 0) in a run-off system.

I think you're still looking at this wrong. Yes, if your initial results are:

A: 40.01%
B: 39.99%
C: 20%

Then yes, it is unlikely the run-off will be close. But here's the crux. The chances, on average, of having elections close in the initial tally are balanced, on average, by the chances of having run-offs where the result is close, even though the initial tally wasn't.

[emailking]

Exactly. In a run-off the first result is replaced whole-cloth. So in the end, with either plurality or run-off, we have the same number of elections that can be close.

[emailking]

I disagree. You assume that all races are equally likely to be close. I assume that races with more than two candidates have a greater likelihood of being close due to the reduced pool of voters for the top two candidates.  With fewer votes the statistical fluctuations that can lead to a close vote have a greater impact. This leads me to fewer close races after a runoff than in a plurality system.

Alright I'll give you that. It's a significant but not a huge difference. Suppose a 3rd party candidate takes 1/4 of the vote.  Then it's something like sqrt(3)/2 times less likely.

(edit, changed 1/3 to 1/4 so it's still a two-way race.)

[emailking]

The possibility for a separate runoff producing an equally close result exists, but it is not probable.

That's not the issue. You're looking at it the wrong way. It doesn't have to be close to begin with. Let's say we use a runoff system. Here's the initial count.

Candidate A: 45%
Candidate B: 30%
Candidate C: 25%

Here, A and B go into a runoff. It turns out that candidate C voters split such that the final result is:

Candidate B: 50.001%
Candidate A: 49.999%

Now we're locked in endless recounts as Candidate A fights to win and everyone is saying how nice it would be if we had just used a plurality system as then the result would have been clear in this case.

More likely the Candidate C voters don't show up for the runoff.

Along with candidate A and candidate B voters who also don't show up since it's a run-off. It's also likely that the result is not razor-thin close. The point is that you are going to see that sort of thing happen occasionally, just like we see really close elections happen now, even though they are unlikely.

[emailking]

The Minneapolis Star Tribune has done their own analysis of the ballots and is predicting Franken to win by 174 votes: http://senaterecount.startribune.com/ballots/

It was 193 when I clicked. Isn't this more of a group conclusion from (possibly biased) readers who go through them all, rather than an analysis by the paper?

[emailking]

What was Nate Silver's prediction, again?

It depends when you asked him.

I still don't understand the idea of objecting to dynamic predictions.  I think that ignoring incoming evidence in favor of maintaining a pre-existing decision is pretty much responsible for half the bad in politics these days.

Ok, I'll hold off until the final count is certified and when it's not the exact number that His Highness predicted, we'll hear what excuses His followers have for me.

The excuse is that you are misrepresenting his prediction. He put large error bars on the 27 votes. Go read that post. It's there. You are acting like the man on the street who sees a poll result +2 and thinks the pollster got it wrong when it ends up -1.

[emailking]

I tend to doubt that since I am not sure anyone knows the vote patterns of absentees "wrongly" rejected.

I think the assumption most people are going on is that they will have a slight Franken edge since (according to the polling) the absentees in general had a slight Franken edge and the wrongful rejection aspect of it shouldn't favor one side over the other since it's a clerical error. But you're correct in that no one actually knows.

[emailking]

I think he should still push for it. Better to count all the votes and lose because of it and lose than to advocate not counting perfectly valid votes...which is what he would be doing if he backs off on the issue at this point.

[emailking]

What Coleman is referring to is the 113 ballots that simply went missing in a Minneapolis precinct. At first it was assumed those ballots were initially just counted twice, but then it was proven the numbers don't add up. So as strange as it sounds, those ballots did simply go missing. The Canvassing Board has basically agreed to just use the original count in that precinct. Coleman however continues to cling to the argument they were double counted. His arguments really are similar to J. J.'s Bradley Effect arguments and just as taken seriously by anyone besides his campaign, but it's his last argument.

I don't think this is correct. It's 2 different issues. Some ballots that can't go through the machines (because they're crumpled or whatever) or are absentee ballots on normal paper (not thick enough) are copied and the copies are sent through the machines. Coleman is arguing over 100 of these were counted twice in the hand recount, once for the original ballot and once for the copy that was sent through the machine on election night.

[emailking]

if the Illuminati were real, they would have ensured that this ballot was counted for Franken.

[emailking]

He NEVER predicted a Jim Martin victory, on election night or otherwise, WTF?

Correct. As I recall he thought it would be close with Obama turnout and said he wouldn't be surprised if Martin won. That's not predicting a Martin victory.

[emailking]

Amusingly, Nate Silver's track record is only good when it comes to predicting Democratic victories.

That doesn't even make any sense. If he predicts Republican victories and tends to get them wrong, he isn't good at predicting democratic victories because he missed many.