« The Arnebeck mystery solved | Main | 30,000 hits »

December 13, 2004

Recount in Washington

Time once again for me to dissent from the Republicans with whom I usually ally. Let none accuse me of blind partisanship.

So as political junkies far and wide know, the state of Washington is currently undergoing an unprecedented statewide manual recount of an election... the gubernatorial race, to be specific, which Republican Dino Rossi led after the initial tally by 261 votes, and by 42 votes after the legally mandated machine recount. State law permits a candidate to demand a manual recount, either statewide or partial, provided he's willing to pay for it. If the result of the recount changes the outcome, the money is refunded (and the tab is picked up by the state), and if the recount was partial the rest of the state is manually recounted, also on the state's dime. Rossi's opponent, Christine Gregoire, demanded a statewide manual recount and forked over the dough.

Republicans have been howling bloody murder and demanding that Gregoire concede. They're infuriated about the recount. I am not.

Look, imagine the shoe on the other foot. Suppose Gregoire was the one who led by 261 votes, and Gregoire's lead dropped to 42 after the machine recount. Would you, Republicans, want a recount in those circumstances? I know I sure as hell would, and I'm betting Rossi would too. In fact, if Rossi conceded in that case, I'd be upset with him.

So I can't be mad at Gregoire for calling for the recount. It's her legal right, and it's what I'd do if I were in her position. And honestly, I find the brouhaha from the Republicans to be a bit embarrassing.

Republicans have also been lambasting Gregoire for calling the race a "tie". Stefan Sharkansky over at Sound Politics, who's been doing a fantastic job covering the race, has been particularly harsh. Stefan writes:

Christine Gregoire continues to insult the intelligence of Washington's voters:

"I've said all along, 42 votes out of 2.9 million is literally a tie," Gregoire told The Associated Press on Friday.

No, 42 votes is a victory as is 261 votes.

Sorry, Stefan, I love your site and I respect you, but I have to disagree.

Yes, 42 votes is as much a victory as a million votes. And if an omniscient being examined the ballots and declared a 42 vote margin of victory, I'd accept it as being perfectly legitimate. The problem is that we don't have omniscient vote counters. We have vote-counting machines, and vote-counting humans. Machines occasionally make errors, and humans do so with somewhat greater frequency. That's why each count has produced a different result.

Vote counting is a statistical process. Each ballot has a certain (slim) chance of being mistallied. It's very likely that the officially certified outcome of any race is not the exact content of the ballots. We're back to normal distributions, with the curve centered on the certified result. For most races, this doesn't matter at all... the bell curve tapers off to insignificance well within the margin of victory. But a margin of 42 votes out of over 2.8 million cast... well, the area under the curve to the left of X-42 is too large to say with any reasonable degree of certainty that the outcome is known.

Look at it this way. 1,372,484 votes were tallied for Rossi, 1,372,442 were tallied for Gregoire. If each ballot had an, oh, 0.05% chance of having been mistallied, what is the probability that Gregoire was the winner? Let's see, the sum of C(1,372,442, N) as N goes from 0 to 1,372,442, multiplied by 0.0005 to the power N and 0.9995 to the power (1,372,442 - N), carry the one... shoot, I don't remember how to do this. You're good with this sort of analysis, Stefan, you tell me. But I'm guessing it's relatively high... below 50% certainly, but probably higher than 10%. Statistically speaking, this race is a tie. Gregoire is right.

And at this point, I think she's likely to win. As of tonight, Gregoire has gained one ballot in the manual recount for every 1,610 tallied for her, while Rossi's gained one ballot for every 1,672 tallied for him. And there are more Gregoire ballots left to tally: 1,140,279 for her, only 1,054,194 for him. If the trend holds, Gregoire can expect to pick up 703 more votes, while Rossi will only pick up 630. Added to the gains they've already made, that's a net 852 for Gregoire, 820 for Rossi, a swing of 32 votes in Gregoire's favor, which would cut Rossi's lead to only 10 votes.

A lead that small will be overwhelmed by the 561 uncounted absentees from King. And yes: if those are valid votes, if real registered voters really cast them, and if they were really rejected because of an error by the county, they really should be counted. Even if it means Rossi loses. And I hope the GOP doesn't fight to keep them uncounted. I'd rather be honest than victorious. Wouldn't you?

And yes, I know that as attorney general, Gregoire opined that valid votes that were nevertheless untabulated in the initial canvass were not eligible to be included in recounts. So she's a first-class hypocrite. That's no surprise. But it's her earlier opinion that was wrong, not her current stance.

I've heard Rossi fans crying fraud. But I haven't seen any evidence. If you have some, please use it. I'd love to see Rossi pull this thing out. But if you don't, quit hurling unfounded accusations. Or go to DU... you'll fit right in.

December 13, 2004 in Election '04 | Permalink

TrackBack

TrackBack URL for this entry:
https://www.typepad.com/services/trackback/6a00d834559cb569e200d834362b5353ef

Listed below are links to weblogs that reference Recount in Washington:

Comments

I agree with your assessment of the WA situation.

BTW, if you wish to see a DU-lite site I posted the URL above. It is sometimes more comical than DU is, if that's possible.

Posted by: SirEverlast | Dec 14, 2004 10:01:07 AM

The comments to this entry are closed.