Gubernatorial Vote Count Post-Mortem

* Congratulations to Dino Rossi and his excellent campaign team!

* It was a big day for Sound Politics too, with nearly 35,000 page views today. These came from 5,400 different IP addresses. The winner is Comcast user 67.161.117.128, who apparently hit the reload button 275 times today!

We're honored to have so many new readers. Please bookmark the site and visit us again soon.

* How were our projections? This table looks at the projections for today's count that I posted last night vs. the actual outcomes.

	Projections			Actual
Group	Estimated Uncounted Votes	Approx. % Lead	Approx. Projected Vote Lead	Counted Votes	Percentage Lead	Vote Lead
Rossi Counties	3,400	+23%	+775	4,244	+19.8%	+840
Gregoire Counties (Excl. King)	1,700	-3%	-45	2,537	-5.99%	-152
King County (non-affidavit)	800	-18%	-144	1,359	32.82%	-446
King County (affidavit)	650		-550
Advantage Rossi			+36			+242

My net projection was a +730 lead for Rossi outside of King County, and we ended up +688. Not bad, I'd say, especially given the inaccuracy in the counties' estimates of uncounted ballots.

* On Monday, the day the AP story appeared in the P-I and other newspapers, I was interviewed on two different radio stations (KVI and KIRO) and a TV station (KIRO) predicting that Rossi would win by "between 1,700 and 3,400" votes. Of course, the final outcome was only 261 votes.

But my prediction was better than it might first appear. It was based on the information that King County had 11,000 uncounted ballots and on the assumption that those ballots would give Gregoire a lead in the range of 17% - 26%. Later on Monday, we learned that King had many more uncounted ballots than they earlier estimated. In fact, King actually reported a total of 19,642 votes this week, which went to Gregoire with roughly a 21% lead. Applying this same lead to the unexpected discovery of 8,642 extra votes, gives us a discrepancy of 1,807 extra Gregoire votes. If you adjust the final outcome for this discrepancy, you arrive at a Rossi lead of 2,068. This is right in the ballpark of my earlier prediction based on available data.

An even better way of looking at this would be to simply plug the 19,642 into the original spreadsheet instead of the 11,000. The model would have then projected a Rossi victory with a margin of between 157 and 1,098 votes.

* How did the Gregoire people do?

TNT, Nov. 12

Gregoire campaign spokesman Morton Brilliant said, “Our model predicts Chris winning.”

I'm relieved that the party that doesn't know how to count won't be in charge of the state's finances for the next four years. Posted by Stefan Sharkansky at November 17, 2004 11:44 PM | Email This