WA: Prediction on King County results, with cool spreadsheet!

[merwin]

OK, so there's one county left to count in the Gregoire/Rossi race for Gov of WA. King county- Gregoire's stronghold.

I took the liberty of collecting the data off of the WA government's detailed voting website and matched counties' manual count vs machine count and added in the manufacturer and equipment used to count the votes.

Guess what? The manufacturer of all 3 counties that had extra votes "found" during the recount process is ES&S (Go figure). And to go a bit further, they were all the EXACT same machine types (ES&S OpTech 4C-400). Snohomish County also used a Sequoia system in some places according to the voting system list. Oh, and did I mention that these were the only counties to be using the OpTech 4C-400 machines?

King County, the only one left on the recount, oddly enough uses ES&S as well (under the name Global Election Systems.... which I believe was bought by ES&S?). Based on the current results, I predict that Gregoire will pick up an additional 200-250 net votes out of the recount.

My prediction of 200-250 net votes goes out the window when you start speculating patches on the OpTech 4C-400 machines though, since King County isn't using that exact machine type. The only thing to do is to wait and see I guess.

Anyways, here's the spreadsheet (and fodder for the tin-foils out there :-) )
2004 WA Recount Results (Minus King County)

Sources:
Voting Systems
Recount Results by County

[claudiajean]

1. Sorry to dispute your results...

...I can see you've done a lot of great number crunching.

But your base premise for King is flawed.

King County uses the AccuVote-OS scanning devices and GEMS tabulation software, both originally manufactured by Global Election Systems. Global was bought by Diebold in January of 2002. There is no connection to ES&S and the systems are very different.

It is an optical scan system, but not a connect the arrow system such as ES&S uses. It is a fill in the bubble system, such as used for SAT's.

Also, I would not recommend any expectation of gaining a great number of votes in King. The additional votes from the machine count to the hand count in the other counties' tallies are due to further inspection of the ballots for the manual recount (most counties did their first machine recount without a re-inspection of the ballots.)

King, however, inspected all of their ballots for voter intent problems before running the ballots for the machine recount. So it is unlikely that many more votes will be harvested in King as a part of the manual recount.

Basically, IMHO, if we don't get to count the 723 ballots that King County Elections screwed up and missed in the first and second counts, we will not be able to bridge the 50 vote difference between Rossi and Gregoire now.

:(

edited for spelling

[merwin]

3. With all that's gone on in King County though...

Out of curiosity, do you have information on where these extra votes came from? I haven't been able to find information on that.

Having the ballots inspected quickly before a machine recount, and making people agree 100% on voter intent are two different things.

Also, to go further into the tin foil land, manipulating the GEMS system isn't particularly hard, and a hand recount would point out any flaws there as well.

But, as I said before, the only thing we can do is wait and see...

[claudiajean]

5. The ballots are absentees, still in their original mailing envelopes...

...unopened.

They were kicked out during the signature verification process becasue the voter had no signature on file in the computer system.

Not the voter's fault -- the computer system was upgraded this year and some data was screwed up, including some voter signatures.

What the procedure was supposed to be was that if the voter's comparison signature is not present on the computer, then the election staff was supposed to go to the original voter registration cards and check the signature on the paper copy.

That never happened. For whatever reason, and it's appearing to be an erroneous order from the inexperienced and frankly unqualified superintendent of elections, these ballots were never checked further and were noted on the computer records as "do not tally".

So they were never even opened, let alone tallied. They have never even been out of their original absentee ballot envelopes.

[merwin]

7. Nonono, not the 723 extra King county votes,

but the extra votes that are showing up in manual recounts?

[claudiajean]

8. Ooooohhhhh...the "found" votes...

Prinicpally, it's a human error problem.

A lot of voters don't understand how an optical scan ballot works (regardless of how much voter education is in place), and they mark their ballots oddly, so the scanner can't read them.

Most often, they circle the candidate instead of connecting the arrow or coloring in the bubble, or they inexplicably write in the name of a candidate on the ballot, rather than just voting for them.

In a manual recount, these "voter intent" marks become obvious and can be counted, although the machine didn't read them. That's why a hand count is always more accurate than a machine count.

SO, that's why not to expect more votes from kIng. Before their machine recount, they went through all of their ballots and found voter intent markings that couldn't be read by the machine, and enhanced the ballot so that it could be read by the machine, and made their count more accurate.

The other counties didn't do that before the machine recount, so that's why they are still finding high numbers of votes.

[merwin]

9. Well, let's hope they did as good of a job fixing voter intent

as they did with figuring out valid absentee ballots :-)

[claudiajean]

10. LOL! That may be the saving factor!

King County incompetence!

:)

[BlueDog2u]

13. Thanks for the information

In an earlier thread I also calculated that, extrapolating from Snohomish and Pierce, Gregoire would win by approximately 1-300 votes. I didn't calculate for the difference in voting technology. I guess its up to the courts to decide --once again -- whether all the votes will be counted.

[claudiajean]

2. If you want to tweak your formula to re-calculate King...

...then use Chelan, San Juan and Klickitat Counties to create your projected change formula.

These three counties use exactly the same balloting system as King.

[merwin]

4. That's not really possible though, since

those counties have an average of about 15,000 people, so you can't really make a decent comparison with the almost 900,000 votes in King

[claudiajean]

6. Actually, although the total numbers are smaller,

the error and mis-read rates from 1998 - 2003 were statistically similar in each of the 4 Washington GEMS counties for this equipment.

The bottom line is that 1 in 10,000 ballots are mis-read in the Accu-Votes. That's an 88 vote mis-read rate for the number of ballots they have in King for this recount.

Assuming with a 60-40 split in the Rossi-Gregoire vote in King, that's 53 "found" votes for Gregoire, and 35 for Rossi, meaning a net gain of 18 for Gregoire.

Rossi is ahead by 50 right now.

So Gregoire still loses by 32 votes in my analysis.


Edited becasue it didn't make sense even to me the first time....

[chapel hill dem]

11. Is King County Finished?

I checked the SoS site and it appeared that Gregoire won the overall election by about 200 votes (mental math/pre-coffee). I cannot get the website to come back up again. I searched DU and AP and found nothing.

Does anyone know any more about this?

http://www.vote.wa.gov/general/recount.aspx

[skids]

12. You must have been looking at the machine recount, not the hand n/t

[chapel hill dem]

16. Early morning recount page

Well, I had the correct page bookmarked for the past week, so maybe they were working on the server at 5 AM (8 AM EST) when I checked. Also, the numbers were different than the machine recount page. Maybe they were loading a test set for King County and I happened to be on their server before they shut it out?

Who know, supposedly this will be settled by the 23rd except for the lawsuits...

[rjbny62]

14. Gregorie is picking up votes at a higher rate, wins w/o the found votes

As of today, with all counties reporting except King, Gregorie has picked up votes at a higher rate than Rossi (.065% vs .056%). So if this rate continues, Gregorie will win by about 82 votes, and that is without counting the 700+ found ballots that the repubs don't want counted.

Machine Recount Total - Gregorie 1,372,442 Rossi 1,372,484
Manual Recount (as of 12/18) - Gregorie 867,167 Rossi 1,021,926
Machine Recount King Total - Gregorie 505,836 Rossi 351,127

Gained Votes (as of 12/18) - Gregorie 561 Rossi 569
Rate of Gained Votes - Gregorie 0.065% Rossi 0.056%

Projected Total Gain - Gregorie 888 Rossi 764

Difference after Machine Recount - Rossi +42
Difference after Hand Count - Gregorie +82

[weeve]

15. Then don't count Rossi's additional "found" votes !

Apparantly Rossi picked up some votes that were found after the fact ( can't recall which Eastern Washington counties ) early in the recount. Why should these be counted , and NOT the ones found in King Co. ??? Yet again, utter hipocracy.

[merwin]

17. To save the democratic process (at least in WA), those 700 MUST count n/t