December 09, 2004
Recount by the numbers

Here are some interesting numbers to mull over:

Counties reported: 10, representing ~4% of the statewide vote
Rossi's lead over Gregoire in the counties that have reported: +683, or a margin of +0.6686%
Net new two-party votes since the machine recount: +57, an increase of 0.06%

In the remaining 29 counties, Gregoire led the machine recount by 640 votes, or a margin of 0.0242%.

In order for Gregoire to win (or tie), she would need to increase that lead of 640 votes to 683 votes, a 6.7% increase. Simply put, either she has to get the same 0.0242% margin on a 6.7% larger pool of votes than there were in the machine recount, or she needs to get a 6.7% larger margin on the same number of ballots. To grow the unrecounted pool by 6.7% requires another 178,000 ballots. That's a little optimistic, even for Paul Berendt, who thinks there are 15,000 valid ballots in the REJECT pile. The reporting counties only grew their pool of two-party ballots by 0.06%, which would translate into 1,475 such ballots from the remaining counties. The machine recount produced 2,300 new ballots in those counties, so you never know. In any event. It's still a small enough increase that she would still need to increase her margin by essentially the same 6.7% to 0.0258%.

Here's where we do some of the math that will only produce horselaughs from those who try to make statistical arguments without using any actual numbers or formulas.

Gregoire's lead in the 29 remaining counties in the first count was 0.0163%. If you look at her leads in the two counts: (0.0163%, 0.0242%) you can calculate sample mean, standard error and a t-statistic of 1.41. This tells us that there is slightly less than a 20% chance for the next count to give Gregoire a lead of 0.0258% or greater. i.e, based on the information we have available today, we can be 80% confident that Rossi will win the 3rd recount.

Of course, this number will be revised (and can change significantly) as new information becomes available. A Supreme Court order to change the way that the recount is conducted could change this number by quite a lot. In the meantime, if you are a Democrat who is contemplating a job in the Gregoire administration, but you have another job offer that you have to decide on by today, you might not want to forfeit a 100% sure thing for a 20% long shot.

Posted by Stefan Sharkansky at December 09, 2004 09:54 PM | Email This
Comments
1. Dang, I love people who can work numbers like that.. Good going and thanks for the analysis. I understand it's all subject to change based on a lot of factors, but I appreciate the effort. I come to this blog for my info on the race now instead of anywhere else.

Posted by: Troy Colley on December 9, 2004 10:10 PM
2. very good article Stefan. Thank you for updating so well

Posted by: JP on December 9, 2004 10:23 PM
3. Very, very nice work, Stefan. I may steal your ideas in the near future. In the meantime, I've posted some of my own number crunching. If the trends continue (hah!), Rossi should increase his lead by 17 votes (give or take a lot).

Posted by: Timothy on December 9, 2004 10:26 PM
4. Thanks for the calculations Stefan.

I just checked over at ManureCentral. Goldy's got quite a treatise there to convince himself that there is no way that Gregoire was the one who wrote that memo to Munro.

But that's the beauty of the memo, while there's no way to prove that it was Gregoire, there is also no way to prove that it was not Gregoire. Goldy writes a lot about whether or not Gregoire herself would have handled this or maybe it was an underling.... blah x 3.

The bottom line is that those at the top of credible organizations take responsibility for their collective actions whether they are aware of those actions or not.

So, in the eyes of the typical Washingtonion, the possibility that it was Gregoire, coupled with her denial is enough to at least raise suspicion.

For a candidate that is in "do anything mode" to win this election, and who happens to be on the same side as the those lawyers who realize that her only real hope is to litigate, the damage is done.

Good luck shoring up public opinion Christine.

Posted by: Jeff B on December 9, 2004 11:36 PM
5. Nice calculations Stefan. But King County is the real wildcard. They added to their ballot count in the machine recount, without any obvious reason. 898,238 ballots increased by 336 additional ballots, for a total of 898,574 in the machine recount. No other county had this kind of unexplained deviation. The other counties that increased their ballot counts by significant amounts all found stacks of ballots that had been overlooked on the first count.

If King County adds another 336 ballots on the manual count, this would be more than enough to overcome Rossi's 42 or so vote lead, since Gregoire is winning in King County by 58% to 40%.

On the other hand, if the extra 336 ballots resulted from some being run through the machine twice on the machine recount, this sort of error would be corrected on a manual recount. There will be stacks for each precinct, and probably they won't end up being counted twice.

So if 336 ballots disappear on the manual recount in King County due to this error being corrected, then that would result in somewhere around 60 net votes in Rossi's favor.

It would be nice to have a list of every single person in King County who voted on November 2, by precinct and type of ballot. Those numbers could be added up, and we would know whether extra ballots have been wrongfully inserted, or whether ballots might actually be missing.

Posted by: Richard Pope on December 9, 2004 11:47 PM
6. Stefan--I appreciate all you folks having fun with these numbers...predictions based on past events & certain underlying assumptions are harmless. But when you take a look at the small margin of difference and the human error factor, which differs by County due to ballot type, prior ballot review etc. etc., it's pretty futile. I am a lowly bean-counter who has worked with plenty of actuaries and frankly...they laugh at some of this. But it is fun...and harmless..and helps relieve some stress...at least while you are number-crunching.
Goldy is oh so serious & oh so self-righteous. He demands perfection in this election. I'll bet he would be mighty tough to live with...kind of the "Felix Unger" of bloggers!!! Why I'll bet Goldy's toilet could never be clean enough for him.

Posted by: Mr. Cynical on December 10, 2004 12:55 AM
7. It all comes down to King County, (where the Demos have the real power to affect things). I am still 80% confident that Gregoire, not Rossi, will be our next governor. The demos have too much to lose if Rossi gets it....

Posted by: Me on December 10, 2004 07:17 AM
8. An interesting point: if there is reason to believe that voter fraud is taking place it means that Gregoire as the current Attorney General and highest law enforcement officer in the state has a duty to investigate -- herself...

Posted by: Lew on December 10, 2004 07:30 AM
9. When is the WA Supreme Court supposed to hear/rule on whether or not previously rejected ballots should be reviewed again in the manual recount? I think the whole race basically boils down to this ruling. If they let them in, Gregoire probably pulls it out. If they don't, Rossi should prevail. Does anybody know legal precedent for this?

Posted by: Marc on December 10, 2004 08:26 AM
10. I believe they hear the case Monday at 1:30 pm PST.

Posted by: Jeff B on December 10, 2004 09:12 AM
11. Rossi picked up another vote in Columbia I think, +2 overall (30 - 28).

Posted by: Marc on December 10, 2004 09:31 AM
12. Holy Cow. Kitsap came up with a ton of votes and they use Optical Scan equipment. The only thing I can think happened is that they did not use the King County method of "enhancing" ballots on the 2nd count and that is where these "extra" votes came from. Just in Kitsap is Rossi +82 and Gregorie +72 out of the 115,000 or so ballots counted. Anybody else surprised by this?

Posted by: Marc on December 10, 2004 09:53 AM
13. Kitsap came in and "found" 154 new votes. Where did these come from?
Net +10 for Rossi, but at this rate King will "find" over 1000 new votes.
Rossi now up 54, but it all will come down to King

Posted by: zorkpolitics on December 10, 2004 09:56 AM
14. So far Dino is gaining votes in counties where he lost votes in the first recount. Not sure why yet but it is an encouraging pattern.

Posted by: Chris Vance on December 10, 2004 10:02 AM
15. Some data on ballots counted:

1st 2nd 3rd
Columbia 2,144 2,144 2,145 (+1)
Ferry 3,385 3,385 3,385
Garfield 1,331 1,331 1,331
Grays Har 28,256 28,256 28,256
Jefferson 18,772 18,772 18,772
Kitsap 119,458 119,459 119,443 (-15)
Lincoln 5,900 5,900 5,900
Mason 25,836 25,839 25,839
Pacific 10,620 10,621 10,621
Pend O. 6,262 6,114 6,268 (+6)
Skamania 5,193 5,193 5,193
Wahkiakum 2,255 2,255 2,255

King 898,238 898,574 xxxxxxx (+336)

The one that sticks out is Pend Oreille. Can that be right? The number of ballots counted from the machine recount to the manual recount changed by over 150 ballots. Kitsap lost 15 ballots?

Posted by: Marc on December 10, 2004 10:14 AM
16. The State Supreme Court does not hear the case until Monday. It seems a few of the justices are indisposed while they hear a 'mock' case regarding the mid-19th century hanging of Chief Leschi. Not that there's anything pending of great importance to the state or anything like that...

Posted by: chunkstyle on December 10, 2004 10:26 AM
17. Shnomish and Yakima counties are using the exact
same methodology in the manual recount as in the
original recount. Expect a variation of near
zero. Likewise, King County offered Gregoire a
manual recount gratis. It should have a near
zero variation in the "manual" recount.

The hanging chads will be in Rossi counties.

If the rules aren't changed,...

Posted by: Dave on December 10, 2004 10:38 AM
18. well well now that mr vance has decided grace us with his presence on the blog. I think a few facts
about the upcoming gop state chair race are in order.


I want everyone here to listen this:Mark hulst
is still running for state chair.chris vance be
afraid be very afraid!!ladies and gentleman
last saturday night i had the good fortune of
attending the seattle republican women's christmas
party.Dino rossi was there and he and i had great
talk. for those you who have had the opportunity
to meet him he is real deal.

anyway during the course of conversation
mark hulst's name came up.Dino informed
me that mark infact was not going to run.
I thought if know dino would.so based
on that information i told a number of
people mark was not running.well i finally
got hold of mark and got the good news that
he still is.It seems what is going on here
is Mr vance told members of dino rossi's
staff that he was the only one running.
that everyone else dropped out.

for the record there are 3 people that are
also running.this should tell all of you what
kind of guy vance is. that he use dino rossi
to perputrate this nonsense says it all.

Posted by: phil spackman on December 10, 2004 10:43 AM
19. I for one am glad that Chris is posting his observations here and besides the subject of this thread has to do with the recount and not the race for state chair. I come here to get the latest info on the recount.

Posted by: Terry Mueller on December 10, 2004 10:51 AM
20. Pend Oreille County does not concern me. They obviously did not "lose" any ballots on the machine recount, even though the number reported to the Secretary of State appears to be 148 lower. Data for the three counts in order: original 6,262; machine 6,114; manual 6,268

It appears on the machine recount that Pend Oreille did not actually report the physical number of ballots counted, but only the number of ballots which contained a vote for Governor (including write-ins). On the manual recount, this was corrected, and now includes all ballots (votes for candidates, votes for write-ins, blank votes, and overvotes).

Douglas and Whitman Counties committed the same error in reporting their machine count totals. Each county showed a loss of several hundred, but almost no change in the number of votes for each candidate. Again, undervotes and overvotes were omitted from the total ballots reported on the machine count.

The loss of 15 or 16 ballots in Kitsap County is interesting. No logical explanation, given that the first two counts, both by machine, were only one ballot difference: original 119,458; machine 119,459; manual 119,443

Maybe when you have a large county, with lots of precincts and ballots, you have to sort those all out by precinct -- which is a real pain for absentees and provisionals, which can come from all over the county. Kitsap County has 185 precincts, by far the largest of any county reporting so far. Maybe in this sorting process, ballots somehow get "lost". If the people involved are honest, then ballots could not get "created".

If ballots get "lost" in the sorting process due to county size, this bodes very well for Rossi. King County has many times more precincts than Kitsap: 2,616 versus 185. Also, most King County precincts are at consolidated polling locations, where ballots of several precincts went into one counting machine, and now have to be sorted. Kitsap County has larger precincts, both geographically and by number of voters, and probably most polling locations had only a single precinct, and those poll ballots did not need to be sorted.

Posted by: Richard Pope on December 10, 2004 11:04 AM
21. Dino has picked up 11 votes in Walla Walla. We also expect a net gain of 6 in Asotin. Net gain of 24 votes. So far, so good!

Posted by: Chris Vance on December 10, 2004 11:24 AM
22. Interesting letter in the Seattle Times today



I voted in the Nov. 2 election. My vote was included in neither the first count nor the second. The Seattle Times now argues that my vote should not be included in the manual recount.


I am a registered permanent absentee voter in Whatcom County. Prior to the election, I lost my absentee ballot. Therefore, I voted at my precinct on Election Day. Because my name appeared on the list of absentee voters, I was required to vote a provisional ballot.


My provisional ballot was incorrectly mailed from Whatcom County to King County, where I was formerly registered to vote. There, my ballot was rejected because my King County registration was canceled when I moved to Whatcom County. I would like my ballot to be returned to Whatcom County for inclusion in the manual recount. Sam Reed, our secretary of state, has ordered that previously rejected votes such as mine are not to be considered in the manual recount.


Anyone who believes in democracy should agree that votes like mine need to be counted in order to determine the true outcome of the election. The Seattle Times editorial page argues that I should remain disenfranchised in the interest of convenience. I beg to differ.


Now why should his vote be disallowed because of the fault of the county?

Posted by: DustinJames on December 10, 2004 11:28 AM
23. This might be the kind of mistake that can and should be corrected, but the Democrats want to go much farther. They want to compel every county to look at every ballot that was rejected, and they want the right to observe and object as that is being done! It would be a chaotic circus, and it is not allowed by law.

Posted by: Chris Vance on December 10, 2004 11:35 AM
24. Sam Reed was interviewed yesterday on TVW yesterday. The 60 minute tape is worth listening to. Here's the link: http://www.tvw.org/search/eventResults.cfm?EvntType=I

He pointed out this is the closest governor's race in US history in percentage terms; only one 1839 race in Massachusetts was closer in numerical terms was closer.

He thinks that the Democrats will lose the suit before the Supreme Court next week because the canvassing boards and county auditors do a good job and there should is no reason to reopen the consideration of ballots.

As far as the charge that Gregoire was involved in an opinion in 1996 that she's now opposing: the Secy of State's office issued an opinion in 1990 when there was a close race that new votes cannot be considered in recounts. That opinion was reissued in 1996 when there was another close race. It's standard practice that the Secy of State's office would discuss that with the Asst. Atty General. It's likely Gregoire was personally uninvolved.

He expects the losing party will resort to litigation, which he feels is unfortunate since Washington has one of the best systems in the nation.

Posted by: Terry Mueller on December 10, 2004 11:36 AM
25. While I certainly hope Rossi wins and don't doubt the expert analysis, what about these numbers.....

So far they have counted roughly 250,000 votes and there has been a net shift of 18 votes (Rossi has gained 18 overall).

That is a shift of roughly .0072%

There are roughly 750,000 votes in King County, a similar shift of .0072% would be 54 votes net.

Assuming that shift is for Gregoire, it could make the race tight. Admittedly right now Rossi would still win (42 votes + 18 votes - 54 votes is a 6 vote lead). But is it really 80% for Rossi at this point ?

Just wondering

Posted by: Pat on December 10, 2004 11:41 AM
26. So far 14 counties have reported results. In the machine count 193 new votes were found and 30 lost, in the manual count 225 votes were found and 9 lost. In 9 counties the net change has been better for Rossi (either he gained more or lost less), in 1 county there was no net change either time, and in only 4 counties has Gregoire done better in the manual recount. If this trend continues it would obviously be good for Rossi.

Posted by: zorkpolitics on December 10, 2004 11:49 AM
27. Phil Sprackman. Pardon me if I'm a bit dubious of a person posting something so far afield of the subject at hand. Your first amendment right is intact, but audience turned off with your rudeness. Shame on you for talking internal party stuff inappopriately.

Besides, I just don't cotton to folks who can't be bothered to capitalize or put at least one space after a period.

It really biases me against your horses--which could be good folks.

Shabby man, shabby.

Posted by: Matt on December 10, 2004 12:00 PM
28. Matt

my apologies for the way it was written.
I got little to carried away and left a
few words out.


I will not take back what I said.
my reason for puting it in this forum
was and is because chris vance posted
here.I don't like it when dino is
used this way.I also take exception
when things are being said about
other people for no other reason
than political gain. I called someone
on it everytime I see it happen.

Posted by: matt on December 10, 2004 12:11 PM
29. First it's "Phil Sprackman" posting, then the poster changes to "matt". I'm confused! Press the right buttons, man. Your message will be accepted a little better.

I personally welcome Chris Vance here as well, and I was glad to get an email directly from him when I contacted him to see if they needed King County observers. Both were good personal efforts to represent us. When I emailed Paul, one of his VERY liberal and angry assistants replied back.

As for the guy who voted by provisional legitimately in Whatcom, for every one of those, there's a lot more that can abuse the system. I don't think anyone here advocates taking away his vote. But he's the exception, and if giving him his vote results in 200 duplicate votes from illegal voters, that's not what should be done. I too would love to see the list of every voter published. If Christine somehow pulls this off, and that list shows a few hundred illegal provisionals from King, I'd consider that the best reason as I've seen to recall her (and I've seen lots of reasons so far).

Posted by: Chris on December 10, 2004 01:01 PM
30. "This might be the kind of mistake that can and should be corrected, but the Democrats want to go much farther."

So then the question becomes, how do we correct problems like these without getting into that legal recanvassing issue?

Posted by: DustinJames on December 10, 2004 01:42 PM
31. By having the Democrats, who always pose as the saviors of the underdog, to take some of those out-of-state doubloons and dinars and shekels from the Kerry campaign and the DNC and - set up classes to teach their pet underdogs first to read, and then to follow instructions, and then to fill out forms (say for instance ballots) correctly. And the "registered permanent absentee" could do the body politic a great favor by showing up at the poll and voting with a real ballot, thereby saving the post office and two counties worth of precinct workers untold work to fix his own careless mistake.

The fake ideal that every piece of paper must be flawlessly handled, no mater how screwed up by the would-be voter, doesn't fit the real world and can't be expected to be honored as an excuse to stop the whole works until the confusion gets smoothed out. Almost perfect (the current system) is good enough in human terms, and those precinct workers deserve our thanks and confidence.

Posted by: Insufficiently Sensitive on December 10, 2004 05:58 PM
32. Insufficiently Sensitive,

Did you take the time to read his letter before hitting the reply button?

You said:

And the "registered permanent absentee" could do the body politic a great favor by showing up at the poll and voting with a real ballot...

He said:

I lost my absentee ballot. Therefore, I voted at my precinct on Election Day. Because my name appeared on the list of absentee voters, I was required to vote a provisional ballot.


You said:

...thereby saving the post office and two counties worth of precinct workers untold work to fix his own careless mistake.

He said:

My provisional ballot was incorrectly mailed from Whatcom County to King County, where I was formerly registered to vote. There, my ballot was rejected because my King County registration was canceled when I moved to Whatcom County.

The provisional ballot was mailed by the pollworkers, not the voter.

Every legal vote not counted really isn't all a big democratic conspiracy, no matter how much you'd like to make it so...

Note that the voter does not say exactly how he voted, either for Dino, Chris, Ruth, or someone else...

So back to the question, how does this voter get his vote counted if recanvassing can't be allowed to happen, as supported by the pubs in their brief?

Posted by: DustinJames on December 10, 2004 07:03 PM
33. I agree with Mr. Cynical. Crunching these numbers
looks like harmless fun, but the assumptions required
make the claimed results meaningless. Different
counties *cough*KING*cough" have a much greater
propensity for (biased) humans to find new ballots,
reinterpret old ones, and so forth - defininetly not
the random process required for use of a t-test.

I remember asking one of my professors (someone
actually relatively famous in the field) about how
well a certain theory worked in the real world. He
replied (getting a good laugh from the class) "don't
give me any of that 'real world' s***."

The outcome depends on the ability of partisans in
King county to manufacture new votes. Period.

Posted by: Anon on December 10, 2004 07:42 PM
34. Anon,

Agreed! However, some of them will also be imprisoned in the next 18 to 24 months. (Guaranteed)! That MAY change a few crooks minds. They are being watched closely.

Posted by: leaddog2 on December 10, 2004 11:54 PM
Post a comment
Name:


Email Address:


URL:


Comments:


Remember info?