To answer Micajah's question posed in a comment on the earlier post about King County's Batch Slip Spreadsheet [1MB zip]:
It looks as though the number of absentee ballots returned was something more than 583,144 (571,284 from the "original count" column and 11,860 from those 65 batches which have no number in the "original count" column).I'm told that RB means "re-batch". i.e. if somebody returns an absentee ballot but, say, forgets to sign it, they'll be notified of the problem and offered an opportunity to correct it. If they cure the problem in time and resubmit the ballot, it will be put into a new batch with similarly resubmitted ballots, a "re-batch". The way the spreadsheet tries to account for this is pretty bogus. But if you treat all batches other than RB as initial submissions and the RB batches as resubmissions, you should come up with this:I wonder if anyone knows where those "RB" batches came from.
| Ballots Initially Returned |
569,884
|
| Ballots initially rejected |
(18,914)
|
| Ballots resubmitted, accepted and rebatched |
13,260
|
| Net Ballots Accepted |
564,230
|
Posted by Stefan Sharkansky at April 14, 2005
05:45 PM | Email This
These sheets are screwed up. I started by sorting the main page on suspicious fields. Sort by comments first.
Found 2 IH-3203 entries - wonder why that is and why one is anonymous???
Look at IH-3200. 95 ballots accepted, 73 went into tabulation????
Sorting on suspect fields is definitely the way to go.
But since I do not understand what in the world these columns mean I might just be randomly sorting data. One thing I would like to know is the relationship between the "Hand Count?" column and the final numbers.
Posted by: AJSTRATA on April 14, 2005 06:46 PMThen on the "Daily Report Total" the whites total up to 563,990?? Are these the numbers that caught your attention early on?
Posted by: AJSTRATA on April 14, 2005 07:02 PMLook at RB 3366. It appears twice: 44 accepted and 45 accepted. The second entry puts 45 ballots into the 564,230 column, but nowhere else, so far as I can tell. The first entry involves ballots tabulated over in the columns to the right, while the second entry only adds 45 ballots to the column it's in.
Would that mean that the number of ballots accepted was 564,230 - 45 = 564,185?
If so, I wonder how they managed to count 564,222 absentee ballots for their initial Nov. 17th certification.
Posted by: Micajah on April 14, 2005 07:28 PMNot a problem. Nothing but professionals and experts handle elections in King County. I'm sure they resolved the situation created by the absence of those seals fully in accordance with the law. ;-)
Can I sue King County for pain and suffering? ;)
It's obvious in the *ommission* of explanation for these ballot number descrepancies...that they were conducting fraud...
Posted by: Deborah on April 14, 2005 10:10 PMThe manual recount counted precisely (?) 564,448 absentee ballot forms. Right there, this is 218 more absentee ballot forms counted than were accepted.
Next, there were 292 people who either returned the wrong election ballot or multiple ballots. These folks were accepted and credited, but their ballots were not counted. So now we have 510 excess ballots.
Finally, there were 94 people whose absentee ballot envelopes were accepted and validated, but whose ballots were left in the envelopes and not counted. This makes 604 excess ballots.
If we look at voter crediting, there are 874 more ballots than voters credited with casting them, using the same method Stefan came up with.
So a logical interpretation of all this is (a) 270 people had their ballots validated and accepted, but somehow weren't credited, (b) maybe some of these 270 people should not have been validated, but who can tell?, and (c) 604 absentee ballots appeared out of nowhere -- maybe they were literally "duplicated", as in both the original AND duplicate getting tabulated.
By the way, Bill Huenekkens told me the only way to know how for sure (?!?) many absentee ballots have been tabulated is to count them by hand. While absentee ballots have a different visual appearance than poll ballots, they have exactly identical coding, and the AccuVote machine cannot tell them apart. The machine counts broke it down into poll, absentee, and provisional because the way the machine was programmed for each batch of ballots fed through it, and not because the machine could tell the difference between the ballot types.
When the initial count showed 564,222 absentee ballots, that actually meant that 564,222 ballots were tabulated when the machine was programmed to put the totals under absentee.
Same thing when the machine recount showed 564,420 absentee ballots. While new ballots could have been invented or added in, it could also have come from ballots being transferred from the poll to absentee stacks.
The only valid absentee ballot count was when they were each visually identified and separated in the manual recount, which result in 564,448 absentee ballots being counted (prior to the correction for the 566 "Larry Phillips" absentee ballots).
Of course, it was quite suspicious that 198 more absentee ballots were counted in the machine recount than in the original count, and another 28 were counted in the manual recount.
But the more valid comparison numbers would be the overall count -- with an increase of 336 for the machine recount, and a further increase of 59 for the manual recount.
While one can understand ballots getting transferred from one stack to another for many different reasons, it is simply not possible to understand why 395 more ballots overall were tabulated in the 12/23/2004 manual recount than in the 11/17/2004 original count.
Those 395 extra ballots truly came out of nowhere.
Posted by: Richard Pope on April 14, 2005 10:22 PMYeah, there did seem to be ~395 ballots that materialized after the initial certification. (Now that I've seen the sloppy records they created, I guess I'll have to get in the habit of putting a tilde in front of virtually every number expressing a conclusion about King County's elections to acknowledge that nothing can be known with certainty.)
That's what makes it so important to figure out what number of ballots they really had as of November 17.
I've looked again at their spreadsheet, and it surely does appear that batch "RB 3366" added 45 supposedly "accepted" absentee ballots to their total. By entering that batch twice in their spreadsheet (the second time only in the "accepted" column), they increased the number of supposedly valid absentee ballots to 564,230.
Their spreadsheet, then, indicates that they counted 8 fewer absentee ballots than had been accepted as valid -- but actually their numbers show they supposedly counted 564,222 when they only had 564,185 valid ballots that could be counted. (And, as you point out, they left 93 absentee ballots in their envelopes after verifying that they were valid and presumably including them in the 564,185 total. I don't know if the one emailed/faxed absentee ballot that they found was included in the 564,185.)
So, they seem to have always counted more absentee ballots than they had available to them -- the first time and the two recounts.
Add that to the mystery Stefan's trying to solve about the number of ballots counted in excess of voters known to have cast them, and it's a mess. There may have been ~649 more ballots counted than accepted and ~881 more ballots counted than voters who cast them.
If those two numbers get any closer together, we may have an "ah ha" moment.
Meanwhile, back at the polling places, ~866 more ballots were counted than there were signatures in the poll books -- and that figure is probably low because of the sloppy job they did with their summary of the "big binder" and with the "big binder" contents themselves.
Eventually, we may have totals from absentee ballots and from polling place ballots that approximate the difference between the number of voters credited with voting in the records created prior to certification and the number of ballots in the counts.
Posted by: Micajah on April 14, 2005 11:14 PMDate No Original otal batches % Batch not
Count counted counted
11/2/2004 12 26 46.15%
11/6/2004 5 36 13.89%
11/8/2004 6 31 19.35%
11/9/2004 7 14 50.00%
11/10/2004 8 21 38.10%
11/11/2004 13 17 76.47%
11/12/2004 2 8 25.00%
11/13/2004 1 8 12.50%
11/15/2004 5 7 71.43%
11/16/2004 8 10 80.00%
11/19/2004 1 1 100.00%
What makes this interesting every one of these no count for original ballots where RB-? But 7 of these batches did have original counts. I dont understand the differences in these batches but not a single RB batch had any ballots rejected. They were always counted. There may be a good reason for this but I dont know why from the data I see.
There's no good reason for omitting the "original count" from the batch slip, but King County doesn't need good reasons for what they do. Bad reasons do just fine for them.
They would virtually all be accepted, since the "re-batched" ballots were those which had failed the signature comparison. The only thing done is ask the voter to send a new signature sample into the elections office. Unless a forgery was done by someone who didn't know how to repeat the forged signature, the newly submitted signature sample could be expected to match.
That's one of the things that makes voting by mail so vulnerable to fraud. The person who submitted the mismatching signature got to change the official records to the new signature without ever setting foot in any office, much less proving who he was.
Posted by: Micajah on April 15, 2005 09:23 AMHow many voters are listed as being credited with voting a provisional ballot in the database you made available for searches?
I tried to make sense of the King County spreadsheet's provisional ballot data and came up with a couple of things that don't seem to make sense.
Their spreadsheet says they only accepted as valid 25,597 provisional ballots and counted the votes on only 25,764.
Does the list of voters you used to make that searchable database available have a total of voters credited with casting valid provisional ballots that is closer to the 27,641 or 28,010 ballots King County reported counting?
Also, when I looked to see if the figures being reported to the public from Nov. 10 through 17 (when they were reporting the numbers of provisional ballots counted) I couldn't get them to match at all. The provisional ballots shown on the spreadsheet as tabulated on Nov. 10, 12, 15, 16 and 17 are always different from the numbers reported to the public (and presumably to the canvassing board) on those days.
Recognizing that the spreadsheet probably contains data entry errors and other stuff that makes it less than perfect, what source of information was King County using to make reports to the canvassing board on the days when they canvassed and accepted absentee and provisional ballots for inclusion in the vote tabulation? If the spreadsheet comes even a little close to the counts reflected in their batch slips, then their reports to the public (and board?) were wildly off the mark.
Posted by: Micajah on April 16, 2005 09:10 AMIn Acc.GEN04, there are two batches that aren't on the other sheet: IH 3217 (with 24 "original count") and IH 3218 (with 14). (It may not mean anything, but batch IH 3216 isn't on either sheet.)
Also, the numbers used in about 24 of the batches that are common to both sheets aren't the same. There are 328 fewer ballots in the "original count" column on the Acc.GEN04 sheet compared to the same batches on the IH Batches sheet.
The net difference is 290, since the 38 ballots in those two extra batches on Acc.GEN04 aren't on the IH Batches sheet at all.
Acc.GEN04 totals 48707.
In-House Batches totals 48997.
Copying and pasting is so easy. Why would the numbers be different on one sheet compared to the other? Did they do this by committee and parcel out the data entry tasks on each sheet to different people? If so, it seems they would have tried to reconcile the discrepancies. Oh yeah--never mind: They don't do reconciliation, do they?
Recognizing that the spreadsheet released on April 14 is full of odd things, it's worth considering what King County's own summary of their internal records shows. I've spent some time trying to figure out what it says.
569,884 absentee ballots returned.
5,699 absentee ballots rejected at the signature verification step.
564,185 absentee ballots accepted and voters supposedly credited at the verification step. (RB 3366 was erroneously entered twice on the spreadsheet, but only one entry contains all info. The second entry contains only the number 45 as the number of ballots accepted. Subtract that 45 from the 564,230 total.)
Subtract the 93 absentee ballots (ignoring the faxed/emailed ballot, since it might not represent an accepted ballot) that were left in their envelopes after the voters were credited.
Subtract the 292 absentee ballots rejected after the voters were credited -- wrong ballots and multiple ballots.
That leaves 563,800 ballots available for them to count.
They claim to have counted 564,222 in the initial count. That's 422 more than their records show were available to be counted.
In the second recount, they claimed to have counted 564,448 absentee ballots (not including the 566 or 564 "Larry Phillips ballots -- 2 provisional ballots were accepted on Dec.23 in addition to the "NSOF" ballots).
They ended up counting 648 absentee ballots more than were available to be counted -- according to their own summary of their internal records.
Posted by: Micajah on April 17, 2005 10:55 PM