After a couple of weeks where some struggled just to understand that better Election Night reconciliation numbers were coming, the release of those data seems to have finally put the 1,860 "voter credit shortfall" tally to bed as the barometer of King's counting accuracy. Both Seattle papers took a crack at what they mean over the weekend, as did the gunpowder-laced fingers of Stefan Sharkansky and Sound Politics, fresh from firing smoke at various pollbooks with surely the most egregious error. It seems SP and the Seattle Times' Keith Ervin are playing off one another here, or were coincidentally led to the same two pollbooks: Bothell Library and Denny Terrace. SP has their dope on the two polling sites here (Bothell) and here (Denny); about an hour prior, Ervin makes these comments:
[The] documents released yesterday also show that the number of provisional ballots believed to have gone through polling-place tabulators is considerably higher than the 348 documented cases.
At the Bothell Regional Library, 31 provisional ballots apparently went through counters but couldn't be definitively matched with specific voters. Thirty provisional ballots were mishandled at Denny Terrace in Seattle.
The election staff made 660 "adjustments" for mishandled provisional ballots and other identified errors. [emph mine]
Ervin concentrates primarily on the subject of provisional voters, certainly not one of Rossi's strongest arguments of contest. Even if Ervin's figure of 660 is the corrected total of mishandled provisional ballots, their mere mishandling doesn't amount to much but an administrative embarassment as far as Washington law is concerned.
If there is any value to the extent of this kind of error, it would be in the reflection of verified ineligible provisional ballots cast. Of the original 348 cited by King County, Ervin reveals* that the ineligible total among them is just 40--yielding an eligibility rate of around 89%. Given the provisional eligibility rate in general was 87%, there's no indication whatsoever that mishandled provisionals reflect any sort of fraud or intentional desire to vote irregularly. Furthermore, such a high eligibility rate draws down Rossi's ability to claim any significant vote gain, based even on the fairly specious concept of "proportionality." From roughly 80 ineligible provisional voters, Rossi would stand to gain no more than a dozen ballots if they voted in the same proportion as the rest of the county.
King does not consider mishandled provisional ballots as discrepancies per se; the election worker usually ends up with an envelope, a signature in the book, but no ballot. It's not a difficult leap to discover that they were put through the machine. Thus, they are not unexplained. But what of the admitted discrepancies, where voters and ballots did not match up perfectly? Both papers agree that the sum of ballots exceeding voters is 216 countywide; the P-I claims precincts totalling 158 more voters than ballots, while the Times says 159.
Neither writer (the P-I story is written by Gregory Roberts) describes the reducibility of the differences to cancel out discrepancies; Roberts cites precinct differences while Ervin refers to "polling sites." It's not clear whether he means polling places or "sites" in the sense that each precinct is a site. This is a relatively important difference; if Ervin is closer to correct then the numbers are probably not very reducible.
The reason is that much of the time, discrepancy error comes when a voter is registered in one precinct but receives a ballot for another precinct at the same polling place. If one precinct has an extra ballot, and another at the same polling place has a missing ballot, it's quite possible there is a match between them. However, if the net discrepancies were figured at the polling place level, those matches would have mostly shaken out at that stage, and the gaps are much less explainable. So theoretically, the true discrepancy might range anywhere from 58 to 375.
That's the pollnight discrepancy, remember--absentee ballots are not counted. Neither writer addresses a figure for that subset of the total. But even so, it's clear that the harping on 1,860 (or 3,700, or any of the other numbers based on voter crediting) as an indicator of poor accuracy was unfounded, and in fact the true discrepancy is much lower. Worse for Rossi, the numbers do nothing to help him, because there is no identification of illegal votes.
There is a discrepancy, perhaps it is unresolved and evidence of less-than-hoped care with the balloting process--but once again, it must be stated that the courts tend smile wanly at haplessness, saving their starkest remedies for provable mischief or calculable neglect. Under the best circumstances, Rossi's provable neglect may buy him a dozen votes from everything discussed here. The felon votes are another story, one with its own set of problems and possibilities. But on balance, if anything the results have furthered darkened Rossi's chances of gaining a set aside election, much less an overturning of the result.
*Update 3pm 3/14--
As chew2 points out in comments, Ervin is not quite so clear. I neglected to provide the link to the PI story, which provides the information from King Executive Huennekens on the number of verified ineligible provisional votes.
--TJ
TJ,
Small Correction:
You said 40 ballots were disqualified from the 348 improperly fed provisional ballots. The Seattle Times article you referred to says **92** were disqualified. Meaning 73% were valid. Logan stated today that 83% of these provisionals were qualified, so the Times article could be wrong.
(He mentioned most of them were not properly registered to vote, but some were registered in other counties and their ballots were sent to those counties for processing.)
Logan today, said that the number of provisionals fed through the vote machines without proper verification could be as high as the 660 vote "adjustement" number, and that the already confirmed number of 348 could be containted within that larger 660 number.
Otherwise I pretty much agree with you, there is very little here to help the Rossi case.
Posted by: chew2 | March 14, 2005 at 13:55
chew2--
I actually meant to refer to the PI story, which I unhelpfully neglected to provide a link to. Here is what I base my statement on:
---
Huennekens said that of those, 252 were filled out by legal voters (although it's not possible to know which ballots were the legal ones); the identity of about 50 of the 348 voters could not be determined; and about 40 of those provisional ballots were cast by persons ineligible to vote.
---
Based on this, it's possible that all 50 others were also ineligible, but we don't know--and it sounds like some of them may have been valid, but in other counties.
I could agree that specific to King, ballots intended for other counties would be ineligible for King races--but for the governor, they would be legitimate. At least, they would have come from legal voters. But the bottom line is that there are only about 11% known illegal votes.
Posted by: Torridjoe | March 14, 2005 at 15:16
Torridjoe,
It seems to me that you're whistling past the graveyard.
Either that, or you just haven't looked closely at the reconciliation report.
The last page of the reconciliation report shows the totals: 216 ballots that came from who-knows-where but somehow got into the voting machines; 660 provisional ballots were inserted into the voting machines, based on the deductions of the canvassers; and another 348 provisional ballots were inserted into the voting machines based on the reports of the polling place officers and the deductions of the canvassers.
That's a total of 1224 ballots which were included in the vote count even though they were not lawfully cast.
And, since the reconciliation report omitted all the precincts that used the St. Benedict's polling place, it isn't a complete report.
The number may well be higher than 1224.
I don't yet know what other polling places may have been omitted from it. (In their usual unhelpful fashion, the King County elections office didn't total the number of votes counted on election night, so there is no ready indication of how many other polling places were mistakenly omitted from the report. The "Election Day MC" and "Add-ons" columns could have been totaled easily using their spreadsheet function, but they didn't do it.)
The fellow who accompanied Logan, named Triplett (I think they said he was Sims' chief of staff) stated in his opening remarks that there were approximately 1200 polling place ballots and 400 absentee ballots that couldn't be matched to any voters. That's the first time I've heard a King County official admit that absentee ballots contributed to the problem of the "voterless ballots."
You need to read more closely, so you can perhaps overcome your desire to see only what you want to see. (Either that, or admit that walking past the graveyard frightens you, and stop the whistling.) Drop by "Croker Sack" and see what we reasonable folks have to say about the information released by King County on Friday afternoon. There are still plenty of puzzles to work out. ;-)
Posted by: Micajah | March 14, 2005 at 22:22
Micajah, Logan's testimony on the provisionals gave 660 as the reconciliation adjustment; the 348 are not in addition; there was overlap.
Your characterization of the 216 ballots without voters is apropos of nothing; you make it seem as if it's magic rather than human error. And unless you have information to supply me with on it, the 216 and the 158/9 are potentially matching discrepancies in many cases.
Further, your conception of provisional ballots as not lawfully cast is more excitist rhetoric. They were not properly cast, but the votes were in fact valid. Whatever total you'd like to use, cut it nearly by a factor of 10 before you talk about unlawful votes.
Two points you might have missed:
*1200 is quite a bit lower than 1860, is it not? Meaning an already overwhelmingly accurate count just got a fair bit more accurate.
*Vote totals are nearly meaningless, especially now that we have upper bounds to our speculation about the extent of any error. By no stretch does the extent represent unusual, preventable or neglectful rates of inaccuracy, as has been shown repeatedly from multiple sources. The question is of what value these mishandled votes are to the election contest. And I'll answer it for you--fewer than 20 Rossi votes, at best, in their wildest fantasies of accepted evidence.
Posted by: Torrid | March 14, 2005 at 22:58
TJ - Great post, thanks for summarizing all that. As I see it, Republicans are now desperately hoping for some "judicial activism" in their election contest.
Usually the GOP rails against any case where a judge has to "interpret" a law to make a ruling. But now, since by any "strict" interpretation of the constitution or established case law the GOP has no case, they want the judge to make some new law in the spirit of "fairness" (or because of "uncertainty", etc.).
Posted by: jcricket | March 16, 2005 at 15:31