SoS Recount Directive 07-30 - It's NOT a RANDOM sample - again!

For understanding, you'll need to access the Recount Directive and read it at http://www.sos.state.oh.us/sos/ElectionsVoter/directives/2007/Dir2007-30.pdf
or at http://www.box.net/shared/ip0gdc03y0

I actually was not going to post this immediately; thought it might be resolved in a professional, easy way. But then I got a response today, which I include near the bottom, before my thirdly included, finally frustrated from trying to work with the SoS office on other matters too - my response back.


I. Sent:
December 2, 2007

To:
The Office of the Secretary of State
• Kellye Pinkleton, Director, The Voting Rights Institute
• cc. David Farrell, Election Director
From:
Adele Eisner, Member of the VRI and a Designated Cuyahoga Election Observer
Re: Recount Directive 07-30

Dear Kellye,

First I'm very pleased to see that the web site is working today. The public's having the mandated access to the Directives, Advisories and Memos is necessary to full participation and oversight of the citizens' election process.

As I just tried your website again today, this was the first time I read the recent Secretary of State's Directives, including 07-30, on recounts.

As I am a designated Observer for the 11/6/07 election, per ORC 3505.221, approved by the county Democratic party, thus observing Cuyahoga's recounts, I would appreciate your facilitating helping me get some quick and clarifying answers to questions I have about this directive. Cuyahoga still has at least four more recounts upcoming this week.
As an Observer, I need to understand the rules under which I am observing.

1. I am particularly flummoxed by # VI. - A. - 10 - a. through f. - regarding how to randomly select of recount samples.

I begin by saying that I appreciate that SoS Brunner is memorializing the fact that while a BOE may recount a 3% sample first, to determine if an entire handcount is necessary, (where the recount includes more than one precinct) that sample of course, by all valid sampling standards, must be randomly selected.

Her re-introducing the concept of random sampling in 07-30, in concept, re-institutes a years long procedure in Ohio recounts, which Mr. Blackwell in his tenure, allowed to be overlooked if BOE's should so choose, thus, also allowing those recounts where boards selectively chose or selectively patterned the precincts' ballots, poll books, etc. to be checked in recount, to become an essentially invalid and thus, an almost useless procedure.

In such a non-random sampling process, any previous manipulation of ballots/results in particular precincts, or recurring anomalies, could most assuredly be kept hidden from observers or unable to be citizen/candidate found or proved, until far too late to assure election justice.
This was a great cause for the still widely consisdered, unanswered questions, non-confidence, and even legal proceedings about the '04 presidential election, as you know.

But I also find that directive section #VI,A,#10 - about how to randomly select - both very difficult to understand, and seemingly bulky and confusing to perform - as opposed to a simple truly random number generating procedure, such as: counties always using "the universe" of all precincts held within it, then after shaking and mixing those precinct numbers in a hat, just picking numbers out until the 3% is reached - or using some similar, random number generating process of all precincts involved in the recount.

Worse, the directive's process, if I read it correctly, seems to in some cases, re-introduce a definite pattern of precincts to be recounted/ inherent NON-randomness.

Most directly as example, consider a one-suburb race in Cuyahoga, lets say a race for a council seat, with 30 precincts. By this directive, by having to choose the lowest numbered and lettered precincts first, then subsequently on up the line, it seems to me, that in every recount in that political subdivision, it would be known before the election begins which precincts would be most likely to be recounted and not, should such a recount occur.

Further, it seems that in a county-wide recount, for example in Cuyahoga, where there are 20+ political subdivisions, (cities, townships, etc.) but where the one city of Cleveland holds by far the largest number of precincts, (and in which the largest number of voting problems have occurred in the past,) by starting with a forcing of the categorization of cities, townships, etc., it makes makes the city of Cleveland (including many wards and precincts,) equally available to be chosen, to say a Bedford Heights with only 5 precincts. And if I'm reading it right, for example, at the start should Cleveland be "randomly" chosen, and Bedford Heights is too, it would always be those cities' same precincts ( the lowest numbered and lettered) that would be recounted.

The SoS process seems to force an inequitable chance for a truly equitable, random selection of the couny's precincts. If I'm reading this right, it forces a not-truly random pattern: this is far from a plain equitable, random selection as I suggested with the hat method above.

Please help me get clarity on this procedure, and a possible rethinking in your office, if what is suggested is still not truly random selection.


2. Then in VI, - C. -2. -b. under "The Recount" (with DRE's )- pg. 8


• First, it directs the BOE's to check the public counters on the voting units, and the poll books and the summary reports at the bottom of the VVPATs to make sure they all match. But it does not say what to do if they do match or don't.

• Further, that also seems to imply that BOE's must bring all machines used for the chosen precincts to the site of the recount, and set them up, so that observers can observe that process too, if it is to have bearing on the process of the recount.
Is that true?

• What are candidates' rights if the poll book signature numbers do NOT match the total of VVPAT summaries for the selected precincts.
They VERY often do NOT match, for any number of reasons from poll worker confusion, to encoders not working so signing in voters, vote on paper ballots.

d. And last , the directive's number VI, - C. -2. -b. -3 in that same section, it says:


What exactly are "all the ballots for the recount" - that are to be counted by hand?
Is it not presumed in this procedure, that once a precinct is selected, all the ballots for that precinct are to be hand counted?

Is that not how the workers found out in the first place, whether any VVPAT summary did not match the hand count for that tape?

I will appreciate answers to the above questions as soon as possible.

Also, to make you aware, as there has already been discussion about this directive among voting rights advocates in Ohio, within and outside of the VRI, I will be posting these questions and hopefully salient answers from your office to a wider group.


Thank you,

Adele Eisner

PS. I would have copied the sections in question in a more readable format had these documents been presented in digitized text form, or in a text accessble PDF format on your site. That would make such discussion easier. For now, if you double click "the pictures" I copied from your larger "picture page", in most applications, my small pictures will enlarge to original size for reading.
____________________________________________
II. The SoS Response today:
(I omit the repetition of my email at the bottom of this response.)

Hi Adele. Thank you for the email.

I am reviewing your email and will be following up with the appropriate staff as soon as possible to provide you a response. In the meantime, it is more efficient to get answers when we have specific questions to pass on for response. To that end, I have extracted, I believe, the questions posed in your email and I have listed them below. Let me know if these are not accurate.

Thank you,

Kellye

-What do you do if the counters, poll books and VVPAT summary reports do not match?

-Are BOE's supposed to bring all machines used for the chosen precincts to the site of the recount, and set them up?

-What are candidates' rights if the poll book signature numbers do NOT match the total of VVPAT summaries for the selected precincts.

-(VI, - C. -2. -b. -3) What exactly are "all the ballots for the recount" - that are to be counted by hand?
Is it not presumed in this procedure, that once a precinct is selected, all the ballots for that precinct are to be hand counted?

Kellye Pinkleton
Director, Voting Rights Institute
Ohio Secretary of State Jennifer Brunner
180 East Broad Street, 15th Floor
Columbus, Ohio 43215
614.995.1619 Executive Assistant
614.752.4360 Fax
kpinklet@sos.state.oh.us

III. My response back today:

(To clarify what may seem to be my non-based assumptions of avoidance and not answering that which is wanted to be avoided, I note that this reflects an ongoing pattern, as only very partially indicated in the content of my email. Also, I spoke at some length with Brian Green, an attorney in the SoS office, yesterday, as he called and indicated he would be sending me the documents I had requested in September (noted in a previous post here.) We had a very pleasant conversation, and I re-iterated to him the public's concerns about the invalidity of the "random" process presented in the directive, and told him I had already written those down and sent them to Kellye. He indicated his understanding of my message, and he would get a copy for consideration.)
_________________________________

In a message dated 12/4/07 12:46:31 kpinklet@sos.state.oh.us, writes:

(I here omit her message seen above)

Kellye -
Those though are some (sic), they, as I think is obvious, are actually the minor points to my previous email.

The main point was in #1 of my email. About the fact that the way the directive demands HOW a "random" sample is derived - makes it NOT RANDOM at all, thus able to predict BEFORE an election which precincts are most likely to be exposed to the public, should a recount occur. It thus makes it easy to "game" /manipulate over 80% of the ballots in an election, knowing no one from the public will ever have a legal chance to see what happened until well after candidates are seated and going full force...

This is as invalid a sample, or actually worse, as Mr. Blackwell's saying your 3% sample does not have to be random. Some honest, diligent BOE's would have gone to a regular picking precincts out of a hat system.
This 07-30 directive demands that they do something you're calling random, but allows anyone so inclined to do election manipulation, to do it in the majority of precincts with the highest number/letter designations and know they can get away with it.

Or as one person explained :

"...there is a difference between statistical confidence levels and telling people ahead of elections that you can make certain that 80% of the precinct can never be included in a recount, oh and here's a list of which precincts those are.
This isn't an exaggerated problem, and even if we assume there are no dishonest people, it allows counties to do a good job with all "A" precincts and guarantee there will never be a full hand count.
Also, don't worry about errors or bad poll workers in the "B"-"G" precinct because our tabulation data is the final word on those.
I agree a 3% of precincts recount is not meaningful from a
statistics point of view because you are not random sampling
3% of a bucket of ballots. But this gives a green light for
fraud that has a very low likelyhood of ever being caught."

Further it is very confusing to understand thus to perform, and bulky to boot.

AND in my PS. is the concern that when your office puts directives and advisories on the web site, instead of putting up a digitized form, they scan them, and in non-text accessible manners/but in whole page "pictures" - so that it's impossible to cut and paste the words the public might want to discuss and reference. Why not use an image file for the letterhead, and signature, in directives, advisories, and memos then make the text digitized-accessible, as other websites do - to make it easier on the public, than having to re-type everything?

So to put it in six simple questions:
1. Who came up with this method you're calling random selection?

2. Exactly why this, and not just picking precinct (on the whole, the most equal voting district breakdowns) numbers from a bowl or hat ?

3. Exactly why are you avoiding addressing the most obvious part of my email, when at least 6 people contacted me directly, and others indirectly, and thanked me for it's clarity of presenting the problems with your "random" : trying to avoid another post-election democracy debacle - of being able to pattern or manipulate elections and results ahead of time, without voters finding out?

4. When will a new recount directive be issued that reinstates a simple, clear, validly random method of random selection of the 3% random sample for a recount - those things that are exposed to the public before election certification - such as a hat or similar method?

5. What exactly is going on there - as your reply, for now, gives me great NON-confidence, especially now also in the testing process, into which the public, - including even the VRI-tech group, (despite our trying to work with your office, and to repeatedly be professional in getting the peer oversight inclusion, ( to even election officials) as was repeatedly inferred in your office's emails, conference calls, talks, as "planned and/or coming"- but your office never delivered - thus into which the public ( who will have to pay for any changes the legislature recommends from your recommendations, and worse who may have to suffer through a whole new round of undemocratic electronic machines and the "results" they yield,) has had no oversight or input? (My inclusion on blog - I know how garbled this one is, but you can get the drift. I was rushed/ have many more things to do - like work! - than go around in these circles with the SoS office.)

6. (And indirectly) Why not use an image file for the letterhead, and signature, in directives, advisories, and memos then make the text digitized-accessible, as other websites do - to make it easier on the public, than having to re-type everything?
_________________________________________

Here is my #1 again:
1. I am particularly flummoxed by # VI. - A. - 10 - a. through f. - regarding how to randomly select of recount samples.
....(and it repeats from the first section, above.)

Maybe someone else wants to try contacting the SoS, to actually get through?

One should NOT have to be an election official working "at the pleasure of the SoS," or on some other "inner circle" to be able to actually present a problem of great importance to fair elections, nor create a huge "hissy fit" or spend months of frustrating time-sinks, to get it considered and resolved for the good of the public.